Arachidonic acid metabolism by 5-lipoxygenase leads to production of the potent inflammatory mediators, leukotriene (LT) B4 and the cysteinyl LT. Relative synthesis of these subclasses of LT, each with different proinflammatory properties, depends on the expression and subsequent activity of LTA4 hydrolase and LTC4 synthase, respectively. LTA4 hydrolase differs from other proteins required for LT synthesis because it is expressed ubiquitously. Also, in vitro studies indicate that it possesses an aminopeptidase activity. Introduction of cysteinyl LT and LTB4 into animals has shown LTB4 is a potent chemoattractant, while the cysteinyl LT alter vascular permeability and smooth muscle tone. It has been impossible to determine the relative contributions of these two classes of LT to inflammatory responses in vivo or to define possible synergy resulting from the synthesis of both classes of mediators. To address this question, we have generated LTA4 hydrolase-deficient mice. These mice develop normally and are healthy. Using these animals, we show that LTA4 hydrolase is required for the production of LTB4 in an in vivo inflammatory response. We show that LTB4 is responsible for the characteristic influx of neutrophils accompanying topical arachidonic acid and that it contributes to the vascular changes seen in this model. In contrast, LTB4 influences only the cellular component of zymosan A-induced peritonitis. Furthermore, LTA4 hydrolase-deficient mice are resistant to platelet-activating factor, identifying LTB4 as one mediator of the physiological changes seen in systemic shock. We do not identify an in vivo role for the aminopeptidase activity of LTA4 hydrolase.
Leukotrienes (LT),3 biologically active metabolites of arachidonic acid (AA), have been implicated in the pathological manifestations of inflammatory diseases, including asthma, arthritis, psoriasis, and inflammatory bowel disease (1). Two subgroups of LT, differing in the metabolic pathway by which they are synthesized and by the inflammatory responses they elicit, are recognized. Synthesis of both groups of LT is initiated by the conversion of free AA to an epoxide intermediate, LTA4, by arachidonate 5-lipoxygenase (5-LO) in the presence of the accessory 5-LO-activating protein (FLAP). LTA4 is an unstable compound that is hydrolyzed by the enzyme LTA4 hydrolase to LTB4 in a reaction that results in the suicide inactivation of the hydrolase (2). Alternatively, LTA4 may be conjugated with glutathione by LTC4 synthase to produce the cysteinyl leukotriene, LTC4. This compound and its metabolites, LTD4 and E4, constitute the slow-reacting substance of anaphylaxis.
The contribution of LT to a specific inflammatory response depends both on the ability of cells present in the inflammatory lesion to produce a particular LT and the response of the tissue to these bioactive lipids. 5-LO is expressed predominantly by cells of myeloid origin, particularly neutrophils, eosinophils, monocytes/macrophages, and mast cells (3, 4). The enzyme has also been detected at low levels in B cell lines (5, 6) and in extrahemopoietic cells types, including keratinocytes (7) and colonic epithelia (8). Similarly, the GST LTC4 synthase is expressed in a limited number of cell types, predominantly eosinophils, monocytes/macrophages, mast cells, and some leukemic cell lines (9).
In contrast to 5-LO and LTC4 synthase, LTA4 hydrolase expression is ubiquitous with high levels of enzyme seen in several tissues. In particular, abundant message is detected in small intestine epithelial cells, lung, and aorta (10). Moderate expression is observed also in leukocytes, particularly neutrophils (11).
Contribution of the AA metabolites produced by the enzymes of the 5-LO pathway to inflammatory pathophysiology is supported by the identification of high levels of LT in both acute and chronic inflammatory lesions and by the demonstration that introduction of LT into normal tissue can elicit a number of the primary signs of inflammation. Such experiments also support a differential function for LTB4 and cysteinyl LT during the inflammatory response. LTB4 stimulates adhesion of leukocytes to vascular endothelia, thus facilitating the migration of these cells into adjacent tissue. Furthermore, it has been shown to initiate PMN degranulation (12). The cysteinyl LT, predominantly secreted by eosinophils, mast cells, and macrophages, cause vasodilation, increase postcapillary venule permeability, and potently stimulate bronchoconstriction and mucous secretion (13).
Interestingly, LTC4 production by endothelial cells (14), vascular smooth muscle, and platelets has been demonstrated (9). Because these cells do not express 5-LO, this finding suggests that exogenous LTA4 can serve as substrate for LTC4 synthase via intercellular transfer.
As with LTC4, LTB4 production has been verified in cells that are unable to synthesize their own LTA4, and transfer of LTA4 into these cells has been described (15, 16). The differential expression of 5-LO, LTC4 synthase, and LTA4 hydrolase, and particularly the presence of LTA4 hydrolase in tissues unable to synthesize LTA4, suggests an important regulatory role for intercellular transfer of LTA4 in LTB4 synthesis. It is possible that this process is important for the amplification, regulation, and/or progression of the inflammatory response.
An alternate explanation for the high levels of LTA4 hydrolase present in tissues that do not express 5-LO became apparent with the isolation of the cDNAs encoding this enzyme (17, 18). Examination of the primary sequence of LTA4 hydrolase identified a region homologous to the active sites of aminopeptidases, enzymes critical for protein maturation, digestion, and regulatory turnover (19). It is possible, therefore, that LTA4 hydrolase is a bifunctional enzyme, because it possesses an in vitro aminopeptidase activity in addition to its epoxide hydrolysis of LTA4 (20, 21). No target for the aminopeptidase function of LTA4 hydrolase has been identified in vivo, although there is some evidence that such a target may be similar or identical to the intercellular target of the lethal factor of Bacillus anthracis (which is also unidentified) (22). A search for naturally occurring peptides that can be cleaved by the peptidase activity of LTA4 hydrolase led to the demonstration that the opioid dynorphin fragment 1–7 is a good substrate for the enzyme (23), suggesting a role for LTA4 hydrolase in neuropeptide homeostasis.
A direct method for determining the dependence of a particular inflammatory response on LT production was made possible by the generation of mouse lines deficient in 5-LO and FLAP (24, 25). Analysis of these mice has shown that LT production is critical for edema formation and neutrophil migration in a number of acute models of inflammation. For example, edema formation and neutrophil migration were largely abolished in topical AA-induced inflammation of the skin. Similarly, the time course of edema formation and neutrophil migration is altered in 5-LO- and FLAP-deficient animals in a model of zymosan A-induced peritonitis.
We describe in this work the generation of mice deficient in the enzyme LTA4 hydrolase and compare the inflammatory responses of these animals with those of wild-type controls and mice deficient in 5-LO. This comparison provides a means of directly assessing the contribution of LTB4 and the cysteinyl LT to inflammatory processes in vivo. Furthermore, any phenotypic differences between LTA4 hydrolase-deficient, 5-LO-deficient, and wild-type mice could provide insight into an in vivo aminopeptidase activity for LTA4 hydrolase.
Materials and Methods
Isolation of the murine LtA4 hydrolase gene
An 882-bp murine LtA4 hydrolase cDNA fragment was generated by RT-PCR (primer 1, 5′-ACGAATTCAACCAGAGGGTTCC-3′; primer 2, 5′-AAGAATTCCTCTTTGGCAGTAACCC-3′) and then used as a probe to screen a genomic DNA library in the Lambda FIX II vector generated from the 129 mouse strain (Stratagene, La Jolla, CA). Six genomic clones were isolated and purified to homogeneity, and one of these was subcloned into the NotI site of pBluescript II SK+ (Stratagene) for mapping analysis. The mapped LtA4 hydrolase clone, λ5111, was found to possess a coding region flanked on either side by several kb of noncoding DNA and was useful, therefore, in the creation of a targeting construct at this locus.
Generation of LTA4 hydrolase-deficient mice
The murine genomic clone λ5111 was used to construct a targeting vector for the LtA4 hydrolase gene. Two genomic fragments, 4.25 and 3.99 kb, respectively, were inserted 5′ and 3′ of the neo gene of pJNS2 (26) such that coding material would be replaced by neo after homologous recombination of the plasmid with the endogenous locus. Sequence analysis of the fragment replaced through this targeting strategy showed that successful targeting would interrupt an exon at the BglII site at position 738 of the LtA4 hydrolase cDNA and would eliminate the remainder of the exon (bases 739–853). This construct was electroporated into E14TG2a embryonic stem (ES) cells established from the 129/Ola mouse line (27), and colonies resistant to neomycin and ganciclovir were identified, as described previously (26). Southern blot analysis was performed on EcoRV-digested genomic DNA to identify correctly targeted resistant ES cell lines. An 871-bp EcoRI/EcoRV fragment downstream of the regions of homology was used to screen for recombinants (Fig. 1 A). Verified recombinant ES cell lines were microinjected into mouse blastocysts, and chimeric mice were generated. Chimeras were bred to 129/SvEv mice, and offspring were genotyped by Southern blot analysis of tail DNA digested with EcoRV, as described above. Heterozygotes in the F1 generation were then mated to give rise to animals homozygous for the disrupted allele in the F2 generation.
LtA4 hydrolase mRNA analysis
Animals of the indicated genotypes were euthanized, and the femurs of each mouse were exposed in a sterile field. Cuts were made through the knee and hip joints to isolate the bone and to expose the marrow. A total of 3 ml serum-free media was flushed through each femur into a petri dish to isolate the marrow. Cells from one leg were centrifuged and used for RNA isolation, as indicated below. Marrow cells collected from the other leg of each animal were processed to isolate bone marrow-derived mast cells (BMMC), according to Rottem et al. (28). Total RNA was extracted from bone marrow cells and BMMC by resuspension of cell pellets in RNAzol B (Tel-Test, Friendswood, TX), according to the protocol supplied by the manufacturer, and was separated by gel electrophoresis (15–20 μg/lane). Northern blot analysis was conducted according to standard techniques using the 882-bp murine LtA4 hydrolase cDNA fragment described above as a probe to detect expression of LtA4 hydrolase mRNA. A 429-bp probe upstream of the disrupted exon was also hybridized with these Northern blots to insure that no spliced or partial message was being produced in the LtA4 hydrolase−/− tissue. Finally, a probe for murine β-actin was used as a positive control (Stratagene).
Inflammatory responses induced by AA
129 strain LTA4 hydrolase-deficient mice were injected i.v. with 0.5% Evans blue dye (10 ml of dye solution/kg of body weight) dissolved in PBS (2.7 mM KCl, 1.2 mM KH2PO4, 138 mM NaCl, 8.1 mM Na2HPO4, pH 7.5) and filtered through a 0.2-μm nitrocellulose membrane. The pinna of the left ear was coated with 20 μl of AA (100 mg/ml) in acetone (Sigma, St. Louis, MO) to induce an inflammatory response, while the right ear received acetone alone on its inner surface. After 1 h for edema assays or 7 h for myeloperoxidase (MPO) analysis, mice were euthanized, and an 8-mm-diameter disc of tissue was obtained from the center of each ear for analysis. For comparison, experiments were conducted in parallel with age- and gender-matched 129 wild-type, 5-lo−/−, and Flap−/− mice.
Inflammatory responses induced by PMA
Mice of the indicated genotypes were injected i.v. with 0.5% Evans blue dye in identical fashion to the AA experiments above. The left ear was then coated with 1 μg of the diacylglycerol analogue PMA in acetone, while the right ear received acetone alone as a control. After 6 h, 8-mm-diameter discs of ear tissue were biopsied, and edema and plasma protein extravasation were assessed.
Quantification of edema and vascular permeability
The wet weight of each ear biopsy was determined to assess edema formation. Each ear disc was then incubated in 1 ml formamide at 55°C for 24 h. Extravasation of Evans blue dye was quantified by spectrophotometric analysis of the formamide extracts at 610 nm (29). Data are expressed as mean differences between experimental and control ears for each animal.
Ear punches (8 mm diameter) were homogenized and analyzed, according to a previously described protocol (24) modified from Bradley et al. (30). MPO present in the lysed cell supernatants was quantified by comparison with a standard curve derived from serial dilution of commercial MPO (Calbiochem, San Diego, CA).
Zymosan A-induced peritoneal inflammation
Peritoneal inflammation was induced in LtA4 hydrolase−/−, Flap−/−, 5-lo−/−, and wild-type animals using a protocol modified from Rao et al. (31). Each mouse received 1 ml zymosan A suspension (1 mg/ml in PBS, pH 7.5) by i.p. injection. After 3 h, mice were killed, and a peritoneal lavage was performed with 4 ml of ice-cold PBS. After lavage fluid was collected, 1 ml was transferred to microcentrifuge tubes and centrifuged at 40,000 × g for 6 min. Supernatants were collected for direct enzyme immunoassay (EIA) analysis of LTC4 and PGE2 levels (Cayman Chemical, Ann Arbor, MI). Aliquots of the collected supernatants were also purified by HPLC (see below), and LTB4 levels were measured in the eluted fractions by EIA (Cayman Chemical). Cell pellets, meanwhile, were resuspended in 1 ml PBS/0.5% hexadecyltrimethylammonium bromide (Sigma), and the cells were lysed by sequential freeze-thaw cycles. Cellular debris was removed by centrifugation, and the presence of neutrophils in the cell pellets was determined by detection of MPO in the lysis buffer using an enzymatic assay, as described above. MPO analysis of cell pellets was also performed in this way at time points of 0, 1, 6, and 12 h following zymosan A treatment.
Total cell counts were made for lavage fluid collected at the 3-h time point using a hemocytometer. Approximately 2.5 × 105 cells were then mounted on glass slides by cytospin centrifugation and were stained with Diff-Quik (Dade Diagnostics, Aguada, PR). The percentage of PMN present in the lavage fluid was determined by microscopy.
In related experiments, plasma protein extravasation into the peritoneal cavity was examined. Mice received 0.5% Evans blue dye (10 ml of dye solution/kg of body weight) in PBS i.v. immediately before injection of zymosan A suspension. Mice were euthanized at time points of 10, 20, 30, 60, 180, and 360 min and underwent peritoneal lavage with 4 ml of cold PBS, as described above. Cells were centrifuged out of the lavage fluid, and Evans blue dye extravasation was assessed by light spectrophotometry of supernatants at 610 nm.
HPLC purification of zymosan A-stimulated lavage fluid
Peritoneal lavage fluid was collected from zymosan A-treated animals after 3 h, as detailed above. Lavage fluid (0.5 ml) was spiked with 100 μg indomethacin in 10 μl methanol as an internal standard with or without LTB4 standard (25 ng; Biomol, Plymouth Meeting, PA), and protein was precipitated by addition of 1 ml acetonitrile, followed by centrifugation at 8000 × g for 10 min. The supernatant was acidified and diluted in water/acetic acid (100/0.1, v/v) containing 1 mM disodium EDTA and passed through a 100 mg BondElut C18 solid phase extraction cartridge (Varian, Harbor City, CA). The column was washed twice with methanol/water (80:20, v/v) and then eluted with 0.5 ml methanol/triethylamine (100:0.1, v/v). The entire eluate was diluted with 4.1 ml water/acetic acid (100:0.1, v/v) containing 1 mM disodium EDTA, and 5 ml was injected into the HPLC column (a Perkin-Elmer (Norwalk, CT) CR-C18 cartridge column, 3 × 0.46 cm, 3 μm). The column was equilibrated with methanol/water/acetic acid (20:80:0.1, v/v/v). After sample injection, a mobile phase gradient was established from 100% methanol/water/triethylamine (20:80:0.1, v/v/v) to 50% methanol/water/triethylamine (20:80:0.1, v/v/v) and 50% methanol over 20 min at a flow rate of 1.5 ml/min. The column eluant was monitored by two Perkin-Elmer LC-95 UV detectors in series at 280 and 235 nm. Eluate fractions were collected every 0.2 min, dried under a vacuum, and redissolved in 250 μl EIA buffer for EIA detection of LTB4 (Cayman Chemical).
Contact hypersensitivity to topical FITC
A method adapted from that of Furue and Tamaki (32) was used to assess the ability of LTA4 hydrolase-deficient mice to mount a contact hypersensitivity response. Briefly, LtA4 hydrolase−/− and wild-type mice of the 129 strain were sensitized by application of a 100 μl topical dose of either 0.5% or 2.5% FITC (Sigma) in 1:1 acetone:dibutyl phthalate (w/v) to the shaved abdomen. After 6 days, a hypersensitivity reaction was elicited by topical application of 20 μl of a corresponding concentration of FITC to the left ear. Nonsensitized animals were also treated with 0.5% and 2.5% FITC as a control. The right ear of each mouse received vehicle alone. Twenty-four hours later, mice were euthanized, and ears were collected and weighed, as described above. The differences in wet weight between experimental (left) and control (right) ears were compared for mice of each genotype.
Passive systemic IgE-induced anaphylaxis
Passive systemic anaphylaxis was induced as previously described (26). LTA4 hydrolase-deficient animals and wild-type controls received 20 μg of a monoclonal mouse anti-mouse DNP IgE in 200 μl of PBS i.v. After 24 h, baseline body temperatures were determined using a rectal probe, and mice received an i.v. injection of 1 mg DNP-human serum albumin (HSA) and 1% Evans blue (w/v) in 200 μl PBS to induce anaphylaxis. Thirty minutes following this injection, body temperature was measured again. Mice were then euthanized, and Evans blue dye extravasation was quantitated in 8-mm-diameter ear punches, as described above.
PAF-induced systemic shock
LTA4 hydrolase-deficient and wild-type control mice received i.v. injections of PAF (50 μg/kg of body weight) in PBS containing BSA (at 2.5 mg/ml) and propranolol (at 0.05 mg/ml). Survival was assessed for 24 h following the injection of PAF.
Survival after endotoxin treatment was determined using a protocol adapted from Ogata et al. (33). LTA4 hydrolase-deficient, 5-LO-deficient, FLAP-deficient, and wild-type mice received an i.p. dose of 5 mg iota (type V) carrageenan (Sigma) in 500 μl PBS. After 24 h, mice were injected i.v. with 30 μg LPS (from Escherichia coli serotype 0111:B4; Sigma) in 300 μl PBS and were monitored for survival for the subsequent 72 h.
Generation of LTA4 hydrolase-deficient mice
Fragments of DNA corresponding to the mouse LtA4 hydrolase gene were isolated from a genomic library. These fragments were used to prepare a plasmid that was expected to disrupt the coding sequence of this gene upon homologous recombination with the endogenous locus (Fig. 1,A). The region of the gene replaced by the neomycin-resistance cassette (neo) upon homologous recombination (see Fig. 1 A) was subcloned and sequenced. This analysis indicated that successful targeting would result in the integration of neo into the BglII site present at bp 738 of the mouse LtA4 hydrolase cDNA. In addition, the recombination event would remove 114 bp of coding sequence immediately 3′ of this restriction site (bases 739–853 of the mRNA).
The LTA4 hydrolase-targeting plasmid was electroporated into E14TG2a ES cells (27), and transfected clones were selected by growth in the presence of G418 and ganciclovir. Resistant colonies were picked, expanded, and screened for the presence of homologous recombination events by Southern blot analysis of EcoRV-digested genomic DNA using a probe corresponding to DNA present 3′ of the DNA used in the preparation of the targeting construct (Fig. 1,A). Because the neo cassette introduces a novel EcoRV site into the recombinant locus, this probe identifies an 8.7-kb DNA fragment from the endogenous LtA4 hydrolase locus and a 5.1-kb DNA fragment from the targeted locus (Fig. 1 B). Two targeted ES cell lines were identified and microinjected into mouse blastocysts to produce chimeric mice capable of transmitting the targeted allele to their offspring. Intercrossing of these heterozygous animals produced LtA4 hydrolase−/− mice in expected numbers based on Mendelian inheritance of the targeted locus. Thus, loss of LTA4 hydrolase expression had no effect on the fetal development or perinatal survival of the animals. The LTA4 hydrolase-deficient animals could not be distinguished from LtA4 hydrolase+/− or wild-type littermates by simple observation or on histological analysis of major organ systems.
Lymphocytic development in the LTA4-deficient mice was examined using two-color FACS analyses of cells isolated from spleen, thymus, and lymph nodes with Abs to murine CD4 and CD8. Similar numbers of CD4+, CD8+, double-positive (CD4+8+), and double-negative (CD4−8−) thymocytes and thymocyte precursors were counted in cells collected from LtA4 hydrolase−/− and control animals. FACS analysis of these cells using B220 and Thy-1.2 Abs also revealed no difference in the ratios of cells bearing these B and T cell markers in the two populations of mice (data not shown). Differential counts of hemopoietic cells in blood smears also failed to identify any differences between the wild-type and LTA4 hydrolase-deficient mice.
The mRNA encoding LTA4 hydrolase is abundant in leukocytes (11). Therefore, to determine whether the homologous recombination event had resulted in a null LtA4 hydrolase allele, total RNA was isolated from bone marrow cells and BMMC obtained from LtA4 hydrolase−/− and wild-type mice, and expression of LtA4 hydrolase was examined. Northern blot analysis using the 882-bp LtA4 hydrolase cDNA fragment used to isolate the initial LtA4 hydrolase genomic clones (Fig. 1 C) indicated high levels of LtA4 hydrolase mRNA in RNA prepared from wild-type animals. In contrast, expression of this gene could not be detected in RNA samples prepared from bone marrow cells and BMMC of LtA4 hydrolase−/− mice. This finding was confirmed using a cDNA probe corresponding to coding regions located entirely outside of the region of the gene altered by the homologous recombination event. Again, no LtA4 hydrolase message was detected in the LTA4 hydrolase-deficient animals (data not shown).
LTB4 production in LTA4 hydrolase-deficient animals
An experimental peritonitis was elicited in LtA4 hydrolase−/− and wild-type mice by injection of zymosan A into the peritoneal cavities of these animals. HPLC-purified fractions of peritoneal lavage fluids that coeluted with known LTB4 standard were analyzed for the presence of LTB4 by EIA. While LTB4 was easily detected in the lavage fluid of wild-type animals, LTB4 could not be detected in samples obtained from the LtA4 hydrolase−/− animals (Fig. 2,A). In contrast, levels of LTC4 in the peritoneal fluid from the LtA4 hydrolase−/− animals were equal, if not higher, than those measured in samples from wild-type animals (Fig. 2,B). LTA4 hydrolase-deficient mice produced ∼3.75 times more LTC4 in this assay (3015 ± 916.3 pg/ml) than wild-type mice (801.2 ± 290.8 pg/ml); however, this increase did not reach statistical significance (p = 0.074). The loss of LTA4 hydrolase expression had no apparent effect on the activity of the cyclooxygenase pathway in this peritoneal inflammatory response, as wild-type and LTA4 hydrolase-deficient mice produced similar levels of PGE2 in this assay (Fig. 2 C).
Role of LTB4 in the vascular and cellular components of AA-mediated acute inflammation
Topical application of AA onto the inner surface of the mouse ear causes an intense acute inflammatory reaction with both vascular and cellular components. This inflammatory response results from the conversion of the applied AA into eicosanoids, primarily metabolites of the 5-LO and cyclooxygenase pathways, with proinflammatory properties. The contribution of LT to both the cellular and vascular components of this response has been demonstrated using animals deficient in expression of 5-LO and FLAP (24, 25, 34). Because 5-LO-deficient animals are unable to synthesize both LTB4 and the cysteinyl LT, comparison of the responses in these mice with those observed in the LTA4 hydrolase-deficient mice should allow determination of quantitative and qualitative differences in the contribution of these two groups of LT to the inflammatory response. Therefore, we treated LtA4 hydrolase−/−, 5-LO-deficient, and control animals with 2 mg of AA, biopsied ear tissue, and examined the impact of loss of LTB4 and cysteinyl LT on both the vascular and cellular components of the inflammatory response. Edema formation was assessed by comparison of the weight of ear biopsies from the ear treated with AA with the vehicle-treated control ear of each animal. To measure alterations in plasma exudation, mice received an i.v. injection of Evans blue dye before treatment with AA. Because this dye binds to serum proteins, quantification of the dye present in ear biopsies provides a reliable means of measuring changes in vascular permeability (29). The cellular component of this inflammatory response was quantitated by measuring the levels of MPO, an enzyme specific to cells of the myeloid lineage, including neutrophils, present in the ear biopsies (35).
The increase in ear weight following topical AA treatment was attenuated in LTA4 hydrolase-deficient mice compared with age- and gender-matched wild-type controls (Fig. 3,A). Although the response of the LTA4 hydrolase-deficient mice was significantly lower than that observed in wild-type animals (p = 1.5 × 10−5), the response remained more vigorous than that seen in age- and gender-matched 5-LO-deficient mice (p = 0.00045). Changes in vascular permeability and plasma protein extravasation in response to AA stimulus paralleled these findings, with LTA4 hydrolase-deficient animals displaying responses intermediate between those of the wild-type and 5-LO-deficient animals. The decrease in protein extravasation into the AA-treated tissue observed in the LTA4 hydrolase-deficient mice compared with wild-type animals was highly significant (p = 0.009). Protein leakage was significantly greater, however, than that measured in the 5-LO-deficient animals (p = 0.0046) (Fig. 3 B). These results indicate that edema formation and plasma protein extravasation in response to AA are dependent in part, either directly or indirectly, on the synthesis of LTB4 by LTA4 hydrolase.
The MPO activity present in the AA-treated ear tissue provides an indirect measurement of the cellular response to this acute inflammatory stimulant. The MPO activity observed in the AA-treated ear tissue of LTA4 hydrolase-deficient mice was compared with that seen in tissue from similarly treated 5-LO-deficient and wild-type animals. As expected, MPO activity was easily detected in cutaneous tissue of normal mice exposed to AA (Fig. 3 C). We have reported previously that this response is significantly reduced in the 5-LO-deficient animals. In contrast to the intermediate vascular response seen on treatment of the LTB4-deficient mice with AA, neutrophil infiltrate in LtA4 hydrolase−/− ear tissue was decreased to the same low level observed in the 5-lo−/− mice. No significant difference was seen between these two groups of animals.
The loss of 5-LO or LTA4 hydrolase did not alter either the vascular or cellular components of the inflammatory response to topical application of PMA (Fig. 3, A–C). Exposure to PMA results in general stimulation of cellular activity mediated by protein kinase C (36) including induction of cytokine production and cyclooxygenase expression in endothelial cells (37). PMA also activates T cells (38), neutrophils, platelets, basophils, and mast cells (39). In the experiments described in this work, the inflammatory properties of PMA are not diminished by the absence of LT.
LtA4 hydrolase−/− mice show blunted PMN recruitment, but normal plasma protein influx in an experimental model of peritonitis
To further examine the roles of LTB4 and the cysteinyl LT in the inflammatory process, we measured plasma protein extravasation and PMN infiltration in response to i.p. injection of zymosan A. To assess extravasation of plasma proteins into the peritoneal cavity, mice were injected with Evans blue dye before treatment with zymosan A. Peritoneal lavage was performed 0, 10, 20, 30, 60, 120, 180, and 360 min after treatment. As shown in Fig. 4 A, plasma protein leakage in response to this irritant was identical in LTA4 hydrolase-deficient and wild-type animals. In contrast, and consistent with previous reports, the rate at which serum proteins accumulated in the peritoneal cavity after initiation of the inflammatory response was altered in 5-LO-deficient animals. These data indicate that cysteinyl LT synthesis alone is sufficient to allow LtA4 hydrolase−/− mice to exhibit a normal rate of plasma protein influx into the peritoneal cavity in response to zymosan A.
PMN infiltration was examined in the zymosan A peritonitis model. Peritoneal lavage was performed 0, 1, 3, 6, and 12 h following zymosan A injection, and cells present in the lavage fluid collected by centrifugation. MPO activity in lysates prepared from these cells present in 1 ml of lavage fluid was used as an indirect measurement of PMN migration into the peritoneal cavity in response to zymosan A. Examination of lysates prepared from lavage fluid collected from wild-type mice indicated a steady accumulation of PMN in the peritoneal cavity after injection of zymosan A, with levels reaching a maximum ∼6 h after treatment. Although the level of MPO increased in lavage fluids collected from the 5-LO-deficient animals, it remained significantly lower than levels seen in wild-type mice at all time points examined after injection of zymosan A (Fig. 4 B). The influx of PMN in the LtA4 hydrolase−/− animals was significantly decreased in comparison with wild-type animals. The level of MPO in the lysates prepared from lavage fluid from these animals, however, indicated that PMN influx into the peritoneal cavity of the LtA4 hydrolase−/− was greater than that observed in the 5-LO-deficient animals. This difference between the LtA4 hydrolase−/− and 5-lo−/− mice was verified by direct enumeration of the numbers of neutrophils present in the lavage fluid obtained from animals of each genotype 3 h after treatment with zymosan A. Consistent with the results obtained on examination of MPO activity, the number of neutrophils present in the LtA4 hydrolase−/− mice was significantly less than that seen in the wild-type mice, but remained higher than that observed in the lavage fluid collected from the 5-lo−/− animals (data not shown).
Dermal contact hypersensitivity to FITC is unaltered in the absence of LTA4 hydrolase
Because high levels of LTA4 hydrolase are found in mature T cells, we examined the possibility that the aminopeptidase activity of this enzyme might contribute to T cell-dependent responses in vivo. Contact hypersensitivity to FITC, a form of delayed-type hypersensitivity reaction, was induced in wild-type and LTA4 hydrolase-deficient animals by topical application of FITC. The T cell-dependent response of the two groups of animals was quantitated by topical application of FITC to the ears of sensitized animals 1 wk later and then measuring the change in wet weight of the ear tissue 24 h after challenge. As shown in Fig. 5, LTA4 hydrolase-deficient mice are able to mount a normal contact hypersensitivity response to Ag.
Systemic shock and anaphylaxis induced by PAF, LPS, and IgE
PAF is a potent mediator of systemic shock characterized by microvascular leakage, vasodilation, contraction of smooth muscle, endothelial adhesion, and activation of neutrophils, macrophages, and eosinophils. The ability of PAF to stimulate LT production and the effectiveness of LT inhibitors in diminishing the physiological changes seen on treatment of animals with PAF suggested that many of the effects attributed to PAF are dependent on the secondary production of leukotrienes. Direct evidence supporting this mechanism has come from the demonstration that mice deficient in FLAP and 5-LO have an increased resistance to PAF (24, 25, 34). To determine whether this change in sensitivity to PAF results from loss of production of LTB4 or the cysteinyl LT, leukotrienes, 13 LtA4 hydrolase−/−, and 14 control mice received i.v. injections of PAF and were monitored for survival over 24 h. Survival of LtA4 hydrolase−/− animals was significantly higher (92.3%) than that observed in the control group, in which 7 of the 14 mice died. Analysis of these results using the χ2 test indicates a difference in susceptibility to PAF-induced shock between the two groups with a significance level of 0.016. This suggests that production of LTB4 contributes to physiological changes leading to death on exposure to high levels of i.v. PAF.
In contrast to results obtained on treatment of 5-LO- and FLAP-deficient animals with PAF, no difference was reported in the sensitivity of LT-deficient mice to LPS-induced shock (34). To verify this and to determine whether loss of the aminopeptidase activity of LTA4 hydrolase might alter the sensitivity of the animals to LPS, wild-type, LtA4 hydrolase−/−, 5-LO-deficient, and FLAP-deficient mice were treated with carrageenan followed by LPS challenge. After 72 h, it was found that mortality following LPS treatment was identical in mice of all four genotypes (data not shown).
Systemic IgE anaphylaxis in the mouse is characterized by bronchoconstriction, plasma extravasation, hypotension with resultant tachycardia, and drop in body temperature. These physiological changes result from the release of proinflammatory mediators stored in mast cells after activation of the high affinity IgE receptor as well as the de novo synthesis of lipid mediators from AA (26). Passive systemic anaphylaxis was induced in LTA4 hydrolase-deficient and wild-type mice by injection of monoclonal mouse anti-mouse DNP IgE. Twenty-four hours later, DNP-HSA and Evans blue were administered. No difference in this response was observed between normal and 5-LO- and FLAP-deficient mice in previous studies. Again, to determine whether the aminopeptidase activity of LTA4 hydrolase contributes to IgE-mediated inflammation, we compared this response in LTA4 hydrolase-deficient and wild-type mice. Changes in body temperature and vascular permeability were similar in these groups of animals after treatment with Ag (Fig. 6).
We have introduced a deletion mutation into the murine gene encoding LTA4 hydrolase by homologous recombination in ES cells. Mice homozygous for this mutation do not produce detectable levels of LTB4 during zymosan A-induced peritonitis, while synthesis of LTC4 remains intact and may be increased. LTB4 deficiency does not affect the survival of mice, and the development of LtA4 hydrolase−/− mice appears normal. In addition, despite the high levels of LTA4 hydrolase in various lymphocyte populations of normal mice, these cells appeared to develop normally in LtA4 hydrolase−/− mice, as they display normal ratios of both B and T lymphocytes and CD4+, CD8+, double-positive, and double-negative thymocytes. As discussed in detail below, however, LtA4hydrolase−/− mice display alterations in several acute inflammatory responses as well as in their sensitivity to PAF-induced shock. Comparison of these responses in LTA4 hydrolase-deficient, 5LO-deficient, and normal control mice allows us to draw conclusions about the relative importance of LTB4 and the cysteinyl LT in these responses. This comparison also allows us to demonstrate the apparent absence of a detectable role for the aminopeptidase activity of LTA4 hydrolase in these responses.
One well-characterized inflammatory model in which LT have been implicated involves topical application of AA to the murine ear. This treatment initiates an inflammatory response characterized by both a cellular and a vascular component. The cellular component of this response consists of leukocyte accumulation at the site of inflammation. The production of both LTB4 (40) and LTC4 (41, 42) has been reported in murine ear tissue following topical AA treatment, each peaking ∼15 min after stimulation. Histological examination and enzymatic analysis for MPO (40, 42) have shown that few neutrophils are present in murine ears 15 min after AA treatment. PMN begin to accumulate 30–60 min post-AA exposure, with the highest numbers seen 4–7 h after initiation of the response (40, 41, 42). It is likely, therefore, that cells resident in the skin produce the LT measured immediately following induction of this inflammatory response. Neutrophils attracted to the inflamed ear by these eicosanoids may then contribute to further accumulation of cells by production of additional LT as well as other chemotactic mediators. Direct evidence for a role for LT in the recruitment of PMN to inflamed tissue in this model has come from the observation that the level of MPO in the ear tissue of 5-LO-deficient mice is significantly lower than that seen in wild-type mice. We show in this study that the reduction in MPO activity in the tissue from LTA4 hydrolase-deficient mice is similar to that seen in the 5-LO-deficient animals. Thus, while both LTB4 and LTC4 typically are produced in ear tissue immediately following AA exposure, only LTB4 appears to contribute to the cellular component of the resulting inflammatory response.
Measurements of edema and plasma extravasation, which represent the vascular component of the AA-induced response, indicate the involvement of both LTB4 and the cysteinyl LT. Formation of edema in the LTA4 hydrolase-deficient mice, although reduced in comparison with wild-type animals, did not reach the low levels measured in 5-lo−/− mice (Fig. 3, A and B). Similar findings have been reported in this model after treatment of mice with a high affinity LTB4 receptor antagonist (43). This in vivo study supports previous reports that have identified distinct mechanisms by which the cysteinyl LT and LTB4 contribute to edema formation. Increases in postcapillary venule permeability in response to the cysteinyl LT have been shown to result from plasma protein leakage through gaps between contracted endothelial cells (44). In contrast, these studies suggested that LTB4 contributes to edema formation by initiating leukocyte adhesion to endothelium and diapedesis into the interstitial space (45, 46). Plasma leakage accompanies this process and is dependent upon leukocyte migration (47). Finally, marginating leukocytes may themselves produce factors to enhance or augment edema formation.
A second inflammatory response that is characterized by production of both LTB4 (31) and LTC4 (48) is elicited in response to peritoneal administration of zymosan A. LTC4 production peaks rapidly between 30 and 60 min after zymosan A exposure and slowly subsides over several hours. LTB4 levels increase in a biphasic manner with an early plateau followed by an additional increase, which reaches a maximum 2–3-h postzymosan A treatment (31). Both peritoneal macrophages and infiltrating neutrophils have been shown to synthesize LTB4 during phagocytosis of zymosan A particles (31). The biphasic pattern of LTB4 production is, therefore, believed to result from an initial burst of synthesis by phagocytic resident macrophages, followed by a further increase in LTB4 synthesis by PMN recruited to the site of inflammation ∼2 h after zymosan injection (31).
Mice deficient in either FLAP or 5-LO show a decreased rate of edema formation and PMN migration into the peritoneal cavity after treatment with zymosan A. Unlike the observation made in AA-induced ear inflammation, a more pronounced decrease in influx of neutrophils was seen in 5-LO-deficient mice than in animals lacking only in LTB4 synthesis. This suggests that in this inflammatory model, LTC4 synthesis can contribute to the influx of neutrophils. To the contrary, however, i.p. injection of LTC4 alone does not result in PMN recruitment into the peritoneal cavity (49). A possible explanation for this finding in LTA4 hydrolase-deficient mice is that cysteinyl LT exert their effects on neutrophil influx by causing endothelial cell contraction, thereby enhancing neutrophil recruitment by any chemotactic factors in the microenvironment, including those released by resident macrophages and mast cells upon stimulation by zymosan A. Because LTB4 is thought to be the predominant chemotactic factor produced in topical AA-induced inflammation, loss of LTA4 hydrolase reduces influx of neutrophils with the same impact as that seen in the 5-LO-deficient animals. LTC4 would be unable to enhance the migration of PMN in the topical AA model because there are no accessory chemotactic factors beyond LTB4 produced in the microenvironment.
Although loss of 5-LO significantly altered the rate of peritoneal edema after treatment with zymosan A, no reduction in edema was seen in the LTA4 hydrolase-deficient animals. This clearly implicates the cysteinyl LT in the vascular permeability mediated by LT in this response. This differs from the observations made in the topical AA model, in which a decrease in edema formation (albeit smaller than that seen in the 5-LO-deficient mice) was observed in the absence of LTA4 hydrolase.
The differential contribution of LTB4 to edema formation in the two models may reflect the complexity of cells and inflammatory mediators each possesses. The numbers and types of inflammatory cells present in the peritoneal cavity certainly vary from those stimulated by epidermal exposure to AA. Peritoneal macrophages and circulating PMN, as well as other leukocytes, are available to act on zymosan granules and produce a variety of bioactive mediators in our model of peritonitis. Again it is important that the milieu of inflammatory mediators produced following zymosan exposure is not likely to be limited to AA metabolites, while eicosanoids probably predominate in the response to topical AA. As mentioned above, measurable levels of neutrophil influx are seen in the 5-LO- and LTA4 hydrolase-deficient mice after i.p. injection of zymosan A, presumably because of production of multiple chemotactic agents in this inflammatory model. This influx of neutrophils may be sufficient to potentiate the effects of LTC4 on vascular permeability in the absence of LTB4-induced chemotaxis. In the topical AA model, the role of LTB4 in potentiation of edema formation is more apparent, because neutrophil influx in this model is dependent on the synthesis of this bioactive lipid.
It is also important to note that the impact of transcellular LTA4 metabolism on the cellular and vascular components of inflammation could differ according to the site of the response. Available LTA4 carries the potential to be shunted to cysteinyl LT or to lipoxin formation in the absence of the hydrolase. The effects of abundant LTA4 on cysteinyl LT or lipoxin concentration and function in an inflammatory focus will again depend on the cells resident in a particular tissue and their ability to use LTA4. Variable LT or lipoxin synthesis could contribute to the differences we have seen in the vascular and cellular components of inflammation in the two models described above.
The third type of inflammatory response in which LT have been implicated involves PAF-induced systemic shock. Although the mechanism by which exposure to high levels of PAF leads to shock and death is poorly understood, direct evidence for a role for LT in this response comes from the observation that both 5-LO- and FLAP-deficient animals display a decreased sensitivity to PAF (24, 25). The high percentage of the LTA4 hydrolase-deficient animals surviving PAF-induced shock in this study suggests that loss of synthesis of LTB4 is responsible for the increased survival of the 5-LO-deficient animals. Thus, stimulation of LTB4 synthesis by neutrophils and pulmonary sequestration of these activated neutrophils may be important factors in determining survival after exposure to PAF. It is also possible that the absence of LTA4 hydrolase shifts the availability of free LTA4 such that production of lipoxins or other potentially protective substances is up-regulated. Neither the LTB4- nor 5-LO-deficient animals are completely protected from the effects of PAF, suggesting that, while alteration of the LT profile contributes to the physiological changes seen on treatment of animals with PAF, other pathways also influence mortality. In contrast with the clear role LTA4 hydrolase plays on PAF-induced shock, no protection was seen in IgE-mediated passive anaphylaxis or LPS-induced systemic shock in the LTA4 hydrolase-deficient animals compared with wild-type. Similar results have been obtained on examination of 5-LO- and FLAP-deficient animals in these assays.
LTA4 hydrolase is a bifunctional enzyme, possessing an in vitro aminopeptidase activity in addition to its role in AA metabolism (20, 21). The significance of this activity in vivo is not known, although some naturally occurring opioid peptides, particularly the dynorphin fragment 1–7, have been shown to be cleaved by the enzyme (23). Such cleavage can inactivate the peptides, making it impossible for them to bind specific receptors (50). A potential role for the neuropeptides, which can cause mast cell degranulation, vascular permeability, LTC4 and thromboxane B2 release, and neutrophil accumulation, has been suggested in inflammatory diseases of the skin (51). Thus, it has been hypothesized that the loss of LTA4 hydrolase could affect the inflammatory process through alteration of opioid peptide catabolism.
We have seen reductions in neutrophil recruitment, edema formation, and plasma protein extravasation in LTA4 hydrolase-deficient animals stimulated by either topical AA or i.p. zymosan A, which appear to correlate with the failure of these mice to synthesize LTB4. In inflammatory responses in which no role for LT could be established using 5-LO- and FLAP-deficient animals, there were also no differences noted in responses of LTA4 hydrolase-deficient mice. For example, despite the high LTA4 hydrolase expression in T cells, no alteration was seen in the contact hypersensitivity response to the Ag FITC. It is possible that future studies using the hydrolase-deficient mice in models of psoriasis and skin disease will reveal a role for opioid peptide cleavage by LTA4 hydrolase in the initiation and maintenance of an inflammatory response. In addition, the LTA4 hydrolase-deficient mice will continue to be useful in distinguishing between the specific functions of LTB4 and the cysteinyl LT in complex inflammatory reactions and as a model to study intercellular transfer and metabolism of LTA4.
We greatly appreciate the expertise of B. Young in help with isolation of the LtA4 hydrolase gene, A. Latour and E. Hicks with culture of ES cells, and B. Garges and K. Brigman for animal husbandry.
This work was supported by National Institutes of Health Grant PO1-DK38108 and by a grant from Pfizer Central Research to B.H.K.
Abbreviations used in this paper: LT, leukotriene; AA, arachidonic acid; BMMC, bone marrow-derived mast cell; EIA, enzyme immunoassay; ES, embryonic stem; FLAP, 5-lipoxygenase-activating protein; HSA, human serum albumin; LO, lipoxygenase; MPO, myeloperoxidase; PAF, platelet-activating factor; PMN, polymorphonuclear leukocyte.