The Cysteine Protease Legumain Is Upregulated by Vitamin D and Is a Regulator of Vitamin D Metabolism in Mice.

Legumain is a lysosomal cysteine protease that has been implicated in an increasing amount of physiological and pathophysiological processes. However, the upstream mechanisms regulating the expression and function of legumain are not well understood. Here, we provide in vitro and in vivo data showing that vitamin D3 (VD3) enhances legumain expression and function. In turn, legumain alters VD3 bioavailability, possibly through proteolytic cleavage of vitamin D binding protein (VDBP). Active VD3 (1,25(OH)2D3) increased legumain expression, activity, and secretion in osteogenic cultures of human bone marrow stromal cells. Upregulation of legumain was also observed in vivo, evidenced by increased legumain mRNA in the liver and spleen, as well as increased legumain activity in kidneys from wild-type mice treated with 25(OH)D3 (50 µg/kg, subcutaneously) for 8 days compared to a control. In addition, the serum level of legumain was also increased. We further showed that active legumain cleaved purified VDBP (55 kDa) in vitro, forming a 45 kDa fragment. In vivo, no VDBP cleavage was found in kidneys or liver from legumain-deficient mice (Lgmn-/-), whereas VDBP was cleaved in wild-type control mice (Lgmn+/+). Finally, legumain deficiency resulted in increased plasma levels of 25(OH)D3 and total VD3 and altered expression of key renal enzymes involved in VD3 metabolism (CYP24A1 and CYP27B1). In conclusion, a regulatory interplay between VD3 and legumain is suggested.

A Novel Porcine Model of Ischemia-Reperfusion Injury After Cross-Clamping the Thoracic Aorta Revealed Substantial Cardiopulmonary, Thromboinflammatory and Biochemical Changes Without Effect of C1-Inhibitor Treatment.

Ischemic injury worsens upon return of blood and innate immunity including the complement system play a central role in ischemia-reperfusion injury (IRI) as in thoracic aortic surgery. Complement component1 inhibitor (C1-INH) has been shown to reduce IRI and is a broad-acting plasma cascade inhibitor. We established a new porcine model of IRI by cross-clamping the thoracic aorta and evaluated the global changes occurring in organ function, systemic inflammatory response and organ damage with or without treatment with C1-INH-concentrate. Twenty-four piglets (8.8-11.1 kg) underwent 45 minutes clamping of the thoracic aorta at the Th8 level. Upfront 12 piglets received human saline and 12 received C1-INH (250 IU/kg) intravenously. Three sham animals received thoracic opening without clamping. Reperfusion lasted 5 hours. We studied ten cardiorespiratory markers, three hematologic markers, eleven inflammatory markers, and twelve organ damage markers over the whole experimental period. Postmortem tissue homogenates from seven organs were examined for inflammatory markers and analysed by two-way repeated-measures ANOVA, area under the curve or unpaired t-tests. By excluding sham and combining treated and untreated animals, the markers reflected a uniform, broad and severe organ dysfunction. The mean and range fold change from before cross-clamp onset to maximum change for the different groups of markers were: cardiorespiratory 1.4 (0.2-3.7), hematologic 1.9 (1.2-2.7), plasma inflammatory 19.5 (1.4-176) and plasma organ damage 2.9 (1.1-8.6). Treatment with C1-INH had only a marginal effect on the IRI-induced changes, reaching statistical significance only for the plasma complement activation product TCC (p=0.0083) and IL-4 (p=0.022) and INF-α (p=0.016) in the colon tissue. In conclusion, the present novel model of porcine global IRI is forceful with regards to central markers and could generally be applicable for pathophysiological studies. C1-INH treatment had no significant effect, but the model allows for future testing of other drugs attenuating IRI globally.

Legumain is upregulated in acute cardiovascular events and associated with improved outcome - potentially related to anti-inflammatory effects on macrophages.

BACKGROUND AND AIMS: We have previously found increased levels of the cysteine protease legumain in plasma and plaques from patients with carotid atherosclerosis. This study further investigated legumain during acute cardiovascular events. METHODS: Circulating levels of legumain from patients and legumain released from platelets were assessed by enzyme-linked-immunosorbent assay. Quantitative PCR and immunoblotting were used to study expression, while localization was visualized by immunohistochemistry. RESULTS: In the SUMMIT Malmö cohort (n = 339 with or without type 2 diabetes and/or cardiovascular disease [CVD], and 64 healthy controls), the levels of circulating legumain were associated with the presence of CVD in non-diabetics, with no relation to outcome. In symptomatic carotid plaques and in samples from both coronary and intracerebral thrombi obtained during acute cardiovascular events, legumain was co-localized with macrophages in the same regions as platelets. In vitro, legumain was shown to be present in and released from platelets upon activation. In addition, THP-1 macrophages exposed to releasate from activated platelets showed increased legumain expression. Interestingly, primary peripheral blood mononuclear cells stimulated with recombinant legumain promoted anti-inflammatory responses. Finally, in a STEMI population (POSTEMI; n = 272), patients had significantly higher circulating legumain before and immediately after percutaneous coronary intervention compared with healthy controls (n = 67), and high levels were associated with improved outcome. CONCLUSIONS: Our data demonstrate for the first time that legumain is upregulated during acute cardiovascular events and is associated with improved outcome.

Sepsis causes right ventricular myocardial inflammation independent of pulmonary hypertension in a porcine sepsis model.

INTRODUCTION: Right ventricular (RV) myocardial dysfunction is a common feature in septic shock. It can worsen outcome, but the etiology is poorly understood. Pulmonary artery hypertension (PAH) plays a part in the pathogenesis of the right heart dysfunction in sepsis but its importance is unknown. In pigs, PAH in sepsis is substantial and the translational value of porcine sepsis models therefore questioned. We hypothesized that porcine sepsis causes a myocardial inflammatory response which leads to myocardial dysfunction independent of PAH. MATERIALS AND METHODS: Sepsis was induced by Escherichia coli-infusion in 10 pigs resulting in PAH and increased right ventricular pressure (RVP). The same degree of RVP was achieved by external pulmonary artery banding (PAB) in a consecutive series of 6 animals. RESULTS: Sepsis, but not PAB, led to increase in endothelial damage marker PAI-1 and cytokines TNF and IL-6 (all p<0.05) in plasma. In myocardium, TNF and IL-6 were significantly elevated in sepsis, TNF in both ventricles and IL-6 mostly in RV, while IL-1ß, IL-18 and C5a were significantly higher in RV compared to LV after PAB (all p<0.05). Myocardial mRNA levels of IL-1ß, IL-6, IL-18, IP-10, E-selectin and PAI-1 were significantly elevated in RV and LV during sepsis compared to PAB, while Caspase-1 was decreased in septic compared to PAB animals (all p<0.05). Cathepsin L activity was increased in RV by PAB, while sepsis inhibited this response. Escherichia coli-induced sepsis caused myocardial inflammation independent of PAH. CONCLUSION: Prominent PAH should therefore not exclude porcine sepsis models to further our understanding of human sepsis.

Mammalian legumain - A lysosomal cysteine protease with extracellular functions?

The cysteine protease legumain (asparaginyl endopeptidase, AEP) plays important roles in normal physiology but is also associated with several disorders, such as atherosclerosis, osteoporosis, cancer and neurodegenerative diseases. The functional roles of legumain have mainly been associated with the presence in lysosomes where legumain is active and mediates processing of multiple proteins, such as the conversion of single to double chain forms of cysteine cathepsins. However, in recent years, a number of studies point to extracellular roles of legumain in addition to the pivotal roles in the lysosomes. In this review, recent knowledge on novel extracellular functions of this protease will be addressed and new discoveries in relation to the diseases mentioned above will be presented.

Lansoprazole inhibits the cysteine protease legumain by binding to the active site.

Proton pump inhibitors (PPIs) are prodrugs used in the treatment of peptic ulcer diseases. Once activated by acidic pH, the PPIs subsequently inhibit the secretion of gastric acid by covalently forming disulphide bonds with the SH groups of the parietal proton pump, that is the H+ /K+ -ATPase. Long-term use of PPIs has been associated with numerous adverse effects, including bone fractures. Considering the mechanism of activation, PPIs could also be active in acidic micro-environments such as in lysosomes, tumours and bone resorption sites. We suggested that the SH group in the active site of cysteine proteases could be susceptible for inhibition by PPIs. In this study, the inhibition by lansoprazole was shown on the cysteine proteases legumain and cathepsin B by incubating purified proteases or cell lysates with lansoprazole at different concentrations and pH conditions. The mechanism of legumain inhibition was shown to be a direct interaction of lansoprazole with the SH group in the active site, and thus blocking binding of the legumain-selective activity-based probe MP-L01. Lansoprazole was also shown to inhibit both legumain and cathepsin B in various cell models like HEK293, monoclonal legumain over-expressing HEK293 cells (M38L) and RAW264.7 macrophages, but not in human bone marrow-derived skeletal (mesenchymal) stem cells (hBMSC-TERT). During hBMSC-TERT differentiation to osteoblasts, lansoprazole inhibited legumain secretion, alkaline phosphatase activity, but had no effects on in vitro mineralization capacity. In conclusion, lansoprazole acts as a direct covalent inhibitor of cysteine proteases via disulphide bonds with the SH group in the protease active site. Such inhibition of cysteine proteases could explain some of the off-target effects of PPIs.

Glycosylation is important for legumain localization and processing to active forms but not for cystatin E/M inhibitory functions.

The asparaginyl endopeptidase legumain and its inhibitor cystatin E/M are endogenously glycosylated. However, little is known about the nature of the carbohydrate groups and whether they affect the functions of these proteins. In this study both glycosylated and unglycosylated forms of legumain and cystatin E/M were studied. HEK293 cell lines stably over-expressing legumain or cystatin E/M, and HCT116 cells were used as cell models, and mature legumain was purified from bovine kidneys. To obtain unglycosylated proteins, cells were treated with tunicamycin, an inhibitor of N-linked glycosylation, whereas PNGase F and Endo H were used to characterize the glycosylation types. Cells were incubated with glycosylated, unglycosylated proteins and/or legumain selective activity-based probe, and legumain and/or cystatin E/M was studied by activity measurement, ELISA or immunoblotting in cell lysates or conditioned media. Legumain and probe in whole cells were studied by immunofluorescence. The carbohydrates on legumain were shown to be of the hybrid or high mannose type, whereas cystatin E/M was characterized as complex mannose-linked. While glycosylated prolegumain was able to autoactivate, the unglycosylated form was not, and addition of glycosaminoglycans did not facilitate autoactivation of unglycosylated prolegumain. Glycosylated prolegumain was internalized and processed to the mature active form, but no internalization of unglycosylated prolegumain was observed. A Cy5-labelled legumain specific activity-based probe (MP-L09) was synthesized and shown to be a novel tool to study intracellular legumain. Also, internalization of mature legumain (36 kDa) was visualized both alone and complexed with probe. Contrary to the importance of legumain glycosylation, both glycosylated and unglycosylated cystatin E/M showed similar capacity to inhibit legumain. In conclusion, glycosylation of prolegumain is necessary for correct processing to active forms and internalization, whereas the inhibitory property of cystatin E/M is independent of the glycosylation status.

Legumain Regulates Differentiation Fate of Human Bone Marrow Stromal Cells and Is Altered in Postmenopausal Osteoporosis.

Secreted factors are a key component of stem cell niche and their dysregulation compromises stem cell function. Legumain is a secreted cysteine protease involved in diverse biological processes. Here, we demonstrate that legumain regulates lineage commitment of human bone marrow stromal cells and that its expression level and cellular localization are altered in postmenopausal osteoporotic patients. As shown by genetic and pharmacological manipulation, legumain inhibited osteoblast (OB) differentiation and in vivo bone formation through degradation of the bone matrix protein fibronectin. In addition, genetic ablation or pharmacological inhibition of legumain activity led to precocious OB differentiation and increased vertebral mineralization in zebrafish. Finally, we show that localized increased expression of legumain in bone marrow adipocytes was inversely correlated with adjacent trabecular bone mass in a cohort of patients with postmenopausal osteoporosis. Our data suggest that altered proteolytic activity of legumain in the bone microenvironment contributes to decreased bone mass in postmenopausal osteoporosis.

Increased levels of legumain in plasma and plaques from patients with carotid atherosclerosis.

BACKGROUND AND AIMS: The cysteine protease legumain has been shown to be up-regulated in unstable atherosclerotic plaques. This study aims to further elucidate legumain in atherosclerosis, by examining legumain in plasma and carotid plaques from patients with carotid stenosis. Furthermore, legumain secretion from monocyte-derived macrophages treated with atherogenic lipids during macrophage polarization was studied. METHODS: Plasma levels of legumain from patients with carotid stenosis (n = 254), healthy controls (n = 91), and secreted from monocyte-derived macrophages were assessed by enzyme-linked-immunosorbent assay. Quantitative PCR and immunoblotting of legumain were performed on isolated plaques and legumain localization was visualized by immunohistochemistry and fluorescence microscopy. Monocyte-derived macrophages polarized to M1 or M2 macrophages were treated with VLDL, oxLDL or cholesterol crystals (CC) and the level of legumain analysed. RESULTS: Patients with carotid stenosis had significantly higher levels of plasma legumain compared with healthy controls (median 2.0 versus 1.5 ng/ml, respectively; p = 0.003), although there was no correlation between the level of legumain and the degree of stenosis, and legumain was not an independent factor to identify patients with carotid plaques. Moreover, patients with symptoms the last 2 months had higher expressions of mature legumain, cystatin C and E/M, and the macrophage markers CD80 (M1) and CD163 (M2). Legumain co-localized with both M1 and M2 macrophages within plaques, whereas legumain mRNA expression was significantly higher (p < 0.0001) in plaques compared to non-atherosclerotic arteries (controls). Furthermore, in vitro studies showed significantly increased secretion of legumain from pro-inflammatory M1 compared to pro-resolving M2 macrophages (p = 0.014), and particularly in M1 treated with CC. In plaques, legumain was localized to structures resembling foam cells. CONCLUSIONS: Legumain is increased in both plasma and plaques of patients with carotid stenosis and might be a new and early biomarker of atherosclerosis.

Legumain expression, activity and secretion are increased during monocyte-to-macrophage differentiation and inhibited by atorvastatin.

Macrophages express several lysosomal cysteine proteases such as cathepsins and legumain. In this study, we assessed the expression, activity and secretion of legumain in cellular models of monocytes/macrophages. Macrophages were derived from M-CSF- or GM-CSF/IFNγ-stimulated human primary monocytes (M2 and M1, respectively), PMA-treated human THP-1 cells, or murine RAW264.7 macrophages. In both primary monocytes and THP-1 cells, monocyte-to-macrophage differentiation caused highly increased cellular expression and activity of legumain. Also, secretion of legumain from macrophages, but not from monocytes, was observed. Notably, M2 macrophages expressed significantly higher levels of active legumain than M1 macrophages, which are not previously reported. Legumain mRNA has been shown to be down-regulated in monocytes isolated from patients treated with the HMG-CoA reductase inhibitor atorvastatin. Interestingly, in our study, the active legumain produced by M2 macrophages was found to be inhibited by atorvastatin, which was reflected in aberrant cellular expression and processing.

Synthesis of a novel legumain-cleavable colchicine prodrug with cell-specific toxicity.

Conventional chemotherapy has undesirable toxic side-effects to healthy tissues due to low cell selectivity of cytotoxic drugs. One approach to increase the specificity of a cytotoxic drug is to make a less toxic prodrug which becomes activated at the tumour site. The cysteine protease legumain have remarkable restricted substrate specificity and is the only known mammalian asparaginyl (Asn) endopeptidase. Over-expression of legumain is reported in cancers and unstable atherosclerotic plaques, and utilizing legumain is a promising approach to activate prodrugs. In this study we have synthesized the legumain-cleavable peptide sequence N-Boc-Ala-Ala-Asn-Val-OH. The peptide was subsequently conjugated to deacetyl colchicine during three steps to produce Suc-Ala-Ala-Asn-Val-colchicine (prodrug) with >90% chemical purity. Several cell lines with different expressions and activities of legumain were used to evaluate the general toxicity, specificity and efficacy of the microtubule inhibitor colchicine, valyl colchicine and the legumain-cleavable colchicine prodrug. The prodrug was more toxic to the colorectal cancer HCT116 cells (expressing both the 36kDa active and 56kDa proform of legumain) than SW620 cells (only expressing the 56kDa prolegumain) indicating a relationship between toxicity of the prodrug and activity of legumain in the cells. Also, in monoclonal legumain over-expressing HEK293 cells the prodrug toxicity was higher compared to native HEK293 cells. Furthermore, co-administration of the prodrug either with the potent legumain inhibitor cystatin E/M or the endocytosis inhibitor Dyngo-4a inhibited cell death, indicating that the prodrug toxicity was dependent on both asparaginyl endopeptidase activity and endocytosis. This colchicine prodrug adds to a legumain-activated prodrug strategy approach and could possibly be of use both in targeted anticancer and anti-inflammatory therapy.

Simvastatin inhibits glucose metabolism and legumain activity in human myotubes.

Simvastatin, a HMG-CoA reductase inhibitor, is prescribed worldwide to patients with hypercholesterolemia. Although simvastatin is well tolerated, side effects like myotoxicity are reported. The mechanism for statin-induced myotoxicity is still poorly understood. Reports have suggested impaired mitochondrial dysfunction as a contributor to the observed myotoxicity. In this regard, we wanted to study the effects of simvastatin on glucose metabolism and the activity of legumain, a cysteine protease. Legumain, being the only known asparaginyl endopeptidase, has caspase-like properties and is described to be involved in apoptosis. Recent evidences indicate a regulatory role of both glucose and statins on cysteine proteases in monocytes. Satellite cells were isolated from the Musculus obliquus internus abdominis of healthy human donors, proliferated and differentiated into polynuclear myotubes. Simvastatin with or without mevalonolactone, farnesyl pyrophosphate or geranylgeranyl pyrophosphate were introduced on day 5 of differentiation. After 48 h, cells were either harvested for immunoblotting, ELISA, cell viability assay, confocal imaging or enzyme activity analysis, or placed in a fuel handling system with [¹4C]glucose or [³H]deoxyglucose for uptake and oxidation studies. A dose-dependent decrease in both glucose uptake and oxidation were observed in mature myotubes after exposure to simvastatin in concentrations not influencing cell viability. In addition, simvastatin caused a decrease in maturation and activity of legumain. Dysregulation of glucose metabolism and decreased legumain activity by simvastatin points out new knowledge about the effects of statins on skeletal muscle, and may contribute to the understanding of the myotoxicity observed by statins.

Autoactivation of prolegumain is accelerated by glycosaminoglycans.

The cysteine protease legumain participates in several biological and pathological processes including tumour invasion and metastasis. Legumain is synthesized as a zymogen and undergoes pH-dependent autoactivation of the proform in order to reach an enzymatically active form. Here we demonstrate that the naturally occurring polyanionic glycosaminoglycans (GAGs) chondroitin 4-sulphate (C4S), chondroitin 6-sulphate (C6S), chondroitin 4,6-sulphate (C4,6S), heparin, heparan sulphate (HS) as well as chondroitin sulphate (CS)-derived decasaccharides accelerated the autocatalytic activation of prolegumain through ionic interactions in a concentration-, size- and time-dependent manner at pH 4.0. In contrast, at pH 5.0 only C4S and C4,6S were able to promote prolegumain activation, while CS-derived decasaccharides, C6S, heparin and HS lost their effect at this pH.

The role of bradykinin and the effect of the bradykinin receptor antagonist icatibant in porcine sepsis.

Bradykinin (BK) is regarded as an important mediator of edema, shock, and inflammation during sepsis. In this study, we evaluated the contribution of BK in porcine sepsis by blocking BK and by measuring the stable BK metabolite, BK1-5, using anesthetized pigs. The effect of BK alone, the efficacy of icatibant to block this effect, and the recovery of BK measured as plasma BK1-5 were first investigated. Purified BK injected intravenously induced an abrupt fall in blood pressure, which was completely prevented by pretreatment with icatibant. BK1-5 was detected in plasma corresponding to the doses given. The effect of icatibant was then investigated in an established model of porcine gram-negative sepsis. Neisseria meningitidis was infused intravenously without any pretreatment (n = 8) or pretreated with icatibant (n = 8). Negative controls received saline only. Icatibant-treated pigs developed the same degree of severe sepsis as did the controls. Both groups had massive capillary leakage, leukopenia, and excessive cytokine release. The plasma level of BK1-5 was low or nondetectable in all pigs. The latter observation was confirmed in supplementary studies with pigs undergoing Escherichia coli or polymicrobial sepsis induced by cecal ligation and puncture. In conclusion, icatibant completely blocked the hemodynamic effects of BK but had no beneficial effects on N. meningitidis-induced edema, shock, and inflammation. This and the fact that plasma BK1-5 in all the septic pigs was virtually nondetectable question the role of BK as an important mediator of porcine sepsis. Thus, the data challenge the current view of the role of BK also in human sepsis.

Autophagic activity measured in whole rat hepatocytes as the accumulation of a novel BHMT fragment (p10), generated in amphisomes by the asparaginyl proteinase, legumain.

To investigate the stepwise autophagic-lysosomal processing of hepatocellular proteins, the abundant cytosolic enzyme, betaine:homocysteine methyltransferase (BHMT) was used as a probe. Full-length (45 kDa) endogenous BHMT was found to be cleaved in an autophagy-dependent (3-methyladenine-sensitive) manner in isolated rat hepatocytes to generate a novel N-terminal 10-kDa fragment (p10) identified and characterized by mass spectrometry. The cleavage site was consistent with cleavage by the asparaginyl proteinase, legumain and indeed a specific inhibitor of this enzyme (AJN-230) was able to completely suppress p10 formation in intact cells, causing instead accumulation of a 42-kDa intermediate. To prevent further degradation of p10 or p42 by the cysteine proteinases present in autophagic vacuoles, the proteinase inhibitor leupeptin had to be present. Asparagine, an inhibitor of amphisome-lysosome fusion, did not detectably impede either p42 or p10 formation, indicating that BHMT processing primarily takes place in amphisomes rather than in lysosomes. Lactate dehydrogenase (LDH) was similarly degraded primarily in amphisomes by leupeptin-sensitive proteolysis, but some additional leupeptin-resistant LDH degradation in lysosomes was also indicated. The autophagic sequestration of BHMT appeared to be nonselective, as the accumulation of p10 (in the presence of leupeptin) or of its precursors (in the additional presence of AJN-230) proceeded at approximately the same rate as the model autophagic cargo, LDH. The complete lack of a cytosolic background makes p10 suitable for use in a "fragment assay" of autophagic activity in whole cells. Incubation of hepatocytes with ammonium chloride, which neutralizes amphisomes as well as lysosomes, caused rapid, irreversible inhibition of legumain activity and stopped all p10 formation. The availability of several methods for selective targeting of legumain in intact cells may facilitate functional studies of this enigmatic enzyme, and perhaps suggest novel ways to reduce its contribution to cancer cell metastasis or autoimmune disease.

CD14 inhibition efficiently attenuates early inflammatory and hemostatic responses in Escherichia coli sepsis in pigs.

Sepsis is a severe infection-induced systemic inflammatory syndrome. Inhibition of downstream inflammatory mediators of sepsis, e.g., TNF-alpha, has failed in clinical trials. The aim of this study was to investigate the effects of inhibiting CD14, a key upstream innate immunity molecule, on the early inflammatory and hemostatic responses in a pig model of gram-negative sepsis. The study comprised two arms, whole live Escherichia coli bacteria and E. coli lipopolysaccharide (LPS) (n=25 and n=9 animals, respectively). The animals were allocated into treatment (anti-CD14) and control (IgG isotype or saline) groups. Inflammatory, hemostatic, physiological, and microbiological parameters were measured. The proinflammatory cytokines TNF-alpha, IL-1beta, IL-6, and IL-8, but not the anti-inflammatory cytokine IL-10, were efficiently inhibited by anti-CD14. Furthermore, anti-CD14 preserved the leukocyte count and significantly reduced granulocyte enzyme matrix metalloproteinase-9 release and expression of the granulocyte membrane activation molecule wCD11R3 (pig CD11b). The hemostatic markers thrombin-antithrombin III complexes and plasminogen activator inhibitor-1 were significantly attenuated. Anti-CD14 did not affect LPS or E. coli DNA levels. This study documents that CD14 inhibition efficiently attenuates the proinflammatory cytokine response and granulocyte activation and reverses the procoagulant state but does not interfere with LPS levels or bacterial counts in E. coli-induced sepsis.-Thorgersen, E. B., Hellerud, B. C., Nielsen, E. W., Barratt-Due, A., Fure, H., Lindstad, J. K., Pharo, A., Fosse, E., Tønnessen, T. I., Johansen, H. T., Castellheim, A., Mollnes, T. E. CD14 inhibition efficiently attenuates early inflammatory and hemostatic responses in Escherichia coli sepsis in pigs.