Pseudotyping the adenovirus serotype 5 capsid with both the fibre and penton of serotype 35 enhances vascular smooth muscle cell transduction.

Ex vivo gene therapy during coronary artery bypass grafting (CABG) holds great potential to prevent excessive smooth muscle cell (SMC) proliferation, neointima formation and graft failure. The most successful preclinical strategies to date have utilised vectors based on the species C adenovirus, Ad5, which engages the Coxsackie and Adenovirus receptor (CAR) as its primary attachment receptor. Profiling receptors on human SMCs demonstrated the absence of CAR but substantial expression of the species B receptor CD46. We performed transduction experiments using Ad5 and the CD46-utilising adenovirus Ad35, and found Ad35 significantly more efficient at transducing SMCs. To evaluate whether transduction could be further augmented, we evaluated chimeric CD46-utilising Ad5/Ad35 vectors comprising the Ad5 capsid pseudotyped with the Ad35 fibre alone (Ad5/F35) or in combination with the Ad35 penton (Ad5/F35/P35). In human smooth muscle cells (hSMCs), Ad5/F35/P35 mediated significantly higher levels of transduction than either parental vector or Ad5/F35. Ex vivo transduction experiments using mouse aortas from CD46 transgenics demonstrated that Ad5/F35/P35 was significantly more efficient at transducing SMCs than the other vectors tested. Finally, ex vivo transduction and immunofluorescent colocalisation experiments using human tissue from CABG procedures confirmed the preclinical potential of Ad5/F35/P35 as an efficient vector for vascular transduction during CABG.

Systemic delivery of scAAV9 in fetal macaques facilitates neuronal transduction of the central and peripheral nervous systems.

Correction of perinatally lethal neurogenetic diseases requires efficient transduction of several cell types within the relatively inaccessible CNS. Intravenous AAV9 delivery in mouse has achieved development stage-specific transduction of neuronal cell types, with superior neuron-targeting efficiency demonstrated in prenatal compared with postnatal recipients. Because of the clinical relevance of the non-human primate (NHP) model, we investigated the ability of AAV9 to transduce the NHP CNS following intrauterine gene therapy (IUGT). We injected two macaque fetuses at 0.9 G with 1 × 10(13) vg scAAV9-CMV-eGFP through the intrahepatic continuation of the umbilical vein. Robust green fluorescent protein (GFP) expression was observed for up to 14 weeks in the majority of neurons (including nestin-positive cells), motor neurons and oligodendrocytes throughout the CNS, with a significantly lower rate of transduction in astrocytes. Photoreceptors and neuronal cell bodies in the plexiform and ganglionic retinal layers were also transduced. In the peripheral nervous system (PNS), widespread transduction of neurons was observed. Tissues harvested at 14 weeks showed substantially lower vector copy number and GFP levels, although the percentage of GFP-expressing cells remained stable. Thus, AAV9-IUGT in late gestation efficiently transduces both the CNS and PNS with neuronal predilection, of translational relevance to hereditary disorders characterized by perinatal onset of neuropathology.

In utero administration of Ad5 and AAV pseudotypes to the fetal brain leads to efficient, widespread and long-term gene expression.

The efficient delivery of genetic material to the developing fetal brain represents a powerful research tool and a means to supply therapy in a number of neonatal lethal neurological disorders. In this study, we have delivered vectors based upon adenovirus serotype 5 (Ad5) and adeno-associated virus (AAV) pseudotypes 2/5, 2/8 and 2/9 expressing green fluorescent protein to the E16 fetal mouse brain. One month post injection, widespread caudal to rostral transduction of neural cells was observed. In discrete areas of the brain these vectors produced differential transduction patterns. AAV2/8 and 2/9 produced the most extensive gene delivery and had similar transduction profiles. All AAV pseudotypes preferentially transduced neurons whereas Ad5 transduced both neurons and glial cells. None of the vectors elicited any significant microglia-mediated immune response when compared with control uninjected mice. Whole-body imaging and immunohistological evaluation of brains 9 months post injection revealed long-term expression using these non-integrating vectors. These data will be useful in targeting genetic material to discrete or widespread areas of the fetal brain with the purpose of devising therapies for early neonatal lethal neurodegenerative disease and for studying brain development.

Efficient gene delivery to the adult and fetal CNS using pseudotyped non-integrating lentiviral vectors.

Non-integrating lentiviral vectors show considerable promise for gene therapy applications as they persist as long-term episomes in non-dividing cells and diminish risks of insertional mutagenesis. In this study, non-integrating lentiviral vectors were evaluated for their use in the adult and fetal central nervous system of rodents. Vectors differentially pseudotyped with vesicular stomatitis virus, rabies and baculoviral envelope proteins allowed targeting of varied cell populations. Efficient gene delivery to discrete areas of the brain and spinal cord was observed following stereotactic administration. Furthermore, after direct in utero administration (E14), sustained and strong expression was observed 4 months into adulthood. Quantification of transduction and viral copy number was comparable when using non-integrating lentivirus and conventional integrating vector. These data support the use of non-integrating lentiviral vectors as an effective alternative to their integrating counterparts in gene therapy applications, and highlight their potential for treatment of inherited and acquired neurological disorders.

Intra-amniotic delivery of CFTR-expressing adenovirus does not reverse cystic fibrosis phenotype in inbred CFTR-knockout mice.

Due to its early onset and severe prognosis, cystic fibrosis (CF) has been suggested as a candidate disease for in utero gene therapy. In 1997, a study was published claiming that to how transient prenatal expression of CF transmembrane conductance regulator (CFTR) from an in utero-injected adenovirus vector could achieve permanent reversal of the CF intestinal pathology in adult CF knockout mice, despite the loss of CFTR transgene expression by birth. This would imply that the underlying cause of CF is a prenatal defect for which lifelong cure can be achieved by transient prenatal expression of CFTR. Despite criticism at the time of publication, no independent verification of this contentious finding has been published so far. This is vital for the development of future therapeutic strategies as it may determine whether CF gene therapy should be performed prenatally or postnatally. We therefore reinvestigated this finding with an identical adenoviral vector and a knockout CF mouse line (Cftr(tmlCam)) with a completely inbred genetic background to eliminate any effects due to genetic variation. After delivery of the CFTR-expressing adenovirus to the fetal mouse, both vector DNA and transgenic CFTR expression were detected in treated animals postpartum but statistically no significant difference in survival was observed between the Cftr(-/-) mice treated with the CFTR-adenovirus and those treated with the control vector.

Lentiviral transduction of the murine lung provides efficient pseudotype and developmental stage-dependent cell-specific transgene expression.

Gene transfer for cystic fibrosis (CF) airway disease has been hampered by the lung's innate refractivity to pathogen infection. We hypothesized that early intervention with an integrating gene transfer vector capable of transducing the lung via the lumen may be a successful therapeutic approach. An HIV-based lentiviral vector pseudotyped with the baculovirus gp64 envelope was applied to the fetal, neonatal or adult airways. Fetal intra-amniotic administration resulted in transduction of approximately 14% of airway epithelial cells, including both ciliated and non-ciliated epithelia of the upper, mid and lower airways; there was negligible alveolar or nasal transduction. Following neonatal intra-nasal administration we observed significant transduction of the airway epithelium (approximately 11%), although mainly in the distal lung, and substantial alveolar transduction. This expression was still detectable at 1 year after application. In the adult, the majority of transduction was restricted to the alveoli. In contrast, vesicular stomatitis virus glycoprotein pseudotyped virus transduced only alveoli after adult and neonatal application and no transduction was observed after fetal administration. Repeat administration did not increase transduction levels of the conducting airway epithelia. These data demonstrate that application at early developmental stages in conjunction with an appropriately pseudotyped virus provides efficient, high-level transgene expression in the murine lung. This may provide a modality for treatment for lung disease in CF.

Persistent episomal transgene expression in liver following delivery of a scaffold/matrix attachment region containing non-viral vector.

An ideal gene therapy vector should enable persistent transgene expression without limitations of safety and reproducibility. Here we report the development of a non-viral episomal plasmid DNA (pDNA) vector that appears to fulfil these criteria. This pDNA vector combines a scaffold/matrix attachment region (S/MAR) with a human liver-specific promoter (alpha1-antitrypsin (AAT)) in such a way that long-term expression is enabled in murine liver following hydrodynamic injection. Long-term expression is demonstrated by monitoring the longitudinal luciferase expression profile for up to 6 months by means of in situ bioluminescent imaging. All relevant control pDNA constructs expressing luciferase are unable to sustain significant transgene expression beyond 1 week post-administration. We establish that this shutdown of expression is due to promoter methylation. In contrast, the S/MAR element appears to inhibit methylation of the AAT promoter thereby preventing transgene silencing. Although this vector appears to be maintained as an episome throughout, we have no evidence for its establishment as a replicating entity. We conclude that the combination of a mammalian, tissue-specific promoter with the S/MAR element is sufficient to drive long-term episomal pDNA expression of genes in vivo.

Clinically applicable procedure for gene delivery to fetal gut by ultrasound-guided gastric injection: toward prenatal prevention of early-onset intestinal diseases.

Targeting gene therapy vectors to the fetal intestinal tract could provide a novel means toward prevention of the early postnatal intestinal pathology of cystic fibrosis and other conditions, such as congenital enteropathy, that cause intestinal failure. Among these conditions, cystic fibrosis is by far the most common lethal genetic disease. It is caused by a functional absence or deficiency of the cystic fibrosis transmembrane conductance regulator and manifests in the gut as meconium ileus. Prenatal treatment of genetic disease may avoid early-onset tissue damage and immune sensitization, and may target cells that are less accessible in the adult. We investigated gene transfer to the fetal gut, using a minimally invasive injection technique. First-generation replication-deficient adenoviral vectors encoding the beta-galactosidase gene and transduction-enhancing agents were injected into the stomach of early-gestation fetal sheep (n = 8, 60 days of gestation; term, 145 days) under ultrasound guidance. Reporter gene expression was observed 2 days after injection in the villi of the gastrointestinal epithelia after 5-bromo-4-chloro-3-indolyl-beta-D-galactopyranoside staining and beta-galactosidase immunohistochemistry of fetal tissues. Expression of beta-galactosidase, as measured by enzyme-linked immunosorbent assay, was enhanced after pretreatment of the fetal gut with sodium caprate, which opens tight junctions, and after adenovirus complexation with DEAE-dextran, which confers a positive charge to the virus. Instillation of the fluorocarbon perflubron after virus delivery resulted in tissue transduction from the fetal stomach to the colon. Using a clinically relevant technique, we have demonstrated widespread gene transfer to the fetal gastrointestinal epithelia.

Proof-of-concept: neonatal intravenous injection of adeno-associated virus vectors results in successful transduction of myenteric and submucosal neurons in the mouse small and large intestine.

BACKGROUND: Despite the success of viral vector technology in the transduction of the central nervous system in both preclinical research and gene therapy, its potential in neurogastroenterological research remains largely unexploited. This study asked whether and to what extent myenteric and submucosal neurons in the ileum and distal colon of the mouse were transduced after neonatal systemic delivery of recombinant adeno-associated viral vectors (AAVs). METHODS: Mice were intravenously injected at postnatal day one with AAV pseudotypes AAV8 or AAV9 carrying a cassette encoding enhanced green fluorescent protein (eGFP) as a reporter under the control of a cytomegalovirus promoter. At postnatal day 35, transduction of the myenteric and submucosal plexuses of the ileum and distal colon was evaluated in whole-mount preparations, using immunohistochemistry to neurochemically identify transduced enteric neurons. KEY RESULTS: The pseudotypes AAV8 and AAV9 showed equal potential in transducing the enteric nervous system (ENS), with 25-30% of the neurons expressing eGFP. However, the percentage of eGFP-expressing colonic submucosal neurons was significantly lower. Neurochemical analysis showed that all enteric neuron subtypes, but not glia, expressed the reporter protein. Intrinsic sensory neurons were most efficiently transduced as nearly 80% of calcitonin gene-related peptide-positive neurons expressed the transgene. CONCLUSIONS & INFERENCES: The pseudotypes AAV8 and AAV9 can be employed for gene delivery to both the myenteric and the submucosal plexus, although the transduction efficiency in the latter is region-dependent. These findings open perspectives for novel preclinical applications aimed at manipulating and imaging the ENS in the short term, and in gene therapy in the longer term.

Evaluation of prenatal intra-amniotic LAMB3 gene delivery in a mouse model of Herlitz disease.

Prenatal gene therapy has been considered for Herlitz junctional epidermolysis bullosa (H-JEB), a lethal genodermatosis caused by the absence of any of the three subunits of laminin-5, resulting from birth in widespread blistering and erosions of skin and mucosae. To investigate this strategy in an animal model, adenovirus type 5- and adeno-associated virus (AAV) type 2-derived vectors carrying a beta-galactosidase reporter gene or LAMB3 cDNA encoding the beta3 chain of laminin-5 were generated, tested for stability in amniotic fluid and evaluated in vitro on murine H-JEB keratinocytes, and in vivo by prenatal injection into the amniotic cavities of laminin-5 beta3-deficient mice. The different vectors were administered individually or combined at maximum doses on day 14 post coitum. Adenoviral vectors infected preferentially the foetal epidermis, whereas AAV delivered the transgene mainly to mucous membranes of the airways and the upper digestive tract. The LAMB3 transgene was expressed in target epithelia of newborn laminin-5 beta3-deficient mice, and the transgenic beta3 chain was shown to assemble with its endogenous partner chains, resulting in detectable amounts of laminin-5 in the basement membranes of skin and mucosae and in a lower extent of tissue separation in the skin. However, only combined delivery of the two vector types led to a minor increase of the life span of H-JEB mice. Failure to rescue diseased animals was, at least in part, due to abandonment of any conspicuous pup by the heterozygous mother. This is the first study of a prenatal gene therapy approach to a heritable blistering disorder. Although our findings indicate that prenatal combined administration of adenoviral and adeno-associated LAMB3 vectors provides therapeutic benefit to H-JEB mice, this animal model appears unsuitable for long-term investigations of the therapeutic concept.

Permanent partial phenotypic correction and tolerance in a mouse model of hemophilia B by stem cell gene delivery of human factor IX.

Immune responses against an introduced transgenic protein are a potential risk in many gene replacement strategies to treat genetic disease. We have developed a gene delivery approach for hemophilia B based on lentiviral expression of human factor IX in purified hematopoietic stem cells. In both normal C57Bl/6J and hemophilic 129/Sv recipient mice, we observed the production of therapeutic levels of human factor IX, persisting for at least a year with tolerance to human factor IX antigen. Secondary and tertiary recipients also demonstrate long-term production of therapeutic levels of human factor IX and tolerance, even at very low levels of donor chimerism. Furthermore, in hemophilic mice, partial functional correction of treated mice and phenotypic rescue is achieved. These data show the potential of a stem cell approach to gene delivery to tolerize recipients to a secreted foreign transgenic protein and, with appropriate modification, may be of use in developing treatments for other genetic disorders.

Gene therapy progress and prospects: fetal gene therapy--first proofs of concept--some adverse effects.

Somatic gene delivery in utero is a novel approach to gene therapy for genetic disease based on the hypothesis that prenatal intervention may avoid the development of severe manifestations of early-onset disease, allow targeting of otherwise inaccessible tissues including expanding stem cell populations, induce tolerance against the therapeutic transgenic protein and thereby provide permanent somatic gene correction. This approach is particularly relevant in relation to prenatal screening programmes for severe genetic diseases as it could offer prevention as a third option to families faced with the prenatal diagnosis of a genetically affected child. Most investigations towards in utero gene therapy have been performed on mice and sheep fetuses as model animals for human disease and for the application of clinically relevant intervention techniques such as vector delivery by minimally invasive ultrasound guidance. Other animals such as dogs may serve as particular disease models and primates have to be considered in immediate preparation for clinical trials. Proof of principle for the hypothesis of fetal gene therapy has been provided during the last 2 years in mouse models for Crigler Najjar Disease, Leber's congenital amaurosis, Pompe's disease and haemophilia B showing long-term postnatal therapeutic effects and tolerance of the transgenic protein after in utero gene delivery. However, recently we have also observed a high incidence of liver tumours after in utero application of an early form of third-generation equine infectious anaemia virus vectors with SIN configuration. These findings highlight the need for more investigations into the safety and the ethical aspects of in utero gene therapy as well as for science-based public information on risks and benefits of this preventive gene therapy approach before application in humans can be contemplated.

Factors influencing adenovirus-mediated airway transduction in fetal mice.

Intra-amniotic injection of adenovirus allows transduction of the fetal airways following natural fetal breathing movements. This administration method is promising for use in gene therapy for cystic fibrosis and other diseases for which the main target for exogenous gene expression is the lung. Here we have investigated factors that may affect the efficacy of gene transfer to the murine fetal lung. We examined marker compound distribution and transgene expression (from a first-generation adenoviral vector) at different stages of development. This demonstrated that fetal breathing movements at 15-16 days of gestation are of sufficient intensity to carry marker/vector into the fetal lungs. These movements can be significantly stimulated by the combination of intra-amniotic theophylline administration and postoperative exposure of the dam to elevated CO(2) levels. However, the most important factor for efficient and consistent pulmonary transgene delivery is the dose of adenoviral vector used, as both the degree of transduction and the percentage of lungs transduced increases with escalating viral dose.

Arginase in glomerulonephritis.

Arginase metabolizes L-arginine to L-ornithine and urea. Two arginase isoforms, AI (liver arginase) and AII (ubiquitously expressed, functions unknown), have been identified. It is clear that arginases potentially have important roles in addition to urea generation for high concentrations are present at inflammatory sites. Regulation occurs through cytokines, substrate competition and products of nitric oxide (NO) metabolism. The functions of arginases at inflammatory sites are unknown, but may include regulation of apoptosis and NO production and generation of structural and cellular protein precursors. In glomerulonephritis there is increased arginase activity in nephritic glomeruli following a pattern similar to that in wound healing. The level can be further increased by NO inhibition suggesting substrate competition. The potential sources in the inflamed glomerulus include infiltrating leucocytes and mesangial cells, and the predominant isoform expressed is AI (AII predominates under physiological conditions). The recent identification of different isoforms of arginase has been an important step towards understanding the significance of arginase activity in glomerulonephritis.

Induced nitric oxide (NO) synthesis in heterologous nephrotoxic nephritis; effects of selective inhibition in neutrophil-dependent glomerulonephritis.

Increased NO synthesis, due to inducible NO synthase (iNOS) activity, is found in macrophage-associated glomerulonephritis. Little is known about NO in neutrophil-dependent immune complex inflammation, and its role remains controversial. We therefore studied early phase heterologous nephrotoxic nephritis (HNTN) induced in rats by nephrotoxic globulin and the effects of selective iNOS inhibition of this model. At 2 h of the model iNOS mRNA was induced and nitrite (NO-2) was generated in glomeruli incubated ex vivo (5.2 +/- 1.0 nmol/2000 glomeruli per 24 h). There were 14.7 +/- 2.2 polymorphonuclear neutrophils (PMN)/glomerulus (normal controls 0.1 +/- 0.1). At 8 h urinary protein was 69 +/- 15.3 (normal controls 0. 6 +/- 0.2 mg/24 h). Peritoneal PMN expressed iNOS and produced significant NO-2 (basal 11.2 +/- 0.3 nmol/106 cells per 24 h). Selective iNOS inhibition with L-N6-(1-iminoethyl)-lysine (L-NIL) in vitro inhibited nephritic glomerular and PMN NO-2 synthesis. In HNTN L-NIL in vivo significantly suppressed elevated plasma NO-2/NO-3 levels (representative experiment: 17 +/- 2 microM, untreated 40 +/- 4 microM, P = < 0.01, normal control 18 +/- 2 microM). This inhibition did not affect leucocyte infiltration into glomeruli or induce thrombosis. There was no consistent effect on proteinuria. This is the first demonstration of glomerular iNOS induction and high output NO production in the acute phase of PMN-dependent acute immune complex glomerulonephritis. Selective iNOS inhibition does not affect the primary mechanism of injury (leucocyte infiltration) in this model.

Arginase activity is modulated by IL-4 and HOArg in nephritic glomeruli and mesangial cells.

Arginase shares a common substrate, L-arginine, with nitric oxide synthase (NOS). Both enzymes are active at inflammatory sites. To understand regulation of arginase and its relationship to nitric oxide (NO) production, we studied effects of NG-hydroxy-L-arginine (HOArg) and interleukin-4 (IL-4) on urea and NO2- synthesis by glomeruli during rat immune glomerulonephritis and compared these with macrophages and glomerular mesangial cells (MC). In nephritic glomeruli, elicited macrophages, and MC stimulated with IL-1 and adenosine 3',5'-cyclic monophosphate agonists, increased arginase and induced NOS activity was found. Urea production was inhibited by HOArg and increased by IL-4. NO inhibition [NG-monomethyl-L-arginine (L-NMMA)] increased arginase activity in nephritic glomeruli and macrophages but not MC. NO2- synthesis was inhibited by L-NMMA and IL-4. It was increased with HOArg under conditions of NO inhibition. In contrast, in normal glomeruli and basal MC, where there was no induced NO synthesis, IL-4 had no effect on arginase activity, whereas HOArg consistently reduced it in glomeruli only. Type II arginase (Arg II) mRNA was detected in normal glomeruli; nephritic glomeruli expressed both Arg I and Arg II mRNAs. This is the first demonstration of arginase modulation in glomeruli and MC and of the expression of arginase isoforms in glomeruli. The differential responses to two endogenous compounds generated by inflammation suggest this may be part of coordinated regulation of arginase and inducible NOS in immune injury, whereby arginase is inhibited during high-output NO production and stimulated with NO suppression. This, together with control of arginase and NOS isoforms, may be important in controlling the balance of inflammatory and repair mechanisms.

Arginase AI is upregulated in acute immune complex-induced inflammation.

Previous studies have shown high arginase activity at inflammatory sites. Arginase converts L-arginine to L-ornithine, sharing a common substrate with nitric oxide synthase. It exists as two isoforms, AI and AII. While the function of liver arginase (AI) in ureagenesis has been defined, the role and isoform of arginase in cells without a complete urea cycle are unknown. We therefore determined arginase isoform mRNA expression in glomerular acute immune complex inflammation, and its cultured constituent cells. AI was induced in nephritic glomeruli, and in mesangial cells stimulated with IL-4 and cAMP, and was present in elicited neutrophils and macrophages. AII was constitutively expressed. Our data strongly suggest that AI, thought to be restricted to the liver, accounts for high arginase activity at inflammatory sites where it may limit high output nitric oxide production and generate polyamines and proline essential for cell proliferation and matrix production. This identification of AI in inflamed tissue is an important step for understanding the consequences of increased arginase activity.

Inducible nitric oxide synthase induction in Thy 1 glomerulonephritis is complement and reactive oxygen species dependent.

Thy 1 glomerulonephritis (GN) is a rat model of complement-dependent immune mesangial injury with induced glomerular nitric oxide (NO) synthesis. To examine mechanisms of inducible nitric oxide synthase (iNOS) induction, we studied the effects of treatment with the antioxidant N-acetyl-cysteine (NAC) and soluble complement receptor 1 (sCR1). Thy 1 GN was induced by intravenous anti-Thy 1 antibody. Glomeruli were isolated and kidney tissue taken from 30 min to 24 h after induction. Nitrite (NO-2) synthesis, luminol chemiluminescence for reactive oxygen species (ROS), and iNOS and cytokine mRNA were assayed in isolated glomeruli. Mesangial injury (mesangiolysis) and leucocyte infiltration were quantitated on tissue sections. NAC (i.p. 1,000 mg/kg, 1 h prior to anti-Thy 1) reduced glomerular NO-2 synthesis (3.5 +/- 0.66 vs. untreated 8.2 +/- 1.1, p = 0.02), and iNOS mRNA expression, and abolished enhanced chemiluminescence. In vitro incubation of nephritic glomeruli with 20 mM NAC also suppressed nitrite production (4.7 +/- 0.8 vs. untreated 12.2 +/- 0. 7 nmol NO-2/2,000 glomeruli/48 h, p = 0.003), and chemiluminescence. In NAC-treated animals, neutrophil infiltration (0.5 +/- 0 vs. untreated 9.6 +/- 1.6 glomerulus, p = 0.0005), and macrophage infiltration (1.7 +/- 0.4 vs. untreated 12.0 +/- 0.1, p = 0.006) were abolished, and mesangiolysis was significantly reduced (45.9 +/- 1.3 vs. untreated 34.4 +/- 2.1 cells/glomerulus, p = 0.009). NAC did not inhibit anti-Thy 1 antibody deposition. C1q was unaffected, but C3 was reduced. sCR1 treatment prevented iNOS mRNA induction, the enhanced chemiluminescence, and the neutrophil infiltration at 1 h. IL-1beta and TNFalpha mRNAs were not affected by either NAC or sCR1. These results show that NAC inhibits iNOS induction and NO synthesis in this model, and suppresses ROS synthesis and injury. They suggest that complement-dependent ROS generation is the critical initiating event that follows fixation of anti-Thy 1 antibody.