Recycling and reuse of kerf-loss silicon from diamond wire sawing for photovoltaic industry.

With the rapid growth of the global photovoltaic (PV) industry, the waste from PV industry cannot be ignored, especially the solid wastes from silicon kerf loss and the used quartz crucibles from silicon casting. The silicon kerf loss during wafer sawing was nearly 160,000 tonnes and the used crucible waste was nearly 70,000 tonnes in 2017. With the transition of wafering technology from the slurry-based wire to diamond wire sawing, recycling and reuse of kerf-loss silicon have become more feasible due to the lower impurity contents. In this paper, we aimed to find a simple approach to recycle the kerf loss and identify the purity for reuse. We first analyzed the contents of the as-received kerf-loss silicon from the industry. Then, suitable acids and refining procedure were proposed. The metals, especially nickel, could be easily reduced to several ppmw, boron and phosphorous to sub-ppmw, and carbon to several hundred ppmw, while oxygen was less than 5 wt%. Although the purity of the recycled silicon was not sufficient for casting feedstock, it had a comparable purity of about 5 N with the commercial silicon nitride releasing agent and crucibles used in silicon casting for solar cells. Because the nitride crucibles could be reused a few times for casting, the used crucible waste could be significantly reduced as well.

Comparing ectopic bone growth induced by rhBMP-2 on an absorbable collagen sponge in rat and rabbit models.

Recombinant human Bone Morphogenetic Protein-2 (rhBMP-2) is currently employed as an autograft replacement for spinal fusion. The morphogen is incorporated onto its carrier, an absorbable collagen sponge (ACS), in the operating room. Although the effectiveness of the rhBMP-2/ACS implant in stimulating bone formation in human subjects has now been well established, further investigations of its use are necessary to deepen our understanding of its performance. The objective of the present study was to determine whether fluid released from the rhBMP-2/ACS implant could induce bone growth in tissue sites away from the implant site. We first measured the amount of protein in the fluid released from the rhBMP-2-soaked ACS during intraoperative handling. Variables included soak time and degree of compression. In the compression group that most closely approximated intraoperative conditions, more than 95% of the rhBMP-2 protein was retained by the ACS following a 15-min. soak time. This in vitro study was followed by an in vivo ectopic implant experiment using rat and rabbit models. The animal investigation compared the amount of bone induced by rhBMP-2 solution alone versus the de novo bone formation induced by rhBMP-2/ACS implants with varying concentrations of rhBMP-2. No ossicles were found at the sites where rhBMP-2 solution was injected in either animal species. Twenty-two of the 24 subcutaneous sites in the rats implanted with the rhBMP-2/ACS constructs displayed the presence of the typical 4- and 12-week ossicle. There were no noticeable differences in the size and shape of the ossicles after 4 and 12 weeks. There was a greater percentage of implant sites without ossicles in the rabbits, compared to the rats.

Modulation spectroscopy study of the effects of growth interruptions on the interfaces of GaAsSb/GaAs multiple quantum wells.

The effects of growth interruption times combined with Sb exposure of GaAsSb/GaAs multiple quantum wells (MQWs) have been investigated by using phototransmittance (PT), contactless electroreflectance (CER) and wavelength modulated surface photovoltage spectroscopy (WMSPS). The features originated from different portions of the samples, including interband transitions of MQWs, interfaces and GaAs, are observed and identified through a detailed comparison of the obtained spectra and theoretical calculation. A red-shift of the interband transitions and a broader lineshape of the fundamental transition are observed from samples grown under Sb exposure compared to the reference sample grown without interruption. The results can be interpreted in terms of both increases in Sb content and mixing of Sb in the GaAs interface layers. An additional feature has been observed below the GaAs region in the samples with Sb treatment. The probable origin of this additional feature is discussed.

Long-term results of coronary artery bypass grafting in patients with dialysis-dependent renal failure.

AIM: Coronary artery disease is the main cause of mortality and morbidity in dialysis-dependent renal failure patients. Both the prevalence and incidence of renal failure are high in Taiwan. However, there were few reports exploring the outcome of coronary aortic bypass grafting (CABG) in these patients. The aim of this study was to determine the survival outcome and risk factors for mortality from CABG in this population. METHODS: The operative, early postoperative and late results of 170 dialysis patients undergoing isolated coronary artery bypass grafting from January, 2000 to January, 2012 were retrospectively reviewed. Operative mortality, long-term survival, and risk factors were analyzed. RESULTS: One hundred and seventeen patients (68.8%) were male, and the mean age was 61.5±10.3 years (range, 34-86 years). Follow-up was 40.3±32.1 months. Operative mortality was 8.2%. Actuarial survival, including operative mortality, was 81±3% at 1 year, 68±4% at 3 years, 58±5% at 5 years and 49±6% at 10 years, better than the natural course of dialysis-dependent renal failure patients. Age, emergent operation, postoperative ventricular tachycardia or fibrillation, postoperative intra-aortic balloon pump insertion, gastrointestinal bleeding, and left internal mammary artery graft were significant predictors of operative or long term mortality. Most causes of late death were due to infection or cardiac events. CONCLUSION: CABG in dialysis patients is associated with a higher incidence of complications, but has acceptable mortality. CABG is beneficial in this population. Internal mammary artery grafting may provide more favorable long term outcomes.

Comparative membrane locations and activities of human monoamine oxidases expressed in yeast.

Human monoamine oxidases A and B were expressed under the control of a galactose inducible promoter in Saccharomyces cerevisiae. The two MAO isoenzymes were found located in the yeast mitochondrial outer membrane, probably in different orientations as suggested by controlled proteolysis experiments. A high level of both human MAO-A or -B activities is measured in intact mitochondria without the need for any detergent solubilisation step. The substrate and inhibitor selectivities of the membrane-bound MAOs are highly similar to those of purified human enzymes. The level of MAO-B activity, however, is selectively lowered when bound to the membrane.

Connective tissue response to tubular implants for peripheral nerve regeneration: the role of myofibroblasts.

The presence of contractile cells, their organization around regenerating nerve trunks, and the hypothetical effect of these organized structures on the extent of regeneration across a tubulated 10-mm gap in the rat sciatic nerve were investigated. Collagen and silicone tubes were implanted both empty and filled with a collagen-glycosaminoglycan (GAG) matrix. Nerves were retrieved at 6, 30, and 60 weeks postoperatively and time-dependent values of the nerve trunk diameter along the tubulated length were recorded. The presence of myofibroblasts was identified immunohistochemically using a monoclonal antibody to alpha-smooth muscle actin. Myofibroblasts were circumferentially arranged around the perimeter of regenerated nerve trunks, forming a capsule which was about 10 times thicker in silicone tubes than in collagen tubes. The nerve trunk diameter that formed inside collagen tubes was twice as large as that inside silicone tubes. In contrast, the collagen-GAG matrix had a relatively small effect on capsule thickness or diameter of regenerate. It was hypothesized that the frequency of successful bridging by axons depends on the balance between two competitive forces: the axial forces generated by the outgrowth of axons and nonneuronal cells from the proximal stump and the constrictive, circumferential forces imposed by the contractile tissue capsule that promote closure of the wounded stumps and prevent axon elongation. Because the presence of the collagen-GAG matrix has enhanced greatly the recovery of normal function of regenerates in silicone tubes, it was hypothesized that it accelerated axonal elongation sufficiently before the hypothetical forces constricting the nerve trunk in silicone tubes became sufficiently large. The combined data suggest a new mechanism for peripheral nerve regeneration along a tubulated gap.

Synovium-like tissue from loose joint replacement prosthesis: comparison of human material with a canine model.

The formation of synovium-like tissue is a biological response to a loose joint replacement prosthesis. Histological examination of this tissue has shown a synovial lining with a predominance of fibroblasts and macrophages, some multinucleated giant cells, and dispersed particles from the implant. Previous studies have reported elevated interleukin 1 (IL-1), prostaglandin E2 (PGE2), and collagenase in this tissue. We developed a canine model for the loose cemented femoral stem. Tissue harvested from the canine model was compared with human tissue retrieved at revision arthroplasty. Histology showed synovium, similar to that observed around loose human prostheses, adjacent to the canine cement sheath. Cells were isolated from this tissue and incubated in culture medium with or without naproxen for 3 days. Aliquots of the conditioned media were tested in the thymocyte proliferation assay to determine IL-1-like activity. IL-1 beta levels in human cell-conditioned media were analyzed by enzyme-linked immunosorbent assay, and PGE2 levels were measured by radioimmunoassay (RIA) using a PGE2 RIA kit (New England Nuclear). Human tissue contained levels of IL-1 beta in the range of 150 to 7,040 pg/mL and PGE2 levels of 82 to 952 ng/mL. The canine specimens contained IL-1-like activity and significant amounts of PGE2 (76 to 1,720 ng/mL). Naproxen decreased PGE2 levels in vitro. This animal model provides the means to investigate the in vivo and in vitro activity of the synovial cells around loose total joint prostheses.

Early peripheral nerve healing in collagen and silicone tube implants: myofibroblasts and the cellular response.

Injuries to peripheral nerves innervating a limb cause paralysis, and can necessitate amputation. The inability of the nerves to regenerate spontaneously and the limitations of autograft procedures led to the development of treatments involving insertion of the nerve ends into prosthetic tubular devices. Previous work showed that 'entubulation' of the nerve ends in a silicone tube containing a specific porous, resorbable collagen-GAG (CG) copolymer, serving as an analog of extracellular matrix, improved regeneration compared to an empty silicone tube. However, long-term treatment with silicone tubes produced constriction that caused partial degradation of the regenerated axons; for this and other reasons, implementation of a nondegradable tube may require a second surgical procedure for removal. In this study the silicone tube was replaced with porous and non-porous collagen tubes in order to produce fully degradable devices. CG-filled collagen tubes and controls (CG-filled silicone tubes and empty collagen and silicone tubes) were implanted in a 10-mm gap in the rat sciatic nerve, with three rats in each group. The regeneration was evaluated after six weeks using light microscope images of cross sections of the nerve that were digitized and analyzed. Histograms of the diameters of the axons were generated and compared. The cellular response to the implanted biomaterials was assessed histologically, and immunohistochemistry was performed using an antibody to alpha-smooth muscle actin in order to determine the presence of myofibroblasts (contractile cells). Axonal regrowth was comparable in porous collagen, non-porous collagen, and silicone tubes filled with a CG matrix. These results support the implementation of a degradable collagen tube in place of a silicone device. Confirming earlier work, regeneration through the silicone and collagen tubes was enhanced by the CG copolymer, compared to empty tubes. A notable finding was a continuous layer of myofibroblasts on the surfaces of all of the six silicone tube prostheses, but on the inner surface of only one of six collagen tubes (Fisher's exact tests; P < 0.01). This is the first report of contractile capsules around silicone tubes, and supports the use of degradable collagen tubes in peripheral nerve regeneration. Macrophages were found bordering both the silicone and collagen tubes, and in the case of the collagen tubes, appeared to be participating in the regulation of the tubes.

Compressive properties of cancellous bone defects in a rabbit model treated with particles of natural bone mineral and synthetic hydroxyapatite.

A rabbit model was developed to evaluate the compressive mechanical properties of cancellous bone defects treated with particles of selected bone graft substitute materials. A novel feature of the model was the precise retrieval of the site of implantation. A notable finding was a 9-fold increase in the modulus of elasticity of the defect implanted with a synthetic hydroxyapatite material after 26 weeks when compared to the modulus of the trabecular bone normally at the site. The compressive modulus of lesions treated with particles of a natural bovine bone mineral (anorganic bovine bone) was closer to the normal modulus of the cancellous bone at the site. While the compressive strength of the anorganic bone particles was less than that of normal bone, the site implanted with the bone mineral particles achieved compressive strength greater than normal after 6 weeks. Moreover, the anorganic bone particles accelerated the increase in strength of the lesion, at 6 weeks exceeding the strength achieved by the untreated defect after 26 weeks. The potential problem associated with the disparity in the compressive modulus between sites implanted with the synthetic HA particles and surrounding bone is discussed.

Chondrocyte-seeded collagen matrices implanted in a chondral defect in a canine model.

The objective of our study was to evaluate reparative tissues formed in chondral defects in an adult canine model implanted with cultured autologous articular chondrocytes seeded in type I and II collagen GAG matrices. Two defects were produced in the trochlea grooves of the knees of 21 dogs, with cartilage removed down to the tidemark. This study includes the evaluation of 36 defects distributed among five treatment groups: Group A, type II collagen matrix seeded with autologous chondrocytes under a sutured type II collagen flap; Group B, type I collagen matrices seeded with chondrocytes under a sutured fascia flap; Group C, unseeded type I collagen matrix implanted under a sutured fascia flap; Group D, fascia lata flap alone; and Group E, untreated defects. All animals were killed 15 weeks after implantation. Six other defects were created at the time of death and evaluated immediately after production as 'acute defect controls'. In three additional defects, unseeded matrices were sutured to the defect and the knee closed and reopened after 30 min to determine if early displacement of the graft was occurring; these defects served as 'acute implant controls'. The areal percentages of four tissue types in the chondral zone of the original defect were determined histomorphometrically: fibrous tissue (FT); hyaline cartilage (HC); transitional tissue (TT, including fibrocartilage); and articular cartilage (AC). New tissue formed in the remodeling subchondral bone underlying certain defects was also assessed. Bonding of the repair tissue to the subchondral plate and adjacent cartilage, and degradation of the adjacent tissues were evaluated. There were no significant differences in the tissues filling the original defect area of the sites treated with chondrocyte-seeded type I and type II matrices. Most of the tissue in the area of the original defect in all of the groups was FT and TT. The areal percentage of HC plus AC was highest in group E, with little such tissue in the cell-seeded groups, and none in groups C and D. The greatest total amount of reparative tissue, however, was found in the cell-seeded type II matrix group. Moreover, examination of the reparative tissue formed in the subchondral region of defects treated with the chondrocyte-seeded collagen matrices (Groups A and B) demonstrated that the majority of the tissue was positive for type II collagen and stained with safranin O. These results indicate an influence of the exogenous chondrocytes on the process of chondrogenesis in this site. In all groups with implants (A-D), 30(50% of the FT and TT was bonded to the adjacent cartilage. Little of this tissue (6-22%) was attached to the subchondral plate, which was only about 50% intact. Remarkable suture damage was found in sections from each group in which sutures were used. Harvest sites showed no regeneration of normal articular cartilage, 18 weeks after the biopsy procedure. Future studies need to investigate other matrix characteristics, and the effects of cell density and incubation of the seeded sponges prior to implantation on the regenerative response.

Healing of chondral and osteochondral defects in a canine model: the role of cultured chondrocytes in regeneration of articular cartilage.

In this study a canine model was developed to investigate the nature of early healing responses to both chondral and osteochondral defects and to evaluate the tissue regenerative capacity of cultured autologous chondrocytes in chondral defects. The healing response to surgically created chondral defects was minor, with little cellular infiltration. In contrast, osteochondral defects exhibited a rapid cellular response, resulting ultimately in the formation of fibrous tissue. The lack of significant cellular activity in chondral defects suggests that an evaluation of the capacity of cultured autologous chondrocytes to regenerate articular cartilage is best studied in chondral defects using the canine model. When dedifferentiated cultured articular chondrocytes were implanted into chondral defects, islands of type II collagen staining were demonstrated in the regenerative tissue within 6 weeks. The relatively early expression of cartilage specific markers by the implanted chondrocytes, coupled with the inability of untreated chondral defects to repair or regenerate, demonstrates the utility of the canine model in evaluating novel materials for cartilage repair and regeneration.

Lapine and canine bone marrow stromal cells contain smooth muscle actin and contract a collagen-glycosaminoglycan matrix.

Lapine and canine marrow stromal cells were found to contain a contractile actin isoform, alpha-smooth muscle actin (SMA), by immunohistochemistry and Western blot analysis. The SMA was found to be incorporated into stress fibers that were prominently displayed by the cells in monolayer culture. The cell content of this actin isoform increased with passage number. The contractility of SMA-expressing stromal cells was demonstrated by their contraction of collagen-glycosaminoglycan analogs of extracellular matrix into which they were seeded. The demonstration that marrow-derived stromal cells express the SMA gene may explain recent findings of this expression in musculoskeletal connective tissue cells including osteoblasts, chondrocytes, and fibrochondrocytes that may be derived from this mesenchymal stem cell. The implications of these findings for tissue engineering strategies employing marrow stromal cells are also discussed.

Transcriptional regulation of the Bacillus subtilis bscR-CYP102A3 operon by the BscR repressor and differential induction of cytochrome CYP102A3 expression by oleic acid and palmitate.

The adjacent yrhI and yrhJ genes were identified by the Bacillus subtilis genome sequencing project. We now report that yrhJ (renamed CYP102A3) encodes a cytochrome P450 and that yrhI (renamed bscR) encodes a repressor that negatively regulates the transcription of the bscR-CYP102A3 operon. The transcriptional initiation site of bscR has been mapped by primer extension analysis. An 18-bp perfect palindromic sequence centered 65.5 bp downstream from the transcriptional initiation site of bscR has been identified as the binding site for BscR by gel mobility shift assays. Base substitutions in the 18-bp inverted repeat resulted in derepression of the bscR-xylE transcriptional fusion in vivo. bscR-xylE fusion studies and Northern blot analysis revealed that oleic acid and palmitate could induce the expression of the bscR-CYP102A3 operon to a considerable extent. However, only oleic acid was capable of preventing the binding of BscR to its operator DNA in vitro, suggesting that the induction of CYP102A3 expression by oleic acid and palmitate in B. subtilis might be mediated through different mechanisms.

The effects of tubulation on healing and scar formation after transection of the adult rat spinal cord.

PURPOSE: The purpose of this study was to characterize the effects of implantation of a collagen tube on healing and scar formation following transection of tbc adult rat spinal cord. METHODS: The spinal cords of adult rats were completely transected at the mid-thoracic level. At 30 days after injury, the cellular and extra-cellular components of repair tissue present within tubulated and non-tubulated (control) wounds were compared using qualitative and quantitative histological techniques. RESULTS: The presence of the tube reduced fibrocollagenous scar invasion into the gap, promoted astrocyte migration, and oriented axonal and connective tissue components of the repair tissue. Tube implants supported the regeneration of a substantial number of myelinated axons. A notable finding was the identification of cells containing a contractile actin isoform in the healing spinal cord. CONCLUSIONS: The tubulation model allows for the study of spinal cord wound healing and axon elongation in a controlled experimental environment within the tube lumen. Using this model, it will be possible to study manipulation of the healing response by the introduction of exogenous agents within the tube.

Tumour necrosis factor-alpha causes an increase in blood-brain barrier permeability during sepsis.

Blood-brain barrier (BBB) permeability during sepsis with Escherichia coli or Streptococcus pneumoniae was examined in a mouse model and measured by a circulating beta-galactosidase tracer. The leakage of brain microvascular vessels during sepsis was confirmed by transmission electron microscopic examination of brain tissues stained with horseradish peroxidase. The increase of BBB permeability induced by E. coli and S. pneumoniae, which was maximal at 3 h and 12 h after injection, respectively, was transient because of rapid clearance of the bacteria from the blood. Tumour necrosis factor-alpha (TNF-alpha) was stained on microvascular vessels of the brain during sepsis and intravenous injection of recombinant TNF-alpha also increased the BBB permeability. The increase in BBB permeability induced by either E. coli or S. pneumoniae could be inhibited by anti-TNF-alpha antibody. It was concluded that circulating TNF-alpha generated during sepsis induced the increase in BBB permeability.

Healing of canine articular cartilage defects treated with microfracture, a type-II collagen matrix, or cultured autologous chondrocytes.

The effects of three different treatments on the healing of articular cartilage defects were compared with use of a previously developed canine model. In the articular surface of the trochlear grooves of 12 adult mongrel dogs, two 4-mm-diameter defects were made to the depth of the tidemark. Four dogs were assigned to each treatment group: (a) microfracture treatment, (b) microfracture with a type-II collagen matrix placed in the defect, and (c) type-II matrix seeded with cultured autologous chondrocytes. After 15 weeks, the defects were studied histologically. Data quantified on histological cross sections included areal or linear percentages of specific tissue types filling the defect, integration of reparative tissue with the calcified and the adjacent cartilage, and integrity of the subchondral plate. Total defect filling (i.e., the percentage of the cross-sectional area of the original defect filled with any type of reparative tissue) averaged 56-86%, with the greatest amount found in the dogs in the microfracture group implanted with a type-II collagen matrix. The profiles of tissue types for the dogs in each treatment group were similar: the tissue filling the defect was predominantly fibrocartilage, with the balance being fibrous tissue. There were no significant differences in the percentages of the various tissue types among dogs in the three groups.

Effects of a cultured autologous chondrocyte-seeded type II collagen scaffold on the healing of a chondral defect in a canine model.

Using a previously established canine model for repair of articular cartilage defects, this study evaluated the 15-week healing of chondral defects (i.e., to the tidemark) implanted with an autologous articular chondrocyte-seeded type II collagen scaffold that had been cultured in vitro for four weeks prior to implantation. The amount and composition of the reparative tissue were compared to results from our prior studies using the same animal model in which the following groups were analyzed: defects implanted with autologous chondrocyte-seeded collagen scaffolds that had been cultured in vitro for approximately 12 h prior to implantation, defects implanted with autologous chondrocytes alone, and untreated defects. Chondrocytes, isolated from articular cartilage harvested from the left knee joint of six adult canines, were expanded in number in monolayer for three weeks, seeded into porous type II collagen scaffolds, cultured for an additional four weeks in vitro and then implanted into chondral defects in the trochlear groove of the right knee joints. The percentages of specific tissue types filling the defects were evaluated histomorphometrically and certain mechanical properties of the repair tissue were determined. The reparative tissue filled 88+/-6% (mean+/-SEM; range 70-100%) of the cross-sectional area of the original defect, with hyaline cartilage accounting for 42+/-10% (range 7-67%) of defect area. These values were greater than those reported previously for untreated defects and defects implanted with a type II collagen scaffold seeded with autologous chondrocytes within 12 h prior to implantation. Most striking, was the decreased amount of fibrous tissue filling the defects in the current study, 5+/-5% (range 0-26%) as compared to previous treatments. Despite this improvement, indentation testing of the repair tissue formed in this study revealed that the compressive stiffness of the repair tissue was well below (20-fold lower stiffness) that of native articular cartilage.

Effects of harvest and selected cartilage repair procedures on the physical and biochemical properties of articular cartilage in the canine knee.

This study utilizes a canine model to quantify changes in articular cartilage 15-18 weeks after a knee joint is subjected to surgical treatment of isolated chondral defects. Clinical and experimental treatment of articular cartilage defects may include implantation of matrix materials or cells, or both. Three cartilage repair methods were evaluated: microfracture, microfracture and implantation of a type-II collagen matrix, and implantation of an autologous chondrocyte-seeded collagen matrix. The properties of articular cartilage in other knee joints subjected to harvest of articular cartilage from the trochlear ridge (to obtain cells for the cell-seeded procedure) were also evaluated. Physical properties (thickness, equilibrium compressive modulus, dynamic compressive stiffness, and streaming potential) and biochemical composition (hydration, glycosaminoglycan content, and DNA content) of the cartilage from sites distant to the surgical treatment were compared with values measured for site-matched controls in untreated knee joints. No significant differences were seen in joints subjected to any of the three cartilage repair procedures. However, a number of changes were induced by the harvest operation. The largest changes (displaying up to 3-fold increases) were seen in dynamic stiffness and streaming potential of patellar groove cartilage from joints subjected to the harvest procedure. Whether the changes reported will lead to osteoarthritic degeneration is unknown, but this study provides evidence that the harvest procedure associated with autologous cell transplantation for treatment of chondral defects may result in changes in the articular cartilage in the joint.

Autologous chondrocyte implantation in a canine model: change in composition of reparative tissue with time.

The objective of the study was to evaluate the tissue types filling 4-mm diameter defects in the canine trochlear groove 1.5, 3, and 6 months after autologous chondrocyte implantation (ACI). Untreated defects served as controls. Periosteum alone controls were also included at the 1.5-month time period. The results were compared with previously published findings obtained 12 and 18 months postoperative. After 3 months the ACI-treated defects contained significantly more reparative tissue than found in the untreated control group, including twice the amount of hyaline cartilage (HC). These findings, however, were the only significant effects of the ACI treatment when compared to the periosteum alone or empty control groups. The benefits of ACI found at 3 months did not persist to longer time periods. An evaluation of the inter-observer error associated with the histomorphometric method indicated that it was generally less than the inter-animal variation in the results.

TNFalpha-induced cyclooxygenase 2 not only increases the vasopermeability of blood-brain barrier but also enhances the neutrophil survival in Escherichia coli-induced brain inflammation.

In Escherichia coli-induced brain inflammation, cyclooxygenase-2 was induced not only on brain arterioles at 3 h, but also on infiltrating neutrophils at 9 h post-intracerebral injection. Intravenous injection of E. coli or recombinant TNFalpha also induced cyclooxygenase-2 expression on arterioles. Cyclooxygenase-2 and TNFalpha were co-localized on the arterioles as well as the infiltrating neutrophils by serial-section staining, indicating that cyclooxygenase-2 was induced by TNFalpha. NS398 (a cyclooxygenase-2 selective inhibitor) not only inhibited the increase of blood-brain barrier permeability, but also enhanced the apoptosis of the infiltrating neutrophils after E. coli stimulation. This suggests that TNFalpha-stimulated cyclooxygenase-2 induction play an important role on E. coli-induced brain inflammation. Its inhibition would help the resolution of neutrophil-mediated brain inflammation.