JOURNAL OF VIROLOGY, Aug. 2002, p. 83358346 0022-538X/02/$04.000 DOI: 10.1128/JVI.76.16.83358346.2002 Copyright 2002, American Society for Microbiology. All Rights Reserved.

Vol. 76, No. 16

Critical Role for Protein Tyrosine Phosphatase SHP-1 in Controlling Infection of Central Nervous System Glia and Demyelination by Theilers Murine Encephalomyelitis Virus
Paul T. Massa,* Stacie L. Ropka, Sucharita Saha, Karen L. Fecenko, and Kathryn L. Beuler
Department of Neurology and Department of Microbiology and Immunology, Upstate Medical University, State University of New York, Syracuse, New York 13210
Received 25 February 2002/Accepted 17 May 2002
We previously characterized the expression and function of the protein tyrosine phosphatase SHP-1 in the glia of the central nervous system (CNS). In the present study, we describe the role of SHP-1 in virus infection of glia and virus-induced demyelination in the CNS. For in vivo studies, SHP-1-decient mice and their normal littermates received an intracerebral inoculation of an attenuated strain of Theilers murine encephalomyelitis virus (TMEV). At various times after infection, virus replication, TMEV antigen expression, and demyelination were monitored. It was found that the CNS of SHP-1-decient mice uniquely displayed demyelination and contained substantially higher levels of virus than did that of normal littermate mice. Many infected astrocytes and oligodendrocytes were detected in both brains and spinal cords of SHP-1-decient but not normal littermate mice, showing that the virus replicated and spread at a much higher rate in the glia of SHP-1decient animals. To ascertain whether the lack of SHP-1 in the glia was primarily responsible for these differences, glial samples from these mice were cultured in vitro and infected with TMEV. As in vivo, infected astrocytes and oligodendrocytes of SHP-1-decient mice were much more numerous and produced more virus than did those of normal littermate mice. These ndings indicate that SHP-1 is a critical factor in controlling virus replication in the CNS glia and virus-induced demyelination. Neurotropic viruses that infect astrocytes and myelin-forming oligodendrocytes often lead to demyelinating disease similar to that seen in multiple sclerosis (11, 43, 56). Demyelination in rodent models for multiple sclerosis results in inefcient saltatory conduction of nerve bers with accompanying motor decits and limb paralysis (61, 62). Recent research has centered on understanding the mechanisms responsible for virusinduced demyelination in these animals and the genetic susceptibility to disease (3, 7, 8, 12, 20, 29, 44, 48). These studies have indicated that damage to oligodendrocytes and myelin may occur by multiple distinct pathways. Depending on the particular virus, these pathways include direct cytopathic effects of the virus in oligodendrocytes, virus-induced inammatory immune responses promoted by infected glia in the white matter, or molecular mimicry between virus and myelin antigens (43, 56, 57). In each of these responses, the activities of proinammatory cytokines, interferons, and virus-induced genes play an important role in promoting or protecting against oligodendrocyte pathology (6, 40, 44, 45, 51, 65). Therefore, the regulation of these activities in central nervous system (CNS) cells may be particularly important in controlling virus replication and virus-induced demyelinating processes. However, many of the host genes that control virus infection and demyelination in the CNS through multiple intracellular signaling pathways have not been identied. Virus-induced genes provide for a rapid innate response to

* Corresponding author. Mailing address: Department of Neurology and Department of Microbiology and Immunology, Upstate Medical University, State University of New York, 750 E. Adams St., Syracuse, NY 13210. Phone: (315) 464-7606. Fax: (315) 464-6402. E-mail: massap@upstate.edu. 8335
control virus replication at the earliest stages of infection. The activities of virus-induced cellular proteins, including interferons, cytokines, and intracellular signaling molecules, are controlled at multiple levels to provide for modulation of the antiviral state and inammation (9, 19, 30, 38, 42, 55). Although these regulatory pathways have been extensively studied, such mechanisms in neural cells have been less well studied and may be unique. For instance, it was recently reported that interferons protected CNS neurons from virus infection but were unable to stimulate the expression of major histocompatibility complex class I genes in these cells (36). Multiple mechanisms likely are responsible for mediating tissue-specic antiviral responses in the CNS, but one such regulatory mechanism appears to involve SHP-1, a cytosolic protein tyrosine phosphatase that controls interferon and virus-induced signaling in the glia (16, 34, 37, 38, 66). SHP-1 has been characterized as a key functional modulator of cytokine responses in hematopoietic and neural cells (9, 17, 19, 34, 42). The physiological ramications of SHP-1 loss in animals have been extensively studied by using two independent strains of mice with natural mutations in the SHP-1 gene (53). Moth-eaten (me/me) mice have a single nucleotide deletion mutation which generates a cryptic mRNA splice donor site, a resulting frameshift, and a complete loss of SHP-1 protein expression (53). Viable moth-eaten (me[supi]v/mev) mice have a T-to-A transversion mutation in a splice donor that leads to the usage of cryptic donors on either side of the mutation. This situation results in an in-frame deletion and an insertion in the mRNAs, which encode a slightly smaller or larger SHP-1 protein with activity reduced to approximately 10% that in normal mice. Moth-eaten animals display a number of well-characterized hematopoietic abnormalities (32, 52);

MASSA ET AL.

J. VIROL.
fetal bovine serum per brain were plated on 60-mm culture dishes and then fed with fresh medium consisting of DMEM containing 10% heat-inactivated horse serum at 5 days after plating. Glial cultures were used at 14 days postplating to analyze TMEV replication. Virus and cell lines. The attenuated strain of TMEV, BeAn 8386, was obtained from the American Type Culture Collection (ATCC), Manassas, Va. (ATCC VR-995, originally contributed by H. L. Lipton) (27, 28). BeAn 8386 was prepared by propagating in BHK-21 cells (ATCC CCL-10) and harvesting in tissue culture supernatant at PFU/ml. Puried virus stock was prepared from the tissue culture supernatant by polyethylene glycol precipitation and sucrose density gradient centrifugation. Briey, virus-containing culture supernatant was claried by centrifugation at 2,500 g for 20 min. Virus was precipitated with 8% polyethylene glycol in 1.6 M NaCl (50). The concentrated viral lysate was then treated with 1% sodium dodecyl sulfate for 10 min and centrifuged over a 20 to 70% continuous sucrose gradient at 160,000 g in a Beckman SW41 rotor. This puried stock contained 7.PFU/ml in BHK-21 cells. Virus inoculation. Weanling mice were anesthetized with methoxyurane and inoculated intracerebrally (i.c.) in the left hemisphere with 1.PFU of BeAn 8386 in a volume of 0.02 ml. Mice were observed on a daily basis for signs of paralysis. Paralyzed moth-eaten mice were euthanatized with age-matched normal littermates, and the brains and spinal cords were removed and stored at 80C until assayed for viral infectious units. Additionally, normal littermate and diseased moth-eaten mice were anesthetized and perfused with 4% paraformaldehyde in phosphate-buffered saline (PBS), and the brains and spinal cords were embedded in parafn for immunohistochemical analysis. For in vitro studies, glial cell cultures grown on 60-mm plates or plated on glass chamber slides were inoculated with puried virus stock (3.PFU [total]/dish) at a multiplicity of infection of 1.0 and incubated for 1 h at 37C. Afterward, the inoculum was removed, and the cultures were rinsed with PBS and refed with DMEM containing 10% normal horse serum. The cultures were then incubated for 3 and 6 days, triplicate cell supernatants and cell lysates were harvested, and virus titers were determined by plaque assays on BHK-21 cells. Virus plaque assays. To quantify the production of TMEV, brains (brain plus brain stem) and spinal cords were prepared separately as 10% homogenates in DMEM and claried by centrifugation as described above. The resulting supernatants were assayed for virus titers. Both in vivo and in vitro samples were assayed for virus titers by standard plaque assays (14). Briey, BHK-21 cell monolayers were grown on 60-mm dishes. Cell monolayers were inoculated with virus-containing samples (tissue homogenates or cell supernatants) and incubated for 1 h at 37C. Inoculated cell monolayers were overlaid with 1% agar in DMEM containing 2% fetal bovine serum. Four days after infection, the agar layer was removed, and the cell monolayers were xed with methanol and then stained with 1% crystal violet in 20% ethanol. Plaques per dish were counted, and PFU per milliliter or per gram were determined. Immunohistochemical analysis. Double immunohistochemical analysis of myelin basic protein (MBP), myelin proteolipid protein (PLP), or glial brillary acidic protein (GFAP) and TMEV antigens was performed by deparafnizing microtome sections of brains and spinal cords, rehydrating the samples in graded ethanols, and blocking the samples in 10% normal horse serum. The spinal cord sections were incubated with monoclonal antibodies to either MBP (rat monoclonal immunoglobulin G [IgG] against MBP; MCA 409; Accurate Chemical and Scientic Corp., Westbury, N.Y.), PLP (mouse monoclonal IgG against myelin PLP; clone plpc1; Oncogene Research Products, Cambridge, Mass.), or GFAP (rat monoclonal IgG; Zymed Laboratories, South San Francisco, Calif.) overnight; this step was followed by rinsing and incubating the sections in goat anti-rat or anti-mouse IgG conjugated to tetramethyl rhodamine isothiocyanate (TRITC; Zymed). The sections were further incubated overnight with rabbit antiserum against TMEV strain BeAn 8386 (provided by H. L. Lipton, Northwestern University) at a 1:2,000 dilution in 10% normal horse serumPBS. After being rinsed, the cultures were incubated in goat anti-rabbit IgG conjugated to uorescein isothiocyanate (FITC; Zymed). Coverslips were mounted with uorescence mounting medium (Dako Corp., Carpinteria, Calif.) and viewed by epiuorescence microscopy with a Zeiss Axioskop microscope. To visualize demyelination and infection in individual adjacent sections, spinal cord sections were stained for MBP or TMEV as described above, but the biotin-avidin-alkaline phosphatase technique with the blue 5-bromo-4-chloro-3indolylphosphate (BCIP)nitroblue tetrazolium product was used for detection. Alternatively, sections were incubated with goat anti-rat IgG conjugated to FITC as a secondary detection reagent. Inammatory inltrates were detected by staining adjacent sections of spinal cords and brain with hematoxylin and eosin (HE). To visualize infection of oligodendrocytes in vitro by double immunouorescence, live infected and noninfected cells were incubated at 4C with mouse monoclonal antibody to oligodendrocyte-specic O1 antigens (4, 54); this

however, the regulatory role of SHP-1 in cells of epithelial origin, including glia, and the pathological consequences of SHP-1 loss in these cells have only recently been investigated (5, 16, 31, 42, 66, 67). Massa and colleagues previously described the expression and functions of SHP-1 in astrocytes and oligodendrocytes, which represent the major macroglial populations of the CNS (34, 37, 38). By examination of glia from SHP-1-decient mice (either moth-eaten or viable moth-eaten mice), they showed that SHP-1 controls gene expression induced by the proinammatory cytokines gamma interferon (38) and interleukin-6 (34), both of which have been implicated in the pathogenesis of virus-induced demyelinating disease. Others have shown that SHP-1 also controls alpha/beta interferon signaling through alpha/beta interferon receptors in hematopoietic cells (9). Similar ndings for oligodendrocytes have been reported elsewhere (P. T. Massa, S. L. Ropka, and S. Saha, abstract Immunology 1216 May 2000, FASEB J. 14:A1084, 2000). Furthermore, it has been shown that SHP-1 controls the direct activation of NF-B in astrocytes by viral mimetic doublestranded RNA (dsRNA) (37), which occurs as a consequence of the activation of the virus-inducible antiviral gene product double-stranded RNA-activated protein kinase (68). Taken together, these observations indicate that SHP-1 may control antiviral signaling pathways in the CNS glia. However, the biological signicance of the regulation of virus-induced responses by SHP-1 has not been demonstrated. It was therefore of interest to determine whether SHP-1 controls the susceptibility of the CNS glia to infection by neurotropic viruses. To do this, we analyzed the susceptibility of SHP-1-decient mice to a paralytic virus-induced demyelinating disease following infection with Theilers murine encephalomyelitis virus (TMEV) (2, 47, 49, 59, 69). We found that astrocytes and oligodendrocytes of mice lacking SHP-1 are extremely susceptible to TMEV infection both in vivo and in vitro and that these mice are highly susceptible to TMEV-induced demyelinating disease. In vitro, astrocytes and oligodendrocytes of SHP-1-decient mice had a higher rate of infection and produced larger amounts of virus. We therefore propose that the susceptibility of astrocytes and myelin-forming oligodendrocytes to TMEV infection is controlled by innate antiviral responses mediated by SHP-1 within the CNS glia.
MATERIALS AND METHODS Mice. SHP-1-decient moth-eaten (me/me) mice (C3HeB/FeJLe-a/a background) and viable moth-eaten (mev/mev) mice (C57BL/6 background) (53) and their phenotypically normal littermates (designated / for either me/ or mev/ and /) were produced from heterozygous breeding pairs obtained from Jackson Laboratories (Bar Harbor, Maine). Strain designations for heterozygous breeders are C3FeLe.B6-a/a Hcphme/ (stock no. 000225) for moth-eaten mice and C57BL/6J Hcphmev/ stock no. 000811) for viable moth-eaten mice. Glial cultures. Glial cultures containing astrocytes and oligodendrocytes were produced from newborn mice as previously described (38). Cerebral hemispheres were used for cultures, and cerebella were used to probe SHP-1 in Western immunoblots to identify either moth-eaten or normal littermate mice. Genomic DNA was isolated from cerebellar tissue for verication of normal and mutant SHP-1 gene structures of moth-eaten mice as previously described (53). Brains from littermates of heterozygous breeding pairs having the moth-eaten mutation of the SHP-1 gene (either moth eaten or viable moth eaten) were dissected and mechanically dissociated for separate cultures. Cells in approximately 10 ml of Dulbecco modied Eagle medium (DMEM) containing 10%

VOL. 76, 2002

SHP-1 CONTROLS TMEV-INDUCED CNS DISEASE
TABLE 1. Incidence of spastic paralysis in normal, moth-eaten, and viable moth-eaten mice infected with TMEV and concurrent oligodendrocyte infection and demyelination
Genotypea No. of mice with spastic paralysis/no. tested % of mice diseased Onset Infection of oligodendrocytesb Spinal cord demyelinationb
/ (C3FeLe.B6-a/a) me/me (C3FeLe.B6-a/a) / (C57BL/6J) mev/mev (C57BL/6J)

19/58 10/10 0/10 8/8

45 wk 510 days 1015 days

No Yes No Yes

/, phenotypically normal heterozygous (me/, and me /) and homozygous (/) at the SHP-1 locus. Infected (/) animals were age matched with paralyzed moth-eaten mice.
step was followed by xation and incubation with goat anti-mouse IgG conjugated to FITC. The cells were then permeabilized with 0.25% Triton X-100, incubated with anti-TMEV antibodies, and nally incubated with goat anti-rabbit antibodies conjugated to TRITC.
RESULTS Increased paralysis in TMEV-infected SHP-1-decient mice. Moth-eaten (me/me) mice and normal littermate mice (C3FeLe.B6 background) infected with attenuated TMEV
strain BeAn 8386 by i.c. injection were monitored daily for clinical signs, including limb paralysis. Moth-eaten (me/me) mice rst appeared lethargic at approximately 3 to 4 days after infection, while normal littermates remained healthy (Table 1). By day 10, a majority of infected moth-eaten mice displayed clinical signs of spastic limb paralysis that rapidly progressed to quadriplegia and a moribund state in all infected animals. In distinction, normal littermates were only partially susceptible, with a third of the animals developing paralytic disease but at
FIG. 1. Focal demyelinating lesions in dorsal cervical spinal cords of moth-eaten (me/me) mice at 5 days after i.c. inoculation with the attenuated BeAn 8386 strain of TMEV. Five-micrometer parafn sections were stained with rat monoclonal antibody to MBP and labeled with FITC-conjugated secondary antibody. Sections were photographed with color lm by double exposure under both FITC (green) and red uorescence lter sets. Green uorescence labels MBP. Red proles above the background in panel D indicate autouorescent red blood cells in focal hemorrhagic lesions in regions of demyelination in the parenchyma of the spinal cord. The dorsal funiculus of the cervical spinal cord is shown in either TMEV-infected (B and D) or sham-infected (A and C) normal littermate mice (A and B) and me/me mice (C and D).

FIG. 2. Demyelination, virus infection, and inammation in the ventral cervical spinal cords of me/me mice 5 days after intracerebral inoculation with TMEV. Immunohistochemical staining of MBP (A) and TMEV (B) and HE staining (C) in adjacent 5-m parafn sections are shown.
much later times after infection (4 to 5 weeks) (Table 1), compared to their moth-eaten littermates. Like moth-eaten mice, viable moth-eaten mice (mev/mev) also showed complete susceptibility to TMEV. However, their normal littermates (C57BL/6 background) were entirely resistant for up to 6 months after infection (Table 1). Infected mev/mev mice generally developed disease a few days later than me/me mice (10 to 15 days after infection). Immunohistochemical examination of demyelination in
moth-eaten mice. The ability of TMEV to produce spastic limb paralysis at early times after infection in moth-eaten mice was surprising, because this type of paralysis is a relatively late event in wild-type susceptible mice (27, 28, 60, 63). Because limb spasticity suggested possible demyelination in the spinal cords of moth-eaten mice (39), the ability of TMEV infection to cause spinal cord demyelination was assessed with immunohistological sections stained for MBP (Fig. 1 and 2). Uninfected moth-eaten mice showed the expected distribution of MBP staining in white matter tracts of the spinal cord (Fig. 1C). In TMEV-infected moth-eaten mice, MBP staining was sharply reduced in both white matter and gray matter, with some white matter regions of both dorsal (Fig. 1D) and ventral (Fig. 2A) tracts displaying extensive areas of focal demyelination. Some lesions were obviously hemorrhagic, with conspicuous red blood cells at the center of the lesions (Fig. 1D), features not seen in normal littermate or uninfected motheaten mice (Fig. 1A to C). In HE-stained adjacent sections, areas of demyelination showed considerable white matter cellular inltrates, especially in the vicinity of blood vessels (Fig. 2C). In contrast to what was seen in the spinal cord, no such large areas of focal demyelination or inammation were detected in sagittal midline sections of the brain (data not shown), indicating that the spinal cord was particularly sensitive to virus-induced demyelination. Immunouorescence identication of infected cells in spinal cords and brains of TMEV-infected moth-eaten mice. Inspection of the most severely demyelinated areas in adjacent sections of the spinal cord by using an antiserum to TMEV showed numerous virus antigen-containing cells in both gray matter and white matter regions (Fig. 2B). Some cells in the gray matter could be morphologically identied as large dorsal horn motoneurons (Fig. 3 and 4). However, many other smaller cells containing TMEV antigens were scattered throughout the spinal cord and could not be morphologically identied. In the brain, many infected cells were detected in the white matter (Fig. 5), but none were detected in distinct neuronal layers in the cerebral cortex, hippocampus, or brain stem nuclei (data not shown). Of particular note, almost all of the TMEV antigen was localized to small cells in white matter tracts, especially in the corpus callosum. Taken together, these observations indicated that TMEV-infected cells in white matter regions occurred in both brains and spinal cords of SHP1-decient animals and that demyelination was extensive in the spinal cord in areas of TMEV infection. To determine whether some of the small infected cells in the spinal cord and brain white matter regions were glial cells, sections were doubly labeled for TMEV and oligodendrocytespecic antigens. While cell bodies in the spinal cord were not discernibly stained with MBP antibodies, many cells in the spinal cord were doubly labeled for both TMEV and oligodendrocyte-specic PLP, especially in areas of diffuse myelin in demyelinating lesions (Fig. 3). In the brain, infected oligodendrocytes were doubly labeled for MBP and TMEV at the interface of the corpus callosum and cerebral cortex, where myelination was sparse enough to allow resolution of the cell bodies in magnied micrographs (Fig. 5B). Nonetheless, not all infected cells in the white matter were labeled with PLP or MBP (Fig. 5). Many of these infected cells were doubly labeled for GFAP in both the spinal cord (Fig. 4A and B) and the brain

FIG. 3. Double immunouorescence of TMEV (A, C, and E; FITC) and PLP (B, D, and F; TRITC) in the ventral cervical spinal cords of me/me (A to D) mice 5 days after inoculation with TMEV. (A and B) Doubly labeled cells (arrows) in the ventral funiculus at the interface between the white matter and the gray matter of the medial nuclei. A large motoneuron in the gray matter also contains TMEV antigen (arrowhead). (C and D) Doubly labeled cells (arrows) in a demyelinated region of the ventral funiculus adjacent to the ventral median ssure ((.)E and F) No TMEV antigens are seen in cervical spinal cords of / animals infected with TMEV.
FIG. 4. TMEV infection of GFAP-positive astrocytes in moth-eaten mouse cervical spinal cords (A and B) and corpus callosa (C and D) 5 days after inoculation with TMEV. Arrows indicate astrocytes doubly labeled for TMEV (FITC) and GFAP (TRITC). The arrowheads in panels A and B indicate TMEV-infected neurons in the spinal cord gray matter that are not stained for GFAP. (E and F) Normal littermate mouse cervical spinal cords show no TMEV-infected cells.
(Fig. 4C and D), indicating that astrocytes were also productively infected in moth-eaten mice. In contrast, no infected cells were detected in the spinal cords or brains of age-matched normal littermate mice that had received TMEV inoculation
(Fig. 3 and 4). Taken together, these data indicated that astrocytes and oligodendrocytes of SHP-1-decient animals were particularly susceptible to TMEV infection and that TMEV rapidly spread from the brain to the glia in the spinal cord.
FIG. 5. Double immunouorescence of MBP (TRITC; red) and TMEV (FITC; green) in the brain 5 days after inoculation with TMEV. Five-micrometer parafn sections were stained with rat monoclonal antibody to MBP and labeled with TRITC-conjugated secondary antibody. TMEV antibodies were detected with FITC-conjugated antibodies. Sections were photographed with color lm by double exposure of the same frame under both FITC and TRITC lter sets. (A) 400 magnication of the corpus callosum (lower) at the interface with the cerebral cortex (upper). (B) Higher magnication (630) of an area similar to that in panel A but centered at the interface, where myelination is sparse, to allow resolution of doubly labeled cells (brownish yellow cell bodies indicated by arrows).

Virus spread and replication in moth-eaten mice. As noted above, TMEV-infected astrocytes and oligodendrocytes were detected in both brains and spinal cords of moth-eaten mice but not in infected normal littermate mice. We reasoned that the replication and spread of TMEV may be much greater in moth-eaten mice. Therefore, we assayed infectious virus in brains and spinal cords of moth-eaten and normal littermate mice after infection. For these studies, moth-eaten (me/me), viable moth-eaten (mev/mev), and normal littermate mice were infected i.c. with 1.PFU/brain. Normal littermates of viable moth-eaten mice had essentially no detectable infectious virus in the brain, on average (0.6 PFU/g of brain); however, diseased viable moth-eaten mice contained an average of 1.PFU/g of brain, constituting approximately a million more virus particles per gram of tissue than the levels found in normal mouse brain (Fig. 6A). Consistent with the latter results, viable moth-eaten mice had nearly 1,000-fold more virus particles per gram of tissue in the spinal cord than did normal littermates (Fig. 6B). Moth-eaten (me/me) mice also contained higher virus titers in brains and spinal cords than did their normal littermates (Fig. 6C and D). However, unlike normal littermates of viable moth-eaten mice (C57BL/6 background), normal littermates of moth-eaten mice (C3FeLe.B6 background) had substantial virus titers in both brains and spinal cords, indicating differences in background susceptibility, in agreement with the data in Table 1. Despite this level of virus
replication, repeated immunohistochemical analysis was not able to detect virus antigen-containing cells in C3FeLe.B6 normal littermates, perhaps due to the lower sensitivity of this assay. Nonetheless, plaque assays indicated that both replication and spread of TMEV were clearly increased in the two strains of SHP-1-decient mice, in accord with their increased susceptibility to clinical disease compared to the status of their normal littermates. Analysis of TMEV infection of moth-eaten mouse glia in vitro. The increased infection of astrocytes and oligodendrocytes and the concomitant demyelination in SHP-1-decient mice suggested a possible alteration in direct virus-oligodendrocyte interactions dependent on SHP-1 in these cells. To test this possibility, we analyzed TMEV replication in glial cell cultures containing astrocytes and oligodendrocytes produced from moth-eaten and normal littermate mice. Glial cell cultures were inoculated with 7.PFU of TMEV/ml (multiplicity of infection, 1.0) and then incubated for 3 days after inoculation. We rst analyzed the numbers and types of glial cells infected by using double immunohistochemical analysis. To analyze oligodendrocyte infection, cultures were stained with oligodendrocyte-specic antibody to O1 antigens and subsequently for intracellular TMEV antigens. O1 antigen-positive oligodendrocytes expressing TMEV antigens were readily identied in both moth-eaten and normal littermate glial cell cultures (Fig. 7A); however, the number of oligodendrocytes

FIG. 6. TMEV titers in brains and spinal cords of infected mev/mev and me/me mice and normal littermate (/ Control) mice. Brains (A and C) and spinal cords (B and D) were harvested from paralyzed viable moth-eaten (mev/mev) (A and B) or moth-eaten (me/me) (C and D) animals along with age-matched sham-infected normal littermate animals (/ Control). Error bars indicate standard errors of the means. Numbers in the histograms are for control PFU per gram of brain or cord where these cannot be read from the ordinate. Differences in the means between normal littermate mice and either moth-eaten or viable moth-eaten mice were signicant (P 0004) for both brains and spinal cords.
infected was much higher (approximately 10-fold) in motheaten mouse cultures (Fig. 7B). Additionally, many O1 antigen-negative cells in moth-eaten mouse cultures contained TMEV antigens and had an astrocytic morphology. To ascertain whether these cells were astrocytes, cultures were doubly labeled for GFAP and TMEV antigens. Many GFAP-positive astrocytes were found to contain TMEV antigens in motheaten mouse cultures, but none were seen in normal littermate cultures (Fig. 8). Taken together, the immunouorescence data showed that TMEV produced much more infection of O1 antigen-positive oligodendrocytes and GFAP-positive astrocytes in moth-eaten than in wild-type mouse glial cell cultures. Virus production in the glia of normal and moth-eaten mice. To ascertain whether moth-eaten mouse glial cells produced higher virus titers than normal littermate glial cells, infectious virus in supernatants and cell lysates from the above-described
infected glial cell cultures was assessed at 3 and 6 days after infection. Mean virus titers (PFU per milliliter of supernatant) were signicantly higher (approximately vefold) in me/me mouse cultures than in normal littermate mouse cultures (Fig. 9). Analysis of lysates of infected cells at 3 days after infection indicated that the majority of the virus was cell associated in both me/me glia and normal littermate glia and that me/me glia contained signicantly more virus particles than normal littermate glia. At 6 days after infection, the ratio between released virus and cell-associated virus was increased in both me/me and normal littermate glial cell cultures, indicating that a higher proportion of virus was being released from the cells as the infection progressed. Nonetheless, the difference in the amounts of released virus and cell-associated virus between me/me glia and normal littermate glia increased over time (Fig. 9). Taken together with the immunohistochemical data, these

FIG. 7. Frequency of oligodendrocyte infection in vitro. (A) Double-immunouorescence analysis of TMEV infection of O1 antigenpositive oligodendrocytes in moth-eaten and normal littermate mice 2 days after inoculation. Individual cells plated on glass chamber slides were labeled for TMEV (TRITC), and O1 antigens (FITC) were photographed at a 630 magnication. (B) Histogram of doubly labeled TMEV-positive, O1 antigen-positive oligodendrocytes counted in random elds at a 120 magnication in /and me/me cultures infected for 2 days with TMEV. O1 antigen-positive oligodendrocytes (TMEV positive plus TMEV negative) were present at the same densities in the two samples. Error bars indicate standard errors of the means.
data indicated that moth-eaten mouse astrocytes and oligodendrocytes sustained a higher level of virus production than did the glia of normal littermates. DISCUSSION In the present report, we have shown that SHP-1 is a critical determinant in controlling virus replication in the glia of the
FIG. 8. Double-immunouorescence analysis of TMEV infection of GFAP-positive astrocytes in moth-eaten and normal littermate mice 2 days after inoculation. Glial cultures of / mice (A to D) or me/me mice (E to H) were either inoculated with TMEV (C, D, G, and H) or sham inoculated (A, B, E, and F). The left panels were stained for TMEV antigens, and the corresponding elds represented in the right panels were stained for GFAP.
CNS. Further, in vitro studies suggested that SHP-1 may control virus replication at least in part at the level of direct virus-cell interactions. However, the way in which SHP-1 may function to control virus replication in these cells and whether
FIG. 9. TMEV titers in vitro. Glial cultures of either me/me mice or normal littermate mice (/ Control) were inoculated with TMEV. Virus was harvested in both supernatants (released) and cell lysates (cell associated) at 3 and 6 days after infection (P.I.) and quantied by plaque assays. Histograms indicate the mean PFU per milliliter in supernatants and PFU per cell based on protein content in cell lysates. Statistical differences between specimens were measured by Students t test (one tailed). Each experiment was performed in triplicate. Error bars indicate standard errors of the means. Differences in the means between normal littermate (/ Control) and moth-eaten (me/me) glial cultures were signicant at each time after infection for both released and cell-associated viruses (P 0.001).

direct virus-cell interactions controlled by SHP-1 fully account for increased virus growth in SHP-1-decient mice in vivo are not known. To address possible alterations in virus-cell interactions, we are currently investigating multiple antiviral pathways that are likely to be affected by SHP-1 in the CNS glia in vitro. For instance, the role of an innate antiviral state including the interferon system has been shown to be critically important for controlling infection by TMEV in the CNS (12). Such studies may be relevant to the possible role of SHP-1 in controlling TMEV infection, because previous studies showed that SHP-1 altered STAT1 activation in response to interfer-
ons (9, 38, 42). One possibility is that increased induction of STAT1 enhances the expression of proapoptotic genes (22), which may increase virus replication in the CNS, as recently described for Sindbis virus infections of neurons (25). However, other distinct antiviral pathways may be directly affected by the loss of SHP-1 in the CNS glia. The role of SHP-1 in controlling NF-B activation by dsRNA in astrocytes was recently described. It is known that NF-B activation by dsRNA is mediated by the antiviral gene product double-stranded RNA-activated protein kinase (68). Additionally, dsRNA is known to affect the expression of other virus-induced tran-

REFERENCES

scription factors and genes (13, 23) that may also be modulated by SHP-1 activity in virus-infected cells. Finally, it was recently shown that SHP-1 is required for the induction of neuronal nitric oxide synthetase (NOS1) activity in nonhematopoietic cells (31), and NOS1 activity has been shown to be a critical antiviral activity for controlling CNS virus infections (21, 46). Future studies will be aimed at determining the role of SHP-1 in regulating these multiple antiviral pathways in the CNS. In a number of models of virus-induced demyelinating disease, infection often involves astrocytes and oligodendrocytes in the white matter in the vicinity of demyelinating lesions (1, 2, 18, 24, 44, 47, 49, 69). Demyelination caused by virus infection in CNS white matter can result from at least two mechanisms that are relevant to the present study. One is direct cytopathic effects of the virus on oligodendrocytes, and the second is an indirect immunopathologic response to the virus or autoantigens involving inammation and oligodendrocyte pathology. Of particular note, the latter is often promoted by proinammatory cytokine secretion and major histocompatibility complex expression induced by viruses in astrocytes (10, 26, 33, 35, 57, 58). The relative contributions of these mechanisms to the demyelinating process in TMEV-infected motheaten mice are presently unknown. Nonetheless, increased virus replication in oligodendrocytes and astrocytes is likely to promote both pathways to myelin degeneration. With respect to possible immunopathology in the demyelinating process, extensive demyelinating lesions in spinal cords showed increased levels of cellular inltrates indicative of an inammatory component. However, inammation may occur as a secondary event in the removal of myelin debris by phagocytic cells following demyelination, such as is seen in toxin-induced models (15). Therefore, the mechanism of virus-induced demyelination in mice lacking SHP-1 remains to be determined but is most likely controlled by both direct and indirect consequences of increased virus replication in astrocytes and oligodendrocytes in the white matter of these mice. The present and previous studies on virus-induced demyelination indicate that mechanisms of demyelination are complex and are controlled by multiple genes that regulate innate, adaptive, and autoimmune responses (11, 41, 56, 64). In the present report, we have focused on a genetic alteration that may act at the level of the CNS glia for controlling TMEV replication. We have found that, compared to normal littermates, mice lacking SHP-1 produce more virus in brains and spinal cords after TMEV infection, succumb to a rapid-onset demyelinating disease, and display early spastic limb paralysis. We believe that disease in mice with a genetic deciency in SHP-1 activity is caused by a specic defect in innate antiviral responses against TMEV in the CNS glia. Our current studies are directed at discovering virus-glia interactions that are altered in the absence of SHP-1 and that lead to increased virus replication, oligodendrocyte pathology, and demyelination.

46. Reiss, C. S., and T. Komatsu. 1998. Does nitric oxide play a critical role in viral infections? J. Virol. 72:45474551. 47. Rodriguez, M. 1985. Virus-induced demyelination in mice: dying back of oligodendrocytes. Mayo Clin. Proc. 60:433438. 48. Rodriguez, M., and C. S. David. 1985. Demyelination induced by Theilers virus: inuence of the H-2 haplotype. J. Immunol. 135:21452148. 49. Rodriguez, M., J. L. Leibowitz, and P. W. Lampert. 1983. Persistent infection of oligodendrocytes in Theilers virus-induced encephalomyelitis. Ann. Neurol. 13:426433. 50. Rozhon, E. J., H. L. Lipton, and F. Brown. 1982. Characterization of Theilers murine encephalomyelitis virus RNA. J. Gen. Virol. 61:157165. 51. Selmaj, K., C. S. Raine, M. Farooq, W. T. Norton, and C. F. Brosnan. 1991. Cytokine cytotoxicity against oligodendrocytes. Apoptosis induced by lymphotoxin. J. Immunol. 147:15221529. 52. Shultz, L. D., D. R. Coman, C. L. Bailey, W. G. Beamer, and C. L. Sidman. 1984. Viable moth-eaten, a new allele at the moth-eaten locus. I. Pathology. Am. J. Pathol. 116:179192. 53. Shultz, L. D., P. A. Schweitzer, T. V. Rajan, T. Yi, J. N. Ihle, R. J. Matthews, M. L. Thomas, and D. R. Beier. 1993. Mutations at the murine moth-eaten locus are within the hematopoietic cell protein-tyrosine phosphatase(Hcph) gene. Cell 73:14451454. 54. Sommer, I., and M. Schachner. 1981. Monoclonal antibodies (O1 to O4) to oligodendrocyte cell surface: an immunocytological study in the central nervous system. Dev. Biol. 83:311327. 55. Song, M. M., and K. Shuai. 1998. The suppressor of cytokine signaling (SOCS) 1 and SOCS3 but not SOCS2 proteins inhibit interferon-mediated antiviral and antiproliferative activities. J. Biol. Chem. 273:3505635062. 56. Stohlman, S. A., and D. R. Hinton. 2001. Viral induced demyelination. Brain Pathol. 11:92106. 57. Sun, N., D. Grzybicki, R. F. Castro, S. Murphy, and S. Perlman. 1995. Activation of astrocytes in the spinal cord of mice chronically infected with a neurotropic coronavirus. Virology 213:482493. 58. Suzumura, A., E. Lavi, S. Bhat, D. Murasko, S. R. Weiss, and D. H. Silberberg. 1988. Induction of glial cell MHC antigen expression in neurotropic coronavirus infections. Characterization of the H-2-inducing soluble factor elaborated by infected brain cells. J. Immunol. 140:20682072. 59. Trottier, M., P. Kallio, W. Wang, and H. L. Lipton. 2001. High numbers of viral RNA copies in the central nervous system of mice during persistent infection with Theilers virus. J. Virol. 75:74207428. 60. Tsunoda, I., C. I. Kurtz, and R. S. Fujinami. 1997. Apoptosis in acute and chronic central nervous system disease induced by Theilers murine encephalomyelitis virus. Virology 228:388393. 61. Waxman, S. G. 1982. Membranes, myelin, and the pathophysiology of multiple sclerosis. N. Engl. J. Med. 306:15291533. 62. Weiner, L. P., S. G. Waxman, S. A. Stohlman, and A. Kwan. 1980. Remyelination following viral-induced demyelination: ferric ion-ferrocyanide staining of nodes of Ranvier within the CNS. Ann. Neurol. 8:580583. 63. Welsh, C. J., P. Tonks, P. Borrow, and A. A. Nash. 1990. Theilers virus: an experimental model of virus-induced demyelination. Autoimmunity 6:105 112. 64. Whitton, J. L., and R. S. Fujinami. 1999. Viruses as triggers of autoimmunity: facts and fantasies. Curr. Opin. Microbiol. 2:392397. 65. Wilt, S. G., E. Milward, J. M. Zhou, K. Nagasato, H. Patton, R. Rusten, D. E. Grifn, M. OConnor, and M. Dubois-Dalcq. 1995. In vitro evidence for a dual role of tumor necrosis factor-alpha in human immunodeciency virus type 1 encephalopathy. Ann. Neurol. 37:381394. 66. Wishcamper, C. A., J. D. Cofn, and D. I. Lurie. 2001. Lack of the protein tyrosine phosphatase SHP-1 results in decreased numbers of glia within the moth-eaten (me/me) mouse brain. J. Comp. Neurol. 441:118133. 67. Xu, Y., D. Banville, H. F. Zhao, X. Zhao, and S. H. Shen. 2001. Transcriptional activity of the SHP-1 gene in MCF7 cells is differentially regulated by binding of NF-Y factor to two distinct CCAAT elements. Gene 269:141153. 68. Zamanian-Daryoush, M., T. H. Mogensen, J. A. DiDonato, and B. R. G. Williams. 2000. NF-B activation by double-stranded RNA-activated protein kinase is mediated through NF-B-inducing kinase and IB kinase. Mol. Cell. Biol. 20:12781290. 69. Zheng, L., M. A. Calenoff, and M. C. Dal Canto. 2001. Astrocytes, not microglia, are the main cells responsible for viral persistence in Theilers murine encephalomyelitis virus infection leading to demyelination. J. Neuroimmunol. 118:256267.

25. 26.

27. 28. 29.

30. 31.

33. 34. 35. 36.
37. 38. 39. 40. 41. 42. 43. 44. 45.
Silberberg, and S. R. Weiss. 1987. Coronavirus mouse hepatitis virus (MHV)-A59 causes a persistent, productive infection in primary glial cell cultures. Microb. Pathog. 3:7986. Levine, B., J. E. Goldman, H. H. Jiang, D. E. Grifn, and J. M. Hardwick. 1996. BCL-2 protects mice against fatal alphavirus encephalitis. Proc. Natl. Acad. Sci. USA 93:48104815. Lieberman, A. P., P. M. Pitha, H. S. Shin, and M. L. Shin. 1989. Production of tumor necrosis factor and other cytokines by astrocytes stimulated with lipopolysaccharide or a neurotropic virus. Proc. Natl. Acad. Sci. USA 86: 63486352. Lipton, H. L. 1975. Theilers virus infection in mice: an unusual biphasic disease process leading to demyelination. Infect. Immun. 11:11471155. Lipton, H. L. 1994. Theilers viruses, p. 14231430. In R. G. Webster and A. Granoff (ed.), Encyclopedia of virology. Academic Press, Inc., San Diego, Calif. Lipton, H. L., R. Melvold, S. D. Miller, and M. C. Dal Canto. 1995. Mutation of a major histocompatibility class I locus, H-2D, leads to an increased virus burden and disease susceptibility in Theilers virus-induced demyelinating disease. J. Neurovirol. 1:138144. Liu, B., J. Liao, X. Rao, S. A. Kushner, C. D. Chung, D. D. Chang, and K. Shuai. 1998. Inhibition of stat1-mediated gene activation by PIASI. Proc. Natl. Acad. Sci. USA 95:1062610631. Lopez, F., G. Ferjoux, P. Cordelier, N. Saint-Laurent, J. P. Esteve, N. Vaysse, L. Buscail, and C. Susini. 2001. Neuronal nitric oxide synthase: a substrate for SHP-1 involved in sst2 somatostatin receptor growth inhibitory signaling. FASEB J. 15:23002302. Lorenz, U., A. D. Bergemann, B. Steinberg, J. G. Flanagan, X. Li, S. J. Galli, and B. G. Neel. 1996. Genetic analysis reveals cell type-specic regulation of receptor tyrosine kinase c-kit by the protein tyrosine phosphatase SHP1. J. Exp. Med. 184:11111126. Massa, P. T., R. Dorries, and V. ter Meulen. 1986. Viral particles induce Ia antigen expression on astrocytes. Nature 320:543546. Massa, P. T., S. Saha, C. Wu, and K. W. Jarosinski. 2000. Expression and function of the protein tyrosine phosphatase SHP-1 in oligodendrocytes. Glia 29:376385. Massa, P. T., A. Schimpl, E. Wecker, and V. ter Meulen. 1987. Tumor necrosis factor amplies measles virus-mediated Ia induction on astrocytes. Proc. Natl. Acad. Sci. USA 84:72427245. Massa, P. T., L. W. Whitney, C. Wu, S. L. Ropka, and K. W. Jarosinski. 1999. Mechanism for selective induction of 2-5 oligoadenylate synthetase, antiviral state, but not MHC class I genes by interferon-beta in neurons. J. Neurovirol. 5:161171. Massa, P. T., and C. Wu. 1998. Increased inducible activation of NF-B and responsive genes in astrocytes decient in the protein tyrosine phosphatase SHP-1. J. Interferon Cytokine Res. 18:499507. Massa, P. T., and C. Wu. 1996. The role of protein tyrosine phosphatase SHP-1 in the regulation of IFN- signaling in neural cells. J. Immunol. 157:51395144. McGavern, D. B., L. Zoecklein, S. Sathornsumetee, and M. Rodriguez. 2001. Assessment of hindlimb gait as a powerful indicator of axonal loss in a murine model of progressive CNS demyelination. Brain Res. 877:396400. McKinnon, R. D., T. Matsui, M. Dubois-Dalcq, and S. A. Aaronson. 1990. FGF modulates the PDGF-driven pathway of oligodendrocyte development. Neuron 5:603614. Miller, S. D., J. K. Olson, and J. L. Croxford. 2001. Multiple pathways to induction of virus-induced autoimmune demyelination: lessons from Theilers virus infection. J. Autoimmun. 16:219227. Min, W., J. Pober, and D. R. Johnson. 1998. Interferon induction of TAP1: the phosphatase SHP-1 regulates crossover between the IFN-alpha/beta and the IFN-gamma signal-transduction pathways. Circ. Res. 83:815823. Olson, J. K., J. L. Croxford, and S. D. Miller. 2001. Virus-induced autoimmunity: potential role of viruses in initiation, perpetuation, and progression of T-cell-mediated autoimmune disease. Viral Immunol. 14:227250. Parra, B., D. R. Hinton, N. W. Marten, C. C. Bergmann, M. T. Lin, C. S. Yang, and StohlmanSA. 1999. IFN-gamma is required for viral clearance from central nervous system oligodendroglia. J. Immunol. 162:16411647. Pouly, S., B. Becher, M. Blain, and J. P. Antel. 2000. Interferon-gamma modulates human oligodendrocyte susceptibility to Fas-mediated apoptosis. J. Neuropathol. Exp. Neurol. 59:280286.

INTERNATIONAL JOURNAL OF

ENERGY AND ENVIRONMENT
Volume 1, Issue 6, 2010 pp.937-952 Journal homepage: www.IJEE. IEEFoundation.org
CFD for hydrodynamic efficiency and design optimization of key elements of SHP
Ana Pereira, Helena M. Ramos
Civil Engineering Department and CEHIDRO, Instituto Superior Tcnico, Technical University of Lisbon, Av. Rovisco Pais, 1049-001, Lisbon, Portugal.
Abstract This paper aims to study how the flow behaves in key elements of small hydropower plants (SHP) which should be well designed in order to achieve properly the best hydraulic and energy efficiency. There are some hydrodynamic and structural fundaments that all hydro circuits design has to follow, and there are other aspects that vary from design to the flow behavior. The variables that influence the hydro systems design are related with performance, technical, operational and environmental aspects. For instance, design discharge, produced energy, intakes and outlets geometry are some of the technical variables. The components of SHP design should be characterized by a balance between hydraulic, structural, operational and environment efficiency and economic issues. To improve the hydraulic efficiency is necessary information concerning with hydrodynamic flow behavior. The knowledge in this area is still insufficient since the hydrodynamic flow patterns, in some key elements of hydraulic circuits of SHP are quite complex. Therefore this paper uses an advanced computational fluid dynamic (CFD) model for flow simulation, with the aim to improve the behavior comprehension enabling the identification of parameters variation which influences the performance efficiency of those components in the design criteria of such SHP. Since the inefficiency and the unsafe operating conditions are normally associated to separated flow zones, vorticity development, macro turbulence intensity, pressure gradients, shear stress increase, this paper intends to analyze causes and consequences of the flow behavior. Among these concerns it is possible to identify induced problems, such as vibrations, resonance effects, ruptures or collapses, cavitation, water column separation, significant friction losses, vortices and regions of reversed flow. Copyright 2010 International Energy and Environment Foundation - All rights reserved. Keywords: CFD analysis, SHP, Design optimization, Hydraulic circuit.
1. Introduction 1.1 Flow control valves Hydraulic systems are composed of a set of pipes, valves and other hydromechanical equipments necessary for adequate operational management, control and safety.
ISSN 2076-2895 (Print), ISSN 2076-2909 (Online) 2010 International Energy & Environment Foundation. All rights reserved.
International Journal of Energy and Environment (IJEE), Volume 1, Issue 6, 2010, pp.937-952
Valves are devices of great importance in the operation of hydro systems, in particular, when is necessary to control the flow [1, 2]. There are different types of valves in order to perform these functions. Depending on the shutter movement, the valves can be classified into two groups: - Valves with linear motion (e.g., globe; wedge; shears, needle, diaphragm); - Valves with angular motion (spherical; butterfly).
Figure 1. Different types of control valves The butterfly valve, (Figure 1a) is often used in water systems, under low hydraulic loads. They are valves suited for emergency shut-off, more specifically, for safety valves with overspeed closing disposal. The diaphragm valves are characterized by having a flexible membrane (diaphragm) whose periphery is fixed in the body of the valve (Figure 1b). As for membrane valve (Figure 1f), it works by pressing one side of the membrane through the actuator, restricting the passage of the flow. This type of valve is used, preferably, in situations of hostile operation. The spherical valves, the wedge (Figure 1c) and shears are the most suitable for the task of stopping the flow. The globe valves (Figure 1d) have a great use in automatic control of pressure and flow. They can present various shutter types and regulation hydraulic systems. Due to the pathway that the liquid makes inside, these valves have a large loss of hydraulic load, even in situations of total openness. The spherical valves (Figure 1e) are, preferably, used at systems with high hydraulic load or for quick flow cuts under high pressure situations. These valves when fully opened induce a low loss of hydraulic load. 1.2 Vortex formation The consequences of vortex formation and development can be the air entrance into the hydraulic circuit, flow circulation, separation zones and pressure and flow velocity variation [1]. There are three different types of vortices, namely forced vortex, free vortex and mixed vortex. On a forced vortex the water has a rotation movement around an axis as a solid body, which is caused by an external force, on which the tangential velocity is proportional to the distance from the axis, where the flow is rotational. When the actuation of the forces finishes, the rotation movement around the axis occurs freely inducing a free vortex, on which the flow velocity is inversely proportional to the distance from the axis, with an irrotational movement. The Euler number that represents the drop pressure by the increasing of the velocity is an adequate parameter to describe the vortex development. The mixed vortex is a combination of a forced vortex near the centre of rotation and a free vortex at the main body. 1.3 Intakes The vortex, which exists at intake pipes, is considered a free vortex with air dragging [2, 3, 4]. The free vortex can be classified a surface or a submerged type. From the stability point of view they can be identified as steady, unsteady or intermittent category and the circulation intensity can be organized in six levels, from weak to strong (Figure 2).

Figure 2. Different types of vortex at a typical SHP intake As main causes for vortex formation can be referred the eccentric orientation of the intake inlet relative to a symmetric approach, asymmetric approach flow conditions, unfavorable effects of obstructions such as offsets, piers or dividing walls, non-uniform velocity distribution caused by boundary layer separation, wind action in the flow surface, wakes or counter currents and the insufficient intake submergence. The consequences of vortex formation at intakes are air, swirl and even solid materials dragged into the intake conveyance hydraulic circuit which in turn induces unfavorable hydrodynamic impact on the operation and performance of turbines, and can cause dangerous hydropneumatic effects, such as noise and vibrations. 1.4 Draft tube and tailrace Another challenge is to understand the hydrodynamic of the flow through a draft tube and a tailrace of a SHP [5, 6]. Operational conditions have significant influence on the turbine efficiency, particularly when those conditions are out of the best efficiency point (BEP).
Figure 3. Change of the runner speed and the frequency of vortex at the draft tube of a Francis turbine One of the most important concerns on turbine runner and blades design is to guarantee a uniform flow to the draft tube entrance in non-disturbed conditions. The draft tube has a geometrical complexity resulting of changes on cross-section shape and direction in order to transform the flow kinetic energy into downstream potential energy position. In this region the flow presents large local pressure gradients, intense longitudinal vortices and regions of reversal flow (Figure 3). Disturbed flow entrance at the draft tube may cause flow reversal downstream of the runner with flow recirculation, formation of rope vortices, cavitation phenomena, which induce considerable efficiency
losses, dangerous pressure fluctuations, which can be propagated into the entire penstock. Thus poor inflow conditions may cause unfavorable hydrodynamic flow behavior. Hence, this paper presents CFD simulations on which the effects on efficiency of SHP at the intake and at the tailrace resulting from changes on solid element configuration are analyzed in order to define the best geometries that can improve the system performance [7]. 1.5 Measures and design criteria The main advantages of reducing the turbulence and vortex intensity are related with the consequent discharge and head increase. This study evaluates effectiveness of using adequate valve design (type, opening degree and diameter), anti-vortex devices such as baffles (vertical walls) or vanes, modifying the shape of flow approach area, to eliminate approach flow non-uniformities creating a good inflow approximation, removing sharp singularities, modifying intake and outlet geometries to lengthen uniform streamlines and to guarantee the minimum submergence or the admissible suction head in reducing the vortex and flow circulation effects at intakes, draftubes and tailraces and turbine operation [8, 9]. To avoid separated flow zones, with non uniform velocity distribution and to minimize head losses, changes on inlet and outlet walls shape design are considered. To decrease the free vortex with air dragging intensity, this study also evaluates the advantages of keep the water level above the critical submergence level, in order to always guarantee the intake inlet submergence. The hydrodynamic flow configuration and the design of special hydraulic structures and devices, such as control discharge structures and valves to control the flow behavior and regulate the pressure are also analyzed. 2. Mathematical approach Although the Navier-Stokes equations have a limited number of known analytical solutions, they are adequate for the flow computational model, by numerical approach of computational fluid dynamics. The CFD model (FloEFD) solves the Navier-Stokes equations, which are formulations of mass, momentum and energy conservation laws for fluid flows. The equations are supplemented by fluid state equations defining the nature of the fluid, and by empirical dependencies of fluid density, viscosity and thermal conductivity on temperature [7, 10, 11]. The Navier Stokes equations are presented by equations (1) for incompressible flows, where these equations are based on differential equations of linear momentum for a Newtonian fluid with constants density and viscosity.

g x g y g z dp du 2u 2u 2u + ( 2 + 2 + 2 ) = dx dt x y z dp dv 2v 2v 2v + ( 2 + 2 + 2 ) = dy dt x y z (1)
dp dw 2w 2w 2w + ( 2 + 2 + 2 ) = dz dt x y z where: : volumetric mass (kg/m3); g : acceleration of gravity (m/s2); p : pressure (Pa); : dynamic viscosity (kg/(ms)); u, v, w : velocity components for each moving fluid particle, depending on x, y, z coordinates for a given t instant (m/s). The incompressible fluid flow behavior is determined by the velocity and pressure variables and their variations in time and space. In equations (1), that allow to get pressure and velocity fields, the velocity components at each point x, y, z are vector fields and the pressures. Most of the flows that occur at hydraulic circuits are turbulent, and this CFD model allows the numerical modeling of both laminar and turbulent conditions. The turbulent flows occur for high values of Reynolds number, given y the equation (2).

941 (2)

UD where: D : conduit diameter (m).
When the flow is turbulent the variables present at each instant random fluctuations. A fluid under to low pressures can reach the vapor pressure at the local temperature leading to the formation of vaporous cavities. The fluid undergoes a phase change and cavities filled with fluid vapor and other dissolved gases are formed. When analyzing areas of flow conditions that leads the occurrence of cavitation, this CFD model, uses an homogeneous equilibrium model of cavitation in water
3. Results analysis 3.1 Hydrodynamic flow behavior through flow control valves The flow was simulated through flow control valves for different valve closure positions [12]. For valves with actuators angular movement (e.g. ball valve) the flow was simulated for different valve opening angles. The angle of valve opening is measured in relation to the position of fully closed valve. For valves with actuators linear movement (e.g. globe valve) the flow was simulated for different opening percentages. The variation of valve head loss coefficient with valve closure position was obtained. This variation shows the energy dissipation induced by the valve in the flow for different valve opening positions.
3.1.1 Ball valve The first step was to build the ball valve geometry model. Two pipe branches of equal length and diameter to the valve size were connected at upstream and downstream of the ball valve geometry model. Concerning to the energy dissipation induced by the ball valve, the values shown in Table 1 and in Graph 1 were obtained. Head losses are associated to the opened valve position. Thus the lower the opening angle the lower the pressure downstream of the valve which may lead to cavitation occurrence. Table 1. Head loss and local head loss coefficient values for different ball valve opening angles Ball valve opening angle (o) H (m) 1300,25 24,28 11,13 2,46 0,48 0,01 Kv (-) 180,80 13,62 8,16 3,17 0,84 0,02

Coeficiente de perda de carga, Kv (-)
1000,00 100,00 10,00 1,00 0,10 0,100
Angle ngulo de abertura ()
Graph 1. Ball valve local head loss coefficient Kv variation with the respective opening angle
From velocity vector distribution, represented in Figure 4, can be concluded that the flow trajectories converge upstream of the valve which can lead to flow separation in the same region and to rotational movement with high turbulence inside the valve. Downstream the valve the velocity vector distribution shows a separation flow zone where occur strong vorticity with high turbulence intensity associated, which leads to local flow energy dissipation. As a result of this dissipation there is negative pressure downstream of the valve which contributes to the cavitation occurrence in this region. The major part of the pressure loss occurs at the closure outlet.
Figure 4. Ball valve opening angle of 20 - pressure distribution in a longitudinal section of a ball valve Figure 5 shows the cavitation occurrence for a ball valve opening angle of 20. Immediately downstream of the valve high vapor volume fraction values and low density of the mixture of water vapor, other dissolved gases in the water body and water values are verified. The pressure values increase again in the pipe downstream of the valve, therefore the vapor volume fraction values decrease again towards downstream and the density values of the mixture of water vapor, other dissolved gases and water increase again in the same direction. Both the water vapor density as the other dissolved gases density is lower than the water density, so that when these gases are dissolved in the water body there is a gas-water mixture of density lower than water density. The vapor volume fraction is the ratio between water vapor and other dissolved gases volume and the water volume in the gas-water mixture. Thus it is concluded that high vapor volume fraction values and low gas-water mixture density values indicate the presence of vapor bubbles in water body that are associated to the cavitation occurrence. The occurrence of this phenomenon in valves has strong influence in valve local head loss coefficient K v () values and in its security.
Figure 5. Cavitation resulting from a ball valve opening angle of 20 - vapor volume fraction values (a) and gas-water mixture density (b) distribution values
The flow through the valve results in the contraction of the liquid vein (Figure 6a) immediately upstream and downstream of the closure and therefore in the flow velocity increase in these regions. What explains the pressure decrease from the region immediately upstream of the actuator towards downstream. This pressure decrease resulting from a ball valve opening angle of 45, but conditions for cavitation occurrence are not created. The representation of flow trajectories, Figure 6b, allows the identification of flow separation, rotational movement inside the valve and vorticity with high turbulence intensity associated, downstream of the closure.

(a) (b) Figure 6. Ball valve opening angle of 45 - velocity vector and pressure distribution (a) and flow trajectories (b) in a longitudinal section of a ball valve
Graph 2. Ball valve opening angle of 45 - velocity (v/vo) (a) and pressure (p/po) profiles Graph 2 shows the layout of the velocity and pressure profiles along stretches immediately upstream and downstream of the valve. For this opening, it shows the rotational flow at downstream of the valve. Due to the convergence of flow paths upstream, the flow has irrotational characteristics as is in a narrowing section. 3.1.2 Globe valve From the 3D geometry of a globe valve were obtained the results regarding to head losses induced. From Table 2 and Graph 3 can be concluded that in the case of a globe valve the local head loss coefficient K v value varies little with the valve opening percentage, and that this valve has higher K v values, concerning to the fully opened valve position, than the other analyzed valves. This can be justified considering the valve geometry much more tortuous for the flow passage than the other valves geometry.
Table 2. Head loss and local head loss coefficient values for different globe valve opening angles Percentagem de abertura da vlvula de globo (%) 100 H (m) 13,60 3,62 2,56 1,71 1,70 Kv (-) 17,25 5,69 4,48 2,82 2,81 The geometry of this valve includes curves, both upstream and downstream of the valve actuator. In the soffit of this curves there is a pressure reduction and a velocity increase. This variation is more evident in the smaller radius curves immediately upstream and downstream of the obturator (Graph 3). The smallest radius curve located immediately downstream of the closure corresponds to a contracted flow section and downstream from it occurs an enlargement of the section that causes the velocity decrease and the flow trajectories divergence.
0,00 0,00 20,00 40,00 60,00 80,00 100,00
Percentagem de abertura da vlvula (%)

Percentage

Graph 3. Globe valve - local head loss coefficient variation with the opening angle As a result is formed a separation flow zone, where the pressure decreases giving rise to the formation of macro vorticity which justifies the energy dissipation induced into the flow due to the globe valve. In turn, this vortex locally blocks the flow section (Figure 7) which causes the flow trajectories contracting and gives rise to new flow separation and thus to energy losses. The formed vortices, which detach and disintegrate towards downstream, cause valve and pipe vibrations and give rise to turbulent wake formation. For larger valve openings the reduction in KV values is low, which can be justified considering that the valve region where the obturator moves always occurs a decrease on pressure and velocity values for any opening degree.

Figure 7. Velocity vector and pressure distribution for an obturator opening of 40% (a) and flow trajectories (b) in a longitudinal section of a ball valve The velocity profile at downstream shows a rapid increase in velocity values, which is due to the concavity of the external borders of the valve, which follows a rapid velocity reduction, explained by the occurrence of flow separation zone, with macro vorticity (Graph 4) with significant recirculation zone along the curvature of the outside of the outlet valve.
Graph 4. Globe valve: (a) velocity (v/vo) and (b) pressure (p/po) profiles 3.2 SHP intake There are different types of intakes with diversion flow to the turbine through the penstock: frontal, lateral, bottom drop and siphon type. It is necessary to design the entrance shape in order to avoid separated zones of the flow and excessive head loss through wing walls and to verify the minimum submergence in order to avoid vortex formation and, consequently, air dragging (Figure 8).
Figure 8. Improved SHP intake: (a) velocity distribution with velocity vectors; (b) streamlines; (c) static pressure distribution The Cauchy-Rieman equations enable the velocity potential to be calculated if stream function is known resulting in the Laplace equation, verifying that streamlines and equipotential lines are mutually perpendicular originating a flow net of streamlines and equipotential lines. When the stream lines converge the velocity increases (Figure 8) and consequently distance between equipotential lines will decrease. So abrupt changes of the outer boundary must be avoided in order to avoid the separation of the streamline from the boundary.

0.45 0.6

101262 101260

101090

0.5 0.35 0.4
Segment AB X-velocity (m/s) Segment CD X-velocity (m/s)

101252

Segment AB Pressure (Pa )

101086

Segment CD Pressure(Pa )
Segment AB Pressure(Pa) Segment CD Pressure(Pa)

101084

Segment AB X-Velocity (m/s) Segment CD X-Velocity (m/s)

101246

101082

101080

0 0.1.2.3.4.Segment AB and CD , Z coordinates(m)
Segment AB and CD , Z coordinates(m) 4 5

101078

101244

0.05 0.1

101242

101076

Graph 5. Velocity (a) and pressure (b) variation along the AB and CD water intake segments In sharp boundaries the velocity at the separation volume will be zero and the fluid trapped there will be stagnant. In convergent the velocity turns away from the fluid, indicating high velocity in the separation with significant rotation. Therefore the assumption of irrotational flow is not valid there. Hence smooth converging has no separation. By analyzing Grapgh 5(a) shape, the conclusion is that the flow is turbulent at the intake entrance and downstream of the trash rack. The Graph 5b is consistent with the Figure 8c and shows the pressure loss across the trash rack. 3.3 Francis turbine The first step is to create the geometry model, using a CAD software, of the hydraulic Francis turbine, represented in Figure 9, in order to simulate the hydrodynamic flow behavior through it. For this model were defined as boundary conditions a inlet volume flow of 6 m3s-1 at the inlet and a static pressure of 121590 Pa at the outlet. For the runner angular velocity two scenarios were considered of 750 rpm and 1000 rpm. The head loss H ( m ) is determined for each scenario considering the equation (3):
(3) where: P0,ups : total pressure at upstream section (Pa); P0,dow :total pressure at downstream section (Pa); :

P0, dow )

water specific weight (N/m3). As upstream sections the inlet model section and the inlet spiral case section were considered. As downstream sections the outlet model section, first and last draft tube bend sections were considered. In order to determine the total pressure P0 ( Pa) at those sections the CFD model considers the equation (4). (4) 2 where: p : static pressure (Pa); : water volume mass (kg/m3); U : average flow velocity at each section (m/s).

P0 = p +

Figure 9. Francis turbine geometric model For the first scenario of 750 rpm the following results related with the pressure head are obtained. Table 3. Pressure Head for the scenario of 750 rpm inlet model section outlet model section inlet spiral case section - first draft tube bend section (net head) inlet spiral case section - last draft tube bend Pressure Head (m) 130
For the second scenario of 1000 rpm the following results related with the pressure head are obtained. Table 4. Pressure Head for the scenario of 1000 rpm inlet model section outlet model section inlet spiral case section - first draft tube bend section (net head) inlet spiral case section - last draft tube bend Pressure Head (m) 186
For this greater angular velocity related with the same runner geometry and the same volume flow rate the obtained pressure head values are greater, so that the turbine net head is also greater enabling greater energy production [13]. The CFD model also provides results enabling the analysis of the hydrodynamic flow parameters distribution on the model's surfaces or on sectioning planes (Figure 10). Analyzing Figures 10 (a) and (b), the conclusion is that the velocity increases from the inlet to the runner were it reaches the maximum value, and diminish from the runner to the outlet. From the runner to the outlet the flow is rotational. At the runner outlet and the draft tubes first stretch and bend, the flow velocity at the inner part is very close to zero and increases from the inner to outer wall.
Figure 10. First scenario of 750 rpm: (a) velocity distribution with velocity vectors; (b) flow velocity trajectories; (c) static pressure distribution
30 Segment AB Z-velocity(m/s) 25 Segment CD Y-velocity (m/s) 5 10

600000

160000

120000

Segment AB Z-velocity (m/s)
Segment CD Y-velocity (m/s)

200000

SEgmentCDPressure(Pa)

SegmentABPressure(Pa)

100000

80000 0.6 60000

0 0.6 0.4 0.0 0.2 0.4 0.6 20
SegmentABPressure(Pa) SegmentCDPressure(Pa) 600000 Segment AB and CD, X coordina tes(m) 0

10 Segment AB a nd CD, X coordina tes(m)
Graph 6. First scenario of 750 rpm: (a) velocity; (b) pressure variation along the AB and CD Francis turbine segments This is in accordance with the vortex developed at the runner outlet, which can be seen by the flow velocity trajectories shape and by the flow velocity vector field. On the Figure 10c the difference between the static pressure at the sections upstream the runner and downstream the runner can be seen and justify the values obtained for the pressure head at Table 3. Analyzing the low pressure values obtained at the runner exit and at the draftube is possible to predict cavitation for this flow conditions [14]. Analyzing Graph 6, at the runner outlet (segment AB) both the velocity and the pressure decreases from the periphery to the center of the segment. This shows that here the flow is rotational. However, unlike the pressure values, the velocity values decreases towards the periphery, providing flow separation, thus
at this segment periphery the flow is irrotational. At the segments CD ends the velocity values are negative, thus the water flows towards the exit (given the y-axis direction). However at the segments CD center the velocity values are positive thus the flow enters the model. This shows that there is a vortex at the draft tubes diffuser which can be consider a reverse flow zone. This is in accordance with the flow velocity trajectories shape and with the flow velocity vector field.
Figure 11. Second scenario of 1000 rpm: (a) velocity distribution with velocity vectors; (b) flow velocity trajectories; (c) static pressure distribution
14 Segment EF velocity (m/s) Segment GH Y-velocity (m/s) 12 15

250000 200000

Segment GH Pressure (Pa )

2155000

2152000 2154000
Segment EF Pressure (Pa )
Segment GH Y-Velocity (m/s) Segment EF Velocity (m/s)

2153000

2151000

0 0.4 0.0.2 0.4 0.6 0.8

0.4 0.2

2150000

Segment EF Pressure(Pa) Segment GH Pressure(Pa)

100000 150000

2149000
2 Segment EF and GH, coordinates(m)
Segment EF and GH , coordinates(m)
Graph 7. Second scenario of 1000 rpm: (a) velocity; (b) pressure variation along the EF and GH Francis turbine segments Analyzing Figures 11 (a) and (b), the conclusions are quite similar to the previous, but in this case the flow velocity on the runner and the pressure head are even greater. The vortex formed at the inner part, from the runner outlet to the model outlet, is in this scenario, even more intense and occupies more space, leading to greater head losses as a result of vortex turbulence. The pressure gradient which is observed at Figure 11b is in accordance with the values obtained on Table 4 and the lower pressure values indicate the cavitation occurrence. The velocity variation at spiral case segment EF (Graph 7) shows that the flow is turbulent. The velocity values increase from the outer wall towards the inner space where the flow direction changes, and the water flows from the model outlet to the draft tubes first stretch, being a reverse flow region.

3.4 Francis turbine outlet and tailrace CFD model has been applied to simulate the highly turbulent flow conditions in turbine and tailrace region, since there are important parts of a hydropower facility that carries water away from the turbines. This analysis is used to study operational and structural possible modifications that could improve the hydrodynamic behavior for different flow and hydromechanical conditions. Critical depths, suction heads and the volume rate of flow can be identified and avoided since such occurrence are limiting factors for a good design, with strongly influence in the hydro systems efficiency.

a) b) c)

Figure 12. Turbine and tailrace: (a) velocity distribution with velocity vectors; (b) streamlines; (c) static pressure distribution Shear stresses are developed when the fluid is in motion, when the particles move relative to each other with different velocities or when the fluid is in contact with a solid boundary. In Figure 12 is visible, due to fluid rotation, a vortex motion in the draft tube where the streamlines form a set of concentric circles and there is a change of total pressure or energy.
Segment AB Y-Velocity(m/s) 5 Segment CD Y-Velocity(m/s) 0.8 0.6 0.4 0.2
Segment AB Y-velocity (m/s)

520000 500000

Segment AB Pressure (Pa)

510000 500000

400000

450000 440000

300000
Segment AB Pressure(Pa) Segment CD Pressure(Pa) 0

430000 0.6 0.8

30 Segment AB and CD, X coordinates(m)
0.0.2 0.4 Segment AB and CD, X coordinates(m)
Graph 8. Turbine outlet and tailrace: (a) velocity; (b) pressure variation along the AB and CD segments In Graph 8 the inversion of the flow velocity in the middle of the draft tube enhance the inversion of the flow due to the influence of the rotational speed and the pressure reduction in that middle zone, confirming the rotational flow type. The influence of the design in the tailrace is quite important in terms of with, length and depth, position of the gate and the uplift of the sill making the transition into the river.
4. Conclusions The advanced CFD model used in this research (FloEFD) solves the Navier-Stokes equations, which are formulations of mass, momentum and energy conservation laws for fluid flows. This CFD model is able of predicting both laminar and turbulent flows. Most of the fluid flows in engineering practice are turbulent, so this model uses the Favre-averaged Navier-Stokes equations, where time-averaged effects of the flow turbulence on the flow parameters are considered, whereas the other, i.e. large-scale, timeISSN 2076-2895 (Print), ISSN 2076-2909 (Online) 2010 International Energy & Environment Foundation. All rights reserved.

dependent phenomena are taken into account directly. Through this procedure, extra terms known as the Reynolds stresses appear in the equations for which additional information must be provided. To close this system of equations, FloEFD employs transport equations for the turbulent kinetic energy and its dissipation rate, the so-called k- model. This paper shows the utility of the CFD numerical simulations as a tool for design and optimization of hydropower performance and flow behavior through hydromechanical devices or hydraulic structures of intake and outlet types. Experimental tests not always are viable because they are very expensive and it is much more difficult to analyze different scenarios and boundaries. The flow of a real fluid in contact with a boundary implies velocity variations, pressures gradients and shear stress development, from which energy losses result, as important factors to take into account in the concept, design, construction, operation and maintenance of hydropower plants or any other type of hydraulic conveyance system.
Acknowledgements To projects HYLOW from 7th Framework Programme (Grant n 212423) and FCT (PTDC/ECM/65731/2006) and (PTDC/ECM/68694/2006) which contributed to the development of this research work in the domain of computational dynamic analyses. References [1] Ramos, H., Non conventional dynamic effects in Pressurised hydraulic systems. Elements to support the course Unsteady Flows and Hydropower and Pumping Systems of Hydraulic MSc Course. IST, DECivil, 2004, (in Portuguese). [2] Ramos, H., Guidelines for Design of Small Hydropower Plants. Book published by WREAN (Western Regional Energy Agency and Network) and DED (Department of Economic Development - Energy Division). Total pp 205. Belfast, North Ireland. ISBN 972-96346-4-5, 2000. [3] Ruprecht, A., Eisinger R., Gde, E.: Innovative Design Environments for Hydro Turbine Components, Bern, HYDRO 2000, 2000 [4] Ramos, H., Hydropower and Pumping Systems. MSc of Hydraulic and Water Resources. IST, DECivil, 2003, (in Portuguese). [5] Douglas, J.F. Gasiorek, J.M., Swaffield, J.A., Fluid Mechanics. 3rd Edition, Longman Group Limited, 1998. [6] Visser, F.C., Brouwers, J.J.H., Jonker, J.B., Fluid flow in a rotating low-specific-speed centrifugal impeller passage. J. Fluid Dynamics Research, 24, pp. 275-292, 1999. [7] MENTOR GRAPHICS, FloEFD - Technical Reference, (EUA), 2008. [8] Lipej, A., Poloni, C.: Design of Kaplan Runner Using Multiobjective genetic algorithm optimization, Journal of Hydraulic Research, Vol. 38, 2000. [9] Mrsa, Z., Sopta, L., Vukovic, S.: Shape optimization method for Francis turbine spiral casing design, ECCOMAS, Athen, 1998. [10] Ramos, H. and Almeida, A. B., Parametric Analysis of Waterhammer Effects in Small Hydropower Schemes. HY/1999/021354. ASCE - Journal of Hydraulic Engineering. Volume 128, 7, pp. 689697, ISSN 0733-9429, 2002. [11] Ramos, H; Almeida, A. B., Dynamic orifice model on waterhammer analysis of high and medium heads of small hydropower schemes. Journal of Hydraulic Research, IAHR, Vol. 39 (4), pp. 429436, ISSN-0022-1686, 2001. [12] Pereira, A., Ramos, H.M., Hydrodynamic analyses in water conveyance components, IX SEREA Seminario Iberoamericano sobre Planificacin, Proyecto y Operacin de Sistemas de Abastecimiento de Agua. Valencia (Espaa), 24-27 de Noviembre de 2009, (in Portuguese). [13] Skotak A.: The CFD Prediction of the Dynamic Behavior of Pump-Turbine, Proc. 11th IAHR WG1 meeting, Stuttgart, 2003. [14] Backman A.G.: CFD Validation of Pressure Fluctuations in a Pump Turbine, Masters Thesis, TU Luela, 2008.

Ana Pereira is in his final year of Civil Engineer MSc at Instituto Superior Tcnico (Technical University of Lisbon Portugal) and has few publications. She is researcher under the scientific domain of water and energy, CFD for SHP and participates in the FCT Project - PTDC/ECM/68694/2006 Vulnerability and behaviour of hydraulic conveyance systems. E-mail address: cardper@gmail.com
Helena M. Ramos has Ph.D. degree with the Aggregation Title and she is Professor at Instituto Superior Tcnico (from Technical University of Lisbon - Portugal) at Department of Civil Engineering. Expert in different scientific domains: Hydraulics, Hydrotransients, Hydropower, Pumping Systems, Leakage Control, Energy Efficiency and Renewable Energy Sources, Water Supply, Vulnerability. More than 250 publications being 1 book in Small Hydro, 52 in Journals with referee and 110 in International Conferences; Supervisor of several post-doc, PhDs and MSc students and author of 8 innovative real solutions in the domain of Civil Engineering - hydropower and hydraulic system control. E-mail address: hr@civil.ist.utl.pt or hramos.ist@gmail.com