Expansion of visceral adipose tissue correlates with the metabolic syndrome and increased cardiovascular risk. Hypertrophied visceral fat becomes inflamed, causing increased lipolysis, decreased triglyceride storage, and lipotoxicity in skeletal muscle and liver resulting in insulin resistance. Perivascular adipose tissue is a normal component of the adventitia of arteries in humans and animals. Whether or not perivascular adipose also becomes inflamed in obesity is an important question, as this may be an additional, direct mechanism by which obesity causes vascular inflammation and disease. Thus, for the first part of my thesis, we asked the question: does perivascular adipose in mice become inflamed with high fat feeding? In contrast to visceral adipose, macrophage gene expression was not increased in perivascular adipose in response to high fat diet, and this correlated with reduced F480 antigen positive cells as seen by immunohistochemistry and flow cytometry. Interestingly, perivascular adipose surrounding the thoracic aorta was similar to brown adipose tissue, a highly thermogenic fat depot, as shown by histology and DNA microarrays. Moreover, inter-scapular brown adipose was also resistant to diet induced inflammation in comparison to visceral adipose. These findings suggest that brown adipose in the perivascular niche may serve to protect the vasculature from diet induced inflammation, or from cold exposure, or both; whether or not brown perivascular adipose tissue exists in humans has yet to be determined. In the second part of my thesis, we evaluated the role of perivascular adipose tissue in the apolipoprotein E knockout mouse, which exhibits severe hyperlipidemia and atherosclerosis, but is resistant to diet induced obesity and glucose intolerance. We tested the hypothesis that in this model of severe atherosclerosis, inflammation of perivascular adipose does occur. However, we were surprised to find that macrophage specific gene expression, as determined by either microarray analysis or quantitative polymerase chain reaction, was not increased in either the perivascular or the visceral adipose of high fat diet fed apolipoprotein E knockout mice. While the visceral adipose of wild type mice had extensive alterations in gene expression in response to high fat diet, in particular, enrichment of inflammatory gene expression and broad down regulation of peroxisome proliferator activated receptor gamma target genes, apolipoprotein E knockout visceral adipose did not. Importantly, the apolipoprotein E knockout visceral adipose instead showed increased expression of genes encoding enzymes in fatty acid oxidation pathways. High fat diet fed apolipoprotein E knockout visceral adipose was also characterized by smaller adipocyte size. We conclude that, 1) inflammation in thoracic perivascular adipose does not occur in conjunction with diet induced obesity in normal animals nor with atherosclerosis in apolipoprotein E knockout mice, 2) thoracic perivascular adipose tissue is essentially identical to brown adipose tissue in mice, thus potentially protecting the vasculature from the cold, and 3) apolipoprotein E knockout mice remain lean on a high fat diet, despite hyperlipidemia and atherosclerosis, and the decreased adiposity correlates with decreased adipocyte size and adipose inflammation but increased oxidation of fatty acids. Consistent with previous work showing apolipoprotein E controls adipocyte uptake and deposition of triglyceride, its absence prevents adipocyte hypertrophy and resultant inflammation of visceral adipose tissue. Thus limiting adipocyte acquisition of fatty acids may be advantageous, provided that compensatory mechanisms to prevent sustained hyperlipidemia and peripheral organ lipotoxicity can be activated.

Very long chain acyl-coA dehydrogenase (VLCAD) is the rate-limiting step in mitochondrial fatty acid oxidation. VLCAD deficient mice and patients’ clinical symptoms stem from not only an energy deficiency but also long-chain metabolite accumulations. VLCAD deficient mice were treated systemically with 1x10 12 vector genomes of rAAV9-VLCAD. Expression was detected in the liver, heart and muscle. Also substantial expression of VLCAD was noted in the brain, where it was expressed across different sections of the brain and in different cell types with different morphologies. Biochemical correction was observed in vector-treated mice beginning two weeks post-injection, as characterized by a significant drop in long chain fatty acyl accumulates in whole blood after an overnight fast. Changes persisted through the termination point around 20 weeks post injection. Magnetic resonance spectroscopy (MRS) and tandem mass spectrometry (MS/MS) revealed normalization of intramuscular lipids in treated animals. Correction was not observed in liver tissue extracts, but cardiac muscle extracts showed significant reduction of long chain metabolites. Disease-specific phenotypes were characterized, including thermoregulation and maintenance of euglycemia after a fasting cold challenge. Internal body temperatures of untreated VLCAD-/- mice dropped below 20°C and the mice became lethargic, requiring euthanasia. In contrast all rAAV9-treated VLCAD-/- mice and the wild-type controls maintained body temperatures. rAAV9-treated VLCAD-/- mice maintained euglycemia, whereas untreated VLCAD-/- mice suffered hypoglycemia following a fasting cold challenge. These promising results suggest rAAV9 gene therapy as a potential treatment for VLCAD deficiency in humans.

The Snail protein functions as a transcriptional regulator to establish early mesodermal cell fate in Drosophila. Later, in germ band-extended embryos, Snail is considered a pan-neural protein based on its extensive expression in neuroblasts. The evidence presented in thesis links snail expression and function in CNS. Cloning and functional characterization of a novel snail homologue, in Drosophila, are also described here. Cloning of this gene, worniu (Chinese for snail), revealed that the neural function of snail is masked by this and another closely related gene escargot. Both Escargot and Worniu contain zinc finger domains that are highly homologous to that of Snail. These three members of Snail protein family are redundantly required for CNS development. Although not affecting formation of neuroblasts, the loss of expression of these three members correlates with disruption of Nb asymmetry and division. Downstream targets of Snail protein family, in these processes, are inscuteable and string. In mutant embryos, which have the three genes deleted, the RNA expression of inscuteable and string is significantly lowered. Consistent with the gene expression defects, the mutant embryos have loss of asymmetric localization of prospero RNA in neuroblasts and nuclear localization of Prospero protein in ganglion mother cells. Transgenic expression of inscuteable and string together, in the snail family deletion mutant, efficiently restores the Prospero expression in GMC, demonstrating that the two genes are key targets of Snail in Nbs. Like in the mesoderm, in CNS Snail function depends on interaction with dCtBP co-repressor. These results suggest that Sna [Snail] family of proteins control both asymmetry and cell division of neuroblasts by activating, perhaps indirectly, the expression of inscuteable and string.

In vivo DNA is compacted tightly, via its association with histones and non-histone proteins, into higher-order chromatin structure. In this state, the DNA is refractory to the cellular factors that require access to DNA. The repressive nature of chromatin is alleviated in part by the action enzymes that modify chromatin structure. There are two major groups of chromatin modifying enzymes: those that post-translationally modify histones by the addition of small chemical moieties and those that utilize the energy derived from ATP hydrolysis to physically disrupt chromatin structure. The SWI/SNF enzyme belongs to this latter group. The SWI/SNF complex was identified originally in yeast. Several of its subunits are required for the expression of a subset of inducible genes. The ATPase activity is provided by the SWI2/SNF2 protein. In mammals, there are two biochemically separable SWI/SNF complexes that contain either BRG1 or BRM, both homologs of yeast SWI2/SNF2. The yeast and mammalian SWI/SNF complexes are able to disrupt the Dnase I digestion pattern of in vitro assembled mononucleosomes and arrays, as well as facilitate the accessibility of restriction nucleases and transcription factors. The mechanism by which SWI/SNF functions has yet to be elucidated. SNF5 is a component of the yeast SWI/SNF complex. It is required for sucrose fermentation and mating type switching. The mammalian homolog of Snf5 is SNF5/INI1. SNF5/INI1 was identified simultaneously by two groups as a protein that shares homology with Snf5 and via a yeast two hybrid assay as a protein that interacts with HIV integrase (INtegrase Interactor). INI1 is a component of all mammalian SWI/SNF complexes purified to date. In humans, mutations and/or deletions in INI1 are associated with a variety of cancers, including malignant rhabdoid tumors, choroid plexus carcinomas, medullablastomas, primitive neuralectodermal tumors, and some cases of leukemia. Furthermore, constitutional mutations within INI1in individuals presenting with these tumors support the role of INI1 as a tumor suppressor. In this thesis, we show that Ini1 also functions as a tumor suppressor in mice. Approximately 20% of mice heterozygous for Ini1 present with tumors. Most of these tumors are undifferentiated or poorly differentiated sarcomas with variable rhabdoid features. All tumors examined to date show loss of heterozygosity at the Ini1 locus. We also show that Ini1 is essential for embryonic development. Mice homozygous-null for Ini1die between days 4 and 5.5 post-fertilization due to an inability to adhere to their substratum, form trophectoderm, and expand their inner cell mass. We further characterize the function of Ini1 in tumor suppression by generating mice heterozygous for both Ini1 and either Rb or p53. While heterozygosity at the Ini1 locus appears to have no effect on the rate of tumorigenesis in Rb-heterozygous mice, many of the tumors arising in compound heterozygous mice present with an altered morphology. This finding suggests that Ini1 may contribute to tumor progression due to loss of Rb. In contrast, mice compound heterozygous for Ini1 and p53 show a marked reduction in the rate of tumorigenesis compared to p53-heterozygous mice. Furthermore, the tumor spectrum is altered in these compound heterozygous mice. These findings suggest that Ini1 may function normally to repress p53 activity. Lastly, we show that expression of the Ini1 tumor suppressor itself is regulated tightly. Tissues and cells heterozygous for Ini1 express roughly equivalent levels of Ini1 protein and mRNA as their wild-type counterparts. We further show that this compensation is mediated by an increase in the rate of transcription from the wild-type Ini1 allele. Moreover, when exogenous Ini1 is introduced into Ini1-heterozygous cells, expression from the Ini1 promoter is reduced. These data indicate that a compensatory mechanism exists to ensure that the steady-state levels of Ini1 are constant. In summary, research detailed in this thesis has contributed to our understanding of the regulation of Ini1 as well as the role this protein plays in mammalian development and tumor suppression.

The overall objective of this thesis work was to develop new insights into the etiology of scleroderma, a human systemic autoimmune disease, by analyzing the autoantibodies to centrosome antigens that develop during the disease. Centrosomes are perinuclear organelles that form microtubule arrays, including mitotic spindles that ensure the faithful segregation of chromosomes during mitosis. These studies used a novel methodology to determine the prevalence of anti-centrosome autoantibodies in patients with scleroderma. Recombinant centrosome antigens were used to determine the antigenic specificity of anti-centrosome antibody subsets by immunoblotting. Centrosome marker antibodies were used in indirect immunofluorescence assays to distinguish centrosomes within the polymorphic staining pattern frequently given by scleroderma sera. We found that 43% of patients are autoreactive to centrosomes, a prevalence higher than has been reported for any other scleroderma autoantigen. Half of the centrosome-positive patients also had autoantibodies against other antigens used in scleroderma diagnosis. However, in the remaining half of these patients, anti-centrosome antibodies represented the sole class of autoantibodies that was detectable. Anti-centrosome antibodies were detected in only a small percentage of normal individuals and patients with other connective tissue diseases. These data suggest that anti-centrosome autoantibodies may represent a new diagnostic tool in scleroderma. Upon examination of anti-centrosome autoantibody development in an animal model, it appeared that this autoantibody specificity may develop in mice as a consequence of an infection. An infectious agent was isolated by plaque-formation from carrier mice. Further characterization of the infectious agent was undertaken to obtain information on its physical, morphological and cytopathological properties. The infectious agent was identified by sequence and unique antigenic properties to be homologous to the pig pathogen Mycoplasma hyorhinis. When reintroduced into naive mice, the murine mycoplasma triggered anti-centrosome autoantibody development. While anti-centrosome autoantibodies of IgM isotype are part of the repertoire of naive unimmunized mice, mycoplasma infection specifically triggered the development of anti-centrosome IgG. Moreover, centrosome autoreactivity was prevented by antibiotic treatment. The autoantibody response evolved to recruit additional specificities, having IgM isotypes, reactive to endoplasmic reticulum-associated autoantigens.

Members of the Runx family of transcription factors play essential roles in the differentiation and development of several organ systems. Here we address the contribution of the osteoblast-related Runx gene, Runx2, to the osteogenic and chondrogenic differentiation of mesenchymal stem cells. Using a transgenic mouse model, we observe Runx2 transcription through one of its two known promoters (designated P1 in pre-cartilaginous mesenchymal condensations as early as E9.5. Runx2 gene activity is later repressed at the onset of cartilage formation, both in vivo and in vitro, necessitating examination of the regulation and function of Runx2 in mesenchymal stem cells. We demonstrate that Runx2 gene activity is repressed by the direct interaction of the homeodomain transcription factor Nkx3.2 with the proximal Runx2 P1 promoter. This repression was found to be required for the progression of BMP-induced chondrogenesis, thereby identifying Runx2 as a modulator of BMP activity in the chondrogenic as well as osteogenic differentiation program. To further understand the regulation of the Runx2 P1 promoter and to determine the contribution of P1-derived gene product, Runx2 Type II, to the formation of mineralized tissue, we have generated a Runx2 Type II-LacZ gene replacement mouse model in which the initial coding sequences and splice donor sites of the Type II isoform are replaced with the LacZ reporter gene. Activity of the endogenous P1 promoter can therefore be monitored by β-galactosidase production. Analysis of Runx2 Type II-LacZ mice demonstrates that the P1 promoter is transcriptionally most active in mature osteoblasts, but its product, Runx2 Type II is dispensable for embryonic skeletal formation. Lastly, we examine the link between growth control and osteogenic differentiation by tissue-specific deletion of the Mdm2 proto-oncogene in developing skeletal tissues of the mouse embryo. Loss of Mdm2 results in impaired bone formation, with skeletal elements exhibiting lower bone mineral content and higher porosity. Ex vivo cultures of calvarial osteoprogenitor cells exhibit severely decreased osteoblastogenesis and bone nodule formation accompanied by a failure to activate Runx2 gene activity. These findings suggest that Mdm2 is required for inhibition of p53 activity that ultimately allows for post-confluent proliferation and induction of Runx2 during maturation of the osteogenic phenotype. Taken together, our findings suggest that Runx2 modulates the commitment of progenitor cells to the osteogenic and chondrogenic lineages, and that Runx2 activity is inextricably linked to mechanisms that control cellular proliferation.

The c-Jun NH2-terminal kinase (JNK) group of kinases include ten members that are created by alternative splicing of transcripts derived from Jnk1, Jnk2 and Jnk3 genes. The JNK1 and JNK2 protein kinases are ubiquitously expressed while JNK3 is expressed in a limited number of tissues. The JNK signaling pathway is implicated in multiple physiological processes including cell transformation. There is growing evidence that JNK signaling is involved in oncogenesis. Nevertheless, the role that JNK plays in malignant transformation is still unclear. The aim of this thesis is to examine the role of JNK in malignant transformation. For this purpose, I used the Bcr/Abl oncogene as a transforming agent. Bcr/Abl is a leukemogenic oncogene that is created by reciprocal translocation between chromosome 9 and 22. The translocation breakpoint is variable and several different Bcr/Abl isoforms have been identified such as Bcr/AblP185 and Bcr/AblP210, whose expression is associated with different types of leukemia. Bcr/Abl activates the JNK signaling pathway in hematopoietic cells and increases AP-1 transcription activity. Furthermore, dominant negative approaches demonstrate that inhibition of c-Jun or JNK prevents Bcr/ Abl-induced cell transformation in vitro. These data implicate the JNK signaling pathway in Bcr/Abl transformation although the role that JNK might have in this process is unclear. Thus, I examined the importance of JNK signaling in Bcr/Abl-induced lymphoid or myeloid transformation. For this purpose I compared Bcr/AblP185- and Bcr/AblP210- induced transformation of wild-type and JNK1-deficient cells using three approaches: in vitro, in vivo and ex vivo. The results obtained with the in vitro approach suggest that both Bcr/AblP185 and Bcr/AblP210 require JNK activity to induce lymphoid transformation. While JNK1-deficiency inhibits Bcr/AblP210 oncogenic potential in lymphoid cells both in vitro and in vivo, pharmacological inhibition of JNK activity (JNK1 and/or JNK2) blocked Bcr/AblP185 induced malignant proliferation in vitro. The differential requirement for JNK observed in the two Bcr/Abl isoforms can be ascribed to the presence in Bcr/AblP210 of the Dbl domain which can activate the JNK pathway in vitro. In the case of Bcr/AblP210, JNK1 is critical for the survival of the ex vivo derived transformed lymphoblasts upon growth factor removal. This result correlates with the fact that mice reconstituted with Bcr/AblP210 transformed Jnk1-l- bone marrow showed normal malignant lymphoid expansion in the bone marrow yet they had reduced numbers of lymphoblast in the bloodstream and lacked peripheral organ infiltration. Thus JNK1 is essential for the survival of the transformed lymphoblast outside the bone marrow microenvironment in Bcr/AblP210induced lymphoid leukemia. Interestingly, while JNK1 is essential for lymphoid transformation, it is dispensable for the proliferation of transformed myeloblasts. Taken together these results indicate that the JNK signaling pathway plays an essential role in the survival of Bcr/AblP210 lymphoblasts and that JNK-deficiency decreases the leukomogenic potential of Bcr/AblP210 in vivo. Thus, cell survival mediated by JNK may contribute to the pathogenesis of proliferative diseases.

The antibody class switch is achieved by DNA recombination between the sequences called switch (S) regions located 5' to immunoglobulin (Ig) heavy chain constant (CH) region genes. This process can be induced in cultured B cells by polyclonal stimulation and switching can be directed to specific antibody classes by certain lymphokines. These stimuli may regulate the accessibility of CH genes and their S regions to a recombinase as indicated by hypomethylation and transcriptional activity. For example, RNAs transcribed from specific unrearranged (germ-line) CH genes are induced prior to switching under conditions that promote subsequent switching to these same CH genes. The function of transcription of these germ-line CH genes is unknown. How stimuli regulate the accessibility of CHgenes is also unclear. I report in this dissertation the structure of the RNA transcribed from the unrearranged Cγ1 gene in mouse spleen cells treated with LPS plus a HeLa cell supernatant containing recombinant interleukin 4 (rIL-4). I will also show that an 150-bp region upstream of the first initiation site of germ-line γ1 RNA contains promoter and enhancer elements responsible for basal level expression and inducibility by phorbol 12-myristate 13-acetate (PMA) and synergy with IL-4 in an IgM+ B cell line, L10A6.2, and an IgG2a+B cell line, A20.3. The germ-line γ1 RNA is initiated at multiple start sites 5' to the tandem repeats of the γ1 switch (Sγ1) region. As is true for analogous RNAs transcribed from other unrearranged genes, the germ-line γ1 RNA has an I exon transcribed from the region 5' to the Sγ1 region.. The Iγ1 exon is spliced at a unique site to the Cγ1 gene. The germ-line γ1 RNA has an open-reading frame (ORF) that potentially encodes a small protein 48 amino acids in length. Elements located within the 150 bp region 5' to the first initiation site of germ-line γ1 RNA are necessary and sufficient to confer inducibility by PMA and synergy with IL-4 to a minimal thymidine kinase (TK) promoter in L10A6.2 cells but are not sufficient to confer this inducibility in A20.3 cells. Linker-scanning mutations demonstrated that these multiple elements function in a mutually dependent manner as indicated by the fact that mutation of any single element will decrease constitutive expression and inducibility by PMA and PMA plus IL-4. This 150-bp region contains several consensus sequences that bind to known or putative transcription factors, including a C/EBP binding site/IL-4 response element (in the promoter for Ia Aαkgene), four CACCC boxes, a PU box, a TGFβ inhibitory element (TIE), an interferon-αβ response element (αβIRE), and an AP-3 site. My results begin to provide a description of the mechanism of regulation of the accessibility of unrearranged germ-line Sγ1-Cγ1 gene. By activating the germ-line γ1 promoter, IL-4 induces transcription of germ-line γ1 RNA, thereby inducing accessibility of the Sγ1-Cγ1 gene. By inhibiting expression of the germ-line γ1 promoter, IFNγ and TGFβ down-regulate transcription of germ-line γ1 RNA, thus reducing the accessibility of the Sγ1-Cγ1 gene. My results also suggest that signaling via the antigen receptor on B cells may be involved in induction of switch to IgG1. Furthermore, this is the first case reported in which multiple functionally interdependent elements are needed to respond to PMA.

The steroid/thyroid hormone receptor superfamily is a large class of ligand-dependent transcription factors that plays a critical role in regulating the expression of genes involved in a broad range of physiological functions, including development, homeostasis, and reproduction. In the absence of cognate hormone, several receptors are able to repress transcription below the basal level via the recruitment of the nuclear receptor corepressors SMRT and NCoR. Upon hormone binding by the receptor, the corepressor complex is dissociated and a coactivator complex is subsequently recruited. This thesis details the mechanisms by which receptor-associated coactivator 3 (RAC3) interacts with nuclear receptors, particularly the vitamin D, estrogen, and retinoid receptors, and modulates their transcriptional activity. It was discovered that these receptors interact with different α-helical LXXLL motifs of RAC3 in vitro. Mutation of specific motifs differentially impairs the ability of RAC3 to enhance transcription by the receptors in vivo. In addition, the intrinsic transcriptional activation function of RAC3 was also characterized. Here, a single LXXLL motif, NR box v, was found to be essential to activation by serving as a binding surface for the general transcriptional integrator CBP/p300. Finally, the cofactor binding pocket of retinoid receptors was characterized. It was demonstrated that, to a large extent, the coactivator pocket of RARα overlaps with the corepressor pocket, with the exception of helix 12, which is required for coactivator, but not corepressor binding. Recruitment of RAC3 or SMRT also correlates directly with the ability of RARα to activate or repress transcription, respectively. Intriguingly, it was discovered that the AF-2 domain of RXRα inhibited cofactor binding to RXRα heterodimers, for deletion of this domain dramatically enhanced RAC3 and SMRT binding. In addition, it was demonstrated that the RXRα cofactor binding pocket contributed minimally to recruitment of cofactors. Conversely, the AF-2 domain of the partnering monomer and its cofactor pocket were required for these interactions. These findings suggest that the partner of RXRα is the primary docking point for cofactors at RXRα heterodimeric complexes. Taken together, this work contributes significantly to the field of nuclear receptor function in detailing the mechanisms by which the coactivator RAC3 is recruited to nuclear receptors and regulates their transcriptional activity.

CD8 T cells recognize complexes of MHC class I and peptide on the surface of target cells. MHC class I antigen presentation is a long pathway, in which proteins are degraded by proteasomes to generating oligopeptides, which may be further trimmed by aminopeptidases in the cytosol. Peptides are transported into the ER, where they may be further trimmed by ER lumenal aminopeptidases and bind to newly-synthesized MHC class I complexes. Proteins degraded by the proteasome are generally tagged with ubiquitin by a combination of ubiquitin-conjugating enzymes and ubiquitin ligases. UBC9 is one ubiquitin conjugating enzyme, which does not conjugate ubiquitin, but instead conjugates small ubiquitin-like molecules (SUMO) to target protein. UBC9 has been found to regulate the functions of many proteins in vivo, most importantly by modifying nuclear transportation and function. Curing [During] my thesis work, I studied the function of UBC9 in MHC class I antigen presentation. UBC9 over-expression in COS cells co-expressing ovalbumin markedly increased presentation SIINFEKL (the immunodominant epitope from ovalbumin in the context of H-2Kb), and UBC9 overexpression increased cell surface H-2Kbin general, suggesting that Ubc9 increased MHC class I antigen presentation by increasing peptide supply. UBC9 did not increase synthesis or degradation of ovalbumin. In transient transfection experiments, Ubc9 increased presentation of SIINFEKL precursors that did, and that did not, depend on proteasomes for processing, as well as SIINFEKL precursors targeted to the ER, bypassing cytosolic processing altogether. However, a C-terminal extended precursor of SIINFEKL, which requires only proteasomal processing before presentation, was the most markedly affected by UBC9 overexpression. This suggested that UBC9 was affecting the pattern of cleavages made by proteasomes in ways that enhance the generation of the C-terminus of SIINFEKL. Because presentation of SIINFEKL itself (which requires no further proteolytic processing) was also enhanced, UBC9 must also affect steps in the class I pathway that occur after the generation of the mature epitopes. UBC9 did not affect the rate of peptide degradation in cytosolic extracts or in intact cells. These findings suggested that UBC9 might have multiple effects on the MHC class I antigen presentation pathway. Immunofluorescent microscopy demonstrated that UBC9 increased the expression of the beta subunits of immunoproteasomes (LMP2, LMP7, and MECL1) as well as of TAP1 and tapasin. In contrast, UBC9 expression did not increase levels of calnexin, calreticulin, ERp57, or Protein disulfide isomerase (PDI). Similarly, levels of leucine aminopeptidase were not increased in UBC9-transfected cells. Therefore, UBC9 overexpression increases the levels of some but not all components of the class I pathway. UBC9 overexpression increased protein levels of MECL1, LMP2 or LMP7 that were under the control of viral promoters, and levels of MECL1 mRNA were similar in control vector and UBC9 transfected cells. Therefore, UBC9 did not increase the level of expression of these subunits through increased transcription. Pulse-chase experiments showed that UBC9 overexpression reduced the degradation of MECL1. Therefore, UBC9 increases the levels of at least some of these components of the MHC class I antigen presentation pathway by increasing their stability. To know the biological significance of UBC9 in MHC class I antigen presentation, I used small interfering RNA (siRNA) to knock down UBC9. Though UBC9 can be successfully knocked down by siRNA, the UBC9-negative cells became very sick, and were not suitable for the study of MHC class I antigen presentation. There are three forms of SUMO molecules in mammalian cells: SUMO-1, SUMO-2 and SUMO-3. My study suggested that SUMO-2 may be involved in UBC9's regulation of MHC class I antigen presentation, since mutant SUMO-2 blocked UBC9's ability to increase H-2Kb-SIINFEKL levels on the cell surface after the cells were loaded with ovalbumin. To further study the function of UBC9, I mutated the active amino acid Cys 93 of UBC9 to Ser (UBC9OH). Unexpectedly, this mutant form (UBC9OH) has very similar effects as wild-type UBC9, increasing Kb-SIINFEKL levels at the cells surface. This suggested that UBC9 protein regulates MHC class I antigen presentation pathway proteins by direct or indirect protein interaction, rather than (or as well as) by SUMO conjugation. Taking account of SUMO-2 results, I propose that wild-type UBC9 (either transfected or endogenous) conjugates SUMO-2 to its substrates, and then UBC9 (wild-type or mutant) interacts with its sumoylated targets, thus affecting protein functions. I also studied heat shock protein Hsp27, which is known to be a substrate for UBC9 in vivo. Hsp27 is expressed in a variety of tissues in the absence of stress, and may regulate actin dynamics. Hsp27 overexpression decreased generation of H-2Kb-SIINFEKL complexes from SIINFEKL precursors that did, and did not, require proteasomes for processing, or that were targeted to the ER. Hsp27 over-expression did not affect protein synthesis, and globally decreased cell surface H2-Kb and H2-Dblevels, but did not affect HLA-A0302 level. Hsp27 overexpression inhibits the presentation of ER-localized SIINFEKL. Taken together, my data suggested that HSP27 may inhibit MHC class I antigen presentation by affecting MHC class I molecules itself rather than peptide supply. After Hsp27 was eliminated with siRNA, the effects were very similar to those seen with Hsp27 overexpression. Levels of H-2Kb-SIINFEKL decreased, and overall cell surface H-2Kb and H-2Db levels decreased. It is possible that when Hsp27 is over-expressed, it acts as a dominant negative form, conferring a similar phenotype to Hsp27 knockdown. These observations suggest that Hsp27 plays an important role in MHC class I antigen presentation.