Transcription initiation is a key step at which gene expression can be regulated. The sigma subunit of RNA polymerase provides the enzyme with the ability to recognize promoter sequences and initiate transcription at specific sites on the chromosome. The molecular basis of sigma function is not well known. It has been suggested that sigma factors confer promoter specificty by making direct contacts to the promoter DNA (Losick and Pero, 1981). To test this idea, suppressors of promoter down mutations were sought that affected the promoter recogniton properties of the σ70 subunit of E. coli RNA polymerase. Four such sigma mutants were obtained, two of which are allele-specific. One of these mutants has a change at a position in the predicted helix-turn-helix DNA binding structure which lies in a conserved region of the protein (region 4). This mutant specifically suppresses promoter down mutations in the -35 region of the promoter. The other mutant has a change at a residue that lies in a predicted α-helix of conserved region 2. This mutant specifically suppresses promoter mutations in the -10 region of the promoter. These data support the idea that regions 2 and 4 of sigma interact with the -10 and -35 regions of the promoter, respectively.

This thesis examines the hypothesis that human erythrocyte net sugar transport is the sum of two serial processes: sugar translocation followed by interaction of newly imported sugar with an intracellular binding complex from which sugar dissociates into the bulk cytosol. This hypothesis suggests that steady-state transport measurements in the human erythrocyte do not accurately reflect the intrinsic catalytic features of the glucose transporter and unless correctly interpreted, may lead to apparent inconsistencies in the operational behavior of the human erythrocyte sugar transport system. Our results support this proposal by demonstrating that although sugar transport measurements in human red blood cells suggest that transport is catalytically asymmetric, ligand binding measurements indicate that transport must be symmetric. In order to examine the serial compartments hypothesis, we set out to determine the following: 1) identify the component(s) of the proposed sugar binding complex, 2) determine whether cytosolic ATP levels and transporter quaternary structure affect sugar binding to the sugar binding complex, and 3) determine whether the sugar binding site(s) are located within or outside the cell. We present findings which support the hypothesis that the sugar binding complex is in fact the sugar transport protein, GLUT1. The number of sugar binding sites and the release of sugar from the GLUT1 complex are regulated by ATP and by GLUT1 quaternary structure. The sugar binding sites are located on a cytoplasmic domain of the GLUT1 complex. We show how these observations can account for the apparent complexity of erythrocyte sugar transport and its regulation by ATP.

A two-element protocol consisting of one donor-specific transfusion (DST) plus a brief course of anti-CD154 mAb greatly prolongs the survival of murine islet, skin, and cardiac allografts. To study the mechanisms involved in the induction of allograft survival, we determined the fate of tracer populations of alloreactive T cell receptor (TcR) transgenic CD8+ T cells circulating in a normal microenvironment. In the first portion of this thesis, we observed that DST plus anti-CD154 mAb prolonged allograft survival and deleted alloreactive TcR transgenic CD8+ T cells. Neither component alone did so. Skin allograft survival was also prolonged in normal recipients treated with anti-CD154 mAb plus a depleting anti-CD8 mAb and in C57BL/6-CD8 knockout mice treated with anti-CD154 mAb monotherapy. We conclude that, in the presence of anti-CD154 mAb, DST leads to an allotolerant state in part by deleting alloreactive CD8+ T cells. Consistent with this conclusion, blockade of CTLA4 and B7-l/2 by CTLA4-Ig, which is known to abrogate the effects of DST and anti-CD154 mAb, prevented the deletion of alloreactive TcR transgenic CD8+ T cells. Also in support of our hypothesis, depletion of CD4+ T cells, which is known to abrogate the effects of DST and anti-CD154 mAb, prevented the deletion of alloreactive TcR transgenic CD8+ T cells. We continued by examining the effects of IFN-γ, IL-10 and IL-4. None of these cytokines had any significant effect on the deletion of alloreactive TcR transgenic CD8+ T cells induced by co-stimulation blockade. The last part of this thesis studied the behavior of alloreactive TcR transgenic CD8+ T cells during the maintenance phase of allograft survival induced by our two-element protocol. Using a hematopoietic TcR transgenic chimera system, our results demonstrated that levels of alloreactive CD8+ T cells remained low throughout the maintenance phase. These results document for the first time that peripheral deletion of alloantigen-specific CD8+ T cells is an important mechanism through which allograft survival can be prolonged by co-stimulatory blockade. We propose a unifying mechanism to explain allograft prolongation by DST and blockade by co-stimulation blockade.

In this dissertation I describe the isolation and characterization of genomic clones for the human synapsin I gene, the establishment of a linkage map for the human synapsin I gene locus, and studies of the possible involvement of this gene in neurological disease. Synapsin I is a neuron-specific phosphoprotein which is concentrated at the presynaptic terminal. Evidence suggests that it plays a fundamental role in the regulation of neurotransmitter release. Altogether 27,500 bp of the human synapsin I gene have been isolated, and the gene structure has been partially determined. DNA sequence comparisons between human and rat genes show a high degree of conservation. Sequenced exons display an 87% identity to each other. The synapsin I genomic clones were employed in the search for a polymorphic marker. A compound (AC)n repeat located 1000 base pairs downstream from the human synapsin I gene and within the last intron of the A-raf-1 gene has been identified. DNA database comparisons of the sequences surrounding the repeat indicate that the synapsin I gene and the A-raf-1 gene lie immediately adjacent to each other, in opposite orientation. Polymerase chain reaction amplification of this synapsin I / A-raf-1 associated repeat using total genomic DNA from members of the 40 reference pedigree families of the Centre d'Etude du Polymorphisme Humaine showed it to be highly polymorphic, with a polymorphic information content value of 0.84 and a minimum of eight alleles. Because the synapsin I gene had been mapped previously to the short arm of the human X chromosome at Xp11.2, linkage analysis was performed with markers on the proximal short arm of the X chromosome. The most likely gene order is: DXS7 - SYN/ARAF1 - TIMP - DXS255 - DXS146 with a relative probability of 5 x 108 compared with the next most likely order. The SynI/Araf marker was next utilized in a linkage study aimed at establishing a more accurate placement of the genetic locus responsible for the ocular disorder Congenital stationary night blindness, which had been mapped previously close to DXS7. Our results confirm this prior localization and also exclude any placement proximal to the SYN/ARAF1 locus. Finally, the inheritance of the different alleles of the SynI/Araf marker in three families with Rett syndrome, a severe neurodegenerative disorder, which has been assigned to the X chromosome, was studied. In at least one of the families in which two half sisters with the same mother suffer from the disease, the inheritance of Rett syndrome was discordant with the inheritance of the same allele for the SynI/Araf marker. Thus, this highly informative repeat has proven already effective in the study of X-linked diseases and should serve as a valuable marker for disease loci mapped to the Xp11 region.

The overall goal of this thesis is to examine how the T cell receptor (TCR) repertoire evolves during and after viral infections. Previous studies had examined TCR usage of selected virus-specific T cell clones, but little was known about how a diverse T cell repertoire changes during the transition between an acute infection and a memory response. It was also unclear how the T cell repertoire evolves under conditions of persistent infections associated with clonal exhaustion. To address these issues I used as a model system the lymphocytic choriomeningitis virus (LCMV) infection of mice, for which the T cell response is well-characterized. LCMV, strain Armstrong (LCMV-ARM), infection induces a strong CD8+ T cell response, which clears the virus and converts to a memory response. In contrast, high doses of LCMV clone 13 leads to persistent infections associated with T cell clonal exhaustion. These two extremes of T cell responses enable one to compare the evolution of the TCR repertoire under conditions where an acute T cell response converts to a memory response with that of T cell clonal exhaustion. In this thesis I analyzed the TCR repertoire usage directly ex vivo by the technique of CDR3 length spectratyping throughout the acute LCMV infection, into memory, after modulation by subsequent heterologous and homologous viral infections, and under conditions of T cell clonal exhaustion. Kinetic studies on the frequencies of precursor cytotoxic T lymphocytes (PCTL) to the three LCMV immunodominant peptides had suggested that the virus-specific T cell repertoire becomes fixed by day 7 postinfection, when the virus is cleared. The pCTL data also showed that a high frequency of the LCMV-specific memory T cells remained stable throughout the lifetime of the mouse. To examine any changes of the TCR repertoire usage that may develop during the acute LCMV infection and into memory, the Vβ8 population was subjected to spectratype analyses, because Vβ8 represented a substantial amount of the LCMV-induced CD8+ T cells recognizing several LCMV-encoded peptides. Analyses of the Vβ8.1 spectratype showed that genetically identical mice generated remarkably different T cell responses, as reflected by different spectratypes and different TCR sequences in same-sized spectratype bands; a conserved CDR3 motif was, however, found within some same-sized bands. This indicated that meaningful studies on the evolution of the T cell repertoire required longitudinal studies within individual mice instead of comparisons between mice. Such longitudinal studies with peripheral blood (PB) samples showed that the virus-induced T cell repertoire changed little after viral clearance and during the silencing phase of the T cell response and that dominant spectratype peaks were preserved into long term memory. To determine the effect of secondary LCMV infection on the spectratype, the recalled LCMV-induced spectratypes were analyzed. Most of the dominant peaks detected in the primary infections remained present in the secondary infection. Some new peaks were also detected for the first time in the secondary infection, suggesting a further selection of the virus-induced T cell repertoire. The spectratype data support the concepts that the LCMV-induced T cell repertoire remains unchanged during the silencing phase after clearance of the virus and that the LCMV infection dramatically skews the host T cell repertoire in the memory state long after the virus is cleared. Studies had shown that high doses of LCMV clone 13 induce a transient anti-viral CTL response followed by clonal exhaustion of T cells. To determine how the TCR repertoire evolves under conditions of persistent infections associated with T cell clonal exhaustion, the Vβ8.1 spectratypes were analyzed at various time points after the infection. In contrast to the stable LCMV-induced spectratype after viral clearance, continuous selection of the T cell repertoire occurred under conditions of persistent infections, as the T cell clones appeared and disappeared at different rates. The T cell repertoire ultimately returned to a Gaussian distribution under conditions of clonal exhaustion, indicating that clonal deletion occurs in the great majority of the virus-induced T cells. To test the stability of the LCMV-induced TCR repertoire under conditions of subsequent heterologous viral infections, the recalled LCMV-induced spectratypes were examined in the presence or absence of intervening heterologous viruses. The results showed that the intervening heterologous viruses disrupted the recalled Vβ8.1-Jβ1.3 spectratype on secondary LCMV infection; this otherwise remained stable in the absence of intervening heterologous viruses. This result supports the hypothesis that subsequent heterologous viral infections disrupt the stable LCMV-induced T cell repertoire. To detennine whether a subset of the memory T cells was deleted by the IFN-induced apoptosis of memory T cells, the LCMV-immune spectratypes were analyzed before and after the injection of the IFN inducer, poly I:C. The LCMV-immune spectratypes remained relatively stable after poly I:C injection, suggesting that there is no selective protection or deletion of discrete memory T cell clones during the IFN-induced apoptosis. In summary, the data in this thesis show that (i) the virus-induced T cell repertoire changes little after viral clearance and during the silencing phase of the T cell response, (ii) the LCMV infection dramatically skews the host T cell repertoire in the memory state, (iii) the evolution of the T cell repertoire occurs during secondary infections and under conditions of clonal exhaustion associated with persistent infections, (iv) genetically identical hosts generate different T cell responses to the same virus, and (v) intervening heterologous viral infections disrupt the recalled LCMV-induced T cell repertoire, but the LCMV-immune repertoire remained relatively stable upon the treatment of the IFN inducer, poly I:C.

Insertion of transposable elements within an operon has been known not only to abolish expression of the gene interrupted by the insertion, but also to exert a strong polar effect on the expression of downstream genes in the same operon. In this dissertation, I have shown that insertions of the transposable ampicillin-resistance element Tn1, either in the polar or nonpolar orientation, in the 3' untranslated region of the bacteriophage P22 antirepressor (ant) operon reduce the rate of upstream ant gene expression; insertions of Tn1 in the nonpolar orientation reduce the rate of ant gene expression more significantly than those in the polar orientation. This effect appears to be due to reduced stability of ant mRNA. Tn1 deletion mutants of one of the nonpolar Tn1 insertion mutations have been isolated. Two classes of Tn1 deletions are obtained. Class I retains a 68 bp Tn1 sequence that shows a potential 14 bp stem and 37 bp loop conformation, while class II retains 147 bp Tn1 sequence that shows a potential 69 bp stem and 6 bp loop conformation. These two classes of Tn1 deletions do not delete any P22 sequences. Class I but not class II Tn1 deletion mutants restore the rate of ant gene expression and ant mRNA stability. Six different Ant+ revertants of the class II Tn1 deletion mutant simultaneously restore the rate of ant gene expression and ant mRNA stability. They all have deletions that remove all or part of the class II Tn1 sequence. In one case, the Tn1 sequence retained shows a potential 15 bp stem and 8 bp loop conformation, in the other cases, no secondary structure is predicted to form. The results of the Tn1 deletion mutants suggest that the stem-and-loop structures and the length of stems potentially formed by the Tn1 sequences in mRNA may affect its stability.

Proteins, which are T cell-dependent antigens, preferentially induce antibodies of the IgG1 class in mouse, whereas LPS, which is a T-independent antigen, preferentially induces IgG3 and IgG2b. Interaction between CD40 on B cells and CD40 ligand (CD40L) on T cells has been shown to mediate T cell contact help for B cell proliferation, differentiation and immunoglobulin isotype switching. In addition, it has been shown that membranes from activated T cells induce germline γ1 transcripts, and that CD40 signaling induces germline γ1 transcripts. These results indicate that T cell contact help mediated by CD40 ligand (CD40L)-CD40 interaction may contribute to this preferential IgG1 isotype selection in response to T-dependent antigens by inducing transcription of germline Ig γ1 transcripts. Here we show that signaling via CD40 increases expression of a transiently transfected luciferase reporter plasmid driven by the germline γ1 promoter in M12.4.1 B lymphoma cells. By linker scanning mutation analysis of the promoter, we have identified a CD40 responsive region (CD40RR) which is able to confer inducibility by CD40L to a minimal c-fos promoter. The CD40RR contains three NF-кB-binding sites, each of which is required for maximal induction of the γ1 promoter activity by CD40L. Binding of the NF-кB/Re1 proteins p50, Re1A, c-Re1 and Re1B to the CD40RR can be induced by CD40 signaling in M12.4.1 cells or in splenic B cells. Co-transfection of expression plasmids for p50 together with Re1A or Re1B, but not p50 alone or p50 and c-Re1, transactivates the CD40RR in transient transfection assays in M12.4.1 cells. These data demonstrate NFкB/Re1 proteins activated by CD40 engagement play an important role in regulation of the germline γ1 promoter. Further support for this conclusion is provided by the finding that treatment of splenic B cells with NF-кB inhibitors prevents induction of germline γ1 transcripts by CD40L. Although LPS also induces NF-кB activation, it poorly induces germline γ1 promoter activity in M12.4.1 cells and it also poorly induces germline γ1 transcripts in splenic B cells and in the mouse B cell line, 1B4.B6. Western blot analyses show that LPS predominantly activates p50 and c-Re1, whereas CD40L induces all NF-кB/Re1 proteins (Re1A, Re1B, c-Re1 and p50). Likewise, in nuclear extracts from LPS-treated cells, p50/cRe1 and p50/p50 dimers are the major NF-кB/Re1 proteins which bind to the promoter for germline γ1 transcripts in electrophoretic mobility shift assays, whereas in nuclear extracts from CD40L-treated cells, p50/Re1A and p50/Re1B dimers are the major complexes. Reporter gene assay by over expressing NF-кB/Re1 fusion proteins indicates that p50/Re1A and p50/Re1B dimers, but not p50/c-Re1 or p50/p50 dimer, can transactivate the germline γ1 promoter. Despite their inability to activate the promoter, p50/c-Re1 and p50/p50 can bind to the promoter and suppress the transactivation activity of p50/Re1A and p50/Re1B. Therefore, the effect of NF-кB activation on the germline γ1 promoter depends on the Relative amounts of transactivating and non-trans activating NF-кB/Re1 dimers. The inability of LPS to induce germline γ1 transcripts can be explained by induction of non-transactivating NF-кB/Re1 dimers and the ability of CD40L to activate the promoter by a greater induction of Re1A and Re1B Re1ative to c-Re1.

Modulation of Ca2+ channels is an important mechanism for regulation of synaptic strength. However, it is clear that some Ca2+ current types are insensitive to inhibitory modulation mediated by heterotrimeric G proteins (G protein inhibition), and among currents which are sensitive to G protein inhibition, there is great variation in the magnitude of Ca2+ current inhibition between cells of different origin. For the experiments in this dissertation, I utilized recently cloned Ca2+ channels to determine the minimal combination of Ca2+ channel subunits which would confer G protein sensitivity to the recombinant channels. In addition, I examined the role Ca2+ channel auxiliary subunits play in regulation of Ca2+ channel sensitivity to inhibitory G proteins, and whether channels which were sensitive to G protein inhibition were regulated equivalently by the auxiliary subunits. Finally, I investigated possible mechanisms by which these auxiliary subunits modulate G protein-mediated inhibition of Ca2+ channels. I found that α1A and α1B Ca2+ currents, when expressed in Xenopus oocytes, were sensitive to modulation by G proteins in the absence of any Ca2+ channel auxiliary subunits, while α1C currents were not modulated under the same conditions. I conclude from this data that Ca2+ channel α1 subunits are differentially sensitive to G protein modulation, and the α1 subunit of the class A and B Ca2+ channels is sufficient for G protein modulation. I also tested the ability of Ca2+ channel auxiliary subunits to modulate the magnitude of G protein-mediated inhibition Ca2+ currents. I found that the Ca2+ channel α2 subunit had no effect on the magnitude of G protein inhibition of α1A and α1B currents. However, the Ca2+ channel β3 subunit eliminated tonic G protein inhibition and sharply reduced the magnitude of muscarinic M2 receptor induced G protein inhibition of both α1A and α1B currents. I found, however, that while the magnitude of α1A and α1B current inhibition was equivalent in the absence of auxiliary subunits, the magnitude of inhibition was greater for the α1B channel after co-expression of the Ca2+ channel β3 subunit. These results indicate that the Ca2+ channel β3 subunit reduces the sensitivity of α1A and α1B Ca2+ channels to voltage-dependent G protein modulation, and does so to a greater extent for α1A channels when compared to α1B Ca2+ channels. I found that M2 receptor induced inhibition of α1B currents was more voltage-dependent after expression of the Ca2+ channel β3 subunit. Additionally, the rate relief of G protein inhibition dramatically increased after co-expression of the Ca2+ channel β3 subunit. I also co-expressed G protein subunits, and determined that inhibition of both α1B and α1Bβ3 currents was mediated by the G protein βγ subunit. Furthermore, the rate of voltage-dependent relief of G protein βγ subunit induced inhibition increased after co-expression of the Ca2+ channel β3 subunit, similar to the increased rate of relief of the M2 receptor induced G protein inhibition. These data, along with data which demonstrates that G protein inhibition results from the binding of the G protein βγ subunit to the Ca2+ channel (De Waard et al., 1997), indicate that the Ca2+ channel β3 subunit subunit reduces the magnitude of G protein inhibition of α1B Ca2+ currents by increasing the rate of dissociation of the G protein βγ subunit, such that moderate depolarizations used to activate the channel also relieve a large portion of the G protein inhibition.

Hox as well as Meis proteins are known to bind DNA as heterodimers with members of the Pbx family, and it is believed that such complexes mediate the in vivo functions of Hox and Meis. To begin exploring the role of hoxb1b and meis3 in vertebrate development, we isolated and characterized a zebrafish pbx cDNA which encodes a novel member of the pbx family, which we called pbx4. In situ analysis revealed that pbx4 RNA is maternally deposited and is detected throughout the zebrafish embryo during blastula stages. It becomes excluded from ventroanterior structures at late gastrula stages and is detected within the developing central nervous system during segmentation stages. pbx4 expression overlaps with that of hoxb1b and meis3, in the region of the presumptive caudal hindbrain during gastrula stages. In vitro binding experiments revealed that Pbx4/Meis3 and Pbx4/Hoxb1b, as well as a novel trimeric complex containing Pbx4, Meis3 and Hoxb1b form in vitro. Thus, protein complexes of different combinations of Pbx4, Meis3 and Hoxb1b form in vitro and importantly pbx4, meis3 and hoxb1b are coexpressed in a domain at the level of the presumptive caudal hindbrain during zebrafish gastrula stages. These findings raised the possibility that similar complexes may exist in vivo and may be involved in the specification of distinct developmental fates. To address this possibility we overexpressed meis3,pbx4 and hoxb1b in zebrafish embryos and we tested for the effect on endogenous gene expression, morphology and neuronal specification. Our results demonstrate that Hoxb1b/Pbx4/Meis3-containing complexes induce extensive expression of several hindbrain genes (hoxb1a, hoxb2, krox20 and valentino) anterior to their normal expression domains, and mediate the transformation of anterior (forebrain and midbrain) fates to posterior (hindbrain) ones, including the formation of excess ectopic Mauthner neurons. Ectopic expression of Hoxb1b/Pbx4/Meis3-containing complexes also leads to truncation of the embryonic axis anteriorly. In contrast, Hoxb1b/Pbx4 expression induces ectopic expression of only hoxb1a (primarily in r2), but does not mediate axial truncations, and Hoxb1b (or its mouse homolog, HoxA1) has been reported to induce an ectopic pair of Mauthner neurons in r2 (Alexandre et al., 1996). Thus, binding of Meis3 to Hoxb1b/Pbx4 generates Hoxb1b/Pbx4/Meis3-containing complexes that have qualitatively (e.g. induction of hoxb2 expression) and quantitatively (e.g. larger number of ectopic Mauthner neurons) different effects than Hoxb1b/Pbx4-containing complexes. These results suggest that Meis3/Pbx4/Hoxb1b-containing complexes may be responsible for specification of hindbrain fates in vivo. In addition to meis3, three other members of the meis/prep gene family are expressed during early embryogenesis in zebrafish. Analysis of gene expression patterns revealed both common as well as unique spatial and temporal expression patterns for each of these genes. This finding raises the question of whether all family members are functionally similar to meis3 or meis3 performs unique functions. To address this question we overexpressed meis1.1, meis2.2 and prep1 in zebrafish embryos and we asked whether they are able to induce hindbrain fates like meis3 does. Overexpression of any Meis protein, or Prep, along with Pbx4 and Hoxb1b resulted in embryos that were truncated anteriorly and exhibited massive ectopic hoxb1a and hoxb2 expression anterior to their normal expression domains. Furthermore, in vitro analysis demonstrated that they are all able to form dimers with Pbx4 in vitro. In addition, analysis of their subcellular localization defined Pbx4 interaction as a prerequisite for nuclear localization of all Meis and Prep proteins. Thus, at least in the overexpression assay there are no functional differences among meis/prep genes. These results raise the question of what is exactly the function of Meis/Prep proteins. Is binding to Pbx proteins and to DNA their only function, or do they have additional roles? To address this question we performed a deletional analysis of Meis3 protein and we tested the requirement of each domain in the overexpression assay. Our experiments revealed that the domain N-terminal to the Pbx-Interaction-Domain (PID) as well as the domain C-terminal to the Homeodomain are not required for the function of Meis3, at least in the overexpression assay. Furthermore, the homeodomain and the domain between the PID and the homeodomain are not required. From our previous analysis (Vlachakis et al., 2001) we know that the PID is required for the Meis3/Pbx4/Hoxb1b synergistic induction of hindbrain fates. Our deletion analysis extended this fmding showing that the PID is also sufficient to provide the Meis3 function in vivo, at least in our overexpression assay. Furthermore, a mutant PID that does not bind Pbx, when fused to the Pbx4 homeodomain induced hindbrain fates upon overexpression along with Hoxb1b. This finding suggests that the PID (motifs M1, M2 and the domain in between them, ID) besides binding to Pbx may also bind another protein that is required for the Meis3/Pbx4/Hoxb1b synergistic induction of hindbrain fates. Taking all our results together, we propose the following roles for Meis proteins in the transcriptional activation complexes. First, they are involved in the nuclear localization of Pbx4. Second, they bind to DNA as heterodimers with Pbx4 facilitating binding of Hoxb1b to Pbx4, which occurs only on DNA. In doing so, they provide the specificity for DNA binding since the Meis3/Pbx4 dimer first recognizes the "hox response element" and then the Hox protein is recruited. Third, they stabilize the binding of Hoxb1/Pbx4 complex on DNA. Fourth, they are responsible for recruiting additional factors to DNA, necessary for activation of target genes. The complicate and dynamic spatial and temporal expression patterns of the meis/prep genes, suggest that they are involved in many different processes during embryogenesis, as well as in the adult organism. We believe that one or more members of the Meis/Prep family execute some of the functions listed above at different times and places during development, although all member are probably capable of executing all these functions.

Type II 5'-deiodinase (D2) catalyzes the conversion of T4 to the transcriptionally active T3. When T4 levels are high, D2 activity levels are low. Conversely when T4 levels are low, D2 catalytic activity is high. Immunocytochemistry and biochemical data from cultured rat astrocytes revealed that physiological concentration of T4 and the non-transcriptionally active metabolite rT3, but not T3, initiates the budding of D2 containing endosomes and their subsequent translocation to the perinuclear space. Further analysis showed that this process required a polymerized actin cytoskeleton but not cellular transcription or translation; however the precise mechanism remained unknown. In this present investigation, we characterized the requirement of an unconventional myosin motor protein, myosin 5a, in the actin-based endocytosis of D2 containing vesicles. We developed an in vitro actin binding assay that exploited the T4 dependent binding of D2 containing vesicles to F-actin, and showed that D2p29:F-actin interactions are calcium, magnesium and ATP-dependent suggesting that a calmodulin (CaM) regulated myosin ATPase is required. Introduction of in vitro transcribed and translated vesicle-binding tail, which lacked the actin binding head, of myosin 5a to the in vitro actin binding assay created a dominant negative inhibitor of D2 binding to the actin cytoskeleton by competing with the native myosin 5a. A replication deficient adenoviral vector expressing the fusion protein of the 29 kDa substrate binding subunit of D2 with a green fluorescent protein reporter molecule enabled us to directly examine T4 dependent regulation of D2 in vitro as well as in living cells. Using immunoprecipitation we showed a T4 dependent association between the vesicle binding tail of myosin 5a and D2 containing vesicles. Biochemical analysis of the interaction of the myosin 5a tail with D2 containing vesicles revealed that the last 21 amino acids of myosin 5a were both necessary and sufficient for the attachment of D2 containing vesicles to the F-actin cytoskeleton. Using rapid acquisition time-lapse digital microscopy in p29GFP expressing rat astrocytes, we showed directed T4 dependent p29GFP movement from the plasma membrane to the perinuclear region. This hormone dependent vesicle movement was not observed in cells treated with T3 or no hormone. Time lapse motion studies allowed for the calculation of the velocity and of the distance traveled for individual fusion protein containing vesicles. The velocity for cells treated with T4 or rT3 was identical to that reported for vesicle-laden myosin 5a in mouse melanophores. In contrast cells treated with T3 or those receiving no hormone treatment had velocities similar to diffusion of proteins within the plasma membrane. Astrocytes constitutively expressing both p29GFP and dominant negative myosin 5a inhibitors failed to show hormone induced centripetal movement. These data demonstrate that myosin 5a is the molecular motor responsible for thyroid hormone dependent actin based endocytosis in astrocytes.