Comprehensive Summaries of Uppsala Dissertations from the Faculty of Medicine 1117
_____________________________ _____________________________
The Role of Stat1 in Retinoic Acid-Induced Myelomonocytic
Differentiation of Human Leukemia Cells
BY
ANNA DIMBERG
ACTA UNIVERSITATIS UPSALIENSIS
Dissertation for the Degree of Doctor of Philosophy (Faculty of Medicine) in Pathology presented at Uppsala University in 2002
Abstract
Dimberg, A. 2002. The role of Stat1 in retinoic acid-induced myelomonocytic differentiation of human leukemia cells. Acta Universitatis Upsaliensis. Comprehensive Summaries of Uppsala Dissertations from the Faculty of Medicine 1117. 55 pp. Uppsala. ISBN 91-554-5224-8
All-trans retinoic acid (ATRA), a biologically active metabolite of vitamin A, is a powerful inducer of terminal differentiation and growth arrest of several myeloid cell lines in vitro.
Although the efficacy of ATRA as an anti-cancer drug has been demonstrated by the successful treatment of acute promyelocytic leukemia (APL), knowledge concerning the molecular mechanisms directing ATRA-induced differentiation and cell cycle arrest of myeloid cells is lacking. Our results show, for the first time, that the complex regulation of cell cycle proteins and myeloid-specific transcription factors induced by ATRA relies on functional Stat1. We found that Stat1 is activated by both tyrosine-701 and serine-727 phosphorylation upon ATRA-induced differentiation of the human monoblastic cell line U-937. Expression of phosphorylation deficient mutants of Stat1 (Stat1Y701F or Stat1S727A) inhibited both ATRA-induced differentiation and cell cycle arrest of U-937 cells, pointing to a requirement of active Stat1 in these processes.
Detailed analysis of the molecular mechanism of ATRA-induced cell cycle arrest and differentiation showed that the onset of cell cycle arrest was associated with a decrease in c-Myc and cyclin E levels and upregulation of p27Kip1 and p21WAF1/CIP1. This was followed by a rapid fall in cyclin A and B and a coordinate dephosphorylation of the retinoblastoma protein (pRb).
The inhibition of ATRA-induced cell-cycle arrest by constitutive expression of Stat1Y701F or Stat1S727A was associated with impaired regulation of these cyclins and p27Kip1, positioning Stat1 activation upstream of these events. To further understand the process of ATRA-induced differentiation, the regulation of myeloid-specific transcription factors was investigated during ATRA-treatment. Notably, ATRA-induced upregulation of Stat2, ICSBP and C/EBP-ε was selectively impaired in sublines expressing Stat1Y701F or Stat1S727A, suggesting an important function of these factors downstream Stat1. Taken together, the work in this thesis clearly demonstrates that Stat1 plays a key role in ATRA-induced terminal differentiation of myeloid cells, through regulation of cell-cycle proteins and myeloid-specific transcription factors.
Key words: Stat1, ATRA, U-937, myeloid, differentiation, cell cycle, growth arrest.
Anna Dimberg, Department of Genetics and Pathology, Rudbeck Laboratory, Uppsala University, SE-751 85 Uppsala, Sweden
© Anna Dimberg 2002 ISSN 0282-7476 ISBN 91-554-5224-8
Printed in Sweden by Eklundhofs Grafiska AB, Uppsala 2002
To my family
This thesis is based on the following papers, which are referred to in the text by their roman numerals:
I. Dimberg, A., Nilsson, K. and Öberg, F. Phosphorylation-deficient Stat1 inhibits retinoic acid-induced differentiation and cell cycle arrest in U-937 monoblasts. Blood. 2000;96:2870-2878.
Copyright American Society of Hematology
II. Dimberg, A., Bahram, F., Karlberg, I., Larsson, L-G., Nilsson, K. and Öberg, F. Retinoic acid-induced cell cycle arrest of human myeloid cell lines is associated with sequential down-regulation of c-Myc and cyclin E and post- transcriptional upregulation of p27Kip1. Blood. 2002 (In press).
Copyright American Society of Hematology
III. Dimberg, A., Karlberg, I., Nilsson, K. and Öberg F. Serine-727 phosphorylated Stat1 is required for ATRA-induced G0/G1 arrest of U-937 cells and the associated regulation of cyclins and p27Kip1. Submitted.
IV. Dimberg, A., Nilsson, K and Öberg F. Inhibition of retinoic acid-induced differentiation of human monoblastic U-937 cells by phosphorylation deficient Stat1 is associated with impaired expression of Stat2, ICSBP and C/EBP-ε.
Manuscript.
TABLE OF CONTENTS _____________________________________________
Abbreviations...6
BACKGROUND Introduction ...7
Hematopoiesis ...8
Regulation of hematopoiesis by the cytokine network Transcription factors involved in hematopoiesis Myeloid differentiation Mammalian cell cycle regulation...13
Retinoids and all-trans retinoic acid (ATRA) ...14
The role of ATRA in myeloid differentiation and leukemia Signaling pathways involved in myeloid differentiation...16
The retinoid receptors Signaling through STAT family proteins Biological function of STATs ...20
STATs and oncogenesis STAT function in myeloid differentiation Activation and gene regulatory function of Stat1 ...21
Stat1 in interferon signaling Stat1 interacting proteins Complex roles of Stat1 in gene regulation Stat1 involvement in growth arrest In vitro models of myeloid differentiation ...25
PRESENT INVESTIGATION Aims of the present investigation...27
Results and Discussion...28
Phosphorylation-deficient Stat1 inhibits retinoic acid-induced differentiation and cell cycle arrest in U-937 monoblasts(Paper I) ...28
Retinoic acid-induced cell cycle arrest of human myeloid cell lines is associated with sequential down-regulation of c-Myc and cyclin E and post-transcriptional upregulation of p27Kip1 (Paper II) ...30
Serine-727 phosphorylated Stat1 is required for ATRA-induced G0/G1 arrest of U-937 cells and the associated regulation of cyclins and p27Kip1 (Paper III) ...32
Inhibition of retinoic acid-induced differentiation of human monoblastic U-937 cells by phosphorylation deficient Stat1 is associated with impaired expression of Stat2, ICSBP and C/EBP-ε (Paper IV) ...34
General Discussion... 36
Conclusions...40
Acknowledgements ...41
References...43
Abbreviations
HSC hematopoietic stem cell CLP common lymphoid progenitor CMP common myeloid progenitor
RA retinoic acid
ATRA all-trans retinoic acid
VitD3 vitamin D3 (1α,25-Dihydroxycholecalciferol) STAT signal transducer and activator of transcription
JAK Janus kinase
IRF interferon regulatory factor C/EBP CCAAT/enhancer binding protein HDAC histone deacetylase
CBP CREB binding protein RAR retinoic acid receptor RXR retinoid X receptor
RARE retinoic acid response element GAS γ-interferon activation site SH2 Src homology region 2
IFN interferon
IL interleukin
M-CSF macrophage-colony stimulating factor G-CSF granulocyte-colony stimulating factor
G-CSFR granulocyte-colony stimulating factor receptor CD cluster of differentiation
CDK cyclin dependent kinase
CKI cyclin dependent kinase inhibitor pRb retinoblastoma protein
APL acute promyelocytic leukemia AML acute myeloid leukemia
Introduction
A multi-cellular organism is commonly composed of a multitude of specialized cells, together constituting the various tissues and organs necessary for all the functions of the body. Both development and homeostasis of this system depend on intrinsic regulatory networks, directing cell growth, differentiation, survival and death. These networks are composed of signals, including soluble molecules and cell-cell contacts, receptors, either membrane bound or intracellular, and signal transducers, for example kinases, phosphatases and transcription factors. A cell can only respond to a stimulus if it expresses the matching receptors and signal transducers. At any given moment, a cell may come into contact with a wide variety of signals, activating numerous intertwined signaling pathways. The total of all signals are integrated in the nucleus of the target cell, activating transcription of genes necessary for modulating cell function.
In most cases, the degree of differentiation negatively correlates to the proliferative potential of a cell. A striking example of the proliferative potential of immature cells is the capacity of hematopoietic stem cells to reconstitute the entire immune system of irradiated mice. In contrast, most terminally differentiated cells, including neurons and plasma cells, are irreversibly growth arrested. This tight coupling of growth and differentiation in normal cells implicates that disruption of pathways affecting differentiation, as well as dysregulation of molecules positively or negatively affecting growth, so called proto-oncogenes and tumor suppressors, can trigger malignant transformation of cells.
Hematopoietic malignancies are often the result of deregulated proliferation of immature blasts, failing to differentiate into mature blood cells. Notably, in acute promyelocytic leukemia (APL), the block in differentiation can be removed by pharmacological concentrations of the vitamin A derivative all-trans retinoic acid (ATRA), resulting in complete remission through differentiation of the leukemic blasts.
Potentially important for the treatment of other types of leukemia, ATRA can induce terminal differentiation and growth arrest of several myeloid cell lines, as well as freshly isolated acute myeloid leukemia (AML) cells, in vitro. Although much research interest has been focused on the biological role of ATRA induced differentiation of myeloid cells, little is known about the signaling networks mediating this process. This thesis is focused on the molecular mechanisms of ATRA-induced differentiation and growth arrest of human myelomonocytic tumor cell lines, and specifically on the role of signal transducer and activator of transcription(STAT)1 in this process.
Hematopoiesis
The hematopoietic system is composed of a vast variety of blood cells with different morphology and functions (Figure 1). These functions include oxygenating the body, blood clotting in response to injury and immune defense. The majority of blood cells are short lived, which necessitates the continuous formation of new cells to maintain homeostasis. It has been estimated that approximately 1012 new blood cells are formed each day from immature progenitors in the bone marrow(Ogawa, 1993). This remarkable turnover of cells requires strict regulatory mechanisms to maintain the correct number and proportion of all different blood cells in the body.
HSC CLP
CMP
T lymphocyte
B lymphocyte
Macrophage
Neutrophil Eosinophil Basophil
Mast cell Erythrocyte
Platelets Monocyte
Megakaryocyte
Figure 1. Hematopoiesis.
All blood cells originate from a common pluripotent progenitor, the hematopoietic stem cell (HSC) (Morrison et al., 1995). HCSs have the capacity to self renew and differentiate into all blood cell lineages(Weissman, 2000). Self renewal can either be achieved through asymmetric cell divisions, where the HSC divides to give rise to one stem cell and one commited daughter cell, or through sequential divisions alternating generation of stem cells and commited progenitors(Knoblich, 2001) (Punzel and Ho, 2001). The HSC can give rise to either the common lymphoid progenitor (CLP) or the common myeloid progenitor (CMP) (Gunsilius et al., 2001). Myeloid progenitors are
commited to differentiate into cells of the macrophage, granulocyte, eosinophil, basophil, mast cell, erythrocyte and megakaryocyte lineages, while lymphoid progenitors give rise to T- and B-cells. Stem cells are believed to randomly differentiate into either myeloid or lymphoid progenitors, while subsequent clonal expansion, maturation and survival can be greatly influenced by signals generated by e.g. infections or injury(Ogawa, 1993) (Hume, 2000). In this way, blood formation is regulated to meet the demand for specific blood lineages. Hematopoiesis may be regulated by soluble factors including cytokines, and hormones, and also by physical interaction with stromal cells and the extracellular matrix(Allen and Dexter, 1984) (Metcalf, 1989).
Regulation of hematopoiesis by the cytokine network
Survival, proliferation and differentiation of hematopoietic cells can be regulated by cytokines, a family of polypeptide ligands(Watowich et al., 1996). The cytokine family can be grouped on the basis of shared subunits that associate with their cognate receptors. This division can also be considered partially functional, since the shared subunit frequently represents the signaling part of the multimeric receptor complex. For example, the IL-6 receptor family, consisting of IL-6R, IL-11R, CNTFR, LIFR and oncostatin M (OM)R, all associate with the gp130 molecule which provides the critical signal transducing capacity. The functional significance of this similarity can be illustrated by the shared ability of IL-6 and LIF to induce differentiation of murine myeloid leukemia M1 cells. Actually, all members of the family share pleiotropic activities on several cell types, with the exception of CNTF, which is exclusively expressed, together with its receptor, in the neural system(Kishimoto et al., 1992).
Similarly, IL-3, IL-5 and GM-CSF, which activate receptors of the IL-3 receptor family and share the KH97 signaling subunit, all cause formation of eosinophil colonies(Miyajima et al., 1992). However, only IL-3 and GM-CSF have an effect on GM colony formation and activation of monocytes. Although most cytokines can act on progenitors and mature cells of several lineages, the effect of a cytokine may vary depending on the target cell. For instance, IL-6 stimulation can induce antibody production in B-cells, platelet formation in megakaryocytes and bone resorption by osteoclasts. In addition, recruitment of early progenitors often require the joint stimulation of several cytokines, each important for specific functions such as cell- cycle entry and survival. On the other hand, different cytokines may have overlapping effects on a single target cell. This redundancy of the cytokine network may be partly explained by shared receptor subunits, leading to activation of the same subset of transcription factors, but may also be functional in allowing synergistic effects by costimulation with several cytokines(Kishimoto et al., 1994).
Transcription factors involved in hematopoiesis
Signals from soluble factors and cell-cell interactions are integrated in the nucleus by the activation of transcription factors that influence the rate of initiation of gene transcription. Regulatory transcription factors are activated by signals transmitted from activated receptors at the cell surface, or in some cases by diffusible molecules directly binding to nuclear receptors. Activation is commonly regulated by phosphorylation of
specific sites, allowing interaction with other transcription factors, nuclear translocation, DNA binding and, importantly, recruitment of the basal transcription machinery. Commonly, this recruitment is achieved through binding to co-activators, such as CBP/p300, which also bind components of the basal transcription machinery.
In eukaryotic cells, the basal transcription machinery is composed of (1) the TATA binding complex TFIID, composed of the TATA binding protein (TBP) and associated factors (TAFs), important for binding to core promoter elements. (2) TFIIB which binds TFIID and is important for specifying the start site of transcription and for recruitment of (3) a complex of RNA polymerase II/TFIIF together with general transcription factors TFIIE and TFIIH, which are required to start transcription(Orphanides et al., 1996). Initiation of transcription is a two-step process in which regulatory and general transcription factors that initially bind DNA recruit acetylases to loosen the chromatin structure, allowing other transcription factors and RNA polymerase II to bind and initiate transcription. Repression of target genes is frequently achieved by recruitment of histone deacetylases (HDACs), which causes chromatin condensation and transcriptional silencing(Kingston and Narlikar, 1999) (Grant and Berger, 1999).
The process of hematopoietic differentiation is directed through the coordinate activation of transcription factors regulating genes important for acquiring a morphologically and functionally mature phenotype. Consequently, the transcription factors that are expressed and activated in response to external differentiation signals determine cell fate(Smithgall, 1998). In striking contrast to the common redundancy of cytokines, several transcription factors have been found to play crucial and non- redundant roles in hematopoiesis (Shivdasani and Orkin, 1996). For example, mice lacking either tal-1/SCL or rbtn2/LMO2 die in utero as a consequence of complete lack of hematopoietic cells, suggesting that these transcription factors are required during the development of immature multipotential progenitors early in hematopoiesis(Shivdasani et al., 1995) (Robb et al., 1995) (Warren et al., 1994).
Targeted disruption of GATA-2 results in a severe and early defect in primitive hematopoiesis, implicating GATA-2 in expansion of early hematopoietic progenitor cells(Tsai et al., 1994). Further, c-myb and AML-1 deficient mice appear to be blocked in definitive hematopoiesis due to a proliferation abnormality of the hematopoietic stem cells in the case of c-myb and block in development of all hematopoietic lineages in the case of AML-1(Mucenski et al., 1991) (Okuda et al., 1996). Lineage specific functions of GATA-1 and PU.1 have been demonstrated by the disruption of erythropoiesis in GATA-1-/- mice and absence of monocytes, granulocytes and B-cells by targeted disruption of PU.1(Pevny et al., 1991) (Scott et al., 1994) (McKercher et al., 1996). Additional transcription factors involved in myeloid differentiation will be discussed below. Several other transcription factors have been implicated in hematopoiesis on the basis of severe hematopoietic defects found in knockout animals(Shivdasani and Orkin, 1996). Recently, the crucial coupling of cytokine function to transcription factor activation was beautifully demonstrated by the hematopoietic defects of transgene mice expressing the G-CSFR mutated at the Stat3 activation site(McLemore et al., 2001).
Many transcription factors important for hematopoietic differentiation were first discovered through recurring chromosomal translocations in human leukemia(Rabbitts, 1994) (Look, 1997). For example, the core binding factor (CBF)α2 (AML-1) is frequently rearranged in acute myelogenous leukemia (AML) and childhood T-cell acute lymphoblastic leukemia (ALL), while dysregulation of both rbtn/LMO2 and tal- 1/SCL has been found in ALL(Lutterbach and Hiebert, 2000) (Rabbitts et al., 1999) (Begley and Green, 1999). Interestingly, targeted disruption of ICSBP in mice results in a hematopoietic disorder resembling chronic myeloid leukemia (CML), most likely reflecting an important function of ICSBP in myeloid maturation(Holtschke et al., 1996). These observations support the widely held view that hematopoietic tumors arise as a consequence of immature blasts failing to differentiate into mature, growth- arrested blood cells, and illustrate the intimate link between developmental biology and tumor biology.
Myeloid Differentiation
All cells that are non-lymphoid are by definition myeloid. However, in this thesis I will use a narrower view of myeloid differentiation and focus on differentiation towards the monocyte/macrophage and granulocytic lineages. The CFU-GM (colony forming unit- granulocyte/macrophage), a bipotential progenitor, originates from the CMP and gives rise to monocytes/macrophages or granulocytes. Several physiologic inducers, including IL-6, LIF, oncostatin M, and M-CSF, can trigger differentiation along the monocytic pathway, while promyelocytes and granulocytes are formed after stimulation with G-CSF and retinoic acid(Figure 2).
G/M
monoblast promonocyte monocyte macrophage myeloblast promyelocyte neutrophil
MZF-1
C/EBP-α PU.1
C/EBP-ε AML1
CBFβ c-Myb
PU.1 Egr-1 C/EBP-β ICSBP
PU .1 Egr-1 C/EBP-β
ICSBP C/EBP-α
IL-6 M-C SF
G-CSF retinoic acid
PU .1 Egr-1 C/EBP-β IL-3
Figure 2. Inducers and transcription factors regulating myelomonocytic differentiaion.G/M: bipotential granulocytic/monocytic progenitor.
Key transcriptional regulators of myelopoiesis include core-binding factors (CBFs), CCAAT/enhancer binding proteins (C/EBPs), c-Myb and Ets factors. These proteins often cooperate in the transcriptional regulation of genes by the combinatorial activation of myeloid-specific promoters (Tenen et al., 1997). Throughout differentiation from the hematopoietic stem cell to the CFU-GM, CBFα2 (AML-1) and its partner CBFβ plays a key role in the commitment to both lymphoid and myeloid lineages, as targeted disruption of either partner is associated with severe developmental defects of all hematopoietic cells(Okuda et al., 1996; Wang et al., 1996a; Wang et al., 1996b). In contrast c-Myb has been suggested to be required for proliferation of early hematopoietic progenitors, due to the severe anemia observed in mice lacking this transcription factor(Mucenski et al., 1991). This is supported by the reported prerequisite of c-Myb expression for self-renewal of multipotent FDCP-mix cells(White and Weston, 2000). In myelopoiesis, PU.1 and C/EBP family member C/EBP-α seem to be important for both granulocyte and monocyte/macrophage differentiation. PU.1 expression is restricted to hematopoietic cells, and targeted disruption of this gene results in defects of both monocytic and granulocytic development(McKercher et al., 1996; Scott et al., 1994). Interestingly, PU.1 has been suggested to be a master regulator of myeloid genes, due to the presence of PU.1 sites in many myeloid-specific promoters(Tenen et al., 1997). The expression of C/EBP-α is limited to myelomonocytic cells in the hematopoietic system and is highly expressed upon induction of differentiation of these cells. Although C/EBP-α is expressed during both monocytic and granulocytic maturation, it is down-regulated during terminal stages of monocytic differentiation(Scott et al., 1992). This may indicate a preferential role of C/EBP-α in differentiation along the granulocytic pathway, which is consistent with the defects in granulocytic differentiation observed in mice lacking this protein(Zhang et al., 1997).
Some transcription factors appear to be functional in only one lineage, and may thus direct differentiation towards either a monocyte/macrophage or granulocytic phenotype. Transcription factors involved mainly in granulopoiesis include the myeloid zink finger protein (MZF)-1 and C/EBP-ε. MZF-1 is preferentially expressed in early granulocytic progenitors (Hromas et al., 1991) and is necessary for granulocyte colony formation in response to G-CSF(Bavisotto et al., 1991). In contrast, the level of C/EBP- ε is preferentially upregulated during granulocytic differentiation(Yamanaka et al., 1997b), and mice lacking this protein exhibit defects in terminal stages of granulocyte differentiation(Yamanaka et al., 1997a). Differentiation along the monocyte/macrophage pathway has been linked to upregulation of transcription factors Egr-1, C/EBP-β and ICSBP. The zink finger transcription factor Egr-1 is upregulated during monocytic, but not granulocytic, maturation in the bipotential myeloid cell line HL-60. Addition of anti-sense Egr-1 oligonucleotides to the growth medium blocked monocytic differentiation, while overexpression of Egr-1 commited HL-60 cells to the monocytic lineage, demonstrating the functional significance of Egr-1 expression in this context (Nguyen et al., 1993). The chicken homologue to C/EBP-β (NF-M) has been shown to regulate myelomonocytic-specific target genes, including the G-CSF
related cytokine cMGF(Katz et al., 1993; Kowenz-Leutz et al., 1994). In addition, C/EBP-β was shown to be upregulated during monocytic differentiation of human HL- 60 and U-937 cells (Natsuka et al., 1992). However, targeted disruption of either Egr-1 or C/EBP-β in mice was not associated with defects in monocytic differentiation, possibly due to a functional redundancy with other Egr-1 or C/EBP family members in these mice(Lee et al., 1996; Screpanti et al., 1995; Tanaka et al., 1995). Instead, mice deficient in C/EBP-β showed some defects in inflammation response that supports a role for this protein in the regulation of cytokine production(Screpanti et al., 1995;
Tanaka et al., 1995). As discussed above, mice lacking ICSBP expression exhibit a block in myeloid differentiation resulting in a development of leukemia reminiscent of human CML(Holtschke et al., 1996). Interestingly, reintroduction of ICSBP in hematopoietic progenitors lacking this protein is associated with commitment to differentiation along the monocytic pathway(Tamura et al., 2000), suggesting a preferential role of ICSBP in monocytic differentiation .
Mammalian cell cycle regulation
Growth and development of all organisms depend on multiplication of cells through cell division. The correct copying and partitioning of the genetic material during this process is central for life, and is accordingly regulated by precise mechanisms. The cell cycle is traditionally divided into four phases (Figure 3) (Johnson and Walker, 1999).
During S-phase, DNA replication is initiated and completed, while separation of the chromosomes and physical division of the cell, giving rise to two daughter cells (mitosis), is carried out in M-phase. During the gap-periods between these phases, known as G0/G1and G2, the cells are preparing and/or waiting for the proper signals to progress to the next phase.
Cell-cycle progression is directed by the regulation of cyclin dependent kinases (CDKs), a family of protein serine/threonine kinases that control phosphorylation of several substrates, including pRb pocket-proteins(Grana and Reddy, 1995). Active, unphosphorylated pRb, binds E2F transcription factors and may also actively repress E2F target genes, important for cell-cycle entry, by the recruitment of HDACs to the E2F-pRb complex(Dyson, 1998) (Brehm et al., 1998). CDK mediated phosphorylation inactivates pRb, and thus allows transcription of E2F target genes necessary for completion of the cell cycle (Figure 3).
CDK activity can be controlled by several distinct mechanisms. Primarily, its activity is positively regulated by binding to cyclins, a family of homologous proteins that are, with the exception of D-type cyclins, differentially expressed during different phases of the cell-cycle(Johnson and Walker, 1999). In mammalian cells, a complex containing CDK7 and cyclin H, termed the CDK-activating kinase (CAK), can additionally increase CDK activity by phosphorylation of a conserved threonine residue(Pines, 1999). The phosphate is removed by the phosphatase KAP after the cyclin partner has been degraded. Negative regulation of CDK activity can be achieved by two different groups of CDK inhibitors (CKIs), namely the Ink4 family, consisting of p15, p16, p18 and p19, and the Cip/Kip family, including p21, p27 and p57(Vidal and Koff, 2000).
The Ink4 family mainly represses CDK4 and CDK6 activity, while Cip/Kip proteins can inhibit most CDKs, and may sometimes be necessary for assembly of CDK/cyclin complexes(Sherr and Roberts, 1999). However, both classes are mainly active in G1, pointing to the central importance of the G1/S checkpoint. Another type of control regulates CDK1 activity, and thus initiation of mitosis. Inhibitory kinases, such as Wee1 can repress CDK1 in early G2 by phosphorylation of inhibitory residues in the ATP-binding site of the CDK, while the cdc25 phosphatase activates CDK1 in late G2
by removal of these same phosphorylations(Pines, 1999). Growth arrest can thus be achieved by several mechanisms, including the down-regulation of cyclins and/or upregulation of CKIs, and appears to be cell-type and inducer-specific.
Cyclins D1, D2, D3
p15 p16 p18 p21
Cyclin E CDK2 CDK4/6
pRb E2F P
Transcription of target genes important for cell cycle progression
G0
G1
p21 p27
Cyclin A CDK2 pRb
P E2F
S M
G2
CyclinB CDK1
CDK1 CyclinB cdc25
pRb P
E2F
pRb E2F
CyclinH CDK7 Wee-1
Figure 3. A simplified view of the mammalian cell cycle.
Retinoids and all-trans retinoic acid (ATRA)
All-trans retinoic acid (ATRA), the main signaling retinoid in the body, is essential during embryonic development, and is important in the adult for maintaining proper function of the skin, lungs, liver, neuronal and immune systems. Retinoids are derived from dietary vitamin A, which is abundant in egg, milk, butter, fish-liver oils and β- carotene. Vitamin A is taken up in the intestinal mucosa, esterified to retinyl esters,
transported in chylomicrons through the blood and finally stored in the liver. The retinyl esters are cleaved to form retinol, which is transported into the circulation bound to retinol binding protein (RBP) (Blomhoff et al., 1990). In the target cell, retinol is oxidized to retinal and retinoic acid (RA)(Napoli, 1999). Only a small part of retinol is converted to ATRA, other biologically active metabolites include 9-cis RA, 13-cis RA and 3,4-didehydro RA. Intracellular retinal- and RA-binding protein (CRBPs and CRABPs) are associated with retinal and RA in the cell and are believed to protect retinal and/or RA from nonenzymatic oxidation, control retinoid metabolism by only releasing their ligands to recognized enzyme, and regulate the cellular concentration of free RA(Napoli, 1996).
The level of RA needs to be tightly controlled during fetal development to ensure proper organogenesis. Vitamin A deficiency is associated with severe malformations, including abnormalities of the heart, CNS, ocular tissues, respiratory system and urogenital system, while high levels of RA is teratogenic, mainly affecting the heart, skull, skeleton, limbs, brain, eyes, CNS and craniofacial structures(Ross et al., 2000).
Increasing evidence suggest that RA acts as a morphogen during embryogenesis, directing differentiation by providing positional information in the developing fetus.
ATRA is also involved in the regulation of differentiation and proliferation of several cell types throughout adult life. This important function has made ATRA a potential drug for the treatment of many types of cancer(Altucci and Gronemeyer, 2001).
Retinoids are already used successfully to treat several malignancies, including breast cancer, and precancerous lesions, such as leukoplakia, actinic keratosis, and cervical dysplasia(Lotan, 1996). In addition, retinoids have been shown to delay the development of skin cancer in patients with xeroderma pigmentosum. Above all, ATRA has proven to be an important agent in the treatment of leukemia, and it is the drug-of-choice in the treatment of acute promyelocytic leukemia (APL) (Chomienne et al., 1996).
The role of ATRA in myeloid differentiation and leukemia
Several lines of evidence implicate ATRA in the regulation of myeloid differentiation.
Firstly, Vitamin A-deficient mice and humans were noted to have defects in hematopoiesis(Wolbach and Howe, 1978) (Hodges et al., 1978). Secondly, retinoids preferentially stimulate granulopoiesis of multipotent hematopoietic cell lines and normal bone marrow cells(Breitman et al., 1980) (Gratas et al., 1993) (Tocci et al., 1996). Thirdly, ATRA is an effective inducer of growth arrest and/or differentiation of several myeloid cell lines and freshly isolated acute myeloid leukemic cells in vitro (Douer and Koeffler, 1982). Finally, disruption of ATRA signaling through expression of a dominant-negative retinoic acid receptor α (RARα) in the multipotent murine FDCP mixA4 cell line switched lineage commitment from granulocyte/monocyte to the mast cell lineage(Tsai et al., 1992). Also, GM-CSF-mediated myeloid differentiation of these cells was blocked at the promyelocyte stage, but could be restored by high doses of ATRA(Tsai and Collins, 1993).
The flagship of ATRA therapy is the efficient treatment of patients with acute promyelocytic leukemia (APL) (Chomienne et al., 1996). The absolute majority of
patients undergoing ATRA treatment achieve complete remission through terminal differentiation, growth arrest and apoptosis of the leukemic blasts. However, treatment with ATRA alone is associated with a high percentage of relapse, often linked to ATRA resistance. ATRA treatment also causes severe side effects, the retinoic acid syndrome, in about 10-15% of the patients. This is associated with increasing leukocyte counts, fever, respiratory distress, weight gain, oedema, pleural or pericardial effusions, hypotension and in some cases renal failure(Altucci and Gronemeyer, 2001). Treatment with ATRA in combination with other agents, including chemotherapy and interferons, is used to reduce the risk of the retinoic acid syndrome, which is fatal in over 10% of the cases, and also to prevent and/or treat relapsed APL. The origin of APL and the basis of the therapeutic efficacy of ATRA are well understood(Melnick and Licht, 1999). APL is in most cases associated with a chromosomal translocation t(15; 17) causing a fusion between the retinoic acid receptor (RAR)α gene on chromosome 17 and the PML gene on chromosome 15. This rearrangement is associated with defect retinoic acid signaling, defect PML signaling, and a block at the promyelocyte stage of differentiation(de The, 1996). Signaling can be restored by pharmacological concentrations of ATRA, allowing differentiation to proceed. In some cases of APL, the RARα gene is instead fused to the promyelocytic leukemia zink finger gene (PLZF), nucleophosphin (NPM), nuclear matric associated gene (NuMA) or signal transducer and activator of transcription (STAT)5α.(Zhang et al., 2000). Retinoic acid signaling and molecular mechanisms involved in APL treatment will be discussed in greater detail below.
Signaling pathways involved in myeloid differentiation
Receptors, signal transducers and transcription factors are necessary for response to any stimulus, for example signals instructing the cell to differentiate or die. The classical example of signal transduction is signaling through tyrosine kinase receptors, which are composed of an extra-cellular ligand binding domain and an intra-cellular kinase domain. Ligand binding induces dimerization of the receptor and activation of the tyrosine kinase domains, triggering binding and tyrosine phosphorylation of molecules that transmit the signal to the nucleus, finally initiating transcription of target genes. Several different signaling pathways representing variations to this theme exist in nature, including signaling through serine/threonine kinase receptors or G-protein coupled receptors, the PI-3 kinase signaling pathway and signaling pathways induced by oxidative stress. However, I will focus on the pathways that are most important for the understanding of this study, namely the retinoid receptors and the STAT signaling pathways.
The retinoid receptors
Growth and differentiation effects of retinoic acid are mediated by a subfamily of nuclear receptors, retinoic acid receptors (RARs) and retinoid X receptors (RXRs).
These are ligand-induced transcription factors that bind to specific DNA binding sites, retinoic acid response elements (RAREs), and induce transcription of target genes when activated (Chambon, 1996). Three subtypes of both RARs and RXRs have been cloned, namely RARα, RARβ, RARγ, and RXRα, RXRβ and RXRγ(Chambon, 1996).
The RAR subtypes and their isoforms can be activated by both ATRA and 9-cis-RA, while the various RXRs exclusively bind 9-cis-RA. RAR homodimers or RAR/RXR heterodimers can signal through RAREs upon ligand binding (Chambon, 1996). In addition, RXR homodimers can bind retinoid X response elements (RXREs) and may also heterodimerize with several other nuclear receptor family members, including thyroid hormone-R, Vitamin D-R, PPAR and NGF1B (Mangelsdorf and Evans, 1995) (Kliewer et al., 1992).
Generation of knockout mice lacking RAR or RXR subtypes, or combinations of several RARs/RXRs have painted a complicated picture of retinoid receptor signaling (Kastner et al., 1995). Mice lacking isotypes of RAR or RXR subtypes appear normal, while the combined disruption of all RARα or RARγ subtypes is associated with some of the defects associated with vitamin A deficiency, including poor viability, growth deficiency and some tissue-specific malformations. Targeted disruption of more than one RAR subtype result in radically reduced viability and malformations resembling the fetal vitamin A deficiency syndrome. While these data show great redundancy in RAR isoform signaling, they also clearly demonstrate that retinoic acid signaling, via RARs, mediate the biological effects of vitamin A during development. RXRα deficient mice die from cardiac failure due to a thin ventricular wall and also show ocular malformations. Combined disruption of RXRα together with different RAR subtypes often increased the severity of these symptoms, suggesting that these effects may reflect the function of RXRα/RAR heterodimer signaling. However, RXRβ null mutants display male sterility, which instead may be due to distruption of RXR/PPARβ heterodimer signaling (Kastner et al., 1996).
Transcriptional control by retinoid receptor signaling is mediated through interaction with corepressors (CoR), such as nuclear corepressor (NCoR) and silencing mediator for retinoid and thyroid hormone receptors (SMRT), and coactivators (CoA), such as the CREB binding protein (CBP) and SRC-1(Minucci and Pelicci, 1999). In the absence of ligand, RARs and RXRs bind to histone deacetylases (HDACs) via CoR and SIN3, which results in silencing of the promoters of target genes through deacetylation of histones and chromatin condensation. Ligand binding to RAR results in conformational changes in the ligand-binding domain, which disrupts CoR binding and instead allows interaction with CoAs. CoAs recruit histone acetyltransferase complexes, leading to chromatin decondensation and derepression of target genes. In addition, a second complex termed TRAP/DRIP/SMCC is believed to bind to the activated heterodimer and establish contact with the basal transcription machinery, thus increasing the frequency of transcription initiation. RAR-PML fusion-proteins in APL are believed to super-repress RARE target genes through oligomerization, leading to recruitment of multiple HDAC-containing complexes. This can be relieved by pharmacological, but not physiological, concentrations of ATRA, explaining the molecular basis of ATRA therapy in APL(Breems-de Ridder et al., 2000). However, the RAR-PLZF fusion-protein is associated with additional binding of HDAC containing complexes to PLZF, which cannot be relieved by ATRA-induction.
Accordingly, APL patients with the RAR-PLZF translocation do not respond to ATRA treatment alone. However, ATRA-resistant APL cells have been shown to respond to combinations of HDAC inhibitors and ATRA in vitro, suggesting the possibility of a
new therapeutic approach to APL treatment (Grignani et al., 1998) (He et al., 1998) (Lin et al., 1998) (Redner et al., 1999) (Marks et al., 2000). The in vivo efficacy of combination of the HDAC inhibitor sodium butyrate was initially reported in one relapsed APL patient (Warrell et al., 1998), but this has not been confirmed in other patients.
Signaling through STAT family proteins
Signal transducers and activators of transcription (STAT) family proteins were first discovered as crucial effectors of interferon signaling, mediating both the transmission of the signal, from the receptor to the nucleus, and the transcriptional activation of target genes(Darnell et al., 1994). Presently, the STAT proteins are known to participate in signaling by most cytokines and several growth factors, and the list of STAT activators is constantly growing. STAT homologues have been identified in evolutionary diverse organisms such as the slime mold Dictyostelium, the nematode C.elegans and the fruit fly Drosophila Melanogaster, suggesting an important function of STATs during development(Schindler, 1999). To date, seven mammalian STATs have been cloned; Stat1, Stat2, Stat3, Stat4, Stat5a, Stat5b and Stat6. Although a specific STAT protein may be activated by several cytokines, mediating a variety of effects, targeted disruption of STAT genes in mice have shown surprisingly specific, non-redundant roles of several STATs(Ihle, 2001). For example, mice deficient in Stat1 show specific defects in interferon signaling and a high susceptibility to viral infections(Durbin et al., 1996) (Meraz et al., 1996), while Stat4-null mice are impaired in IL-12 signaling and Th1 development(Kaplan et al., 1996) (Thierfelder et al., 1996).
STAT proteins share several structural domains (Figure 4) (Bromberg and Darnell, 2000). The C-terminal domain of the protein contains a Src-homology 2 (SH2) domain, which is a common structural motif that has been shown to mediate protein-protein interaction by binding to specific phospho-tyrosines. The SH2 domain is critical for recruitment of STATs to the activated receptor, interaction with janus kinases (JAKs) and for dimerization with other STAT family proteins(Ihle, 1996). Additionally, a conserved tyrosine residue in the C-terminal is important for the reciprocal binding to the SH2 domain of other STATs(Shuai et al., 1994). N-terminal of the SH2 domain is the linker domain (LD), which is highly conserved and shown to be important for transcriptional responses to IFN-γ, but not IFN-α(Yang et al., 1999). Transcriptional activation is mediated by the extreme C-terminal domain, and may be further regulated by the phosphorylation of a conserved serine residue present in Stat1, Stat3, Stat4, Stat5a and Stat5b(Decker and Kovarik, 2000). A coiled-coil domain in the N-terminal is required for protein-protein interactions(Chatterjee-Kishore et al., 2000a). While the DNA binding domain is essential for binding to specific sites in the promoters of target genes(Chen et al., 1998), the far N-terminal of the protein has been shown to be functional in oligomerization of STAT dimers bound to adjacent sites on DNA(Xu et al., 1996), and also to play a role in nuclear translocation and deactivation(Strehlow and Schindler, 1998). Additional modification of STAT function is achieved by the expression of splice variants, lacking the C-terminal transactivation domain(Ihle, 1996). STATs homo- or heterodimerize in response to activation, and may also interact
with other transcription factors, including the glucocorticoid receptor, in the case of Stat5, and interferon responsive factor (IRF) family member p48, in the case of Stat1(Bromberg and Darnell, 2000).
The classical mode of STAT activation is through ligand binding to cytokine receptors (Figure 5). Cytokine receptor signaling differs from receptor tyrosine kinase signaling in that the cytokine receptor family lacks an intrinsic tyrosine kinase domain, and instead relies on associated protein tyrosine kinases (PTKs) for signal transduction(Taniguchi, 1995). Briefly, Janus activating kinases (JAKs) bind to the cytoplasmic portion of the receptors and are activated by cross-phosphorylation upon ligand binding. JAK activation causes tyrosine phosphorylation of the receptor, creating a binding site for STAT family proteins. Subsequently, the STATs are phosphorylated by the JAKs on the conserved tyrosine residue in the C-terminal, which cause dimerization, translocation to the nucleus, DNA binding and transcription of STAT target genes(Schindler and Darnell, 1995) (Liu et al., 1998b). This chain of events was initially named ”the JAK-STAT” pathway due to the apparent dependence on JAKs for STAT activation. However, later studies have shown that STAT proteins may be activated through additional pathways, including growth factor receptors with intrinsic tyrosine kinases and non-receptor tyrosine kinases such as Src and Abl(Bowman et al., 2000). In addition, three groups of proteins have been shown to negatively regulate cytokine signaling(Greenhalgh and Hilton, 2001). Firstly, phosphatases such as SHP-1 have been shown to suppress signals mediated by several cytokine receptors. In the case of Epo-signaling, this is achieved through dephosphorylation of JAK2, but SHP-1 is also believed to act through direct dephosphorylation of activated cytokine receptors. Secondly, suppressors of cytokine signaling (SOCS) family protein have an SH2-domain and can inhibit cytokine signaling either through binding to JAKs or through competitive binding to activated receptors. Thirdly, the protein inhibitor of activated STATs (PIAS) family proteins suppress STAT signaling by direct binding to STATs. Thus the balance between activating kinases and negatively acting proteins will determine the strength of STAT activation in any given cell.
Stat1α (750 aa)
P P Y701 S727
N-Term Coiled-Coil DBD LD SH2 TAD
Figure 4. Structural domains of Stat1 showing the N-terminal domain, the coiled-coil region, the DNA binding domain (DBD), the linker domain (LD),the SH2 domain and the transcription activating domain (TAD).
Biological function of STATs STATs and oncogenesis
The ability of oncogenic tyrosine kinases, including v-Src and v-Abl, to phosphorylate STATs suggests a role for this signaling pathway in malignant transformation. Indeed, constitutive STAT activation, mainly involving Stat1, Stat3 and/or Stat5, is a common feature of many tumors, for example breast cancer, lung cancer and various hematopoietic malignancies(Bowman et al., 2000). Specifically, in vitro evidence links Stat3 activation to cell transformation. Experimental approaches coexpressing a dominant-negative Stat3 mutant together with the Src oncogene blocked Src-induced transformation, suggesting that Stat3 activation is a crucial step in oncogenic transformation by v-Src(Bromberg et al., 1998b) (Turkson et al., 1998). In addition, expression of a constitutively active Stat3 mutant was sufficient to cause transformation of rodent fibroblast cells(Bromberg et al., 1999). Expression of putative Stat3 target genes, including the prosurvival protein Bcl-XL, cell-cycle regulators cyclin D1 and p21 and the proto-oncogene c-myc, was elevated in these cells and suggested to be important for the process of transformation. Similarly, Stat5 has been suggested to contribute to transformation by BCR-Abl through the upregulation of Bcl- XL and cyclin D1(de Groot et al., 1999) (de Groot et al., 1999). The expression of a constitutively active mutant of Stat5 induced factor-independent growth of an IL-3- dependent mouse pre-B cell line, an important step in oncogenisis(Nosaka et al., 1999).
On the other hand, Stat1 activation has mainly been associated with negative regulation of cell-growth and induction of apoptosis (see below). Therefore, the functional importance of Stat1 activation in tumor formation is not easily envisaged.
STAT function in myeloid differentiation
The importance of cytokine signal transduction in hematopoiesis strongly suggests that STAT proteins may participate in this process. Primarily two STAT family members, Stat3 and Stat5, have been implicated in myeloid differentiation in response to distinct cytokines. For example, Stat3 has been shown to be crucial for IL-6 induced differentiation of murine myeloid M1 cells to macrophages, as expression of either an IL-6R mutant unable to activate Stat3 or dominant-negative Stat3 blocked M1 differentiation(Yamanaka et al., 1996) (Minami et al., 1996) (Nakajima et al., 1996).
Similarly, introduction of dominant-negative forms of Stat3 in murine 32D cells, or in a mouse myeloid cell line exogenously expressing the G-CSFR, blocked G-CSF induced differentiation (Shimozaki et al., 1997) (de Koning et al., 2000). In striking contrast to these data, Stat3 has been shown to be important for LIF-induced maintenance of the undifferentiated phenotype in ES cells. The role of Stat5 in erythropoietin (EPO)- induced differentiation remains controversial. Stat5 has been shown to be important either for the EPO-induced proliferation or differentiation in different myeloid cell lines, but yet another report show that PI-3 kinase activation is sufficient for both these processes(Chretien et al., 1996) (Iwatsuki et al., 1997) (Klingmuller et al., 1997).
Dominant-negative Stat5 blocked both IL-3-dependent proliferation and G-CSF-
induced differentiation in murine myeloid 32D cells, suggesting that Stat5 is involved in the biological response to these cytokines(Ilaria et al., 1999). Additionally, Stat5 was activated in response to several inducers of both monocytic and granulocytic differentiation in the human myelomonocytic cell lines U-937 and HL-60, and also during myelomonocytic growth factor-induced differentiation of chicken myeloid cells(Woldman et al., 1997). Although several studies implicate Stat3 and Stat5 in myeloid differentiation, neither Stat5-deficient mice nor the recent report analyzing conditionally Stat3-deficient neutrophils and macrophages in Stat3 cre-lox mice confirm a necessity of these signal transducers during myeloid or lymphoid development(Coffer et al., 2000) (Teglund et al., 1998) (Takeda et al., 1999). Likewise, even though Stat1 is activated in response to several cytokines that control hematopoiesis, Stat1 deficient mice show normal lymphoid and myeloid development and an isolated impairment of interferon signaling(Durbin et al., 1996) (Meraz et al., 1996). It is possible that redundancy of either differentiation signals or signal tranducers may be invoked in these knockout mice during development, while specific signaling events, such as the IFN signaling pathway, requires specific STATs.
Interestingly, recent data from us and others have implicated Stat1 in ATRA-induced differentiation of myeloid cells.
Activation and gene regulatory function of Stat1
Originally discovered for its role in interferon signaling, Stat1 is activated by multiple cytokines and growth factors, and has lately been shown to be upregulated in response to ATRA-treatment in myeloid cell lines(Gianni et al., 1997; Matikainen et al., 1997;
Pelicano et al., 1997). Interestingly, several studies have demonstrated that combinatorial treatment with interferons and ATRA has a strong synergistic effect on growth and differentiation of several cell-types, and may also be used clinically to enhance anti-tumor activity and reduce the risk of the retinoic acid syndrome(Gaboli et al., 1998; Lippman et al., 1997). A possible mechanism for this synergy was suggested by the observation that ATRA upregulates the levels of several interferon-induced genes, including important transcription factors such as Stat1, Stat2, IRF-1 and p48, thus increasing the interferon response(Gianni et al., 1997; Kolla et al., 1996;
Matikainen et al., 1997; Pelicano et al., 1997). However, it is also plausible that increased Stat1 activation, through interferon-treatment, may boost ATRA-induced differentiation per se, by enhancing differentiation-associated signaling pathways downstream of Stat1. The role of Stat1 in ATRA signaling is the main subject of this thesis, and will be extensively discussed further on.
Stat1 in interferon signaling
The majority of the investigations aimed at understanding the molecular mechanisms underlying Stat1 activation and transcriptional control have been focused on the interferon signaling pathway(Figure 5)(Darnell et al., 1994). In response to IFN-γ, Stat1 binds to the activated interferon receptor via the SH-2 domain, and is phosphorylated on the conserved Tyr-701 by Jak1 and Jak2. Tyrosine phosphorylation results in homodimerization, translocation to the nucleus and binding to IFN-γ target
genes via gamma activated sequence (GAS) elements. The IFN-α receptor is instead associated with Jak1 and Tyk2, and activation induces the formation of the ISGF3 complex containing Stat1, Stat2 and the interferon responsive factor (IRF) family member p48/ISGF3γ. The ISGF3 complex translocates to the nucleus and binds to interferon-stimulated response elements (ISREs) in the promoters of IFN-α responsive genes. The biological response to IFN-γ, but not IFN-α, additionally requires Stat1 phosphorylation on Ser-727(Bromberg et al., 1996; Wen et al., 1995), through a process that depends on Jak2 and the proline rich tyrosine kinase (Pyk)2 (Takaoka et al., 1999). The serine kinase involved may vary depending on the cell type, as different reports have implicated either extracellular signal-regulated kinase (ERK)2 or the stress-related p38 MAP kinase in IFN-γ induced serine phosphorylation of Stat1, using different model systems (Takaoka et al., 1999) (Goh et al., 1999). However, there are several contradicting reports on this subject, and a recent paper instead suggests that IFN-γ induced serine-phosphorylation is dependent on a pathway downstream phosphatidylinositol 3-kinase (PI-3K) and its effector kinase Akt (Nguyen et al., 2001).
It is clear that Stat1 tyrosine and serine phosphorylations are independently regulated(Zhu et al., 1997) (Kovarik et al., 1998), and may be differentially induced by distinct stimuli, leading to diverse transcriptional responses. Interestingly, Stat1 Ser- 727 phosphorylation is necessary for full transcriptional activation of a subset of IFN-γ
JAK2 JAK1 α α β
GAS motif P P Stat1
IFN-γ
P P P P
Figure 5. The IFN-γ signaling pathway.
target genes(Kovarik et al., 2001), and is important for interaction with some transcriptional coactivators, including the mini-chromosomal maintenance (MCM) family protein MCM-5 and the tumor suppressor BRCA1(Zhang et al., 1998) (Ouchi et al., 2000).
STATSTAT Pias1
STAT STAT
Nmi STAT
STAT p300/CBP
MCM5
JAKs Serine
Kinases
Tyrosine Phosphatases
BRCA1
cytoplasm nucleus PKR
DNA
Figure 6. Stat1 interacting proteins.
Stat1 interacting proteins
Adding to the complexity of this system, Stat1 has been demonstrated to interact with several different proteins, including STAT family members(Chatterjee-Kishore et al., 2000a), the protein kinase PKR(Wong et al., 1997), STAT inhibitor PIAS1(Liu et al., 1998a), and transcriptional coactivators, e.g. CBP and p300(Zhang et al., 1996)(Figure 6). In addition to forming complexes with Stat2 and p48, Stat1 can also heterodimerize with Stat3 in response to IL-6 family cytokines(Schindler and Darnell, 1995). Although targeted disruption of Stat1 does not appear to affect the biological response to cytokines and growth factors other than interferons, Stat1 activation may be functional in modulating gene expression in response to these inducers. The functional relationship between Stat1 and PKR appears to be somewhat complex. PKR is required for Stat1 serine phosphorylation(Ramana et al., 2000b), but does not directly phosphorylate Stat1. Although this implicates PKR in positive regulation of Stat1, a Stat1 mutant unable to interact with PKR displayed enhanced anti-proliferative and anti-viral activities(Wong et al., 2001). Interaction with PIAS1 suppresses Stat1 activity(Liu et al., 1998a), while binding to coactivators such as CBP/p300 enhance
transcriptional responses through histone acetylation and contacts with the general transcription machinery(Zhang et al., 1996). Association with the CBP/p300 coactivators , and transcriptional activation, is enhanced through Stat1 interaction with the c-myc binding protein Nmi(Zhu et al., 1999).
Complex roles of Stat1 in gene regulation
Strikingly, Stat1 activation may result in either positive or negative transcriptional regulation of target genes . Transcriptional activation requires a STAT-binding element in the promoter, and is likely to involve recruitment of co-activators and cooperation with both specific and general transcription factors and the core transcriptional machinery(Ramana et al., 2000a). Less is known about Stat1-dependent negative regulation of genes. A consensus GAS site in the promoter region of some genes, including the proto-oncogene c-myc(Ramana et al., 2000b) and the heparin sulfate proteoglycan perlecan(Sharma and Iozzo, 1998), is necessary but not sufficient for repression, pointing to a possible involvement of other promoter regions. The promoters of other genes that are repressed by interferons, for example cell-cycle regulator cyclin A(Sibinga et al., 1999), do not have STAT-binding sites. However, it remains to be shown if the inhibitory effect on transcription is mediated directly by Stat1 in these cases. Although the mechanism of Stat1-dependent repression is still unclear, it is likely to include interaction with co-repressors and/or modification of co- activators within the complex that drives transcription. In addition, Stat1 has been implicated in constitutive activation of several genes, a function that appears to be independent of either phosphorylation site. For example, Stat1-dependent expression of caspases can be restored in Stat1-negative U3a cells by introduction of mutated Stat1 lacking either the tyrosine or serine phosphorylation site(Kumar et al., 1997). Along the same line, tyrosine mutated Stat1, in complex with IRF1, can support transcription of the LMP2 gene(Chatterjee-Kishore et al., 2000b).
Stat1 involvement in growth arrest
While Stat3 and Stat5 have been implicated in both proliferative responses and cell cycle arrest, depending on the inducer, Stat1 appears to function mainly in the negative regulation of growth(Bowman et al., 2000; Coffer et al., 2000). Specifically, Stat1 is crucial for interferon and EGF induced growth arrest, and has been demonstrated to regulate the expression of the cell-cycle inhibitory protein p21 in response to these inducers, providing a direct link between Stat1 and growth control(Bromberg et al., 1998a; Bromberg et al., 1996; Chin et al., 1996). Interferon induced growth arrest involves a coordinate up- and down-regulation of several cell-cycle regulators, many of which have previously been suggested to be regulated by STATs, and it is probable that Stat1 plays a key role in this process.
Another mechanism of Stat1-dependent growth arrest has been suggested by the observation that interferon-induced Stat1 down-regulates expression of the proto- oncogene c-myc, by a mechanism that requires both Tyr-701 and Ser-727 phosphorylation(Ramana et al., 2000b). The myc family proteins are nuclear transcription factors that act through dimerization with the related protein max,
activating transcription of target genes by binding to E-boxes in the promoters of target genes and recruiting transcriptional co-activators(Amati et al., 2001). Conversely, the related mad-family proteins also form heterodimers with max, and are believed to inhibit transcription from E-boxes through association with co-repressors(Baudino and Cleveland, 2001). It is well established that myc family proteins are involved in proliferative responses(Grandori et al., 2000; Henriksson and Luscher, 1996). The expression of c-myc is increased upon mitogenic stimulation and is maintained at a basal level in proliferating cells, while differentiation and growth arrest is associated with a decrease in c-myc expression. Consequently, over-expression of Myc has been found to induce a block differentiation and retained proliferative capacity in several cell-types, demonstrating a central role of myc in these linked processes. Consistent with this, amplifications of either the c-myc gene, or myc family member N-myc, is frequently observed in several tumors, and often correlates to poor prognosis(Field and Spandidos, 1990). On the other hand, the mad family proteins are regulated in the opposite manner, most highly expressed in differentiated cells, and are suggested to play a role in differentiation rather than growth(Zhou and Hurlin, 2001). The exact mechanism by which the myc-max-mad network regulates proliferation is not clear but several cell-cycle regulators have been suggested to be myc target genes(Nasi et al., 2001). Thus, Stat1-induced c-myc downregulation is expected to efficiently inhibit growth by shifting the balance from myc/max complexes to mad/max complexes.
Interestingly, IFN-γ treatment has been shown to additionally induce destabilization of the myc/max heterodimer, but the possible role of Stat1 in this process has not been determined(Bahram et al., 1999).
In vitro models of myeloid differentiation
Investigations aimed at uncovering the molecular mechanisms involved in malignant transformations of hematopoietic cells have been greatly facilitated by the establishment of leukemia and lymphoma cell lines. These cell lines represent clonal expansions of tumor cells that are usually arrested at different stages of differentiation.
Although malignant transformation is frequently associated with multiple genetic changes, the ability of several established cell lines to differentiate in response to specific inducers in vitro indicates that at least part of the signaling networks regulating differentiation is still functional in these cells(Sachs, 1987). Several studies have shown that, despite the genetic changes connected with transformation, leukemic cell lines can be successfully used as model systems of hematopoietic differentiation.
In the U-937 cell line, established from a patient with histiocytic lymphoma, cells are arrested at the monoblastic stage of differentiation(Sundstrom and Nilsson, 1976). The well-documented ability of U-937 cells to terminally differentiate in response to TPA, ATRA and VitD3 makes this cell line an excellent model system for myeloid/monocytic differentiation(Nilsson K, 1980) (Olsson et al., 1983; Olsson and Breitman, 1982). In addition, U-937 cells are extensively studied, viable, easy to grow, fairly homogenous, and can readily be transfected with expression vectors coding for different gene-constructs.
ATRA and VitD3 induction of U-937 cells result in terminally differentiated, G0/G1
arrested cells with monocytic characteristics such as phagocytocis, NBT reduction, and expression of the αX subunit (CD11c) associated with β1 integrins(Olsson et al., 1983;
Olsson and Breitman, 1982) (Miller et al., 1986). However, these agents each induce different phenotypes, associated with distinct expression of mature surface markers(Figure 7). ATRA-treatment alone upregulates the α6 subunit associated with β1 and β4 integrins (CD49f), the bilary glycoprotein (CD66a) and the low affinity receptor for IgE (CD23), and give rise to slightly granulocytic cells with lobulated nuclei(Botling et al., 1995) (Oberg et al., 1993). In contrast, VitD3-induction is associated with upregulation of the LPS receptor (CD14), downregulation of CD23 and acquisition of a monocytic phenotype with kidney-shaped nuclei(Oberg et al., 1993).
These observations suggest that different pathways for terminal differentiation, involving separate gene-programs, are functional in these cells.
U-937 Differentiation
ATRA VitD3
CD11c CD14
CD11c G-CSFR CD49f CD66
Proliferating cells G0/G1 arrest CD23
Figure 7. Inducer-specific differentiation of U-937 cells.
The majority of the experiments in this thesis were done using U-937 cells. However, two related ATRA-responsive human myeloid cell lines, HL-60 and NB-4, were utilized to determine the general importance of the U-937 data(Collins et al., 1977) (Lanotte et al., 1991). HL-60 is a widely used bipotential cell line, arrested at an early stage of myeloid differentiation, which can be induced to differentiate along the granulocytic or monocytic pathways by either retinoic acid or DMSO treatment(Collins, 1987). In contrast, the NB-4 cell line, established from a patient with APL, expresses the PML-RAR fusion-protein and is arrested at the promyelocytic stage of differentiation. In accordance with the molecular basis of APL, these cells require ATRA to overcome the differentiation block invoked by the PML-RAR fusion-protein, and differentiate to granulocytic cells upon ATRA-treatment(Lanotte et al., 1991).
Aims of the present investigation
The general objective of this thesis was to gain further insight into the molecular mechanisms directing ATRA-induced growth arrest and differentiation of myeloid cells, and particularly, the role of the transcription factor Stat1 in this process.
The specific aims were to:
Determine the importance of Stat1 activation in ATRA-induced differentiation and cell-cycle arrest of human monoblastic U-937 cells (I)
Investigate ATRA-induced events leading to G0/G1 arrest during terminal differentiation of human myeloid cells, and the function of Ser-727 and Tyr-701 phosphorylated Stat1 in this process. (II) and (III)
Study the role of Stat1 in regulation of myeloid-specific transcription factors during ATRA-induced differentiation of U-937 cells (IV)
RESULTS AND DISCUSSION Paper I
The goal of paper I was to determine if Stat1 was functionally important for ATRA- induced differentiation of myeloid cells. Prior to this study, Stat1 was shown to be up- regulated in response to ATRA-treatment in several myeloid cell lines, but the biological function of Stat1 in this context had not been investigated(Kolla et al., 1997b) (Matikainen et al., 1997) (Pelicano et al., 1997) (Gianni et al., 1997). ATRA- induced regulation of Stat1, and other interferon-induced genes, was suggested to mainly be important for the synergistic differentiation-inducing effects of ATRA together with interferons. Our study provides the first evidence that activation of Stat1 is a necessary and fundamental component of ATRA signaling in mediating the process of differentiation.
Stat1 is activated by ATRA-treatment in human monoblastic U-937 cells
Using the U-937 cell line as a model of monocytic differentiation, we found that Stat1 was both upregulated and activated by ATRA-treatment in these cells. Stat1 activation by phosphorylation on Tyr-701 was accompanied by nuclear translocation, leading to active Stat1 in the nucleus of ATRA-treated cells. This was also associated with transcriptional activation of a luciferase reporter gene driven by an interferon- responsive promoter (GBP) containing a GAS site. Previous studies have demonstrated that IFN-α is produced during ATRA-induced differentiation of the related myeloid cell lines HL-60 and NB-4, suggesting that an autocrine loop may mediate Stat1 activation(Pelicano et al., 1997) (Gianni et al., 1997) (Pelicano et al., 1999). However, we found no evidence of IFN-α production in U-937 cells, either by use of RT-PCR or sandwich ELISA. Therefore, our data suggests an alternative and possibly more direct mechanism of activation, consistent with the comparatively rapid kinetics of Stat1 tyrosine phosphorylation(Gianni et al., 1997).
Establishment of U-937 sublines constitutively expressing phosphorylation-deficient Stat1
To address the role of Stat1 activation during ATRA-induced differentiation, we used dominant negative Stat1 to interfere with the part of the signaling pathway that involves Stat1 function. For this purpose, HA-epitope-tagged phosphorylation deficient Stat1 (Stat1Y701F) was constructed by PCR-based site directed mutagenesis. Transient transfections of U-937 cells with a vector allowing the expression of the Stat1Y701F construct suppressed ATRA-induced GBP-luciferase reporter activity, indicating that Stat1 signaling complexes were inhibited. Finally, sublines of U-937 constitutively expressing wild-type Stat1, HA-tagged Stat1, HA-tagged Stat1Y701F or empty vector (pCIneo), were established by G-418 selection. Although exogenous levels of Stat1 in these sublines were considerably higher than the endogenous level in vector-transfected control cells, they did not exceed those normally induced by interferons, thereby avoiding unspecific side-effects associated with vast over-expression of proteins.
