In this section and the next, we discuss two large classes of receptors that activate protein tyrosine kinases. Protein tyrosine kinases, of which there are about 90 in the human genome, phosphorylate specific tyrosine residues on target proteins, usually in the context of a specific linear sequence of amino acids in which the tyrosine is embedded. The phosphorylated target proteins can then activate one or more signaling pathways. These pathways are noteworthy because they regulate most aspects of cell proliferation, differentiation, survival, and metabolism.
There are two broad categories of receptors that activate tyrosine kinases: (1) those in which the tyrosine kinase enzyme is an intrinsic part of the receptor’s polypeptide chain, called the receptor tyrosine kinases (RTKs), which we discuss in Section 16.3, and (2) those, such as cytokine receptors, in which the receptor and kinase are separate polypeptides, encoded by different genes, yet are bound tightly together. In cytokine receptors, the tightly bound kinase is known as a JAK kinase. Both classes of receptors activate similar intracellular signal transduction pathways (Figure 1). We begin with the cytokine receptors, since they mainly employ a short signal transduction pathway called the JAK/STAT pathway: a STAT transcription factor binds to the activated receptor, becomes phosphorylated by the JAK kinase, moves to the nucleus, and directly activates transcription.
Fig1. Overview of signal transduction pathways triggered by receptors that activate protein tyrosine kinases. Both receptor tyrosine kinases (RTKs) and cytokine receptors activate multiple signal transduction pathways that ultimately regulate transcription of genes. (a) In the most direct pathway, mainly employed by cytokine receptors, a STAT transcription factor binds to the activated receptor, becomes phosphorylated, moves to the nucleus, and directly activates transcription. (b) Binding of one type of adapter protein (GRB2 or Shc) to an activated receptor leads to activation of the Ras/MAP kinase pathway (see Section 16.4). (c, d) Two phosphoinositide pathways are triggered by recruitment of phospholipase Cγ and PI-3 kinase to the membrane. Elevated levels of Ca2+ and activated protein kinase B modulate the activity of transcription factors as well as of cytosolic proteins that are involved in metabolic pathways or cell movement or shape.
Cytokines Influence the Development of Many Cell Types
The cytokines form a family of relatively small, secreted signaling molecules (generally containing about 160–200 amino acids) that control growth and differentiation of specific types of cells. One large family of cytokines, the interleukins, are essential for proliferation and functioning of the T cells and antibody-producing B cells of the immune system. Another family of cytokines, the interferons, are produced and secreted by certain cell types following viral infection and act on nearby cells to induce enzymes that render those cells more resistant to viral infection.
Growth hormone (GH), as its name implies, is a 191-amino-acid protein that stimulates proliferation of many types of body cells; it is made and secreted by cells in the anterior pituitary gland in response to another hormone, growth hormone–releasing hormone, that is made by the part of the brain termed the hypothalamus. GH was one of the first protein drugs to be made by recombinant DNA; it is used clinically to treat growth disorders in children and GH deficiency in adults. The bovine version is used to increase milk production in dairy cows. During pregnancy, a related hormone, the cytokine prolactin, induces epithelial cells lining the immature ductules of the mammary gland to differentiate into the acinar cells that produce milk proteins and secrete them into the ducts.
GH and prolactin have three-dimensional structures very similar to those of several cytokines that induce the formation of important types of blood cells. All blood cells are derived from hematopoietic stem cells, which form a series of progenitor cells that then differentiate into the mature blood cells. For instance, the cytokine granulocyte colony–stimulating factor (G-CSF) induces a granulocyte progenitor cell in the bone marrow to divide several times and then differentiate into granulocytes, the type of white blood cells that inactivate bacteria and other pathogens. A related cytokine, thrombopoietin, stimulates a different progenitor cell to divide and differentiate into megakaryocytes, huge cells that fragment into the platelets that are essential for blood clotting.
A structurally related cytokine, erythropoietin (Epo), triggers production of erythrocytes (red blood cells) by inducing the proliferation and differentiation of erythroid progenitor cells in the bone marrow (Figure 2). Erythropoietin is synthesized by certain kidney cells. A drop in blood oxygen, such as that caused by loss of blood from a large wound, signifies a lower than optimal level of erythrocytes, whose major function is to transport oxygen complexed to hemoglobin. The transcription factor HIF-1α is degraded in ambient oxygen levels. The kidney cells respond to low oxygen by preventing HIF-1α degradation; HIF-1α transcribes the erythropoietin gene, and the cells synthesize more erythropoietin and secrete it into the blood. As the level of erythropoietin rises, more and more erythroid progenitors are induced to divide and differentiate; each progenitor pro duces 30 to 50 erythrocytes in only a few days. In this way, the body can respond to the loss of blood by accelerating the production of erythrocytes.
Fig2. Erythropoietin and formation of red blood cells (erythrocytes). Erythroid progenitor cells, called colony-forming units erythroid (CFU-E), are derived from hematopoietic stem cells, which also give rise to progenitors of other blood cell types (see Figure 21-18). In the absence of erythropoietin (Epo), CFU-E cells undergo apoptosis (programmed cell death). Binding of Epo to its receptors on a CFU-E cell induces transcription of several genes whose encoded proteins prevent apoptosis, allowing the cell to survive. Other Epo-induced proteins trigger a developmental program of three to six terminal cell divisions, induction of hemoglobin and many other erythroid- important genes, reduction in cell and nuclear size, and finally, loss of the cell nucleus. If CFU-E cells are cultured with Epo in a semisolid medium (e.g., containing methylcellulose), daughter cells cannot move away, and thus each CFU-E cell produces a colony of 30–100 erythroid cells; hence its name. See M. Socolovsky et al., 2001, Blood 98:3261.
GH, prolactin, G-CSF, thrombopoietin, and Epo undoubtedly evolved from a common ancestral protein, since all of these cytokines have a similar tertiary structure consisting of four long conserved α helices folded together.
Both Epo and G-CSF are produced commercially by recombinant expression in cultured mammalian cells. Patients with kidney disease, especially those undergoing dialysis, frequently are anemic (have a low red blood cell count) and therefore are treated with recombinant Epo to boost red cell levels. Epo and G-CSF are used as adjuncts to certain cancer therapies because many cancer treatments affect the bone marrow and reduce production of red cells and granulocytes.
Binding of a Cytokine to Its Receptor Activates One or More Tightly Bound JAK Protein Tyrosine Kinases
GH, prolactin, G-CSF, thrombopoietin, and Epo all have similar structures, and they activate receptors of similar structure by forming dimers of two identical cytokine receptor proteins, a process termed receptor homodimerization. Each of these receptor proteins then activates JAK2, the JAK kinase to which it is bound. The extracellular domains of these cytokine receptors are constructed of two subdomains, each of which contains seven conserved β strands folded together in a characteristic fashion.
Many other cytokines, including the interleukins that activate immune-system cells and the interferons that induce expression of proteins that trigger resistance to viral infection, bind simultaneously to two or more different cytokine receptors, a process called receptor hetero-oligomerization. Generally these cytokine receptors bind to and activate two members of the JAK kinase family. Nonetheless, the signaling pathways activated by all cytokine receptors are broadly similar (see Figure 1), and receptor dimerization induced by hormone binding is common to many other receptor types that activate tyrosine kinases.
Interleukin 2 (IL-2) is essential for the formation of functional T cells, an essential component of the immune system. The IL-2 receptor is an oligomer of three different subunits termed alpha, beta, and gamma. The gamma chain is also an essential subunit of the receptors for several other interleukins, including IL-4, IL-7, IL-9, IL-15, and IL-21, all cytokines that are essential for formation of the antibody-producing B cells and other types of immune-system cells. Severe combined immunodeficiency (SCID) is a genetic disease in which neither T nor B cells are produced. People with SCID cannot cope with any bacterial or viral infection, so they must be kept in a sterile environment (like the famous “bubble boy”). Many cases of SCID are due to a deficiency in the IL-2 receptor gamma chain. These children can now be cured by gene therapy: a viral vector is used to introduce a functional gamma chain gene into the hematopoietic stem cells that generate all immune-system cells.
Here we focus on the simpler case of receptor homodimerization. The interaction of one erythropoietin molecule with two identical erythropoietin receptor (EpoR) proteins, as depicted in Figure 3, exemplifies the binding of this group of cytokines to their receptors. The structure of the GH-GH receptor complex is similar, and detailed mutagenesis studies have shown that only eight amino acids in growth hormone (green) contribute 85 percent of the energy that is responsible for tight receptor binding; these amino acids are distant from one another in the primary sequence but adjacent in the folded protein (Figure 4). Because of the multiple weak, noncovalent forces (i.e., ionic, van der Waals, and hydrophobic interactions) involved in this binding as well as molecular complementarity between the interacting surfaces of the receptor and ligand, the binding is extremely tight; a dissociation constant (Kd) of about 10−10 M is characteristic of most cytokine receptors. Thus very low concentrations of GH and most other cytokines—about 1 microgram per liter (roughly one part in a billion)—are sufficient to activate cytokine receptors.
Fig3. Structure of erythropoietin bound to an erythropoietin receptor. Like other cytokines, erythropoietin (Epo) contains four conserved long α helices that are folded in a particular arrangement. The activated erythropoietin receptor (EpoR) is a dimer of identical subunits; the extracellular domain of each monomer is constructed of two subdomains, each containing seven conserved β strands folded in a characteristic fashion. Side chains of residues on two of the α helices in Epo, termed site 1, contact loops on one EpoR monomer, while residues on the two other Epo α helices, termed site 2, bind to the same loop segments in a second receptor monomer, thereby stabilizing the dimeric receptor in a specific conformation. [Data from R. S. Syed et al., 1998, Nature 395:511, PDB ID 1eer.]
Fig4. Growth hormone binds to its receptor through multiple weak, noncovalent forces. (a) As deter mined from the three-dimensional structure of the 1 growth hormone:2 growth hormone receptor complex, 28 amino acids in the hormone are at the binding interface with one receptor molecule. To determine which amino acids are important in ligand-receptor binding, researchers mutated each of these amino acids, one at a time, to alanine and measured the effect on receptor binding. From this study, it was found that only 8 amino acids on growth hormone (green) contribute 85 per cent of the energy that is responsible for tight receptor binding; these amino acids are distant from one another in the primary sequence, but adjacent in the folded protein. Similar studies showed that two tryptophan residues (blue) in the receptor contribute most of the energy responsible for tight binding of growth hormone, although other amino acids at the interface with the hormone (yellow) are also important. (b) As with the Epo receptor, binding of growth hormone to one receptor molecule is followed by (c) binding of a second receptor (purple) to the opposite side of the hormone; this binding involves the same set of yellow and blue amino acids on the receptor, but different residues on the hormone. See B. Cunningham and J. Wells, 1993, J. Mol. Biol. 234:554, and T. Clackson and J. Wells, 1995, Science 267:383. [Data from A. M. de Vos, M. Ultsch, and A. A. Kossiakoff, 1992, Science 255:306, PDB ID 3hhr.]
Cytokine receptors do not possess intrinsic enzyme activity. Rather, a JAK kinase is tightly bound to the cytosolic domain of the receptor (Figure 5). Each of the four members of the JAK family of kinases contains an N-terminal receptor-binding domain, a C-terminal kinase domain that is normally poorly active catalytically, and a middle “pseudokinase” domain that regulates kinase activity by an unknown mechanism. (JAKs are so named because when they were first cloned and characterized, their function was unknown; they were termed just another kinase.) These kinases become activated after ligand binding and receptor dimerization (Figure 5, step 1).
Fig5. General structure and activation of cytokine receptors. The cytosolic domain of a cytokine receptor binds tightly and irreversibly to a JAK protein tyrosine kinase. In the absence of ligand (step 1), two receptors form a homodimer, but the JAK kinases are poorly active. Ligand binding causes a conformational change that brings together the JAK kinase domains, which then phosphorylate each other on a tyrosine residue in a region called the activation loop, activating the kinases (step 2 ). The active JAK kinases then phosphorylate multiple tyrosine residues in the receptor cytosolic domain (step 3). The resulting phosphotyrosines function as docking sites for signal-transducing proteins, including the STAT proteins.
Some cytokine receptors, such as the Epo receptor, are homodimers in the absence of ligand; others dimerize only in the presence of ligand. In both cases, ligand bind ing triggers a conformational change in the JAKs such that they can phosphorylate each other on a critical tyrosine in a region called the activation loop (Figure 16-10, step 2). As with many other kinases, phosphorylation of the activa tion loop leads to a conformational change in the kinase that enhances the affinity of the enzyme for ATP or the substrate to be phosphorylated, thereby increasing kinase activity (Figure5, step 3). One piece of evidence for this activation mechanism comes from study of a mutant JAK2 in which the critical tyrosine is mutated to phenyl alanine. The mutant JAK2 binds normally to EpoR, but cannot be phosphorylated and is catalytically inactive. In erythroid progenitor cells, expression of this mutant JAK2 in greater than normal amounts totally blocks EpoR signaling because the mutant JAK2 binds to the majority of EpoR proteins, preventing binding and functioning by the wild type JAK2 protein. This type of mutation, referred to as a dominant-negative mutation, causes loss of function even in cells that carry copies of the wild-type gene because the mutant protein prevents the normal protein from functioning.
Phosphotyrosine Residues Are Binding Surfaces for Multiple Proteins with Conserved Domains
Once the JAK kinases become activated, they first phosphorylate several tyrosine residues on the cytosolic domain of the cytokine receptor (see Figure 5). Several of these phosphotyrosine residues then serve as binding sites for signal-transducing proteins that have conserved phosphotyrosine-binding domains. One such phosphotyrosine-binding domain is called the SH2 domain. The SH2 domain derived its full name, the Src homology 2 domain, from its homology with a region in the proto typical Src cytosolic tyrosine kinase encoded by the src gene. (Src is an acronym for sarcoma, and a mutant form of the cellular src gene was found in chickens with sarcomas.) The three-dimensional structures of the SH2 domains in different signal-transducing proteins are very similar, but each binds to a distinct short sequence of amino acids surrounding a phosphotyrosine residue. The unique amino acid sequence of each SH2 domain determines the specific phosphotyrosine residues it binds (Figure 6). Variations in the hydrophobic socket in the SH2 domains of different proteins allow them to bind to phosphotyrosines adjacent to different sequences, accounting for the differences in their binding partners. The SH2 domain of the Src tyrosine kinase, for example, binds strongly to any peptide containing a critical four-residue core sequence: phosphotyrosine–glutamic acid–glutamic acid–isoleucine. These four amino acids make intimate contact with the peptide-binding site in the Src SH2 domain. Binding resembles the insertion of a two pronged “plug”—the phosphotyrosine and isoleucine side chains of the peptide—into a two-pronged “socket” in the SH2 domain. The two glutamic acids fit snugly onto the sur face of the SH2 domain between the phosphotyrosine socket and the hydrophobic socket that accepts the isoleucine residue. This specificity plays an important role in determining which signal- transducing proteins bind to which receptors and thus what pathways are activated.
Fig6. Surface model of an SH2 domain bound to a phosphotyrosine-containing peptide. The peptide bound by this SH2 domain from Src tyrosine kinase (blue backbone with red oxygen atoms) is shown in stick form. The SH2 domain binds strongly to short target peptides containing a critical four-residue core sequence: phosphotyrosine (Tyr0 and OPO3 −)–glutamic acid (Glu1)–glutamic acid (Glu2)–isoleucine (Ile3). Binding resembles the insertion of a two-pronged “plug”—the phosphotyrosine and isoleucine side chains of the peptide—into a two-pronged “socket” in the SH2 domain. The two glutamate residues are bound to sites on the surface of the SH2 domain between the two sockets. [Data from G. Waksman et al., 1993, Cell 72:779, PDB ID 1sps.]
SH2 Domains in Action: JAK Kinases Activate STAT Transcription Factors
To illustrate how binding of SH2 domains to specific phosphotyrosine residues induces specific signaling pathways, let’s discuss the straightforward mechanism by which all JAK kinases, and some RTKs, directly activate members of the STAT family of transcription factors. All STAT proteins contain an N-terminal DNA-binding domain, an SH2 domain that binds to one or more specific phosphotyrosines in a cytokine receptor’s cytosolic domain, and a C-terminal domain with a critical tyrosine residue. Once a monomeric STAT is bound to a phosphotyrosine in the receptor via its SH2 domain, the C-terminal tyrosine is phosphorylated by the associated JAK kinase (Figure 7a). This arrangement ensures that in a particular cell, only those STAT proteins with an SH2 domain that can bind to a particular receptor will be activated, and only when that receptor is activated. For example, the erythropoietin receptor, as well as the receptors for GH, prolactin, G-CSF, and several more cytokines, activates STAT5, but not STATs 1, 2, 3, or 4; those STATs are activated by other cytokine receptors. A phosphorylated STAT dissociates spontaneously from the receptor, and two phosphorylated STAT proteins form a homodimer in which the SH2 domain on each binds to the phosphotyrosine in the other (Figure 7b). Because dimerization involves conformational changes that expose the nuclear-localization signal (NLS), the STAT dimers move into the nucleus, where they bind to specific enhancers or promoters (DNA regulatory sequences) controlling target genes (see Figure 7a) and thus alter gene expression.
Fig7. Activation and structure of STAT proteins. (a) Phosphorylation and dimerization of STAT proteins. Step 1: Fol lowing activation of a cytokine receptor (see Figure 16-10), the SH2 domain of an inactive monomeric STAT transcription factor binds to a phosphotyrosine in the receptor, bringing the STAT close to the active JAK associated with the receptor. The JAK then phosphorylates the C-terminal tyrosine in the STAT. Steps 1 P P SH2 domain P DNA-binding domain 2 P 3 N C DNA 2 and 3: Phosphorylated STATs spontaneously dissociate from the receptor and spontaneously dimerize. Because the STAT homodimer has two phosphotyrosine-SH2 domain interactions, whereas the receptor-STAT complex is stabilized by only one such interaction, phosphorylated STATs tend not to rebind to the receptor. Step STAT 4: The STAT dimer moves into the nucleus, where it can bind to promoter sequences and activate transcription of target genes. (b) Ribbon diagram of the STAT1 dimer bound to DNA (black). The STAT1 dimer forms a C-shaped clamp around DNA that is stabilized by reciprocal and highly specific interactions between the SH2 domain (purple) of one monomer and the phosphorylated tyrosine residue (yellow with red oxygens) on the C-terminal segment of the other. The phosphotyrosine-binding site of the SH2 domain in each monomer is coupled structurally to the DNA-binding domain (magenta), suggesting a potential role for the SH2- phosphotyrosine interaction in the stabilization of DNA interacting elements. [Part (b) data from X. Chen et al., 1998, Cell 93:827, PDB ID 1bf5.]
Because different cell types have unique complements of transcription factors and unique epigenetic modifications of their chromatin, the genes that are available to be activated by any STAT are different in different cell types. For example, in mammary gland cells, STAT5, the same STAT that is activated by EpoR in erythroid progenitor cells, becomes activated following prolactin binding to the prolactin receptor and induces transcription of genes encoding milk proteins. In contrast, when STAT5 becomes activated in erythroid progenitor cells following binding of Epo to EpoR, it induces expression of the protein Bcl-xL. Bcl-xL prevents the programmed cell death, or apoptosis, of these progenitors, allowing them to proliferate and differentiate into red blood cells. More generally, it is thought that in each type of cell, activated STAT proteins, like the Smads discussed earlier, bind only to DNA sites in open chromatin and mainly to sites that have master transcription factors or other cell-specific gene regulatory proteins bound at adjacent sites. This combinatorial diversity allows a relatively limited set of receptors, JAK kinases, and STAT proteins to control a vast array of cellular activities.
Multiple Mechanisms Down-Regulate Signaling from Cytokine Receptors
In the last chapter, we saw several ways in which signaling from G protein–coupled receptors is terminated. For instance, phosphorylation of receptors and downstream signaling proteins suppresses signaling, and this suppression can be reversed by the controlled action of phosphatases. Here we discuss two mechanisms by which cytokine receptor signaling is regulated; other mechanisms for down-regulating signaling by protein tyrosine kinases are detailed in the next section because they have been investigated mainly with RTKs.
Phosphotyrosine Phosphatases Phosphotyrosine phosphatases are dephosphorylating enzymes that specifically hydro lyze phosphotyrosine linkages on specific target proteins. An excellent example of how phosphotyrosine phosphatase enzymes function to suppress the activity of protein tyrosine kinases is provided by SHP1, a phosphatase that negatively regulates signaling by several types of cytokine receptors. Its role was first identified by analysis of mice lacking this protein, which died because of excess production of several types of blood cells, including erythrocytes.
SHP1 dampens cytokine signaling by binding to a cytokine receptor and inactivating the associated JAK protein, as depicted in Figure 8a. In addition to a phosphatase catalytic domain, SHP1 has two SH2 domains. When cells are in the resting state, unstimulated by a cytokine, one of the SH2 domains in SHP1 physically binds to and masks the catalytic site in the enzyme’s phosphatase domain. In the stimulated state, however, this blocking SH2 domain binds to a specific phosphotyrosine residue in the activated receptor. The con formational change that accompanies this binding unmasks the SHP1 catalytic site and also brings it adjacent to the phosphotyrosine residue in the activation loop of the JAK associated with the receptor. By removing this phosphate, SHP1 inactivates the JAK, so that it can no longer phosphorylate the receptor or other substrates (such as STATs) unless additional cytokine molecules bind to cell-surface receptors, initiating a new round of signaling.
SOCS Proteins In a classic example of negative feedback, among the genes whose transcription is induced by STAT proteins are those encoding a class of small proteins termed sup pressor of cytokine signaling (SOCS) proteins, which terminate signaling by cytokine receptors. All SOCS proteins contain an SH2 domain and another domain, called the SOCS box, that recruits components of E3 ubiquitin ligases. The SOCS SH2 domain binds to specific phosphotyrosines on an activated receptor (Figure 8b); as a result, the receptor itself, as well as the associated JAK kinase, becomes polyubiq uitinylated (a polymer of ubiquitins is covalently attached to the side chain of a lysine) and is then degraded in proteasomes, thereby permanently turning off all JAK2 mediated signaling pathways until new receptors and JAK2 proteins can be made. The observation that proteasome inhibitors prolong JAK2 signal transduction supports this mechanism. One SOCS protein, SOCS-1, also binds to the critical phosphotyrosine in the activation loop of an activated JAK2 kinase, thereby inhibiting its catalytic activity.
Fig8. Two mechanisms for terminating cytokine signal transduction as exemplified by the erythropoietin receptor (EpoR). (a) Short-term regulation: SHP1, a phosphotyrosine phosphatase, is present in an inactive form in the cytosol of unstimulated cells. Binding of an SH2 domain in SHP1 to a particular phosphotyrosine in the activated receptor unmasks its phosphatase catalytic site and positions it near the phosphorylated tyrosine in the activation loop region of JAK2. Removal of the phosphate from this tyrosine inactivates the JAK kinase. See S. Constantinescu et al., 1999, Trends Endocrin. Met. 10:18. (b) Long-term regulation: SOCS proteins, whose expression is induced by the STAT5 protein in erythropoietin-stimulated erythroid progenitor cells, inhibit or permanently terminate signaling over longer periods. Binding of SOCS to specific phosphotyrosine residues on EpoR or JAK2 blocks binding of other signaling proteins (left). The SOCS box also targets the receptor as well as JAK2 for degradation by the ubiquitin-proteasome pathway (right). Similar mechanisms regulate signaling from other cytokine receptors. See B. T. Kile and W. S. Alexander, 2001, Cell. Mol. Life Sci. 58:1627.
Studies with cultured mammalian cells have shown that the receptor for growth hormone, which belongs to the cytokine receptor family, is down-regulated by another SOCS protein, SOCS-2. Strikingly, mice deficient in SOCS-2 grow significantly larger than their wild-type counterparts; these mice have long bones and proportionate enlargement of most organs. Thus SOCS proteins play an essential negative role in regulating intracellular signaling from the receptors for erythropoietin, growth hormone, and other cytokines.
