MOLECULAR REPRODUCTION AND DEVELOPMENT 27:4f%53 (1990)
Regulation of EGF Receptor Expression and Function GORDON N. GILL Department of Medicine, University of California, Sun Diego
INTRODUCTION The work I will discuss resulted from a collaboration of three groups. The principle scientists involved include those from my lab: Cherri Lazar, Gordon Walton, and Barney Welch; those from Geof Rosenfeld’s lab: William Chen, Chia-Ping Chang, and Allen Wells; and those from Steve Wiley’s lab: Kirk Lund and Brenda Welsh. Two general types of growth factor receptors have been identified in mammalian cells. The first type contains intrinsic protein tyrosine kinase activity in its cytoplasmic domain. These receptors respond to environmental signals via their specific cognate ligands. The second type of growth factor receptor belongs to the group of classical hormone receptors and include those that span the membrane seven times. The mas oncogene (the angiotensin receptor) and the serotonin receptor belong to this second class of receptors that affect cell proliferation. Aberrant regulation of either type of receptor can result in cell transformation. Although we experiment with cells in culture by limiting their growth with the removal of only one factor, cells normally proliferate in response to a group of factors. This requires communication and coordination between various types of signaling pathways. Receptors that contain tyrosine kinase activity must communicate with receptors that work through other mechanisms such a s G proteins. A number of events happen after the environmental signal interacts with the receptor. These events include increases in intracellular Ca+ and pH, increases in gene transcription, and ultimately increased rates of initiation of DNA synthesis and cell replication. +
TYROSINE PROTEIN KINASE RECEPTORS Much has been learned about signal transduction by growth factors from studies of natural mutations and “cloning” that were done by retroviruses. Various retroviruses picked up parts of genes that encode components of normal growth controlIing pathways. In virally transformed cells these oncogenes are constitutively expressed such that normal regulation is abrogated. Oncogenes are components of normal signal transducing pathways that include the tyrosine kinases, G proteins, and nuclear proteins, which are the ultimate targets of proliferative signal transduction
0 1990 WILEY-LISS, INC.
pathways that regulate transcription and other cellular functions. Here I will concentrate on the family of tyrosine kinase growth factor receptors. The members of this family have certain common features that include a n extracellular domain that recognizes the ligand and sense signals from the environment. These receptors have a single membrane spanning domain, a n intracellular tyrosine kinase domain, and a cytoplasmic tail. Fifty amino acids separate the inner side of the membrane from the beginning of the ATP binding sites in the kinase domain. The kinase domain is highly conserved within this family. The EGF receptor on which I shall concentrate is the prototype of this family. I t has two cysteine-rich regions in the ligand binding domain, a kinase domain, and a cytoplasmic tail. The insulin receptor and the c-erb-Blneu are other tyrosine protein kinase receptors with the general features of a large recognition domain, a transmembrane domain, and a cytoplasmic tyrosine kinase, which is the essential biological motor. Another group of tyrosine protein kinase receptors have a series of reiterated cysteines instead of the two cysteine-rich domains of the EGF receptor. This second group has a n immunoglobulinlike domain arrangement in the extracellular portion and a n insert in the kinase domain.
EGFRECEPTORANDRELATEDONCOGENES Figure 1gives a closer look at the EGF receptor and its several domains. The extracellular domain, which is the ligand-binding domain, is separated from the kinase, or activity domain, by a single transmembrane sequence. At the carboxyl terminus is a regulatory domain. The oncogene v-erbB was derived from the EGF receptor by insertional mutagenesis resulting in the loss of its N-terminal regulatory sequences. The v-erbB protein is a truncated version of the EGF receptor and lacks N-terminal sequences a s well as some C-terminal sequence. It is constitutively active and transforms cells. The c-erbB21neu oncogene is closely related in primary structure to the EGF receptor; however, c-erbB2 is encoded by a different gene from the EGF receptor and has a different activating ligand. Based on the results of kinetic studies, analogies with other proteins, and mutagenesis studies, we pro-
REGULATION OF EGF RECEPTOR EXPRESSION AND FUNCTION
Llgand Binding Domain
I
47
Activity Domain I
I
I
I
I
"Tra nsmembr a ne" Domain
i
I Regulatory Domain
Fig. 1. Linear view of the 1,186 amino acid EGF receptor protein. Functional domains are indicated. Cysteine-rich regions are indicated by zigzag lines. [Reprinted with permission from Gill et al. (1988), Cold Spring Harbor Symposia on Quantitative Biology, Vol. LIII, p.467, Cold Spring Harbor Press, New York.]
pose a model to explain the working of the EGF receptor (Fig. 2). The extracellular domain with the two cysteine-rich regions is assumed to have a n open conformation and is predicted to form a type of beta barrel. The results of NMR studies predict that EGF has a n anti-parallel beta pleated sheet structure. By analogy with what is known from the crystal structure of other proteins, one could imagine that when EGF binds the receptor adopts a closed conformation. This structural alteration would signal the activation of the tyrosine kinase domain in the cytoplasm. Data indicates there is one binding site in the rather large extracellular domain of the EGF receptor. The action of the cytoplasmic tyrosine kinase, which is activated by binding of EGF to the extracellular domain, involves the binding ATP to a cytoplasmic site on the receptor. The results of kinetic studies and studies of mutated receptors argued that the C-terminus, which contains the sites of self phosphorylation on tyrosine, inhibits the enzyme while it is in the basal state. When the enzyme is activated, the C-terminal phosphorylation sites would be phosphorylated. Once phosphorylated the inhibitory action of the C-terminus would be relieved and substrates would have access to the substrate binding site and thus be covalently modified by transfer of the y phosphate of ATP to tyrosine residues.
TYROSINE KINASE ACTIVITY REQUIREMENT IN THE ACTION OF THE EGFRECEPTOR An important initial question was whether tyrosine kinase activity was essential for the diverse biological responses that occur after EGF interacts with its receptors on the cell surface. One approach to answering this question was to mutate the EGF receptor to render the kinase inactive and to determine the ability of this receptor to mediate biological responses to EGF. We mutated lysine 721 that is predicted to form a salt bridge to the beta phosphate of ATP, reasoning that this lysine should be essential for kinase activity. We chose to replace the lysine with a methionine because
the methionine might not perturb the structure of this region very much but would inactivate i t as a kinase. We made a single base change in the cDNA to convert lysine to methionine at position 721 and then transfected this gene in a plasmid with the selectable marker, dihydrofolate reductase, into either B82 Lcells or CHO cells. Receptor expression was confirmed by Western blot analysis. K721 denotes the wild-type receptor with lysine at position 721 and a n active kinase domain. M721 has a methionine a t position 721 and is the kinase-deficient receptor. Both proteins are expressed as cell surface glycoproteins and both bind EGF. The differences in their kinase activities can be demonstrated by the ability of EGF to stimulate phosphorylation in intact cells. By using a n exogenous substrate we also demonstrate the M721 receptor has no EGF-dependent tyrosine kinase activity, whereas the holo K721 receptor is active in this assay. The M721 receptor that is expressed and glycosylated binds EGF normally but has no kinase activity. It has lost all biol-ogicalfunctions and its ability to mediate any responses to EGF. For example, Roger Tsien used FURA-2 to track free C a + + in individual cells. The results of these studies showed that when EGF was added to cells t h a t express the K721 (active kinase) EGF receptor, cytoplasmic Ca2+ rose. However, when EGF was added to cells bearing the M721 (inactive kinase) EGF receptor, there was no change in intracellular free Ca2+.These results showed that this very early response to EGF requires the kinase activity of the receptor. The mitogenic effect of EGF also required the kinase activity of the receptor. When cells expressing the mutant and holo-EGF receptors were grown in serum-free medium and treated with EGF, the only cells that increased their growth rate were the cells expressing the K721 holo-receptor. EGF halved the doubling time of these cells. The cells containing the M721, kinase-inactive receptor did not grow in response to EGF. Increased gene transcription in response to EGF also requires the protein kinase activity of the EGF receptor. To show this the firefly luciferase was used as a
Fig. 2. Conceptual model of activation of the EGF receptor. Ligand binding to a single site in the extracellular domain is proposed to cause a n allosteric change resulting in activation of the intracellular cytoplasmic protein tyrosine kinase. Self-phosphorylation of the regulatory C-terminus removes an alternate substrateicompetitive in-
hibitor permitting access of cellular substrates. This conformational change exposes sequences in the regulatory C-terminus that dictate interaction with coated pits for ligand-induced internalization and for coupling to mechanisms that increase cytosolic [Ca2 1.
reporter gene behind the fos enhancer. The fos gene responds to EGF with increased transcriptional activity. EGF increased the expression of this gene when it was introduced in cells that contain the K721 holoreceptor, but, in the cells that contain the mutant M721 receptor lacking kinase activity, there is no effect of EGF on gene transcription. Therefore elimination of the kinase activity by the lysine to methionine mutation abolished all responses to EGF, including early responses of increased calcium, intermediate responses of gene transcription, and the late proliferative response. Although kinase activity was expected to be required for these responses, the observation that removing the kinase activity also abolished the ability of the EGF receptor to be internalized and down-regulated was not quite so expected.
timately fuse with lysosomes. The receptor and the ligand are then degraded in lysosomes. Therefore, ligand-induced internalization results in loss of cell surface receptors. We found that the ability of the receptor to be internalized was also lost when the kinase activity was abolished. Cells that expressed the M721 or K721 receptors were treated with EGF and then assayed for residual EGF receptors on the surface by adding radioiodinated EGF after stripping away the receptor-bound EGF using acidic conditions. Pretreatment of cells expressing the K721 holo-receptor with EGF results in a decrease in the number of receptors on the surface; however, pretreatment with EGF did not change the number of receptors on the surface of cells expressing the M721 mutant receptors. Thus the mutant receptors bound EGF but did not undergo endocytosis. Western blot analysis indicated t h a t down-regulation of the EGF receptor actually represents a loss of mass not just a loss of surface exposure of the receptors. Whereas down-regulation occurs in the K721 holo-receptor transfected cells, the cells expressing the kinaseinactive, M721 receptor show no change in the mass of
TYROSINE KINASE ACTIVITY REQUIREMENT FOR EGF RECEPTOR INTERNALIZATION EGF receptors are normally diffusely distributed on the surface of the cell. Upon ligand binding they cluster in coated pits and are endocytosed in vesicles that ul-
+
REGULATION OF EGF RECEPTOR EXPRESSION AND FUNCTION EGF receptor after incubation with EGF. This same conclusion can be reached from immunocytochemical studies using antibodies against the EGF receptor. In the basal state, the wild-type EGF receptor is expressed diffusely on the transfected cell surfaces. Shortly after adding EGF, the cell surface has essentially been cleared of receptors and they have accumulated in large endocytic vesicles. At earlier times, the endocytic vesicles are smaller. When the kinase-inactive M721 mutant receptors are examined in expressed cells, they are found to be diffusely distributed on the cell surface similar to the holo-receptor. However, these mutant receptors are not cleared from the cell surface in response to EGF. Thus the metabolism of these receptors is blocked by removing the kinase activity. The question of whether the tyrosine protein kinase activity was required for receptor trafficking was also examined in some elegant studies by John Glenney who micro-injected a monoclonal antibody that recognizes phosphotyrosine into cells that expressed the kinase-active receptor. Cells that contained the monoclonal antibody were identified using a second antibody and another antibody was utilized to recognize the EGF receptor. The two antibodies were identified by rhodamine and fluorescein labels. In the cells that had been micro-injected with monoclonal antibody against phosphotyrosine the EGF receptor remained diffusely distributed on the cell surface instead of being endocytosed. By contrast, in uninjected cells the EGF receptor was endocytosed and localized in intracellular vesicles. The conclusion from these experiments is that removal of the tyrosine kinase activity by a n antibody or by mutational changes in the receptor not only abolished the biological effects of EGF but also abolished the metabolism and cell trafficking of the EGF receptor. These results raise the question of how signaling of all the complex cellular events occur through this cell surface receptor? If the kinase activity is required for endocytosis and internalization of the receptor, then removal of the kinase activity may abolish biological effects secondarily. Does the kinase-deficient mutant receptor not signal because it has not entered the cell where it needs to be located to function or is i t inactive because it has no kinase activity? Is the real answer some combination of the two possibilities? A secondary question concerns confirmational change of the receptor, which results in activation of the protein. Selfphosphorylation is part of the allosteric conformational change that results in a n active enzyme. If the receptor can not undergo self-phosphorylation, then it might fail to be endccytosed because either 1)i t failed to become phosphorylated, which is essential for endocytosis or 2) because it was not able to assume the correct conformation for the interactions necessary for endocytosis.
49
THE REGULATORY AND INTERNALIZATION DOMAINS OF THE EGF RECEPTOR One approach to answering these questions was to study the regulatory region at the C-terminus that contains the sites of self-phosphorylation. A series of mutations were made in the EGF receptor, which resulted in progressive truncations starting at the C-terminus. Each of the receptors was made in the context of either the kinase active (K721) or the kinase inactive (M721) receptor. Permanent cell lines were established with mutant receptors amplified using the dihydrofolate reductase gene. The number of receptors that the cells expressed were determined by Scatchard analysis of 1251EGF binding. Activity and expression level was assessed by Western blotting. Cells t h a t had been treated without or with EGF were lysed and displayed on a Western blot. Activity was measured with a n iodinated anti-phosphotyrosine antibody to identify cell substrates and a n anti-EGF receptor antibody was used to quantitate the EGF receptor. In the absence of EGF the holo-receptor showed no. activity. After addition of EGF tyrosine phosphorylated proteins could be detected. The EGF receptor was the primary phosphotyrosine-containing protein but some other phosphoproteins were also detected. Mobility of the receptor was retarded, characteristic of phosphorylated proteins. When a n EGF receptor that was truncated a t residue 1052 to remove the three known tyrosine self-phosphorylation sites was analyzed, i t was found to phosphorylate more cellular substrates in response to the holo-receptor. As expected it is a slightly smaller protein than the holo-receptor. A truncation was made at position 1022 inside the calpain hinge. Cleavage at the calpain hinge resulted in a shift to a 150 kDa receptor from the 170 kDa holo-receptor. As seen with the 1052 truncation, the C' 1022 receptor was quite active as a n EGF-stimulated tyrosine kinase. A mutant receptor that is truncated to residue 991 is also more active than the holo-enzyme in phosphorylating cellular substrates. The receptors with truncations to 973 and 957 are also quite active as EGF-dependent tyrosine kinases in vivo. The EGF receptor gene has 27 exons. The 957 position is located at the end of exon 22 a t a n exonintron junction. Further truncation into exon 22, to residue 944, produced a protein that was inactive. Therefore, the boundary of the EGF receptor kinase domain is somewhere near residue 957, i.e., near a n exon-intron junction. We measured down-regulation of these truncated cell surface receptors in the kinase active vs. inactive background. The holo-receptor was down-regulated, whereas the kinase inactive holo-receptor was not. Interestingly, whereas the kinase-inactive holo-receptor was not down-regulated, when it was truncated to residue 1022, inside the calpain hinge, it was partially down-regulated in response to EGF. This result sug-
50
G.N. GILL
gested that removal of the inhibitory tail, or a significant part of it, exposes sequences in the receptor that are involved in endocytosis and down-regulation of the receptor. Upon truncation to 991, kinase-inactive receptors are not internalized. This suggested that by incrementally truncating the molecule a region was exposed that allowed endocytosis but further truncation removed these sequences and created a receptor that could no longer be endocytosed and down-regulated. We were encouraged by these results because they showed that a kinase-inactive receptor could enter the cell and be down-regulated. This suggested that there was a domain in the cytoplasmic regulatory region of the receptor involved in ligand-dependent internalization. Removal of this region should create a kinse-active but internalization-deficient receptor. This mutant could be used to determine the biological effect of receptors that would remain at the cell surface. The boundaries of the internalization domain were determined by studies of various truncated forms of the receptor. The results showed that position 973 appeared to be the inner (C-terminal) boundary of the sequences responsible for down-regulation. The same results were obtained when the receptors were expressed in mouse 3T3 cells.
EFFECT OF MUTATIONS ON DOWN-REGULATION OF THE EGF RECEPTOR Steve Wiley has developed ways of determining the affinity of the receptor for coated vesicles and the endocytic apparatus. These methods allow one to measure endocytic rate constants for receptor-mediated internalization. Using these methods we demonstrated that cells that express the wild-type EGF receptor internalize the ligand as a function of the surface binding. After a brief lag the receptor enters with linear kinetics during which time the wild-type receptor is internalized a t a rate of about 35% of the occupied receptors per minute. The kinase-inactive mutant, which did not down-regulate, enters five- to seven-fold more slowly through what is essentially a constitutive pathway. This mutant does not decrease its receptor mass on the cell surface. For the C’ 1022 truncation, which is down regulated but has no kinase activity, the endocytic rate is intermediate (about 17% per minute). The C’ 1022 truncation is not endocytosed as well as the kinase active holo-receptor, but these results show clearly that removing some inhibitory C’-terminal residues exposes sequences that allow the receptor to internalize. These results are not a function of the host cell but a function of the receptor. The transferrin receptor was endocytosed a t the same rate in all of the transfected cells that carried different EGF receptors. Also, the uptake of iodinated polyvinyl pyrrolidine or iodinated plastic was invariant in these cells. The altered rates of endocytosis are thus not a function of the cell but are due to changes in the receptor.
Steve Wiley has also developed a method t o analyze the affinities of receptors for coated pits as a function of saturation of the receptors on the surface. The analysis is similar to a n Eadie-Hofstee or Scatchard analysis. Using this analysis, it was found that as the number of occupied receptors on the surface increased, the specific internalization rate diminished. This observation defines a saturable endocytic system for ligand-activated receptors, which has a definite capacity. The system is half occupied at about 50,000 receptors. Above 50,000 receptors per cell the system is saturated and ligand bound receptors enter through a constitutive pathway. The holo-receptor or the C’ 1022 deletion mutant show high receptor occupancy and are saturable. The kinase-inactive C‘ 1022-truncated receptor does not display saturable endocytic kinetics but instead a relatively straight line. The half-maximal capacity of this system is about 10 x lo6 receptors per cell. This higher capacity system is the constitutive uptake system. Receptors that did not down-regulate and that were truncated to remove the ligand-induced endocytic domain (e.g., the 973 or the 957 truncations) also show only constitutive endocytosis. In summary, when measured at low ligand concentrations where you measure the affinity of the system, the holo-receptor has the maximum endocytic rate. The affinity of the receptor is diminished by truncation to 1052 or by replacing the self-phosphorylated tyrosines with phenylalanines. The affinity rises when the truncation is a t 1022 and then i t begins to decrease when the truncation is at 991. When truncated to 973 the receptor is endocytosed near the basal constitutive rate. The receptor mutant that shows a different behavior from the rest is the C’ 1022 truncation, which has a much higher endocytic rate than expected from the rest of the series by the position of the truncation. It seems to be entering through the constitutive system but a t a much higher rate than the other receptors. The results of these studies show that the receptor’s structure determines its ability to interact with a saturable high-affinity endocytic system. Removing kinase activity abolishes the ability of the receptor to interact with the endocytic system. If kinase activity is retained but the C-terminus is truncated, the information that is necessary for the receptor to interact with the endocytic apparatus is lost. There is also a constitutive endocytic system and there are features of the receptor that determine its ability to interact with this endocytic system. This is shown by the difference between the rate of endocytosis of the C’ 1022 truncation and other truncated receptors that have lost the ability to interact with the high-affinity endocytic system.
STRUCTURAL ASPECTS OF THE RECEPTOR NECESSARY FOR BIOLOGICAL ACTIVITY With this background one can address the question of what structural and functional aspects of the EGF receptor are necessary for its biological activity. Mu-
REGULATION OF EGF RECEPTOR EXPRESSION AND FUNCTION tants were constructed that contained the domain defined as important for endocytosis (either the entire C-terminus from 944 to 1186 or the 944 to 1022 sequence) but did not contain the kinase domain. The resulting receptors were down-regulated in response to ligand. These mutants behaved similarly to the kinaseinactive receptor t h a t is truncated at position 1022 but that can still be endocytosed. By immunocytochemistry the mutant receptors were shown to enter cells. Biologically, there was no growth in response to stimulating endocytoses of kinase-inactive receptors. We tested the ability of the kinase-inactive, internalizing receptor to mediate induction of genes in response to EGF by using the luciferase reporter gene downstream from two different promoters that are inducible by EGF acting through the holo-receptor. The promoters were not induced in cells expressing endocytosiscompetent but kinase-inactive receptors. These receptors also do not mediate changes in intracellular Ca2+ These results show t h a t internalization of the kinase inactive receptor into the cell induces no biological response. Sequences on the receptor that mediate endocytosis can be transferred from one molecule to another. In contrast to the kinase-inactive receptors that are internalized, the kinase-active receptor that is not internalized is biologically active. It mediates EGF-induced gene transcription as measured by the activity of luciferase-linked promoters in transiently transfected cell lines. EGF induces gene expression equally well in cells that express the receptors that are not endocytosed but have kinase activity as in cells that express the holo-receptor. In fact, the endocytosis-defective but kinase-active receptor has a stronger effect on cell growth than the holo-receptor. 3T3 cells transfected with the holo-receptor grow into a monolayer in the presence of EGF. Cells that are transfected with the C’ 1022 truncation, which is a very active kinase and does not have the inhibitory domain, look a little different in the presence of EGF but still form a monolayer. Cells that have been transfected with the C’ 973 truncation appear transformed in the presence of EGF even at low cell densities. At high densities they lose contact-inhibition and form morphologically transformed foci. This is the expected result if the cells express a high number of EGF receptors. Cells are transformed by constitutively expressing TGF-alpha or EGF. Also, placing the EGF receptor behind a very strong promoter or amplifying the gene results in cell transformation. Here we demonstrate another mechanism for transformation. Failure to clear activated receptors from the cell surface also yields a transformed phenotype. From these results we deduce that the major signaling by the EGF receptor occurs at the cell surface and that the major effect of endocytosis and down-regulation is to appropriately attenuate the biological response to the ligand. If the receptor is not down-regulated then the cells display a transformed phenotype. We have not ruled out the possibility that there is a
51
small leakage in the system and that a n endocytosed kinase-active receptor may have some specific biological function. But overall, these results suggest that endocytosis functions to attenuate the signal delivered a t the cell surface.
CALCIUM INTERNALIZATION (CAIN) DOMAIN IN THE EGF RECEPTOR These mutants were used to evaluate the ability of EGF to increase intracellular calcium using indicator dyes. EGF increases intracellular calcium in cells expressing the holo-receptor and the receptor with the C’ 1022 truncation. The calcium response is somewhat decreased by truncation to 991 and is essentially abolished by truncation to 973. The ability to increase intracellular calcium is lost in all of the kinase-inactive receptors regardless of whether they are capable of being endocytosed. The cells transfected with the kinaseinactive receptors are still capable of responding to fetal calf serum with increases in intracellular calcium. Within the C-terminal kinase inhibitory domain is a domain that is required for endocytosis and for raising intracellular calcium, which we call the ca1cium:internalization (CAIN) domain (Fig. 3). The CAIN domain has two parts that are defined by the kinetics of response to EGF of the various mutants in internalization and in changing the intracellular free calcium levels. The region from 991 to 1022 was required for lowaffinity internalization and the part from 973 to 991 was required for high-affinity internalization and calcium regulation. When secondary structural predictions for the CAIN domain were made using the Chou Fasman and Garnier programs a predicted turn-helixt u r n motif with nine of the amino acids in the helix being acidic was identified. Such a turn-acid helix-turn motif is present in the related c-erbB2Ineu receptor. Although the primary amino acids are not the same, a similar CAIN domain is present, with variations, in the insulin receptor, the IGF-1 receptor, and with lesser certainty in other members of the family. Thus, it appears that the CAIN domain may be a defined structure in the C-terminus of the EGF receptor, which is responsible for the ability of the receptor to both undergo ligand-induced endocytosis and down-regulation and to raise intracellular free calcium. It also appears t h a t a t least some form of the CAIN domain may be present in the C-terminus of other members of the family of tyrosine kinase growth factor receptors. SUMMARY From the results of these studies of the activities of the various EGF receptor mutants we were able to disassociate the ability of EGF to increase intracellular calcium from its ability to induce genes and to cause morphological transformation and growth. These results lead us to the following concept. The kinase domain has a C-terminal border a t about residue 957. The remainder of the C-terminus is regulatory. The 164 amino acids from residue 1022 to 1186 constitute
52
G.N. GILL
A Kinase domain
"Caln" domain 957 973 991
Inhibitory domain
1022
RALMDEEDMDDVVDADEYLIPQQGFFSSPSTSRTPLLSSLSATSNNST
Required for
I
I
I
Required for low affinity internalization
high affinity
internalization & Ca'+ regulation
I
EGFreceptor
P S P
I
T
D S N F Y R A L M I , f s j M m V
V m A m Y L I P Q Q G F S S P S T S R
(9Y71
erb B-2/neu
(1 038)
I
A S P L D S T F Y R S L Lm
M
G O L
V mA
m
Y L V P Q Q G F
C
P D P A P G
1257 I
lnsulinreceptor
F P E
v s
F F H S
~
K
P H S ~ L O M B Fv ~P LMD sBs HNc Q R
A N
1252
F R E V S F Y Y S
E E 1293
I
IGF-lreceptor
1300 I
I
m
N K L P H P m L m L H P H N M H S V P L D S A S S S S L P
Fig. 3. Functional anatomy of the regulatory C-terminus of the EGF receptor: the CAIN domain. [Reprinted with permission from Chen et al., Cell 59:33-43, 1989, Cell Press, Cambridge, Massachusetts.]
a n inhibitory region for the kinase. It contributes to ligand-induced internalization because this is reduced in a mutant receptor truncated to residue 1052. Proximally within the C-terminus kinase inhibitory domain is a domain that is required for endocytosis and for raising intracellular calcium that we call the calcium internalization (CAIN) domain. In summary, we have found that the kinase activity of the EGF receptor is required for its function even when all of the self-phosphorylation sites have been removed. The EGF receptor has several distinct cytoplasmic domains that are important for its activity to regulate gene expression, DNA synthesis, and the intracellular calcium level. Biological signaling occurs from the cell surface via essential protein tyrosine kinase activity with ligand-induced internalization serving to abbrogate the biological signal.
QUESTIONS AND ANSWERS Q: Will the minus 973 deletion mutant grow in nude mice? A: We have not tested the mutant in nude mice but it will produce colonies in soft agar. Q: Do you see any differences in the amount of phosphotyrosine or the number of substrates that are phosphorylated by the various deletion mutants?
A: I cannot answer your question with absolute certainty but it is a very good question. If I can put it in context, the question is: If you remove the inhibitory domain, does the kinase become much more permissive in terms of the substrates that it will phosphorylate or the extent to which it will phosphorylate its substrates? We attempted to address this question by exposing the gels containing the phosphorylated receptor preparations for a long time to detect substrates other than the receptor itself. We did see a lot more bands than I have shown you but we were unable to determine from one-dimension gels whether they are all exactly the same in preparations from holo- and mutant receptors. My guess is that the answer is probably a little of both. We clearly see more phosphate incorporated into all of the protein substrates except the receptor itself in the extracts from cells expressing mutant receptors compared with the holo-enzyme. There are also probably some new substrates phosphorylated by the mutants that we do not see on the autoradiograms. Q: What is the function of the autophosphorylation sites? A: I think that the auto-phosphorylation sites have a rather classical function. There are many examples of kinases that are normally inhibited by a portion of
REGULATION OF EGF RECEPTOR EXPRESSION AND FUNCTION their own structure t h a t resembles a substrate. Selfphosphorylation is a way of removing this inhibitory domain. The cyclic AMP-dependent protein kinase is a good example of this. The regulatory subunit interacts with the catalytic unit and inhibits it through this kind of mechanism. I think i t is a mechanism that many kinases use. Q: Do you see a different phenotype in cells transfected with a receptor with single amino acid replacement a t the self-phosphorylation site in the C-terminal inhibitory domain? A: We certainly do not see a transformed phenotype. Endocytosis is interfered with a bit when you put a phenylalanine there. I think that when you replace the tyrosine with a phenylalanine the interaction with the C-terminus is not as tight as when tyrosine is present. The C-terminus does not make as good a fold or does not sit a s well in the site. But on the other hand it does not come out as well either. Thus, with this mutation the enzyme does not work quite normally but it is still capable of phosphorylating its substrates and initiating growth. Q: Are the sites of phosphorylation different in the mutants compared with the holo-receptor? A: No. There are five self-phosphorylation sites. They are all in the C-terminus and are all therefore regulatory. There is no phosphorylation in the kinase
53
domain. It is not like the insulin receptor where there is a phosphorylation site in the kinase domain, which is involved in its activation. This is more a classic regulatory domain. Q: Are early events other than calcium fluxes also not observed in response to the interaction of EGF with the receptors that are mutated in the CAIN domain? A: I cannot really say. My own prejudice is that the effects that we observed of the mutant receptors on transcription were accurately reflected by the reporter genes that we used. We have put different EGF-responsive promoters in front of the reporter gene and they all behave in the same way. I cannot answer the question with respect to the N a + / H + antiporter because the B82 cells did not lend themselves to those measurements. We have now put the receptors into some other cells that will perhaps allow us to ask that question. Q: What about the relationship between the CAIN domain and a n early event that is not membrane associated such as S6 phosphorylation? A: That is a good question. I do not know the answer. Q: Do you have a reagent, perhaps related to the inhibitory peptide that can be used to inhibit the function of the receptor? A: No, we do not. We are making the peptide; then we hope to use i t to make a n antibody.
Regulation of EGF Receptor Expression and Function GORDON N. GILL Department of Medicine, University of California, Sun Diego
INTRODUCTION The work I will discuss resulted from a collaboration of three groups. The principle scientists involved include those from my lab: Cherri Lazar, Gordon Walton, and Barney Welch; those from Geof Rosenfeld’s lab: William Chen, Chia-Ping Chang, and Allen Wells; and those from Steve Wiley’s lab: Kirk Lund and Brenda Welsh. Two general types of growth factor receptors have been identified in mammalian cells. The first type contains intrinsic protein tyrosine kinase activity in its cytoplasmic domain. These receptors respond to environmental signals via their specific cognate ligands. The second type of growth factor receptor belongs to the group of classical hormone receptors and include those that span the membrane seven times. The mas oncogene (the angiotensin receptor) and the serotonin receptor belong to this second class of receptors that affect cell proliferation. Aberrant regulation of either type of receptor can result in cell transformation. Although we experiment with cells in culture by limiting their growth with the removal of only one factor, cells normally proliferate in response to a group of factors. This requires communication and coordination between various types of signaling pathways. Receptors that contain tyrosine kinase activity must communicate with receptors that work through other mechanisms such a s G proteins. A number of events happen after the environmental signal interacts with the receptor. These events include increases in intracellular Ca+ and pH, increases in gene transcription, and ultimately increased rates of initiation of DNA synthesis and cell replication. +
TYROSINE PROTEIN KINASE RECEPTORS Much has been learned about signal transduction by growth factors from studies of natural mutations and “cloning” that were done by retroviruses. Various retroviruses picked up parts of genes that encode components of normal growth controlIing pathways. In virally transformed cells these oncogenes are constitutively expressed such that normal regulation is abrogated. Oncogenes are components of normal signal transducing pathways that include the tyrosine kinases, G proteins, and nuclear proteins, which are the ultimate targets of proliferative signal transduction
0 1990 WILEY-LISS, INC.
pathways that regulate transcription and other cellular functions. Here I will concentrate on the family of tyrosine kinase growth factor receptors. The members of this family have certain common features that include a n extracellular domain that recognizes the ligand and sense signals from the environment. These receptors have a single membrane spanning domain, a n intracellular tyrosine kinase domain, and a cytoplasmic tail. Fifty amino acids separate the inner side of the membrane from the beginning of the ATP binding sites in the kinase domain. The kinase domain is highly conserved within this family. The EGF receptor on which I shall concentrate is the prototype of this family. I t has two cysteine-rich regions in the ligand binding domain, a kinase domain, and a cytoplasmic tail. The insulin receptor and the c-erb-Blneu are other tyrosine protein kinase receptors with the general features of a large recognition domain, a transmembrane domain, and a cytoplasmic tyrosine kinase, which is the essential biological motor. Another group of tyrosine protein kinase receptors have a series of reiterated cysteines instead of the two cysteine-rich domains of the EGF receptor. This second group has a n immunoglobulinlike domain arrangement in the extracellular portion and a n insert in the kinase domain.
EGFRECEPTORANDRELATEDONCOGENES Figure 1gives a closer look at the EGF receptor and its several domains. The extracellular domain, which is the ligand-binding domain, is separated from the kinase, or activity domain, by a single transmembrane sequence. At the carboxyl terminus is a regulatory domain. The oncogene v-erbB was derived from the EGF receptor by insertional mutagenesis resulting in the loss of its N-terminal regulatory sequences. The v-erbB protein is a truncated version of the EGF receptor and lacks N-terminal sequences a s well as some C-terminal sequence. It is constitutively active and transforms cells. The c-erbB21neu oncogene is closely related in primary structure to the EGF receptor; however, c-erbB2 is encoded by a different gene from the EGF receptor and has a different activating ligand. Based on the results of kinetic studies, analogies with other proteins, and mutagenesis studies, we pro-
REGULATION OF EGF RECEPTOR EXPRESSION AND FUNCTION
Llgand Binding Domain
I
47
Activity Domain I
I
I
I
I
"Tra nsmembr a ne" Domain
i
I Regulatory Domain
Fig. 1. Linear view of the 1,186 amino acid EGF receptor protein. Functional domains are indicated. Cysteine-rich regions are indicated by zigzag lines. [Reprinted with permission from Gill et al. (1988), Cold Spring Harbor Symposia on Quantitative Biology, Vol. LIII, p.467, Cold Spring Harbor Press, New York.]
pose a model to explain the working of the EGF receptor (Fig. 2). The extracellular domain with the two cysteine-rich regions is assumed to have a n open conformation and is predicted to form a type of beta barrel. The results of NMR studies predict that EGF has a n anti-parallel beta pleated sheet structure. By analogy with what is known from the crystal structure of other proteins, one could imagine that when EGF binds the receptor adopts a closed conformation. This structural alteration would signal the activation of the tyrosine kinase domain in the cytoplasm. Data indicates there is one binding site in the rather large extracellular domain of the EGF receptor. The action of the cytoplasmic tyrosine kinase, which is activated by binding of EGF to the extracellular domain, involves the binding ATP to a cytoplasmic site on the receptor. The results of kinetic studies and studies of mutated receptors argued that the C-terminus, which contains the sites of self phosphorylation on tyrosine, inhibits the enzyme while it is in the basal state. When the enzyme is activated, the C-terminal phosphorylation sites would be phosphorylated. Once phosphorylated the inhibitory action of the C-terminus would be relieved and substrates would have access to the substrate binding site and thus be covalently modified by transfer of the y phosphate of ATP to tyrosine residues.
TYROSINE KINASE ACTIVITY REQUIREMENT IN THE ACTION OF THE EGFRECEPTOR An important initial question was whether tyrosine kinase activity was essential for the diverse biological responses that occur after EGF interacts with its receptors on the cell surface. One approach to answering this question was to mutate the EGF receptor to render the kinase inactive and to determine the ability of this receptor to mediate biological responses to EGF. We mutated lysine 721 that is predicted to form a salt bridge to the beta phosphate of ATP, reasoning that this lysine should be essential for kinase activity. We chose to replace the lysine with a methionine because
the methionine might not perturb the structure of this region very much but would inactivate i t as a kinase. We made a single base change in the cDNA to convert lysine to methionine at position 721 and then transfected this gene in a plasmid with the selectable marker, dihydrofolate reductase, into either B82 Lcells or CHO cells. Receptor expression was confirmed by Western blot analysis. K721 denotes the wild-type receptor with lysine at position 721 and a n active kinase domain. M721 has a methionine a t position 721 and is the kinase-deficient receptor. Both proteins are expressed as cell surface glycoproteins and both bind EGF. The differences in their kinase activities can be demonstrated by the ability of EGF to stimulate phosphorylation in intact cells. By using a n exogenous substrate we also demonstrate the M721 receptor has no EGF-dependent tyrosine kinase activity, whereas the holo K721 receptor is active in this assay. The M721 receptor that is expressed and glycosylated binds EGF normally but has no kinase activity. It has lost all biol-ogicalfunctions and its ability to mediate any responses to EGF. For example, Roger Tsien used FURA-2 to track free C a + + in individual cells. The results of these studies showed that when EGF was added to cells t h a t express the K721 (active kinase) EGF receptor, cytoplasmic Ca2+ rose. However, when EGF was added to cells bearing the M721 (inactive kinase) EGF receptor, there was no change in intracellular free Ca2+.These results showed that this very early response to EGF requires the kinase activity of the receptor. The mitogenic effect of EGF also required the kinase activity of the receptor. When cells expressing the mutant and holo-EGF receptors were grown in serum-free medium and treated with EGF, the only cells that increased their growth rate were the cells expressing the K721 holo-receptor. EGF halved the doubling time of these cells. The cells containing the M721, kinase-inactive receptor did not grow in response to EGF. Increased gene transcription in response to EGF also requires the protein kinase activity of the EGF receptor. To show this the firefly luciferase was used as a
Fig. 2. Conceptual model of activation of the EGF receptor. Ligand binding to a single site in the extracellular domain is proposed to cause a n allosteric change resulting in activation of the intracellular cytoplasmic protein tyrosine kinase. Self-phosphorylation of the regulatory C-terminus removes an alternate substrateicompetitive in-
hibitor permitting access of cellular substrates. This conformational change exposes sequences in the regulatory C-terminus that dictate interaction with coated pits for ligand-induced internalization and for coupling to mechanisms that increase cytosolic [Ca2 1.
reporter gene behind the fos enhancer. The fos gene responds to EGF with increased transcriptional activity. EGF increased the expression of this gene when it was introduced in cells that contain the K721 holoreceptor, but, in the cells that contain the mutant M721 receptor lacking kinase activity, there is no effect of EGF on gene transcription. Therefore elimination of the kinase activity by the lysine to methionine mutation abolished all responses to EGF, including early responses of increased calcium, intermediate responses of gene transcription, and the late proliferative response. Although kinase activity was expected to be required for these responses, the observation that removing the kinase activity also abolished the ability of the EGF receptor to be internalized and down-regulated was not quite so expected.
timately fuse with lysosomes. The receptor and the ligand are then degraded in lysosomes. Therefore, ligand-induced internalization results in loss of cell surface receptors. We found that the ability of the receptor to be internalized was also lost when the kinase activity was abolished. Cells that expressed the M721 or K721 receptors were treated with EGF and then assayed for residual EGF receptors on the surface by adding radioiodinated EGF after stripping away the receptor-bound EGF using acidic conditions. Pretreatment of cells expressing the K721 holo-receptor with EGF results in a decrease in the number of receptors on the surface; however, pretreatment with EGF did not change the number of receptors on the surface of cells expressing the M721 mutant receptors. Thus the mutant receptors bound EGF but did not undergo endocytosis. Western blot analysis indicated t h a t down-regulation of the EGF receptor actually represents a loss of mass not just a loss of surface exposure of the receptors. Whereas down-regulation occurs in the K721 holo-receptor transfected cells, the cells expressing the kinaseinactive, M721 receptor show no change in the mass of
TYROSINE KINASE ACTIVITY REQUIREMENT FOR EGF RECEPTOR INTERNALIZATION EGF receptors are normally diffusely distributed on the surface of the cell. Upon ligand binding they cluster in coated pits and are endocytosed in vesicles that ul-
+
REGULATION OF EGF RECEPTOR EXPRESSION AND FUNCTION EGF receptor after incubation with EGF. This same conclusion can be reached from immunocytochemical studies using antibodies against the EGF receptor. In the basal state, the wild-type EGF receptor is expressed diffusely on the transfected cell surfaces. Shortly after adding EGF, the cell surface has essentially been cleared of receptors and they have accumulated in large endocytic vesicles. At earlier times, the endocytic vesicles are smaller. When the kinase-inactive M721 mutant receptors are examined in expressed cells, they are found to be diffusely distributed on the cell surface similar to the holo-receptor. However, these mutant receptors are not cleared from the cell surface in response to EGF. Thus the metabolism of these receptors is blocked by removing the kinase activity. The question of whether the tyrosine protein kinase activity was required for receptor trafficking was also examined in some elegant studies by John Glenney who micro-injected a monoclonal antibody that recognizes phosphotyrosine into cells that expressed the kinase-active receptor. Cells that contained the monoclonal antibody were identified using a second antibody and another antibody was utilized to recognize the EGF receptor. The two antibodies were identified by rhodamine and fluorescein labels. In the cells that had been micro-injected with monoclonal antibody against phosphotyrosine the EGF receptor remained diffusely distributed on the cell surface instead of being endocytosed. By contrast, in uninjected cells the EGF receptor was endocytosed and localized in intracellular vesicles. The conclusion from these experiments is that removal of the tyrosine kinase activity by a n antibody or by mutational changes in the receptor not only abolished the biological effects of EGF but also abolished the metabolism and cell trafficking of the EGF receptor. These results raise the question of how signaling of all the complex cellular events occur through this cell surface receptor? If the kinase activity is required for endocytosis and internalization of the receptor, then removal of the kinase activity may abolish biological effects secondarily. Does the kinase-deficient mutant receptor not signal because it has not entered the cell where it needs to be located to function or is i t inactive because it has no kinase activity? Is the real answer some combination of the two possibilities? A secondary question concerns confirmational change of the receptor, which results in activation of the protein. Selfphosphorylation is part of the allosteric conformational change that results in a n active enzyme. If the receptor can not undergo self-phosphorylation, then it might fail to be endccytosed because either 1)i t failed to become phosphorylated, which is essential for endocytosis or 2) because it was not able to assume the correct conformation for the interactions necessary for endocytosis.
49
THE REGULATORY AND INTERNALIZATION DOMAINS OF THE EGF RECEPTOR One approach to answering these questions was to study the regulatory region at the C-terminus that contains the sites of self-phosphorylation. A series of mutations were made in the EGF receptor, which resulted in progressive truncations starting at the C-terminus. Each of the receptors was made in the context of either the kinase active (K721) or the kinase inactive (M721) receptor. Permanent cell lines were established with mutant receptors amplified using the dihydrofolate reductase gene. The number of receptors that the cells expressed were determined by Scatchard analysis of 1251EGF binding. Activity and expression level was assessed by Western blotting. Cells t h a t had been treated without or with EGF were lysed and displayed on a Western blot. Activity was measured with a n iodinated anti-phosphotyrosine antibody to identify cell substrates and a n anti-EGF receptor antibody was used to quantitate the EGF receptor. In the absence of EGF the holo-receptor showed no. activity. After addition of EGF tyrosine phosphorylated proteins could be detected. The EGF receptor was the primary phosphotyrosine-containing protein but some other phosphoproteins were also detected. Mobility of the receptor was retarded, characteristic of phosphorylated proteins. When a n EGF receptor that was truncated a t residue 1052 to remove the three known tyrosine self-phosphorylation sites was analyzed, i t was found to phosphorylate more cellular substrates in response to the holo-receptor. As expected it is a slightly smaller protein than the holo-receptor. A truncation was made at position 1022 inside the calpain hinge. Cleavage at the calpain hinge resulted in a shift to a 150 kDa receptor from the 170 kDa holo-receptor. As seen with the 1052 truncation, the C' 1022 receptor was quite active as a n EGF-stimulated tyrosine kinase. A mutant receptor that is truncated to residue 991 is also more active than the holo-enzyme in phosphorylating cellular substrates. The receptors with truncations to 973 and 957 are also quite active as EGF-dependent tyrosine kinases in vivo. The EGF receptor gene has 27 exons. The 957 position is located at the end of exon 22 a t a n exonintron junction. Further truncation into exon 22, to residue 944, produced a protein that was inactive. Therefore, the boundary of the EGF receptor kinase domain is somewhere near residue 957, i.e., near a n exon-intron junction. We measured down-regulation of these truncated cell surface receptors in the kinase active vs. inactive background. The holo-receptor was down-regulated, whereas the kinase inactive holo-receptor was not. Interestingly, whereas the kinase-inactive holo-receptor was not down-regulated, when it was truncated to residue 1022, inside the calpain hinge, it was partially down-regulated in response to EGF. This result sug-
50
G.N. GILL
gested that removal of the inhibitory tail, or a significant part of it, exposes sequences in the receptor that are involved in endocytosis and down-regulation of the receptor. Upon truncation to 991, kinase-inactive receptors are not internalized. This suggested that by incrementally truncating the molecule a region was exposed that allowed endocytosis but further truncation removed these sequences and created a receptor that could no longer be endocytosed and down-regulated. We were encouraged by these results because they showed that a kinase-inactive receptor could enter the cell and be down-regulated. This suggested that there was a domain in the cytoplasmic regulatory region of the receptor involved in ligand-dependent internalization. Removal of this region should create a kinse-active but internalization-deficient receptor. This mutant could be used to determine the biological effect of receptors that would remain at the cell surface. The boundaries of the internalization domain were determined by studies of various truncated forms of the receptor. The results showed that position 973 appeared to be the inner (C-terminal) boundary of the sequences responsible for down-regulation. The same results were obtained when the receptors were expressed in mouse 3T3 cells.
EFFECT OF MUTATIONS ON DOWN-REGULATION OF THE EGF RECEPTOR Steve Wiley has developed ways of determining the affinity of the receptor for coated vesicles and the endocytic apparatus. These methods allow one to measure endocytic rate constants for receptor-mediated internalization. Using these methods we demonstrated that cells that express the wild-type EGF receptor internalize the ligand as a function of the surface binding. After a brief lag the receptor enters with linear kinetics during which time the wild-type receptor is internalized a t a rate of about 35% of the occupied receptors per minute. The kinase-inactive mutant, which did not down-regulate, enters five- to seven-fold more slowly through what is essentially a constitutive pathway. This mutant does not decrease its receptor mass on the cell surface. For the C’ 1022 truncation, which is down regulated but has no kinase activity, the endocytic rate is intermediate (about 17% per minute). The C’ 1022 truncation is not endocytosed as well as the kinase active holo-receptor, but these results show clearly that removing some inhibitory C’-terminal residues exposes sequences that allow the receptor to internalize. These results are not a function of the host cell but a function of the receptor. The transferrin receptor was endocytosed a t the same rate in all of the transfected cells that carried different EGF receptors. Also, the uptake of iodinated polyvinyl pyrrolidine or iodinated plastic was invariant in these cells. The altered rates of endocytosis are thus not a function of the cell but are due to changes in the receptor.
Steve Wiley has also developed a method t o analyze the affinities of receptors for coated pits as a function of saturation of the receptors on the surface. The analysis is similar to a n Eadie-Hofstee or Scatchard analysis. Using this analysis, it was found that as the number of occupied receptors on the surface increased, the specific internalization rate diminished. This observation defines a saturable endocytic system for ligand-activated receptors, which has a definite capacity. The system is half occupied at about 50,000 receptors. Above 50,000 receptors per cell the system is saturated and ligand bound receptors enter through a constitutive pathway. The holo-receptor or the C’ 1022 deletion mutant show high receptor occupancy and are saturable. The kinase-inactive C‘ 1022-truncated receptor does not display saturable endocytic kinetics but instead a relatively straight line. The half-maximal capacity of this system is about 10 x lo6 receptors per cell. This higher capacity system is the constitutive uptake system. Receptors that did not down-regulate and that were truncated to remove the ligand-induced endocytic domain (e.g., the 973 or the 957 truncations) also show only constitutive endocytosis. In summary, when measured at low ligand concentrations where you measure the affinity of the system, the holo-receptor has the maximum endocytic rate. The affinity of the receptor is diminished by truncation to 1052 or by replacing the self-phosphorylated tyrosines with phenylalanines. The affinity rises when the truncation is a t 1022 and then i t begins to decrease when the truncation is at 991. When truncated to 973 the receptor is endocytosed near the basal constitutive rate. The receptor mutant that shows a different behavior from the rest is the C’ 1022 truncation, which has a much higher endocytic rate than expected from the rest of the series by the position of the truncation. It seems to be entering through the constitutive system but a t a much higher rate than the other receptors. The results of these studies show that the receptor’s structure determines its ability to interact with a saturable high-affinity endocytic system. Removing kinase activity abolishes the ability of the receptor to interact with the endocytic system. If kinase activity is retained but the C-terminus is truncated, the information that is necessary for the receptor to interact with the endocytic apparatus is lost. There is also a constitutive endocytic system and there are features of the receptor that determine its ability to interact with this endocytic system. This is shown by the difference between the rate of endocytosis of the C’ 1022 truncation and other truncated receptors that have lost the ability to interact with the high-affinity endocytic system.
STRUCTURAL ASPECTS OF THE RECEPTOR NECESSARY FOR BIOLOGICAL ACTIVITY With this background one can address the question of what structural and functional aspects of the EGF receptor are necessary for its biological activity. Mu-
REGULATION OF EGF RECEPTOR EXPRESSION AND FUNCTION tants were constructed that contained the domain defined as important for endocytosis (either the entire C-terminus from 944 to 1186 or the 944 to 1022 sequence) but did not contain the kinase domain. The resulting receptors were down-regulated in response to ligand. These mutants behaved similarly to the kinaseinactive receptor t h a t is truncated at position 1022 but that can still be endocytosed. By immunocytochemistry the mutant receptors were shown to enter cells. Biologically, there was no growth in response to stimulating endocytoses of kinase-inactive receptors. We tested the ability of the kinase-inactive, internalizing receptor to mediate induction of genes in response to EGF by using the luciferase reporter gene downstream from two different promoters that are inducible by EGF acting through the holo-receptor. The promoters were not induced in cells expressing endocytosiscompetent but kinase-inactive receptors. These receptors also do not mediate changes in intracellular Ca2+ These results show t h a t internalization of the kinase inactive receptor into the cell induces no biological response. Sequences on the receptor that mediate endocytosis can be transferred from one molecule to another. In contrast to the kinase-inactive receptors that are internalized, the kinase-active receptor that is not internalized is biologically active. It mediates EGF-induced gene transcription as measured by the activity of luciferase-linked promoters in transiently transfected cell lines. EGF induces gene expression equally well in cells that express the receptors that are not endocytosed but have kinase activity as in cells that express the holo-receptor. In fact, the endocytosis-defective but kinase-active receptor has a stronger effect on cell growth than the holo-receptor. 3T3 cells transfected with the holo-receptor grow into a monolayer in the presence of EGF. Cells that are transfected with the C’ 1022 truncation, which is a very active kinase and does not have the inhibitory domain, look a little different in the presence of EGF but still form a monolayer. Cells that have been transfected with the C’ 973 truncation appear transformed in the presence of EGF even at low cell densities. At high densities they lose contact-inhibition and form morphologically transformed foci. This is the expected result if the cells express a high number of EGF receptors. Cells are transformed by constitutively expressing TGF-alpha or EGF. Also, placing the EGF receptor behind a very strong promoter or amplifying the gene results in cell transformation. Here we demonstrate another mechanism for transformation. Failure to clear activated receptors from the cell surface also yields a transformed phenotype. From these results we deduce that the major signaling by the EGF receptor occurs at the cell surface and that the major effect of endocytosis and down-regulation is to appropriately attenuate the biological response to the ligand. If the receptor is not down-regulated then the cells display a transformed phenotype. We have not ruled out the possibility that there is a
51
small leakage in the system and that a n endocytosed kinase-active receptor may have some specific biological function. But overall, these results suggest that endocytosis functions to attenuate the signal delivered a t the cell surface.
CALCIUM INTERNALIZATION (CAIN) DOMAIN IN THE EGF RECEPTOR These mutants were used to evaluate the ability of EGF to increase intracellular calcium using indicator dyes. EGF increases intracellular calcium in cells expressing the holo-receptor and the receptor with the C’ 1022 truncation. The calcium response is somewhat decreased by truncation to 991 and is essentially abolished by truncation to 973. The ability to increase intracellular calcium is lost in all of the kinase-inactive receptors regardless of whether they are capable of being endocytosed. The cells transfected with the kinaseinactive receptors are still capable of responding to fetal calf serum with increases in intracellular calcium. Within the C-terminal kinase inhibitory domain is a domain that is required for endocytosis and for raising intracellular calcium, which we call the ca1cium:internalization (CAIN) domain (Fig. 3). The CAIN domain has two parts that are defined by the kinetics of response to EGF of the various mutants in internalization and in changing the intracellular free calcium levels. The region from 991 to 1022 was required for lowaffinity internalization and the part from 973 to 991 was required for high-affinity internalization and calcium regulation. When secondary structural predictions for the CAIN domain were made using the Chou Fasman and Garnier programs a predicted turn-helixt u r n motif with nine of the amino acids in the helix being acidic was identified. Such a turn-acid helix-turn motif is present in the related c-erbB2Ineu receptor. Although the primary amino acids are not the same, a similar CAIN domain is present, with variations, in the insulin receptor, the IGF-1 receptor, and with lesser certainty in other members of the family. Thus, it appears that the CAIN domain may be a defined structure in the C-terminus of the EGF receptor, which is responsible for the ability of the receptor to both undergo ligand-induced endocytosis and down-regulation and to raise intracellular free calcium. It also appears t h a t a t least some form of the CAIN domain may be present in the C-terminus of other members of the family of tyrosine kinase growth factor receptors. SUMMARY From the results of these studies of the activities of the various EGF receptor mutants we were able to disassociate the ability of EGF to increase intracellular calcium from its ability to induce genes and to cause morphological transformation and growth. These results lead us to the following concept. The kinase domain has a C-terminal border a t about residue 957. The remainder of the C-terminus is regulatory. The 164 amino acids from residue 1022 to 1186 constitute
52
G.N. GILL
A Kinase domain
"Caln" domain 957 973 991
Inhibitory domain
1022
RALMDEEDMDDVVDADEYLIPQQGFFSSPSTSRTPLLSSLSATSNNST
Required for
I
I
I
Required for low affinity internalization
high affinity
internalization & Ca'+ regulation
I
EGFreceptor
P S P
I
T
D S N F Y R A L M I , f s j M m V
V m A m Y L I P Q Q G F S S P S T S R
(9Y71
erb B-2/neu
(1 038)
I
A S P L D S T F Y R S L Lm
M
G O L
V mA
m
Y L V P Q Q G F
C
P D P A P G
1257 I
lnsulinreceptor
F P E
v s
F F H S
~
K
P H S ~ L O M B Fv ~P LMD sBs HNc Q R
A N
1252
F R E V S F Y Y S
E E 1293
I
IGF-lreceptor
1300 I
I
m
N K L P H P m L m L H P H N M H S V P L D S A S S S S L P
Fig. 3. Functional anatomy of the regulatory C-terminus of the EGF receptor: the CAIN domain. [Reprinted with permission from Chen et al., Cell 59:33-43, 1989, Cell Press, Cambridge, Massachusetts.]
a n inhibitory region for the kinase. It contributes to ligand-induced internalization because this is reduced in a mutant receptor truncated to residue 1052. Proximally within the C-terminus kinase inhibitory domain is a domain that is required for endocytosis and for raising intracellular calcium that we call the calcium internalization (CAIN) domain. In summary, we have found that the kinase activity of the EGF receptor is required for its function even when all of the self-phosphorylation sites have been removed. The EGF receptor has several distinct cytoplasmic domains that are important for its activity to regulate gene expression, DNA synthesis, and the intracellular calcium level. Biological signaling occurs from the cell surface via essential protein tyrosine kinase activity with ligand-induced internalization serving to abbrogate the biological signal.
QUESTIONS AND ANSWERS Q: Will the minus 973 deletion mutant grow in nude mice? A: We have not tested the mutant in nude mice but it will produce colonies in soft agar. Q: Do you see any differences in the amount of phosphotyrosine or the number of substrates that are phosphorylated by the various deletion mutants?
A: I cannot answer your question with absolute certainty but it is a very good question. If I can put it in context, the question is: If you remove the inhibitory domain, does the kinase become much more permissive in terms of the substrates that it will phosphorylate or the extent to which it will phosphorylate its substrates? We attempted to address this question by exposing the gels containing the phosphorylated receptor preparations for a long time to detect substrates other than the receptor itself. We did see a lot more bands than I have shown you but we were unable to determine from one-dimension gels whether they are all exactly the same in preparations from holo- and mutant receptors. My guess is that the answer is probably a little of both. We clearly see more phosphate incorporated into all of the protein substrates except the receptor itself in the extracts from cells expressing mutant receptors compared with the holo-enzyme. There are also probably some new substrates phosphorylated by the mutants that we do not see on the autoradiograms. Q: What is the function of the autophosphorylation sites? A: I think that the auto-phosphorylation sites have a rather classical function. There are many examples of kinases that are normally inhibited by a portion of
REGULATION OF EGF RECEPTOR EXPRESSION AND FUNCTION their own structure t h a t resembles a substrate. Selfphosphorylation is a way of removing this inhibitory domain. The cyclic AMP-dependent protein kinase is a good example of this. The regulatory subunit interacts with the catalytic unit and inhibits it through this kind of mechanism. I think i t is a mechanism that many kinases use. Q: Do you see a different phenotype in cells transfected with a receptor with single amino acid replacement a t the self-phosphorylation site in the C-terminal inhibitory domain? A: We certainly do not see a transformed phenotype. Endocytosis is interfered with a bit when you put a phenylalanine there. I think that when you replace the tyrosine with a phenylalanine the interaction with the C-terminus is not as tight as when tyrosine is present. The C-terminus does not make as good a fold or does not sit a s well in the site. But on the other hand it does not come out as well either. Thus, with this mutation the enzyme does not work quite normally but it is still capable of phosphorylating its substrates and initiating growth. Q: Are the sites of phosphorylation different in the mutants compared with the holo-receptor? A: No. There are five self-phosphorylation sites. They are all in the C-terminus and are all therefore regulatory. There is no phosphorylation in the kinase
53
domain. It is not like the insulin receptor where there is a phosphorylation site in the kinase domain, which is involved in its activation. This is more a classic regulatory domain. Q: Are early events other than calcium fluxes also not observed in response to the interaction of EGF with the receptors that are mutated in the CAIN domain? A: I cannot really say. My own prejudice is that the effects that we observed of the mutant receptors on transcription were accurately reflected by the reporter genes that we used. We have put different EGF-responsive promoters in front of the reporter gene and they all behave in the same way. I cannot answer the question with respect to the N a + / H + antiporter because the B82 cells did not lend themselves to those measurements. We have now put the receptors into some other cells that will perhaps allow us to ask that question. Q: What about the relationship between the CAIN domain and a n early event that is not membrane associated such as S6 phosphorylation? A: That is a good question. I do not know the answer. Q: Do you have a reagent, perhaps related to the inhibitory peptide that can be used to inhibit the function of the receptor? A: No, we do not. We are making the peptide; then we hope to use i t to make a n antibody.
