The platelet-derived growth factor receptor (PDGFR) family of receptor tyrosine kinases allows cells to communicate with the environment to regulate diverse cellular activities. Here, we highlight recent data investigating the structural makeup of individual PDGFRs upon activation, revealing the importance of the whole receptor in the propagation of extracellular ligand binding and dimerization. Furthermore, we review ongoing research demonstrating the significance of receptor internalization and signal attenuation in the regulation of PDGFR activity. Interactions with internalization machinery, signaling from endosomes, receptor degradation and receptor recycling are physiological means by which cells fine-tune PDGFR responses to growth factor stimulation. In this review, we discuss the biophysical, structural, in silico and biochemical data that have provided evidence for these mechanisms. We further highlight the commonalities and differences between PDGFRα and PDGFRβ signaling, revealing critical gaps in knowledge. In total, this review provides a conclusive summary on the state of the PDGFR field and underscores the need for novel techniques to fully elucidate the mechanisms of PDGFR activation, internalization and signal attenuation.
Receptor tyrosine kinases (RTKs) are a group of single-pass transmembrane proteins which respond to stimulation by growth factors, leading to tyrosine kinase activity, intracellular signaling and cellular activities such as migration, proliferation and differentiation. All RTKs exhibit a structural framework consisting of an extracellular ligand-binding domain, a single transmembrane domain and an intracellular domain harboring a catalytic tyrosine kinase [1,2]. Class III RTKs, including the platelet-derived growth factor receptors (PDGFRs), are defined by the combination of an extracellular region composed of five immunoglobulin-like domains (D1–D5) and a split intracellular catalytic tyrosine kinase domain (TK1–TK2) (Figure 1) [3,4]. The PDGFR family consists of two receptors: an α and a β receptor [5,6]. Upon extracellular binding by one of five dimeric, platelet-derived growth factor (PDGF) ligands, PDGF-AA, PDGF-BB, PDGF-CC, PDGF-DD or PDGF-AB, two PDGFRs dimerize, resulting in the formation of PDGFRα homodimers, PDGFRα/β heterodimers or PDGFRβ homodimers [7–9]. While ligand binding is necessary for PDGFR activation, the ligand–receptor complex is stabilized through direct contacts between the dimerized receptors (Figure 1) . Through methods not fully understood, dimerization of two PDGFRs propagates a conformational change through the transmembrane domains to the kinase domains to allow for trans-autophosphorylation of intracellular tyrosine residues (Figure 1) . This results in the recruitment of adaptor proteins and/or signaling molecules containing phosphotyrosine recognition motifs [1,2] and the subsequent activation of various downstream intracellular signaling pathways .
PDGFRβ homodimer structure and internalization.
PDGFRα and PDGFRβ have a low amino acid similarity of 31% in the extracellular domain, indicative of potentially unique ligand–receptor and receptor–receptor contacts . Conversely, PDGFRα and PDGFRβ have a high amino acid similarity of 85% and 75% in the N-terminal and C-terminal kinase domains, respectively . This suggests that the amino acid sequence composition of these receptors may result in both similar and distinct activation and signal refinement mechanisms.
Following activation, RTKs are internalized from the plasma membrane via endocytosis and are trafficked throughout the cell . Though receptors may be internalized through multiple methods, the most prominent and well studied for PDGFRs is clathrin-mediated endocytosis (CME), which occurs through clathrin-coated pits (Figure 1) . Internalized RTKs are then sorted throughout the cell’s endosomal processing system . Distinct intracellular amino acid sequences in the receptors or specific post-translational modifications provide cues as to the route the dimerized receptors will follow . Receptors are ultimately designated for degradation or, in some cases, recycled to the plasma membrane for additional signaling (Figure 1) [14–16].
In this review, we discuss various studies that have explored the processes of PDGFR activation, internalization and signal attenuation, highlighting potential similarities and differences between the two receptors. Unless otherwise noted, all studies cited focus on human PDGFRs. The data presented here provide a cohesive summary of the current state of the field, while illuminating the necessity for new, innovative tools and experimental approaches to further investigate the mechanism and function of signaling through the PDGFRs.
There remains many questions regarding the mechanisms by which PDGFR dimerization and activation occur. A barrier to understanding the activation of PDGFRα, in particular, is the relative lack of complete structural data. Some structural inference about the ligand-bound state of PDGFRα homodimers can be made, though, from recombinant protein data [17,18]. These studies revealed that PDGF-AA ligand binding predominantly occurs through contacts with PDGFRα D2 and D3, corresponding to data obtained using alternative methods .
Much more in-depth analyses have been performed on the structure of PDGFRβ. Crystallography, negative-stain electron microscopy (EM) and mutagenesis studies have all confirmed that PDGF-BB ligand binds to PDGFRβ homodimer D2–D3 in a symmetric manner, similar to what was observed for PDGF-AA binding to PDGFRα described above (Figure 1) [9,19,20]. EM further revealed that detergent-extracted and solubilized PDGFRβ remains monomeric even at a high density of PDGFRβ, fueling speculations that the primary purpose for ligand binding is to cross-link the receptors and increase their local concentration at the membrane [20,21]. While not directly involved in ligand–receptor contacts, D1 may confer some structural rigidity to the ligand-bound PDGFRβ homodimer by functioning as a ‘cap’ for the ligand-binding activities of D2–D3 (Figure 1) .
More C-terminal, D4 of both PDGFRα and PDGFRβ was previously believed to be the primary ‘dimerization domain’ that mediates receptor–receptor interactions critical for the stabilization of the ultimate ligand–receptor 2:2 complex . Monoclonal antibodies directed against D5 have shown that the inhibition of D5 on either PDGFRα or PDGFRβ did not impact receptor dimerization . However, recent EM data for PDGFRβ homodimers indicated that, in fact, there exists close homotypic receptor–receptor interactions in D5 as well as in D4 of PDGFRβ (Figure 1) . Furthermore, mutagenesis experiments demonstrated that two amino acid residues in PDGFRβ D4 that mediate D4 homotypic interactions and are conserved among Class III RTKs, Arg385 and Glu390, are, surprisingly, not required for receptor dimerization but are critical for receptor activation . This suggests that PDGFRβ extracellular domain-based dimerization alone is necessary but not sufficient for activation of the tyrosine kinase domains.
In total, despite the lack of in-depth structural data on PDGFRα, recombinant protein and monoclonal antibody data for PDGFRα are consistent with PDGFRβ crystallography and biochemical data demonstrating that PDGF ligand binds D2–D3 of the receptors and that dimerization occurs through D4–D5. Overall, these data suggest a relatively conserved mechanism of dimerization in the extracellular domain for the two PDGFRs.
Recent studies have looked beyond receptor interactions in the extracellular domain to further understand how ligand binding and dimerization propagate to the intracellular kinase domain. One such study used computer modeling to predict PDGFR dimeric states from transmembrane amino acid sequences . Surprisingly, when the sequence for the PDGFRα transmembrane domain was queried, the resulting computer simulations suggested the possibility for two transmembrane domain dimerization states . The first is predicted to occur in the absence of ligand, is inactive and includes near-parallel transmembrane domains. The second occurs in the presence of ligand and involves receptor–receptor contacts that lock the transmembrane domains in a right-handed conformation to transmit the signal to the intracellular tyrosine kinase domains and activate the receptors . These data, while intriguing, have not yet been confirmed beyond an in silico setting.
Corresponding to the second dimerization state, the negative-stain EM-resolved PDGFRβ homodimeric data also suggest potential dimerization between the transmembrane domains of the two receptors following ligand binding and extracellular domain dimerization (Figure 1) . The EM density in the 3D reconstruction of the PDGF-BB-bound PDGFRβ homodimer thins to a funnel shape C-terminal to D5 near the extracellular surface of the membrane and gradually widens through the membrane. These findings predict that the two transmembrane domains can be held close together at their N-termini, which, interestingly, is similar to the right-handed conformation suggested for PDGFRα homodimers [20,22]. These data align with a study investigating murine PDGFRβ using peptides containing only the transmembrane domain and part of the cytoplasmic juxtamembrane domain, which found that the transmembrane domains of two receptors can oligomerize via a leucine–zipper interaction . Additional NMR-based findings predict the presence of a transmembrane dimer between two PDGFRβ receptors . Importantly, PDGFRα and PDGFRβ have 48% amino acid similarity in this region, suggesting both similar and unique functions for this region between the two receptors . Taken together, these data propose a transmembrane helix:helix interaction in a PDGFRβ homodimeric complex (Figure 1) and potentially a PDGFRα homodimeric complex. This interaction could theoretically serve as a transducer of the joint signals of ligand binding and dimerization in the extracellular domains to kinase activation in the cytoplasmic domains of the receptors.
Autophosphorylation and activation
Following ligand binding and dimerization, autophosphorylation occurs on 11 and 13 conserved tyrosine residues within the intracellular domains of PDGFRα and PDGFRβ (Figure 1), respectively [10,25]. This autophosphorylation likely results from the close proximity of the two kinase domains following dimerization and/or a conformational change resulting in increased kinase activity . It has been suggested that a single amino acid residue in the cytoplasmic juxtamembrane domain of PDGFRβ, which is conserved in PDGFRα, may be critical to prevent activation of the receptor in the absence of ligand [26,27]. In fact, a crystal structure of the PDGFRα kinase domain suggests such a mechanism via extensive hydrophobic interactions between the juxtamembrane domain and the kinase fold . Furthermore, a glutamic acid/proline repeat motif in the C-terminal tail of PDGFRβ is required for an autoinhibitory role for the receptor in the absence of ligand binding (Figure 1) . However, it must be considered that there is low amino acid similarity between PDGFRα and PDGFRβ in the region C-terminal to TK2 (28%) , so if a similar autoinhibitory mechanism is at play for PDGFRα, it must involve a distinct motif .
The kinase activity mediated by the cytoplasmic domain is thought to be highly conserved, owing to the receptors’ high amino acid similarity in this region . Most of the structural data on this topic have been obtained through studies performed exclusively on PDGFRβ. Truncated PDGFRβ homodimeric intracellular domains have the ability to form a stable complex . Furthermore, 2D averaging of negative-stain EM PDGFRβ homodimer data reveals that the tyrosine kinase domain density exists as a flexible, asymmetric dimer . Together, these data confirm the concept of a PDGFRβ homodimerization state that propagates dimerization through the whole structure of the complex, ultimately resulting in the activation of the tyrosine kinase domains . It is important to consider, though, that while many of the aforementioned studies have garnered insight into the dynamics of individual PDGFRs and the homodimeric complexes they form, very little is known regarding the dimerization and activation dynamics of PDGFRα/β heterodimers.
Signals for internalization
PDGFR internalization following ligand binding and dimerization is critical to the regulation of PDGFR signaling. Before a dimeric complex embarks on its route within the interior of the cell, a suite of post-translational modifications of the receptors or motifs embedded in the amino acid sequence of the receptor complex must signal to the cell that it has been activated and is ready for internalization. One of these markers is the post-translational modification ubiquitylation, whereby Cbl, a really interesting new gene (RING)-finger E3 ubiquitin ligase, is responsible for ubiquitylating activated murine and human PDGFRs through binding to specific phosphotyrosine sites on the receptors [30–32]. This ubiquitylation acts as a signal to recruit endosomal machinery [14,33]. It was previously thought that PDGFR kinase activity was essential for receptor internalization . However, subsequent data probing the internalization of a kinase-dead PDGFRβ revealed that this is not the case . Surprisingly, these data further revealed that while dimerization is the driving force for internalization, additional amino acid motifs on the intracellular domain are equally important and also serve as a signal for internalization . One such motif, an ‘internalization motif’ of 14 amino acid residues containing a high number of hydrophobic residues, and in particular a di-leucine motif, is required for proper internalization of PDGFRβ (Figure 1), consistent with previous findings of decreased internalization of PDGFRβ lacking the majority of its intracellular domain [34–36]. This internalization motif is not widely conserved among RTKs and must be compatible between two receptors in a dimer . This is particularly intriguing in the context of PDGFRα/β heterodimers where it can be presumed that the internalization motifs of the two PDGFRs, which are not highly conserved, must themselves be ‘compatible’ for PDGFRα/β heterodimer internalization.
Routes of internalization
Another major consideration in the process of internalization is the route by which receptors are internalized. The most common route of PDGFR internalization is via CME (Figure 1) [13,38]. Adaptor proteins, such as AP-2, recognize and bind to ubiquitin molecules or di-leucine motifs on PDGFRs to recruit endocytic cargo, including clathrin . Subsequently, a large GTPase, dynamin, pinches off clathrin-coated vesicles containing PDGFRs from the membrane . Importantly, other routes of internalization, both clathrin-independent and dynamin-independent, have also been reported to occur under specific conditions (Figure 1). For example, low concentrations of PDGF-BB ligand result in murine PDGFR internalization via CME, whereas high concentrations result in a clathrin-independent route of endocytosis (CIE), specifically occurring through lipid raft domains in the membrane . These routes further result in distinct physiological outputs . Importantly, PDGFRs can be activated within or outside of lipid raft domains with equal likelihood . This, along with evidence that PDGFRs localize in a variety of membrane lipid domains in mammalian cells, suggests the possibility that both CME and CIE routes are utilized [42–45]. Correspondingly, experiments tracking PDGFRβ homodimer internalization revealed that internalization can occur via dynamin-dependent and dynamin-independent routes .
Additional studies have gone further to suggest other internalization routes dependent on actin dynamics and related effectors [46–48]. This exemplifies the complexity of PDGFR internalization. It is also important to note that the majority of these studies were performed using PDGFRβ, and much is presumed to be true for both receptors. However, it must be considered that the two receptors may have divergent internalization mechanisms.
Signaling and internalization
Biochemical data have provided some indication that activation of specific signaling pathways and PDGFR internalization are reliant on one another. In particular, studies have shown that phosphorylation at tyrosine residues that mediate the interaction of the receptor with phosphoinositide-3 kinase (PI3K) is both necessary and sufficient to internalize PDGFRβ . As another example, the activity of signal transducer and activator of transcription 3 (STAT3), which is activated via Janus kinase (JAK) activity downstream of PDGFRβ , is critically dependent on dynamin activity, implying some requirement of PDGFRβ internalization via a dynamin-dependent pathway for STAT3 signaling . Furthermore, recent data suggest that the ubiquitous dynamin protein, dynamin II, may have roles not only in receptor internalization but also PDGFRβ dimerization and activation . Importantly, though, these findings contradict many other studies, highlighting potential differences due to cell lines and/or culture conditions [39,47,51]. Overall, while strides have been made to uncover the mechanisms of PDGFR internalization, much remains unknown about how these processes proceed for each dimer and whether they differ between cell types and experimental conditions.
Sustained signaling from endosomes
PDGFRβ can continue to signal within endosomes through the recruitment of adaptor proteins, such as Shc and growth factor receptor-bound protein 2 (Grb2), and signaling molecules, such as PI3K and phospholipase Cγ (PLCγ) . These endosome-based signaling platforms result in the activation of downstream signaling pathways such as mitogen-activated protein kinase (MAPK), which was previously shown to be activated from late endosomes , and Akt, among others . Furthermore, biochemical studies have revealed that the full activation of STAT3 requires the intracellular accumulation of PDGFRβ, suggesting that PDGFRβ-mediated activation of STAT3 requires receptor signaling from endosomes . Together, these studies validate continued signaling from endosomes as a mechanism by which PDGFRs refine their signaling capabilities.
PDGFRs are believed to be predominantly degraded following receptor activation and internalization (Figure 1) [15,39]. While the dominant model, developed via studies on both murine and human PDGFRs, suggests that the receptors are degraded in the lysosome, there also exists evidence for proteasomal degradation [30,31,54–56]. A large component of this debate relies on the ubiquitylation of the receptors. One study claims that PDGFRs are monoubiquitylated, resulting in lysosomal degradation , and others point to data suggesting polyubiquitylation, a potential signal for proteasomal degradation [31,55]. Nonetheless, the end result remains the same: as shown for murine receptors, both PDGFRs are ubiquitylated and sorted for degradation, leading to signal attenuation [57,58]. However, the timing of signal attenuation remains an important consideration. Because PDGFRs can continue to signal from endosomes, the time frame from activation to degradation remains relatively undefined. Yang et al.  describe that after treatment with cycloheximide, half of human PDGFRβ is degraded within 90 min when stably expressed in mouse embryonic fibroblasts, aligning with similar endogenous human PDGFRβ expression studies . It is likely that the time frame for degradation is context-specific.
To date, only a handful of studies have demonstrated recycling of PDGFRs. However, some evidence for recycling of PDGFRβ under distinct conditions has been uncovered. Karlsson et al.  showed that the loss of T-cell protein tyrosine phosphatase (TC-PTP) induced receptor recycling for murine PDGFRβ homodimers and PDGFRα/β heterodimers, but had no effect on PDGFRα homodimers, upon PDGF-BB ligand treatment. Beyond demonstrating receptor recycling, these data provided the first indication of dimer-specific PDGFR trafficking outcomes dependent on ligand treatment . Additionally, the loss of TC-PTP led to hyperphosphorylation of tyrosine 1021 on murine PDGFRβ, a binding site for signaling molecule PLCγ, resulting in the activation of protein kinase Cα (PKCα) downstream signaling . When PKCα was inhibited in the absence of TC-PTP, murine PDGFRβ was no longer recycled, but instead degraded, indicating that PDGFRβ recycling is dependent on PKCα signaling downstream of PDGFRβ activation . Taken together, these findings demonstrate that there is a physiologically-relevant mechanism by which PDGFRs are sorted and degraded and/or recycled (Figure 1) .
As demonstrated by the studies presented here, each PDGFR exhibits unique qualities as well as commonalities. The activation mechanisms reviewed suggest the importance of the whole receptor molecule in the propagation of extracellular ligand binding and dimerization. While these studies confirm the structure and dimerization properties of the PDGFRs through multiple experimental approaches, much remains to be elucidated to corroborate these studies between both PDGFRs. For example, intriguing computer modeling suggests that PDGFRα may exist in an inactivated dimerization state, while biochemical studies showed that PDGFRβ remains monomeric even at high receptor density in the absence of ligand [20,22]. Furthermore, data suggest that the tyrosine kinase domains of PDGFRβ form an asymmetric dimer upon ligand binding and activation . Interestingly, both inactivated dimers and asymmetric tyrosine kinase domain dimers are states known to occur for the KIT receptor, another Class III RTK, and the epidermal growth factor receptor (EGFR), a member of the ErbB family of Class I RTKs [59–61]. Likewise, the findings that PDGFRs contain C-terminal hydrophobic motifs that signal for internalization and may be internalized through opposing routes dependent on ligand concentration are known mechanisms for EGFR [34,35,40,62,63]. To take this a step further, the first evidence of PDGFR recycling indicated that receptor degradation and recycling occur following receptor internalization, both of which are demonstrated outcomes for EGFR [15,16,63,64]. The studies on PDGFRs presented here have further demonstrated distinct modes of activation, internalization and signal attenuation for PDGFRs among Class III RTKs. It is particularly fascinating that multiple studies have begun to suggest the idea of potential dimerization of the transmembrane domain, which is previously unreported for any other Class III RTKs, but has been demonstrated for ErbB receptors [20,22–24,65,66]. Other intriguing qualities of PDGFRs reviewed here include their dominant use of CME over other routes , the distinct dependencies between their internalization and signaling activities [39,47,49,51] and their proposed PKCα-dependent mechanism for initiating receptor recycling [15,16], among others.
Despite advances in the field, it is clear that there remain many questions regarding PDGFRs. A lack of structural data has provided barriers to understanding the mechanism(s) of activation. Furthermore, the extent of dimer-specific differences in activation, internalization and signal attenuation for the various PDGFRs remains incompletely understood. Thus, novel techniques will likely be required to dissect these dynamics and to fully comprehend the role of PDGFR signaling in both homeostatic and disease settings.
Understanding the fundamental mechanisms by which PDGFRs signal is critical to gaining insight into how cells communicate with the environment to regulate behavior. This knowledge can provide perspective into wide-ranging developmental processes and disease states.
PDGFRs are activated by ligand binding and dimerization through various receptor–receptor contacts, resulting in the activation of intracellular signaling cascades. Subsequent receptor internalization and intracellular trafficking deliver receptors to the interior of the cell for degradation or, less frequently, result in recycling of the receptors back to the cell surface for continued signaling.
The majority of research on this topic focuses on PDGFR homodimers, with very little information regarding the activation, internalization and signal attenuation dynamics of PDGFRα/β heterodimers. The PDGFR field thus exists at a crossroads where new, innovative tools are essential to further the understanding of the emerging complexity of these receptor dimers.
epidermal growth factor receptor
growth factor receptor-bound protein 2
mitogen-activated protein kinase
platelet-derived growth factor
platelet-derived growth factor receptor
protein kinase Cα
really interesting new gene
receptor tyrosine kinase
signal transducer and activator of transcription 3
T-cell protein tyrosine phosphatase
M.A.R. and K.A.F. conceptualized the manuscript. M.A.R. wrote the original draft of the manuscript, which was revised in an iterative process with K.A.F.
Work in the Fantauzzo laboratory is supported by National Institutes of Health/National Institute of Dental and Craniofacial Research (NIH/NIDCR) grants R01DE027689 and K02DE028572 (K.A.F).
We thank our laboratory colleagues for their helpful discussions and comments on this manuscript. We apologize to authors whose work we were unable to cite due to space limitations.
The authors declare that there are no competing interests associated with the manuscript.