Inhibiting receptor tyrosine kinases has been a cornerstone of cancer therapeutics for decades. Treatment strategies largely involve small-molecule kinase inhibitors and monoclonal antibodies. For receptors activated by constitutively dimeric ligands, another potential mechanism of inhibition exists: developing monomeric ligands that prevent receptor dimerization. In a recent issue of the Biochemical Journal, Zur et al. [Biochem. J. (2017) 474, 2601–2617] describe the details of creating such an inhibitor directed toward the macrophage colony-stimulating factor receptor, c-FMS. In the process of teasing apart the ligand dimer, they also uncover a potential cryptic regulatory mechanism in this receptor subfamily.
c-FMS, perhaps better known as colony-stimulating factor 1 receptor (CSF1R) or macrophage colony-stimulating factor receptor (M-CSFR), is a receptor tyrosine kinase (RTK) in the PDGFR subfamily that includes platelet-derived growth factor receptor (PDGFR), Kit and Flt3 . This receptor family is characterized as having a series of extracellular immunoglobulin domains that bind to their cognate homodimeric ligands. Binding by the dimeric ligand effectively cross-links two receptors and leads to the activation of their intracellular kinases.
One can envision several potential ways of inhibiting activation of these receptors extracellularly: raising an antibody directed against the ligand or receptor , creating a soluble decoy receptor (ligand sink) [3,4], or designing a ligand that no longer stimulates the receptor but maintains binding to the receptor (dominant-negative). The authors sought to develop this latter type of inhibitor, utilizing the receptor's natural agonist (M-CSF) as a template to design a ligand that maintains receptor binding but is unable to self-associate and drive receptor dimerization. Beginning with a mutant that disrupts the covalent nature of the ligand dimer, M-CSFC31S, they determined that the three-dimensional crystal structure was unaltered from wild type — and even remained dimeric at high concentration. Then guided by energy calculations, the authors chose a secondary mutation M27R to sterically hinder the dimer interface without affecting the fold or receptor interactions. The resulting monomeric double-mutant M-CSFC31S/M27R retained receptor-binding affinity and was capable of inhibiting the activation induced by wild-type ligand.
Surprisingly, the M-CSFC31S single mutant was more active than the wild-type ligand in both phosphorylation and differentiation assays. The authors do not speculate on why the M-CSFC31S is more potent. So what mechanism might underlie these results? In addition to M-CSF, c-FMS has a recently identified alternate agonist, IL34 . IL34 has been shown to bind with higher affinity than M-CSF to c-FMS and to produce more transient and higher amplitude signals . Unlike M-CSF, IL34 does not have a covalent intermolecular disulfide bond between monomers, and one might predict that this leads to greater dimer flexibility . Conformational plasticity between ligand monomers has recently been proposed to be critical for M-CSF to engage and dimerize its receptor [8,9]. Might the enhanced activity seen for M-CSFC31S be a consequence of increased flexibility between monomers — functioning in a fashion more similar to IL34? Of note, other non-covalent ligand dimers in this RTK subfamily (e.g. stem cell factor (SCF) and flt3 ligand (FL)) seem to also have rapid kinetics and high amplitude signals , again consistent with the thought that an intermolecular disulfide restricts conformational plasticity and relatively dampens wild-type M-CSF agonist activity .
The authors clearly demonstrate that monomeric M-CSFC31S/M27R functions as the desired dominant-negative. But should the ultimate goal be simply to inhibit receptor activation? Or is there an opportunity for more controlled activation? Just recently, Garcia's group at Stanford has developed a weakened SCF dimer that acts as a partial agonist for its receptor c-Kit . The ability of controlled or partial agonism to lead to biased signaling in RTKs and cytokine receptors is just becoming apparent [12–14]. For SCF, a non-sterically disruptive mutation F63A was introduced to diminish the dimer interface. And like the M-CSFC31S mutant, SCFF63A dimerizes after each monomer engages its receptor — presumably only after high local concentrations in the plane of the membrane are achieved. Can a similar partial agonist be created for c-FMS? Or does wild-type M-CSF with its disulfide restraint represent this already?
In the course of designing an inhibitor to c-FMS, Zur et al.  seem to have uncovered key insights into the basic nature of this receptor's ligand controlled activation. The fact that removing a covalent bond between the ligand monomers is ‘enhancing’ rather than inhibitory is at first counter-intuitive. This result will help inform future endeavors in creating similar types of inhibitors and perhaps is even a new starting point for subtle and controlled activation in this family of RTKs. Indeed, if M-CSFC31S relieves conformational constraints leading to maximal signaling, and the M-CSFC31S/M27R double mutant completely abolishes dimerization, then perhaps different ‘Goldilocks-type’ double mutants of M-CSFC31S/M27X (where X produces less of a steric clash) could create analog control akin to SCFF63A — potentially balancing the activation in osteoclasts but decreasing chances for signal-dependent metastases. On the other hand, could an engineered disulfide in IL34 or SCF be an alternate strategy to create similar partial agonists? More research will be required to answer these questions — but when Zur et al. divided M-CSF in half, many intriguing possibilities were unexpectedly uncovered.
The Author declares that there are no competing interests associated with the manuscript.