Stichting tegen Kanker (https://www.kanker.be) estimates that by the end of this year, approximately 18 million individuals worldwide will be confronted with a cancer diagnosis. The American Cancer Society expects more than 21.000 of those novel patients will be diagnosed with acute myeloid
leukemia (AML), the most common type of leukemia in adults. Although advances in therapy have resulted in significant improvements in outcomes for younger patients, the prognosis of the population with the highest incidence rate remains dramatic: almost 50% of all new patients will be over 65 years
old, and as much as 85% of them will die of their disease within 2 years of diagnosis.
Studies show that in 20-30% of all patients, intracellular mutations in FLT3 are at the root of the constitutive proliferation signal of the AML blasts. Such mutations are associated with an unfavorable prognosis in which a fast relapse rate after initial remission, is translated into an increased risk of death. Given that activating mutations in FLT3 are among the most common genetic lesions found in AML, considerable efforts have been made to specifically target this receptor in a clinical setting. Although administration of most of these tyrosine kinase inhibitors (TKIs) as monotherapies
often results in remission, all patients eventually relapse within a few weeks. Last year’s approval of midostaurin, the first TKI treatment of newly-diagnosed FLT3-mutated AML, added to a classical chemotherapy regimen, promised to mark a new era of targeted agents for the treatment of AML. The modest increase of the 4-year survival rate from 44% to 51%, illustrates the road ahead in decreasing AML related deaths.
FLT3 is a transmembrane protein featuring an extracellular ligand-binding domain (ECD), a single-pass transmembrane domain (TM) and an intracellular split kinase domain (TKD). This particular modular organization is characteristic for the Class-3 of receptor tyrosine kinases (RTK-III), a family harboring CSF-1R, KIT receptor, PDGFRα and PDGFRβ. Dormant RTK-III members reside at the cell membrane as
monomers, in which the juxtamembrane region maintains the inactive state of the kinase domain.
Activation of the receptor is initiated by binding of the ligand to the membrane-distal domains of 2 receptors. Although all RTK-III will eventually transition into an activated state by this event, its divergent molecular mechanism in doing so, profiles FLT3 as a notable exception. Indeed, structural and biophysical studies have shown that FLT3 is the only receptor for which no ligand-induced receptor-receptor contacts between the membrane-proximal domains could be detected.
Consequently, ligand-induced homotypic contacts are devoid from the FLT3 activation paradigm, which proclaims this ligand-receptor assembly as the odd man out.
Simultaneously, the molecular mechanism behind two oncogenic mutations in the ECD remains difficult to marry with this FLT3 activation paradigm, even with the crystallographic model of the ECD to our disposal. In analogy with activating point mutations in the ECD of KIT receptor, it was hypothesized that these mutations result in a ligand-independent receptor dimerization, nucleated by an increased reciprocal affinity of membrane-proximal domains – even though such interactions were conspicuously absent for wild-type receptor activation.
This doctoral dissertation is centered around FLT3 and oncogenic variants thereof. It starts with a brief introduction into hematopoiesis in Chapter A, detailing the role of FLT3 in healthy individuals. This is continued by a more detailed discussion of the FLT3 activation paradigm, explaining how ligandbinding at the extracellular regions is translated into an activation of the intracellular kinase activity.
This knowledge allows elucidation of the molecular principles behind transforming mutations in the TKD, and how such oncogenic mutations are addressed in a clinical context. As the observed issues accompanying the main strategy in the therapeutic targeting of FLT3 provide the rationale behind
presented experiments in Chapter C, we close the introduction with an overview of clinical results of FLT3-targeting drugs.
In Chapter B, we sought to clarify the apparent dichotomy in homotypic receptor contacts in the wildtype versus oncogenic context of FLT3. We approached this problem by rationally engineering two novel FLT3 ligands, which we envisioned to unmask elements of cooperativity in the assembly of the
extracellular FL-FLT3 complex.
Since results of ITC experiments featuring these two engineered ligands would be measured up against the thermodynamic parameters of wild-type ligand binding, we acknowledged the need for a full confidence that in-house produced FLWT reflects the effect of endogenous FL stimulation. We therefore assayed the bioactivity of recombinant FL in three orthogonal assays, showing that FL indeed potently mimics the effect of endogenous FL stimulation in vitro, ex vivo and in vivo.
We continue by explaining the applied strategy to engineer two novel FLT3 ligands, monomeric FLL27D and heterodimeric FLWT/KO. We show that both ligands after production and purification feature expected physicochemical properties, and confirm that they only recruit 1 receptor into complex
formation. Crystallization experiments featuring FLL27D, alone and in complex with the extracellular domain of FLT3, confirm that the structural adaptation in FLL27D is limited to the dimeric interface and does not introduce conformational changes in the receptor-binding epitope. Aside from validating
FLL27D as ideally suited for the envisioned analysis of cooperativity, these experiments provided us for the first time with a view on the receptor’s conformation in absence of an activating ligand. Arguably the most remarkable feature of this observed compact conformation is the interaction between the
most membrane-distal domain and most membrane-proximal domain. Considering this interaction as an extracellular autoinhibitory mechanism that prevents illegitimate receptor activation at the cell membrane, we endowed this tethered conformation with a prominent role in the proposed revision
of the activation mechanism.
Prior to the ITC experiments, we detail the efforts that have been put into the generation of a statistical framework that allows analysis and statistical comparison of all derived thermodynamic parameters. Finally, the subsequent set of ITC analysis show, for the first time, that assembly of ternary
ligand-receptor complex is a multi-step reaction featuring two levels of cooperativity. The first level is provided by an intramolecular reorientation of the ligand, whilst ligand-induced receptor interactions provide the second and largest cooperative increase in affinity.
The second section of Chapter B attempts to rationalize the oncogenic character of two clinically identified oncogenic FLT3 mutations in a structural-mechanistic framework. Unexpectedly, the threonine-343 to isoleucine point mutation does not seem to affect ligand-binding in the context of a
construct comprising only the extracellular domain of FLT3. It did however result in a higher-resolution X-ray crystallographic model of FLT3D1-D5, in which domain 5 could for the first time confidently be modeled into an electron density map. This model furthermore allowed mapping of the location of the
second oncogenic point mutation, a substitution of serine-451 to a phenylalanine. Comparison with structural models of KIT receptor and oncogenic variants thereof, fortified our hypothesis that such point mutation could indeed increase the reciprocal affinity of two FLT3 oncovariants.
Having both wild-type and monomeric FL at our disposal, we were finally able to perform a full thermodynamic dissection of ligand binding onto this oncogenic receptor variant. We can show that the presence of this mutation alters all thermodynamic parameters of the interaction with FLWT, but
does not seem to affect binding of the monomeric FLL27D. The latter observation, in combination with the observed reduction in entropic penalty upon binding of FLWT, confirms that Phe451 is buried in a reciprocal receptor interaction upon ligand binding. The detrimental effect on the enthalpic
component of the FLWT:FLT3S451F interaction, implies that this mutation alters an existing interaction site rather than creating a new one.
The impact of all these observations are the subject of the discussion section at the end of Chapter B, and are crystalized into a proposal for a novel FLT3 activation paradigm, including the autoinhibitory extracellular conformation and with a higher emphasis on the ligand-induced receptor interactions that reunite FLT3 mechanistically with the RTK-III family.
After this structural and biophysical dissection of extracellular FLT3 ligand binding, we switched gears in Chapter C and focused on the ligand-modulated membrane behavior of FLT3 in a cellular context.
Motivation for the presented experiments were given by two observations made elsewhere: the effect
of TKIs on the sub-cellular localization of oncogenic FLT3 variants on the one hand, and the observation
of a ligand-dependent TKI-resistance on the other.
Our preliminary flow cytometry results seem to confirm that membrane presentation of an oncogenic FLT3 variant, FLT3D835Y, only reaches wild-type levels in presence of a TKI. The inhibited state notwithstanding, stimulation with the activating ligand surprisingly elicits a strong internalization of
both the wild-type and oncogenic receptor. Providing arguments that internalization is a direct consequence of receptor activation, allows to infer that a constitutively activated receptor broadcasts a constitutive internalization signal that prevents translocation to the cell membrane. Consequently,
inhibiting the oncogenic autophosphorylation using TKIs, re-establishes normal cellular trafficking and restores membrane expression. Given that FL seems to outcompete TKIs for inhibition of the receptor, such increased membrane expression allows this oncogenic receptor to be activated by the presence of its extracellular ligand.
As debated in the discussion section of Chapter C, if correct, the impact of this hypothesis can arguably have far-reaching consequences for the current strategies in clinical targeting of FLT3 via TKIs.
Fortunately, we can simultaneously provide possible solutions, and make suggestions for strategies that allow addition of TKIs into chemotherapy regimen.