Philadelphia-like B cell acute lymphoblastic leukemia (B cell ALL) is a somewhat unique entity. Most B cell ALL subtypes are molecularly defined by a distinct structural variant, such as the ETV6::RUNX1 fusion (resulting from the t(12;21) translocation) that is generally associated with standard risk B cell ALL. In contrast, Philadelphia-like (Ph-like) B cell ALL is characterised by a gene expression profile that closely resembles Philadelphia-positive (BCR::ABL1 positive) B cell ALL, but without the BCR::ABL1 gene fusion. Unlike Ph-positive B cell ALL, Ph-like B cell ALL is not characterized by a single driver event, but by several possible structural variants (commonly fusions) that have a common feature, the constitutive activation of receptor tyrosine kinase (RTK) signalling.
In almost all cases, Ph-like aberrations conform to a recurrent structure. DNA breaks fuse together two separate genes into a new, in-frame product. The 5’ fusion partner typically consists of the first few exons of a gene that is highly expressed in the B cell lineage, and which encodes an oligomerization domain. This is in-frame with the tyrosine kinase domain of an RTK. Thus, the paradigm is that the oligomerization domains from the N-terminal partner drive dimerization of the chimeric fusion protein, transphosphorylation of the kinase domains and ligand-independent RTK signalling.
A very unusual RTK “fusion”
In our letter to Leukemia Journal (https://rdcu.be/c53dO) we report a novel variant from a paediatric patient with Ph-like ALL. This fusion activates signalling from the kinase domain of Platelet-derived Growth Factor beta (PDGFRB) but does so in a surprising way. The first clue was that this CD74::PDGFRB fusion, sequenced and cloned from the patient, had no apparent open reading frame to link CD74 through to the kinase domain of PDGFRB. Yet somehow, in cells from the patient we could detect a PDGFRB protein species. Expression of the CD74::PDGFRB fusion cDNA in progenitor B cell line models further showed that this protein was active, and sufficient to drive transformation. Importantly, the cells expressing CD74::PDGFRB were highly sensitive to the tyrosine kinase inhibitors (TKIs) imatinib and dasatinib. To understand how a functional protein could be produced from the ‘out-of-frame’ CD74::PDGFRB sequence, we performed sequential deletion and mutation of possible translation start sites within the fusion. This revealed that the translation of this driver event started from within PDGFRB itself, and not from the fusion partner, CD74.
This left us with the intriguing question as to how this “loose” kinase domain is constitutively activated in the absence of a 5’ partner. AlphaFold2 structural modelling showed that the juxtamembrane domain of PDGFRB which normally occupies the ATP binding site, resulting in a negative regulatory loop, is lost in the CD74::PDGFRB fusion. As a result, the usual requirement for a protein-protein interaction domain to drive enforced dimerization and overcome auto-inhibition, is bypassed. Using cytosolic recombinant protein generated from the full length PDGFRB and the truncated PDGFRB resulting from the CD74::PDGFRB fusion, we confirmed that truncated PDGFRB protein has almost double in vitro kinase activity to PDGFRB containing the intact juxtamembrane domain.
Why this is important
With the introduction of kinase inhibitor drugs such as imatinib and dasatinib, Ph-like B cell ALL has a far better clinical outlook than was once the case. Nonetheless, this is highly reliant on rapid and accurate identification of targetable driver lesions. In our study, we have functionally dissected a fusion variant that was predicted from sequence analyses to be out-of-frame but was in fact the targetable driver lesion in this cancer. This raises the possibility, that similar fusions may be overlooked in routine sequencing-based fusion identification pipelines. This case also encourages an expanded view as to the type of structural variants that constitute Ph-like B cell ALL.
There are other important implications. We have described an example that involves PDGFRB, but might there be unknown cases that involve other members of the PDGF receptor family: Platelet Derived Growth Factor Receptor alpha (PDGFRA), KIT Proto-Oncogene, Colony Stimulating Factor 1 Receptor (CSF1R) or Fms-like Receptor Tyrosine Kinase 3 (FLT3)? Can this mechanism of kinase activation drive other cancer types? The more tumors that are sequenced, the better able we will be answer such questions. Ultimately, the goal must be for patients who can benefit from TKIs to receive these drugs in a timely fashion.
Figure 1. Unusual PDGFRB fusion reveals novel mechanism of kinase activation in Ph-like B-ALL. (A) The EBF1-PDGFRB is representative of a typical PDGFRB fusion driving Ph-like B-ALL. The in-frame structural variant in the ALL cells gives rise to a chimeric RNA transcript and then a protein which contains part of EBF1 encoding a protein-protein interaction domain (PPID) linked to the kinase domain of PDGFRB. The PPID in the N-terminal partner drives fusion dimerization, kinase transphosphorylation and activation. (B) We report a novel unusual PDGFRB fusion between CD74 and PDGFRB. The CD74::PDGFRB transcript is not in-frame however, the translation of the resulting oncogene begins within PDGFRB itself at a cryptic ATG site (ATG2). The result is high expression of a ‘loose’ kinase domain. In this scenario, PDGFRB is constitutively activated by the structural loss of the juxtamembrane (JM) region (autoinhibitory), as indicated by the cartoon model. In EBF1::PDGFRB and like fusions, the autoinhibitory nature of the JM domain is overcome by enforced dimerization driven by the PPID of the N-terminal fusion partner.