MPL membrane domain sequencing goes deep

In this issue of Blood, Bridgford et al¹  present the first systematic approach to uncover all possible activating single point mutations within and around the transmembrane domain (TMD) of the thrombopoietin receptor (TpoR; also known as MPL). Several canonical activating MPL mutations (S505N and W515K/L/A/R) have been discovered through structure-function studies and patient sequencing,^2-6  but there remain “triple-negative” essential thrombocythemia and primary and secondary myelofibrosis cases, as well as hereditary thrombocytosis cases, without established driver mutations (ie, no JAK2, CALR, or MPL variants). The TpoR is a cytokine receptor that can be easily activated by dimerization in several different orientations,^7,8  which is unusual, but can explain the multitude of mutations that lead to an active or partially active phenotype. However, which specific variants can activate the receptor and drive disease?

Bridgford et al address this question with deep mutational scanning to evaluate all possible single amino acid substitutions in the human TpoR TMD for their ability to confer cytokine-independent growth in Ba/F3 cells. The authors undertook an incredibly efficient examination of all possible 19 amino acid residues (saturation mutagenesis) at each of the 29 positions of the target (TpoR) juxtamembrane-transmembrane region (residues 488 to 516). In all, 580 mutants were screened (29 positions, 20 codons each including the stop codon) representing the 19 remaining amino acids at each position. An enrichment was detected for the classical S505N and W515 mutations, but also for novel mutations (S493C/L, L498W, V501S, L502S/C, L508Q) (see Figure 1A in Bridgford et al).

This strategy was also used to answer another important question: whether the constitutive signaling of a canonical MPL S505N mutant could be enhanced by another mutation in cis. The deep sequencing approach applied to a template of MPL S505N instead of wild-type MPL led to the novel finding that a number of mutations in cis at H499 and other positions significantly augmented the constitutive signaling of the canonical activating MPL S505N mutation. The power of this saturation deep sequencing approach is impressive and has allowed the rapid identification and testing of novel double mutants.

The results highlight the specificity and conformational requirements for activating mutations. Generally, only 1 or 2 substitutions exerted an enhancing effect at a specific position. Most, but not all, of the single activating mutations were positioned on 1 helical face of the TMD, possibly using polar interactions to promote dimerization, although further studies are required to establish the precise structural requirements for activation. More than 90 different second-site mutations enhanced the activity of MPL S505N. Among these were novel but weak or nonactivating mutations at H499, G503, G509, and S493, and canonical activating mutations at W515 and other mutations around the W515.

Subtle differences were revealed whereby V501A was stronger at enhancing S505N, whereas V501S was more effective as a single mutation. Of great interest, although not developed extensively in this report, the deep sequencing approach on MPL S505N also uncovered second-site mutations that antagonize S505N activation. These data might offer leads to specific inhibitory strategies for MPL S505N.

Careful examination of databases containing published and unpublished data from exon 10 sequencing of 2452 MPN patients showed that some of the 7 novel identified variants and several double mutations had previously been detected in patients. These single and double mutants⁹  were created at the complementary DNA level and tested in Ba/F3 cells, where they confirmed an active phenotype.

Random mutagenesis around the TMD of the erythropoietin receptor coupled to functional selection in Ba/F3 cells has been accomplished previously, but without the next-generation sequencing before and after selection,¹⁰  which in this study allowed for testing of all possible mutants, with rigorous control of similar representation in the library, thus avoiding potential biases that might otherwise result. The deep mutagenesis and next-generation sequencing used here could be extended to any region of MPL and to other cytokine receptors, or indeed to any other signaling protein, either wild type or already mutated. The approach, in principle, allows selection for activators and inhibitors of signaling, provided selection is available. One limitation is that it does not control for defects in cell surface localization or expression of a particular mutant at the protein level. Nevertheless, it does offer a novel and powerful way to readily predict which mutations might lead to oncogenic activity. Last, but not least, the identification of a mutation in MPL in patients that allows constitutive signaling is very likely to mean the identification of a driver mutation, and such mutations need to be confirmed in functional assays. All in all, this work provides a ready reference for possible activating mutations in MPL’s TMD, which is likely to prove useful both for diagnostic purposes and for further understanding of the structural requirements that lead to MPL activation.