Defining the requirements for the pathogenic interaction between mutant calreticulin and MPL in MPN

Recurrent mutations in calreticulin (CALR), an endoplasmic reticulum (ER) resident chaperone protein, represent the second most common mutation in patients with myeloproliferative neoplasms (MPNs) after JAK2V617F.^1-4 CALR mutations in MPN occur as a heterogeneous set of indel mutations in exon 9 of CALR that all result in a +1 bp frameshift in the CALR reading frame.^5,6  Although the mutant CALR C-terminal alterations vary, all CALR mutations lead to a loss of most of the C-terminal acidic domain and a concomitant gain of a novel C terminus consisting of 36 amino acids that are enriched for positively charged residues. Mechanistic studies have demonstrated that mutant CALR (CALR^MUT) binds to the thrombopoietin receptor, MPL, to activate the MPL-JAK-STAT signaling axis,^7-10  and that the positive charge of the CALR^MUT C terminus is required to mediate this interaction.⁹ 

Several unanswered questions remain regarding the molecular and functional basis of CALR^MUT oncogenic activity. In this report, we use a series of mutagenesis experiments to address some of these questions, focusing on the structural determinants of CALR^MUT and MPL that are necessary for hematopoietic cell transformation.

Ba/F3 cells expressing CALR^MUT or MPL variants were generated by retroviral transduction and assayed for cytokine-independent growth as previously described.⁹ 

V5-tagged purified recombinant wild-type CALR (CALR^WT), CALR^MUT, CALR^MUT-D135L, and CALR^MUT-D317A proteins were incubated with purified recombinant ERp57 or MPL for 30 minutes at 4°C, and washed with N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid acetate buffer. Bound proteins were eluted from beads and analyzed by sodium dodecyl sulfate polyacrylamide gel electrophoresis.

To resolve which amino acids within the CALR^MUT C terminus are required for CALR^MUT activity, we generated CALR^MUT variants harboring serial truncations of the mutant C-terminal tail in blocks of 8 to 10 amino acids (Figure 1A) and tested their ability to associate with MPL in pull-down assays. All CALR^MUT variants examined retained the ability to bind MPL (Figure 1B). However, the most severely truncated form (CALR^MUTΔ36) failed to activate JAK-STAT signaling (Figure 1C), stimulate phosphorylation of MPL (Figure 1D), and transform Ba/F3-MPL cells to interleukin-3 independence (Figure 1E). Further truncation of an 11 amino acid positively charged stretch (QRTRRMMRTKM) that is also present in CALR^MUT protein species generated by the 52-bp deletion (CALR^MUTΔ47; see supplemental Figure 1A , available on the Blood Web site) led to loss of MPL binding (supplemental Figure 1B) and inability to transform Ba/F3-MPL cells (supplemental Figure 1C). A CALR^MUT variant in which these 11 residues are deleted but the distal 36 mutant-specific amino acids are retained (CALR^MUTΔ37-47) was still able to bind to MPL and transform Ba/F3-MPL cells (supplemental Figure 1A-C), suggesting that the QRTRRMMRTKM stretch is sufficient but not strictly required for MPL binding. Our data also suggest that although CALR^MUTΔ36 can bind to MPL, this binding is insufficient to activate MPL signaling. To our knowledge, these data provide the first evidence that physical interaction between CALR^MUT and MPL is not ipso facto sufficient to activate MPL. Rather, our data argue for a model whereby different thresholds of positive charge in the CALR^MUT C terminus are required to enable binding of CALR^MUT to MPL and to activate MPL signaling. Because MPL phosphorylation is dependent on homodimerization of single MPL chains, these data may suggest that CALR^MUTΔ36 retains the ability to interact with single MPL chains but is unable to induce homodimerization, which is required for receptor activation. Further studies are warranted to fully resolve the 3-dimensional structure of the CALR^MUT-MPL interaction.

We next sought to elucidate the regions of MPL that are essential to support CALR^MUT activity. We observed that CALR^MUT binds to full-length MPL and to the extracellular and transmembrane fragment of MPL, but not to the intracellular and transmembrane fragment (Figure 1F). Furthermore, an MPL variant in which the thrombopoietin (TPO) binding site is mutated (D235A/L239A)¹¹  was still able to bind to CALR^MUT (supplemental Figure 1D) and could support CALR^MUT-mediated cytokine-independent growth in Ba/F3 cells (supplemental Figure 1E). This suggests that CALR^MUT does not occupy the same binding pocket as TPO and is consistent with CALR^MUT-driven hematopoietic transformation being a TPO-independent process.⁷ 

Analysis of the intracellular portion of MPL reveals 3 tyrosine residues (Y591, Y626, and Y631) that may also be important for CALR^MUT-MPL signaling.¹²  MPL variants were therefore generated where these residues were systematically mutated to phenylalanine individually (FYY, YFY, YYF), in tandem (FFY, FYF, YFF), or altogether (FFF). As expected, all MPL variants were able to physically interact with CALR^MUT in FLAG pull-down assays (Figure 1G). However, we found differences in their ability to support CALR^MUT signaling. MPL variants harboring intact Y626 (MPL-FYY, MPL-YYF, MPL-FYF) supported robust Stat5 phosphorylation (Figure 1H) in association with cytokine-independent growth (Figure 1I), but not MPL variants in which the Y626 is mutated (MPL-YFY, MPL-YFF, MPL-FFY, MPL-FFF). These data indicate that Y626 plays a more prominent role than Y591 and Y631 in MPL signaling downstream of CALR^MUT, which is consistent with a previous report identifying Y626 as the major signaling tyrosine in canonical TPO-MPL signaling.¹²  Our data therefore highlight the importance of tyrosine-mediated MPL signaling as a pathway coopted by CALR^MUT to effect cytokine-independent proliferation. Our studies do not rule out a role for nontyrosine residues of MPL, which may be required for CALR^MUT-mediated activation of other downstream signaling pathways (eg, extracellular signal-regulated kinase)¹³ ; future studies to explore this question are warranted.

Finally, we sought to gain functional insights into how CALR^MUT interacts with MPL to confer cytokine-independent growth. Wild-type calreticulin is an ER-resident chaperone that interacts with glycoproteins by binding to Glc1Man9GlcNAc2 oligosaccharides and the polypeptide backbone to facilitate proper protein folding. We therefore created variants of CALR^MUT harboring mutations in critical residues implicated in 3 key functionalities of wild-type CALR: (1) polypeptide binding, (2) chaperone activity, and (3) lectin activity. We then tested their capacity to bind to MPL and confer cytokine independence (Figure 2A).

We observed that both polypeptide binding-deficient variants of CALR^MUT (CALR^MUT-P19K/V21E and CALR^MUT-W244G) and chaperone-deficient variants of CALR^MUT (CALR^MUT-H153G and CALR^MUT-EEDE)^14,15  retained MPL binding ability (Figure 2B) and conferred cytokine-independent growth (Figure 2C). Consistent with the nonessentiality of chaperone functionality in CALR^MUT oncogenic activity, CALR^WT exhibits strong, direct binding to the ERp57 cochaperone, whereas CALR^MUT does not (Figure 2D). In contrast, lectin-deficient CALR^MUT variants harboring mutations in Asp-135 and Asp-317 (CALR^MUT-D135L and CALR^MUT-D317A)^16,17  were both unable to bind to MPL or confer cytokine independence (Figure 2B-C). In accordance, we also found that only recombinant CALR^MUT protein directly binds to recombinant MPL in an in vitro binding assay, whereas neither CALR^WT nor lectin-deficient CALR^MUT do (Figure 2E). These data explain the previously reported essential role for Asp-135 in mediating CALR^MUT-driven STAT5 activation⁸  as being from a requirement for Asp-135 in mediating binding between CALR^MUT to MPL. Finally, to determine the requisite glycosylation status of MPL that enables CALR^MUT binding, we tested CALR^MUT binding to MPL mutants where either 2 (2xNQ = N117/178Q) or all 4 (4xNQ = N117/178/298/358Q) glycosylation sites in the extracellular domain of MPL were abolished. We found that CALR^MUT can still bind to MPL-2xNQ but not to MPL-4xNQ (Figure 2F). These data are consistent with a previous report that demonstrated that MPL variants devoid of the same 4 N-glycosylation sites failed to support STAT5 activation by CALR^MUT8  and suggests that this defect is due to an inability of unglycosylated MPL to bind CALR^MUT.

In conclusion, our data provide additional insights into the molecular mechanism by which CALR^MUT interacts with MPL to induce MPN (supplemental Table 1). Specifically, we (1) uncouple the binding of CALR^MUT to MPL from MPL activation, (2) define the key properties of MPL required for CALR^MUT binding and for its activation, and (3) decipher the key functionalities of CALR^MUT required for its oncogenic activity.

This work was supported by grants from the National Institutes of Health, National Heart, Lung, and Blood Institute (NHLBI; grant R01HL131835) (A.M.), Damon Runyon clinical investigator award (A.M.), the Starr Cancer Consortium (A.M.), Leukemia and Lymphoma Society (LLS) (A.M.), Wellcome Trust (E.C.), Academy of Medical Science Springboard Award (E.C.), Leuka John Goldman Fellowship (E.C.), and Gabrielle’s Angel Foundation for Cancer Research (G.B.). S.E. is a recipient of a T32 molecular hematology training award (NHLBI) and an LLS Special Fellow Award.

Contribution: S.E., N.S.A., E.C., and A.M. designed the study; G.B. provided guidance on experimental design; S.E., N.S.A., A.J.B., D.B., G.B., J.F.R., A.K., and N.F. performed experiments and collected the data; S.E., N.S.A., A.J.B., and D.B. analyzed the data; and S.E., N.S.A., E.C., and A.M. wrote the manuscript.

Conflict of interest disclosure: The authors declare no competing financial interests.

Correspondence: Ann Mullally, Harvard Institutes of Medicine Building, Room 738, 77 Ave Louis Pasteur, Boston, MA 02115; e-mail: amullally@partners.org; or Edwin Chen, School of Molecular and Cellular Biology, Faculty of Biological Sciences, University of Leeds, Woodhouse Ln, Leeds LS2 9JT, United Kingdom; e-mail: e.chen@leeds.ac.uk.