Molecular and Functional Analysis of Calcium Binding by a Cancer-linked Calreticulin Mutant

Reviewer #1 (Public review):

The authors attempted to compare calcium calcium-binding properties of wildtype calreticulin with calreticulin deletion mutant (CRTDel52) associated with myeloproliferative neoplasms.

The researchers conducted their study using advanced techniques. They found almost no difference in calcium binding between the two proteins and observed no impact on calcium signaling, specifically store-operated calcium entry (SOCE). The study also noted an increase in ER luminal calcium-binding chaperone proteins. Surprisingly, the authors selected flow cytometry as a technique for measurements of ER luminal calcium. Considering the limitations of this approach, it would be better to use alternative approaches. This is particularly important as previous reports, using cells from MPN patients, indicate reduced ER luminal calcium and effects on SOCE (Blood, 2020). This issue matters because earlier research with MPN patient cells reported reduced ER luminal calcium levels and altered SOCE (Blood, 2020). How do the authors explain the difference between their results and previous findings about lower ER luminal calcium and changed SOCE in MPN patient cells expressing CRTDel52? Other studies have found that unfolded protein responses are activated in MPN cells with CRTDel52 calreticulin (see Blood, 2021), and increased UPR could account for higher levels of some ER-resident calcium-binding proteins observed here. Overall, it remains unclear how this work improves our understanding of MPN or clarifies calreticulin's role in MPN pathophysiology.

Reviewer #2 (Public review):

Summary:

Tagoe and colleagues present a thorough analysis of the calcium (Ca2+) binding capacity of calreticulin (CRT), an endoplasmic reticulum (ER) Ca2+-buffer protein, using a mutant version (CRT del52) found in myeloproliferative neoplasms (MPNs). The authors use purified human CRT protein variants, CRT-KO cell lines, and an MPN cell line to elucidate the differing Ca2+ dynamics, both on the level of the protein and on cell-wide Ca2+-governed processes. In sum, the authors provide new insights into CRT that can be applied to both normal and malignant cell biology.

First, the authors purify CRT protein and perform isothermal titration calorimetry to quantify the Ca2+ binding capacity of CRT. They use full-length human CRT, CRT del52, and two truncations of CRT (1-339 and 1-351, the former of which should lead to the entire loss of low-affinity Ca2+ binding). While CRT del52 has previously been shown to lead to a decrease in Ca2+ binding affinity in other models, the ITC data show that this is retained in CRT del52.

Next, the authors utilize a CRT-KO cell line with subsequent addition of CRT protein variants to validate these findings with flow cytometric analysis. Cells were transfected with a ratiometric ER Ca2+ probe, and fluorescence indicates that CRT del52 is unable to restore basal ER Ca2+ levels to the same extent as CRT wild-type. To translate these findings to MPNs, the authors perform CRT-KO in a megakaryocytic cell line, where reconstitution with either CRT variant did not cause a difference in cytosolic calcium levels. The authors further test store-operated calcium entry (SOCE), an important process for maintaining ER Ca2+ levels, in these cells, and find that CRT-KO cells have lower SOCE activity, and that this can be slightly recovered with CRT addition.

Finally, the authors ask whether other effects of CRT-KO/reconstitution can affect the cellular Ca2+ signaling pathway and levels. RNASeq analysis revealed that CRT-KO leads to an increase in various chaperone protein expressions, and that reconstitution with CRT del52 is unable to reduce expression to the same extent as reconstitution with CRT wildtype.

Strengths:

The authors provide new insights into CRT that can be applied to both normal and malignant cell biology.

Weaknesses:

(1) The authors should consider discussing the high-affinity Ca2+ binding site more in the introduction. Can they show a proof-of-concept experiment that validates that incubation of recombinant CRT reduces the function of that high-affinity Ca2+ binding site?

(2) For Figure 2B, do you have an explanation for why the purified proteins run higher than predicted (48-52kDa) - are these proteins still tagged with pGB1?

(3) The MEG-01 cell line has the BCR::ABL1 translocation, while CRT mutations are strictly found in BCR::ABL1 negative MPNs. Could these experiments be repeated in these cells treated with imatinib to decrease these effects, or see if basal MEG-01 Ca2+ levels/activity are changed with or without imatinib?

Author response:

Public Reviews:

Reviewer #1 (Public review):

The researchers conducted their study using advanced techniques. They found almost no difference in calcium binding between the two proteins and observed no impact on calcium signaling, specifically store-operated calcium entry (SOCE). The study also noted an increase in ER luminal calcium-binding chaperone proteins. Surprisingly, the authors selected flow cytometry as a technique for measurements of ER luminal calcium. Considering the limitations of this approach, it would be better to use alternative approaches.

The flow cytometric assay shows good responsiveness to conditions expected to alter ER calcium levels (Figure 4C), is high throughput compared to microscopy, and allows for averaging of signals across a large number of cells. This was thus our original method of choice.

This is particularly important as previous reports, using cells from MPN patients, indicate reduced ER luminal calcium and effects on SOCE (Blood, 2020). This issue matters because earlier research with MPN patient cells reported reduced ER luminal calcium levels and altered SOCE (Blood, 2020). How do the authors explain the difference between their results and previous findings about lower ER luminal calcium and changed SOCE in MPN patient cells expressing CRTDel52?

We thank the reviewer for asking for these clarifications. The referenced study (Di Buduo et al. Blood, 135(2):133-143, 2020) first showed that thrombopoietin induces spontaneous cytosolic calcium spikes in cultured megakaryocytes, which is dependent on store operated calcium entry (SOCE). In parallel, STIM1-ORAI interactions were induced by thrombopoietin. On the other hand, the addition of thrombopoietin caused the dissociation of STIM1-calreticulin interactions, based on proximity ligation assays. The implication is that signaling via the thrombopoietin receptor (TPOR/MPL) activation induces the dissociation of calreticulin-STIM1 complexes, and the formation of STIM1-ORAI complexes, which contribute to the measured spontaneous cytosolic calcium spikes. Different MPN mutations induced spontaneous calcium spikes in a thrombopoietin-independent manner, including the JAK2V617F mutations and the CALR type I and type II mutations. The study found that the number of megakaryocytes exhibiting spontaneous calcium spikes was enhanced in the context of both type I and type II CALR mutations compared to the JAK2V617F mutant. Correspondingly, the calreticulin-STIM1 interactions/cell were more significantly reduced for type I and type II CALR mutations compared to the JAK2V617F mutant. It was suggested that defective interactions between mutant calreticulin, ERp57, and STIM1 activated SOCE and generated spontaneous cytosolic calcium spikes. However, based on the findings with thrombopoietin, the spontaneous calcium spikes could simply result from thrombopoietin-independent MPL activation by the mutant calreticulin and JAK2V617F and downstream signaling. Importantly, the referenced studies did not directly measure ER luminal calcium. A number of undefined factors could account for the measured differences between the megakaryocytes from patients with calreticulin mutations vs. JAK2V617F. These include the relative mutant allele burdens, the extent of MPL activation, as well as genetic differences unrelated to calreticulin. Different from these experiments, through the use of purified proteins, our studies show that the Del52 mutant has calcium binding characteristics resembling that of the wild type protein. Additionally, through genetic manipulations in cell lines, our studies directly address the effects of calreticulin KO and its Del52 mutation upon ER luminal and cytosolic calcium levels, and cellular SOCE signals. We did not measure significant differences in any of these parameters between the KO cells and those reconstituted with wild type calreticulin or the Del52 mutant. As noted by the editors, these results show that Ca2+ binding by calreticulin and store-operated Ca2+ entry in a cell are not fundamentally impacted by the type I deletion mutation. On the other hand, in primary megakaryocytes, when co-expressed with MPL, the Del52 mutant, through its known ability to bind and activate TPOR/MPL, is expected to induce SOCE and calcium fluxes similar to those induced by thrombopoietin. These points will be clarified in the revised discussion.

Other studies have found that unfolded protein responses are activated in MPN cells with CRTDel52 calreticulin (see Blood, 2021), and increased UPR could account for higher levels of some ER-resident calcium-binding proteins observed here.

Multiple studies have suggested the induction of the unfolded protein response (UPR) in cells expressing MPN mutants of calreticulin. We don’t know the specific signals that cause the upregulation of various calcium binding proteins in calreticulin-KO cells and cells expressing the Del52 mutant. Indeed, these could result from increased protein misfolding in cells with wild type calreticulin deficiency. Alternatively, the sensing of cellular calcium perturbations could induce their expression. Regardless of the precise mechanisms underlying the expression changes in calcium binding proteins, the upregulated factors are predicted to compensate for calreticulin deficiency and contribute to the maintenance of the overall cellular calcium homeostasis. These points will be clarified in the revised discussion.

Overall, it remains unclear how this work improves our understanding of MPN or clarifies calreticulin's role in MPN pathophysiology.

The points discussed above as well as their implications for the understanding of calreticulin’s role in MPN pathophysiology will be clarified in the revised manuscript.

Reviewer #2 (Public review):

Tagoe and colleagues present a thorough analysis of the calcium (Ca2+) binding capacity of calreticulin (CRT), an endoplasmic reticulum (ER) Ca2+-buffer protein, using a mutant version (CRT del52) found in myeloproliferative neoplasms (MPNs). The authors use purified human CRT protein variants, CRT-KO cell lines, and an MPN cell line to elucidate the differing Ca2+ dynamics, both on the level of the protein and on cell-wide Ca2+-governed processes. In sum, the authors provide new insights into CRT that can be applied to both normal and malignant cell biology.

First, the authors purify CRT protein and perform isothermal titration calorimetry to quantify the Ca2+ binding capacity of CRT. They use full-length human CRT, CRT del52, and two truncations of CRT (1-339 and 1-351, the former of which should lead to the entire loss of low-affinity Ca2+ binding). While CRT del52 has previously been shown to lead to a decrease in Ca2+ binding affinity in other models, the ITC data show that this is retained in CRT del52.

Next, the authors utilize a CRT-KO cell line with subsequent addition of CRT protein variants to validate these findings with flow cytometric analysis. Cells were transfected with a ratiometric ER Ca2+ probe, and fluorescence indicates that CRT del52 is unable to restore basal ER Ca2+ levels to the same extent as CRT wild-type. To translate these findings to MPNs, the authors perform CRT-KO in a megakaryocytic cell line, where reconstitution with either CRT variant did not cause a difference in cytosolic calcium levels. The authors further test store-operated calcium entry (SOCE), an important process for maintaining ER Ca2+ levels, in these cells, and find that CRT-KO cells have lower SOCE activity, and that this can be slightly recovered with CRT addition.

Finally, the authors ask whether other effects of CRT-KO/reconstitution can affect the cellular Ca2+ signaling pathway and levels. RNASeq analysis revealed that CRT-KO leads to an increase in various chaperone protein expressions, and that reconstitution with CRT del52 is unable to reduce expression to the same extent as reconstitution with CRT wildtype.

Strengths:

The authors provide new insights into CRT that can be applied to both normal and malignant cell biology.

We thank the reviewer for the recognition that this study is important for our understanding of both normal and malignant cell biology.

Weaknesses:

(1) The authors should consider discussing the high-affinity Ca2+ binding site more in the introduction. Can they show a proof-of-concept experiment that validates that incubation of recombinant CRT reduces the function of that high-affinity Ca2+ binding site?

In a previous study (Wijeyesakere et al. 2011 J. Biol Chem, 286 8771-8785), we showed that at a starting calcium concentration of 0 mM and with 3.3 mM injections of CaCl₂, the measured K_D value was 16.6 mM for calcium binding to wild type murine calreticulin, (which has ~95% % sequence identity with human calreticulin), corresponding to the high affinity site. On the other hand, at a starting calcium concentration of 50 mM and with 33 mM CaCl₂ injections, the measured K_D value for calcium binding to wild type murine calreticulin was 590 mM (corresponding to the low affinity sites). Thus, we did not measure the high affinity sites when the starting calcium concentration was 50 mM. This point will be clarified in the revised manuscript.

(2) For Figure 2B, do you have an explanation for why the purified proteins run higher than predicted (48-52kDa) - are these proteins still tagged with pGB1?

Yes, the purified proteins shown in Figure 2B retained a GB1 tag. This point will be clarified in the revised manuscript.

(3) The MEG-01 cell line has the BCR:ABL1 translocation, while CRT mutations are strictly found in BCR:ABL1 negative MPNs. Could these experiments be repeated in these cells treated with imatinib to decrease these effects, or see if basal MEG-01 Ca2+ levels/activity are changed with or without imatinib?

Thank you for the important point. We will assess cytosolic calcium levels in MEG-01 cells with or without imatinib.