Background
Using exome sequencing, calreticulin (CALR) frameshift mutations were discovered in myeloproliferative neoplasms (MPN) and shown to be restricted to essential thrombocythemia (ET) and primary myelofibrosis (PMF) [
1,
2]. Both neoplasms involve striking abnormalities of megakaryocytes [
3,
4]. CALR, Janus kinase 2 (JAK2) V617F, and thrombopoietin receptor (MPL) mutations were mutually exclusive, suggesting that they all activate the JAK2-STAT signaling pathway to transform hematopoietic stem cells. While only exceptional cases of CALR-mutant polycythemia vera (PV) have been described in cases that were negative for JAK2V617F [
5], 15–24 % and 25–35 % of patients with ET and PMF, respectively, carry a CALR mutation [
3]. CALR frameshift mutations are classified according to the length of the somatic deletion or insertion in exon 9 of the
CALR gene. Until now, 36 types of CALR mutants have been observed in MPN [
2]. All of these mutants lead to a 1-bp frameshift and loss of the KDEL sequence and the original
CALR stop codon [
2]. The most frequent variants, the type 1 (c.1092_1143del) and type 2 (c.1154_1155insTTGTC) mutations, representing either a 52-bp deletion (p.L367fs*46; del52) or a 5-bp insertion (p.K385fs*47; ins5), respectively, account for approximately 80 % of all CALR mutations [
1,
2]. Type 1 and 2 CALR mutations have been shown to carry prognostic relevance [
6], but this was not found by all groups [
7].
CALR is a chaperone which is localized in the endoplasmic reticulum (ER) and exhibits an N-terminal ER-signal sequence, a N-, P-, and C-domain, and the ER retrieval sequence KDEL [
8]. CALR function regulates protein folding and quality control processes [
9]. Furthermore, CALR strongly affects calcium (Ca
2+) homeostasis in the ER/cytoplasm and thus Ca
2+-dependent signaling through its P-domain (low Ca
2+ capacity; high Ca
2+ affinity) and C-domain (high Ca
2+ capacity; low Ca
2+ affinity) [
8]. The modified C-terminus in CALR frameshift mutants comprises several additional triplets that were formerly part of the 3′UTR in wild-type (WT) CALR. Importantly, a large proportion of negatively charged amino acids in the C-domain of WT CALR converts into positively charged amino acids, abolishing proper Ca
2+-binding [
10].
While the function of CALR mutants in ET and PMF has remained unclear, recently, Marty et al. and Chachoua et al. have highlighted the necessity of the thrombopoietin (TPO) receptor MPL and its N-glycosylation to be essential for cellular transformation [
11,
12]. Marty et al. established a retroviral mouse model of del52 and ins5, closely reflecting an ET phenotype and, in the case of CALR del52, also the progression to myelofibrosis [
12]. Furthermore, two research groups have shown physical interaction of CALR mutants and MPL and the necessity of the positive electrostatic charge of the novel C-terminus for this interaction [
13,
14]. Araki et al. presented a model by which the P-domain in WT CALR blocks MPL interaction [
13]. This inhibitory function of the P-domain is abolished by the novel C-terminus in mutant CALR, thus enabling the N-domain to interact with the extracellular domain of MPL and leading to its dimerization and activation.
In the present study, we investigated the impact of CALR mutants on megakaryocytic transcription factors implicated in endogenous Mpl and CD41 expression. Moreover, we assessed CALR-mutant protein stability and secretion. We further confirmed MPL-dependence of CALR mutant-driven cell transformation and protection from apoptosis, as well as activation of critical signaling proteins including STAT5, STAT3, AKT, and ERK1/2. Collectively, our findings extend our understanding of CALR frameshift mutants’ cellular characteristics involved in pathogenesis and suggest that CALR mutants support megakaryocytic differentiation by MPL-dependent and MPL-independent mechanisms.
Methods
Patient samples and cDNA
RNA from patients carrying WT CALR or the ins5 mutant was isolated from the peripheral blood of MPN patients after written informed consent and ethics committee approval (EK2127/12). Complementary DNA (cDNA) from a patient with CALR del52 mutant was provided by Prof. S. Schnittger and Prof. T. Haferlach (Munich). The patient gave written informed consent to research studies, and the study was approved by the local ethics committee (05117) and adhered to the tenets of the Declaration of Helsinki. The wild-type and mutant CALR cDNA fragments used for vector cloning were obtained from patients’ RNA by reverse transcription polymerase chain reaction (RT-PCR) with random primers.
Reagents and antibodies
The proteasome inhibitor MG132, tunicamycin, and brefeldin A (BFA) were purchased from Sigma-Aldrich (St. Louis, MO, USA). Ruxolitinib (LC Labs, Woburn, MA, USA), spautin-1 (Selleckchem, Houston, TX, USA), and tunicamycin were dissolved in DMSO. BFA was dissolved in 100 % methanol. TransIT-LT1 (Mirus, Madison, WI, USA) was used to transfect HEK293T cells according to the manufacturer’s instructions. Antibodies used in our study included polyclonal rabbit anti-mouse/human phospho-STAT5 (Tyr694), polyclonal rabbit anti-mouse/human phospho-STAT3 (Tyr705), monoclonal rabbit anti-mouse/human phospho-AKT (Ser473) (193H12), polyclonal rabbit anti-mouse/human phospho-p44/42 MAPK (Erk1/2) (Thr202/Tyr204), polyclonal rabbit anti-mouse/human p44/42 MAPK (Erk1/2), monoclonal rabbit anti-mouse/human LC3B (3868s) and monoclonal rabbit anti-mouse/human STAT3 (D3Z2G), which were obtained from Cell Signaling/New England Biolabs (Frankfurt, Germany). The mouse monoclonal HA-probe antibody (sc-7392), polyclonal goat anti-mouse/human AKT1/2 (sc-1619), monoclonal mouse anti-mouse/human NF-E2 (sc-365083), monoclonal mouse anti-mouse/human GAPDH (sc-32233), and polyclonal rabbit anti-mouse/human DNMT3B antibody (sc-20704) were ordered from Santa Cruz Biotechnology (Santa Cruz, CA, USA). Monoclonal rabbit anti-mouse/human calreticulin antibody (EPR3924) from Merck Millipore (Darmstadt, Germany) was used for total calreticulin detection. The mouse monoclonal antibody CAL2 (Dianova, Hamburg, Germany) against the novel C-terminal of the CALR mutants was used for CALR mutant (CALR mut) detection. The mouse monoclonal anti-flag antibody M2 (F3165) was ordered from Sigma-Aldrich. The antibody against STAT5A/B was a kind gift from Richard Moriggl (Ludwig Boltzmann Institute for Cancer Research (LBI-CR), Vienna, Austria) and was originally generated by Eurogentec, Cologne, Germany. The CD41-APC-eFluor®780 (MWReg30) antibody and its isogenic control (eBRG1) were obtained from eBioscience (San Diego, CA, USA). Polyclonal goat anti-rabbit immunoglobulins/HRP (P0448), polyclonal goat anti-mouse immunoglobulins/HRP (P0447), and polyclonal rabbit anti-goat immunoglobulins/HRP (P0160) antibodies were purchased from DAKO (Hamburg, Germany).
DNA constructs and vectors
The flag-tagged cDNA of WT CALR, del52, and ins5 were cloned into the pMSCV-IRES-GFP vector and pcDNA5/FRT/TO vector (Life Technologies – Thermo Fisher Scientific, Paisely, UK) by Gateway cloning system (Life Technologies – Thermo Fisher Scientific). In case of the WT CALR, the KDEL sequence was cloned behind the flag-tag sequence. The cDNA of human MPL in the pMSCV-neo vector was kindly provided by Rebekka Schneider-Kramann. The C-terminally yellow fluorescent protein (YFP)-tagged cDNA of WT CALR, del52, and ins5 were cloned into the pMSCV-IRES-puromycin vector, and the KDEL sequence was cloned following the YFP-tag only in the WT CALR expression vector.
Cell culture and retroviral transduction
32D cells were cultured in RPMI 1640 medium (Life Technologies – Thermo Fisher Scientific) supplemented with 10 % fetal bovine serum (FBS), 25 U/ml penicillin/streptomycin (Life Technologies – Thermo Fisher Scientific) and 10 % Walter and Eliza Hall Institute (WEHI) supernatant as source of interleukin-3 (IL-3). HEK293T cells were cultured in Dulbecco’s Modified Eagle’s Medium (DMEM, Biochrom, Berlin, Germany) supplemented with 10 % FBS and 25 U/ml penicillin/streptomycin. The HL60e cell line stably expressing ecotropic receptor was a gift from Herbert Strobl (Vienna) generated by Bradley Fletcher (FL, USA). The retroviral transduction was performed as previously described [
15]. Briefly, Plat-E-packaging cells were transfected with pMSCV-IRES-GFP or pMSCV-IRES-neo/puro vectors containing the genes of interest and supernatants were collected after 24 and 48 h. Stable 32D cell lines were generated by three rounds of retroviral spin onto RetroNectin-coated (Takara Bio Europe/Clontech, France) six-well plates followed by selection by means of flow cytometry for GFP or G418/puromycin (InvivoGen, CA, USA) treatments.
Proliferation assay
32D cells were plated in triplicate at a density of 2 × 105 cells/ml and were cultured in WEHI-free RPMI medium. Living cells were manually counted every 24 h in a standard hemocytometer, excluding dead cells by trypan blue staining.
Immunophenotypic analysis
Staining of 32D CALR cells for the presence of membrane-localized CD41 was performed using a Gallios flow cytometer (Beckman Coulter, Krefeld, Germany). Data were evaluated using FlowJo data analysis software (OR, USA).
Preparation of cell lysates, SDS-PAGE, and immunoblotting
Cell lysates were produced with RIPA buffer containing 50 mM Tris pH 7.4, 150 mM sodium chloride, 1 mM EDTA, 1 % Triton-X, 15 % glycerol, 0.5 % sodium deoxycholate, and protease/phosphatase inhibitors. Denaturing of protein lysates was done at 65 °C for 5 min and separated by SDS-PAGE and transferred to polyvinylidene difluoride (PVDF) membrane (GE Healthcare, Frankfurt, Germany). Western blotting was performed overnight in Towbin transfer buffer (3 g Tris, 14.4 g glycine, 5 % methanol per liter ddH
2O) at 100 mA. Ten percent BSA in TBS-I buffer (20 mM Tris–HCl, pH 7.6, 137 mM NaCl, 0.05 % IGEPAL) was used for membrane blocking. The primary antibody (1:1000) was incubated overnight at 4 °C, and the secondary antibody conjugated to HRP (1:2000) for 45 min with three times washing in between. Proteins were detected via chemoluminescence (Fusion SL, PeqLab). ImageJ software was used for protein quantification analysis [
16].
Ubiquitination assay
HEK293T cells were transiently transfected with 3 μg pcDNA5 vector expressing CALR WT and CALR del52 together with or without 4 μg pcDNA3-ubi-HA vector using TransIT-LT1 reagent (Mirus, Madison, USA). After 24 h, the medium was discarded and replaced by fresh DMEM containing 10 μM MG132 for 20 h, and cell lysates were generated. For the immune precipitation (IP) assay, 40 μl Protein G Sepharose (GE Healthcare, Freiburg, Germany) was mixed with 2 μg anti-flag antibody for 4 h at 4 °C and subsequently incubated with 1 mg lysate in 1 ml RIPA buffer overnight, rocking at 4 °C with three times washing using cold PBS in between. The sepharose was then washed three times with RIPA buffer and resuspended in Laemmli/RIPA solution. After denaturation, bound proteins were separated and prepared for Western blotting.
MTT assay
32D cells were plated in triplicate at a density of 3 × 104 cells/well in a 96-well plate after washing twice with PBS. The JAK inhibitor ruxolitinib was added to the medium (max. vol. of 100 μl) at the concentration of 1 μM. Controls were treated with DMSO. Cells were cultured in WEHI-free RPMI medium. The measurement of cell viability was performed 48 h later using 10 μl 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) (5 mg/ml H2O) per well. The 96-well plate was incubated in the dark at room temperature for 4 h, and 100 μl isopropanol-HCl solution per well was added. Samples were analyzed with a microplate reader at a wavelength of 550 nm (Kayto, RT-2100C).
RT-qPCR
RNA isolation was performed by TRIzol/chloroform extraction (Trizol, Life Technologies, Darmstadt, Germany) using 6 × 106 cells. One microgram RNA was used for cDNA synthesis. Quantitative RT-PCR was performed using the 7500 Fast Real-time PCR System (Applied Biosystems by Life technologies, Paisley, UK) with the SYBR Select Master Mix for CFX (Applied Biosystems). The sequences of primers used for RT-qPCR were as follows: GCGTAACAAAGGCAGCAGAG (CALR for), CGTCGTCGTCCTTGTAGTC (CALR-flag rev), GTGAGTCCCCTAGCTTGCTG (mu c-Mpl for), TAGCAGGTGTGAACGACAGG (mu c-Mpl rev), CCAATACGGCCAAATCCG (mu Gapdh for), CCAATACGGCCAAATCC (mu Gapdh rev), ACAGGTGCCTGAAAGGTTGC (mu Nfe2 for), and ACCCTGCAGCTCAGTAATGG (mu Nfe2 rev). The mRNA expression level of the target gene is determined in percentage of Gapdh using 2^-ΔCT.
Concentration of 32D supernatant
32D MPL transduced cells expressing CALR WT or del52 were starved in FBS and WEHI-free RPMI medium at the density of 2 × 106 cells/ml after washing twice with PBS. After starvation for 6 h, the cell viability was analyzed by cell analyzer CASY TTC (OLS, OMNI Life Science, Bremen, Germany) and the supernatant was collected and filtered (0.45 μM filter) to the concentrator Vivaspin 20 (Sartorius Stedim Biotech, Goettingen, Germany) followed by spinning at the speed of 4000×g at 4 °C for 30 min. Around 3.5 or 12 μg protein of the supernatant was used for the analysis of paracrine signaling, respectively.
Statistical analysis
Statistical analysis was performed with the GraphPadPrism software using the two-tailed Student’s
t test. Significant differences were determined by *
P < .05, **
P < .01, and ***
P < .001; mean and standard deviation (SD) are indicated. The results are representative of at least three independent experiments.See Additional file
1 for supplementary material & methods.
Discussion
In this study, we demonstrate the capability of the most frequent CALR type 1 and type 2 frameshift mutants to induce an increase of endogenous
Mpl,
CD41, and
Nfe2 expression in 32D cells, suggesting that these mutants may prime hematopoietic stem cells for megakaryocytic differentiation. Endogenous MPL, CD41, and NF-E2 mRNA and protein expression was increased by CALR del52 even in the absence of ectopic human MPL expression as well as in spontaneously outgrown 32D del52 CALR cells (Figs.
2 and
3; Additional file
2: Figure S2d). Although we cannot fully exclude a CALR del52-independent effect in the outgrown cells, the fact that spontaneous outgrowth occurred in our own experiments as well as in those published by Klampfl et al. [
2] suggests that CALR del52 was responsible for the outgrowth of 32D cells. The increase of
Mpl expression and stabilization of CALR del52 protein in the outgrown cells suggests that cytokine independence and ruxolitinib sensitivity were induced by similar mechanisms as in the CALR mutant-transduced 32D cells ectopically overexpressing MPL. Using Sanger sequencing, we excluded activating mutation in the
Jak2 (e.g.,
V617F) or
c-mpl (e.g.,
W506L)
gene in the outgrown 32D cells. Nevertheless, we cannot exclude the possibility that cells with spontaneously higher MPL or NF-E2 levels were selected in the presence of the CALR mutants, as opposed to direct
Mpl and
Nfe2 induction by the CALR mutants.
There were differences between del52- and ins5-mutant expressing cells: MPL expression was inconsistent in CALR ins5 mutant expressing 32D cells (Fig.
3a and Additional file
2: Figure S2e), and GATA-1 mRNA levels did not correlate with protein except for 32D CALR ins5 cells (Additional file
2: Figure S2 c). In general, the effect of ins5 was less prominent than that of del52 except for CD41 upregulation. No outgrowth of 32D CALR ins5 cells was observed, potentially due to low protein expression. However, the weaker effect of ins5 mutants may also be due to negatively charged amino acids in the C-terminus which are missing in del52. Together, these differences may also at least in part explain clinical data that have associated ins5 mutants with a less severe phenotype than CALR del52 [
12].
Our data suggest that CALR del52 and ins5 increase megakaryocytic differentiation by the alteration of the transcriptional program, although we can only illustrate the initializing components of megakaryocytic differentiation in our model system. These data are in line with observations from Marty et al. in a retroviral mouse model of CALR mutants resulting in an ET and post-ET-MF phenotype [
12] and are supported by a recent study of Nivarthi et al. [
33]. It is known that megakaryocytes arise from bipotent megakaryocytic-erythroid progenitors (MEP) [
20]. Therefore, the reduction of
EpoR mRNA together with the increase of
Mpl that we observed in the outgrown 32D del52 cells suggests a differentiation shift from the erythrocytic to the megakaryocytic lineage.
How the CALR mutants del52 and ins5 led to the upregulation of megakaryocytic markers in 32D cells is not clear, but we detected an increase of AKT phosphorylation in del52 and ins5 expressing 32D cells in the absence of ectopic MPL expression (Fig.
4e). It was reported that active AKT positively influences megakaryopoiesis [
34,
35]. In addition, the knockout of the negative regulator of PI3K/AKT activity PTEN results in a significant increase of platelets [
36]. Therefore, active AKT could be an important mediator of CALR mutant-induced effects in our cell system and in hematopoietic stem cells/megakaryocytes.
CALR del52 and ins5 proteins showed low abundance, and we were able to exclude ubiquitin-triggered proteasomal degradation (Fig.
4b, c). In addition, our data suggests that autophagy or lysosome-dependent degradation [
37] were not involved (Fig.
4d). Furthermore, decreased protein stability was substantiated by our novel YFP fusion CALR mutants (Fig.
5c). Experiments from Chachoua et al. using cycloheximide had already suggested decreased protein stability [
11], which seems to be due to the presence of the new C-terminal domain [
17]. We demonstrated that degradation was partially antagonized by overexpression of MPL, suggesting a stabilizing effect of MPL on CALR mutants, together with reduced CALR mutant protein secretion due to strong MPL interaction, described here and by Araki et al. [
13]. This resulted in enhanced CALR mutant protein detection in cellular lysates (Fig.
4e; Additional file
4: Figure S4e). Similar stabilization has been observed in the outgrown 32D del52 cells (Fig.
1b).
Chachoua et al. had mutated all four N-glycosylation sites in the extracellular domain of MPL and shown strong reduction of STAT5 transcriptional activity in the presence of CALR mutants [
11]. Therefore, it is tempting to speculate that binding of CALR mutants to the glycosylated TPO receptor in the lumen of the ER, Golgi, and transport vesicles may trigger receptor dimerization and JAK activation. Importantly, and in line with this hypothesis, we have observed a significant increase of CALR mutant protein upon BFA treatment of 32D MPL cells, while no increase of WT CALR was detected. BFA blocks the anterograde transport from the ER to the Golgi apparatus, leading to the collapse of Golgi stacks and accumulation of proteins in the ER [
38,
39]. Consequently, it prevents protein secretion [
40]. Tunicamycin treatment had no effect on del52 detection, maybe due to tunicamycin-induced unfolded protein response triggering ER-associated protein degradation (ERAD) [
41].
We detected higher amounts of secreted CALR del52 mutant in the supernatant in comparison to whole cell lysates, which was blocked by BFA treatment (Fig.
5b). It is commonly accepted, albeit still a matter of debate, that WT CALR mediates cellular function from the cell surface and extracellular matrix, although the exit channels from the ER are still unknown [
9]. Importantly, ER depletion of Ca
2+ led to increased secretion and membrane localization of WT CALR [
42,
43]. The occurrence of mutant CALR alters calcium homeostasis [
10,
32], which could be one of the reasons for the observed increase of CALR mutant secretion. In addition, the KDEL sequence of WT CALR needs to be masked, negligible in CALR mutants. Although CALR mutants lack the ER retention signal KDEL, their localization is still primarily in the ER and Golgi apparatus, due to the ER-signal sequence in the N-terminus, where the MPL presumably gets activated [
1,
11]. Nevertheless, activation of glycosylated membrane-standing MPL by CALR mutants cannot be excluded.
A paracrine function of CALR mutants in neighboring cells was hypothesized, as discussed already for WT CALR [
9]. Nevertheless, in our experiments as well as in the work of Araki et al., no paracrine signaling of mutant CALR was detected [
13]. As this absence of paracrine signaling could be due to low CALR mutant concentration in the supernatant, generation of recombinant CALR-mutant protein would be of interest. In addition, in vivo secretion of CALR mutants into the extracellular matrix of bone marrow and spleen would provide further ways of action. Thus, intracellular trafficking as well as enhanced secretion may be the major reasons for the low protein availability of mutant CALR, most likely in combination with proteasome-independent CALR mutant degradation.
Competing interests
Steffen Koschmieder: Research funding (Novartis and Novartis Foundation). Consultancy and Advisory Boards (Ariad, AOP, Baxalta, Bristol-Myers Squibb, CTI, Novartis, Pfizer, Sanofi). Honoraria and Travel grants (Ariad, Alexion, AOP, Baxalta, Bristol-Myers Squibb, Celgene, CTI, Novartis, Pfizer, Sanofi, Shire). Tim H. Brümmendorf: Research funding (Novartis, Pfizer). Consultancy and Honoraria (Ariad, Bristol-Myers Squibb, Pfizer, Novartis). Susanne Isfort: Consultancy (Pfizer). Honoraria (Bristol-Myers Squibb, Pfizer). Travel grants (Novartis, Pfizer, Roche, Mundipharma, Hexal, Amgen). The remaining authors declare no conflict of interest.
Authors’ contributions
LH designed the research, performed the experiments, analyzed the data, and wrote the paper. CS and JK performed the experiments and analyzed the data. SI and TB contributed the research material and analyzed the data. MS analyzed the data and corrected the manuscript. NC and SK designed the research, analyzed the data, and corrected the manuscript. All authors approved the final version of the manuscript.