Hematopoietic expression of a chimeric murine-human CALR oncoprotein allows the assessment of anti-CALR antibody immunotherapies in vivo

Myeloproliferative neoplasms (MPNs) are characterized by a pathologic expansion of myeloid lineages. Mutations in JAK2 , CALR and MPL genes are known to be three prominent MPN disease drivers. Mutant CALR (mutCALR) is an oncoprotein that interacts with and activates the thrombopoietin receptor (MPL) and represents an attractive target for targeted therapy of CALR mutated MPN. We generated a transgenic murine model with conditional expression of the human mutant exon 9 (del52) from the murine endogenous Calr locus. These mice develop essential thrombocythemia like phenotype with marked thrombocytosis and megakaryocytosis. The disease exacerbates with age showing prominent signs of splenomegaly and anemia. The disease is transplantable and mutCALR stem cells show proliferative advantage when compared to wild type stem cells. Transcriptome profiling of hematopoietic stem cells revealed oncogenic and inflammatory gene expression signatures. To demonstrate the applicability of the transgenic animals for immunotherapy, we treated mice with monoclonal antibody raised against the human mutCALR. The antibody treatment lowered platelet and stem cell counts in mutant mice. Secretion of mutCALR did not constitute a significant antibody sink. This animal model not only recapitulates human MPN but also serves as a relevant model for testing immunotherapeutic strategies targeting epitopes of the human mutCALR.


| INTRODUCTION
The identification of CALR gene mutations has been a breakthrough towards understanding the molecular basis and diagnosis of patients with MPN. Type 1 (del52) and type 2 (ins5) mutations are the most recurrent mutations in MPN patients. 1,2 Mutant CALR (mutCALR) proteins induce ligand-independent activation of the thrombopoietin receptor (encoded by the MPL gene). [3][4][5][6][7] The mutCALR protein acts as a rogue chaperone and is known to stabilize the dimeric form of MPL. 8 Also, mutCALR cannot be recycled in endoplasmic reticulum (ER) as its wild type counterpart because of the absence of the KDEL residues at the C-terminus. Notably, mutCALR physically interacts Sarada Achyutuni and Harini Nivarthi contributed equally to this study. with the MPL receptor and is presented on the cell surface as a complex with MPL and it was also shown to be secreted into the extracellular milieu. 4,9 The mutant C-terminus of the protein serves as a bona fide neo-antigen, offering the possibility of developing immunotherapeutic strategies to target the oncoprotein. 10 The alternative reading frame used by the human mutCALR oncoprotein displays a 60-70% homology to the murine Calr exon 9 alternative reading frame. Thus, potential therapeutic antibodies raised against the mutant C-terminus of human mutCALR may not necessarily recognize the mouse frameshifted sequence. Therefore, we aimed to generate a chimeric murine-human transgenic model that can be used for studying immunotherapeutic interventions targeting human mutCALR. To preserve the epitope composition of the mutant C terminus, we conditionally knocked-in the human exon 9 carrying the del52 mutation into the endogenous locus of the mouse Calr gene. The resulting mouse-human CALR-del52 chimeric oncogene recapitulated the MPN phenotype in vivo. Here in, we provide a proof of concept that the mutCALR murine transgenic model can be used to assess the in vivo efficacy of immunotherapeutic interventions using antibodies or their derivatives.

| Generation of conditional knock-in CALR-del52 transgenic mice
The CALR-del52 transgenic mice were generated by homologous recombination of embryonic stem cells in C57BL/6 mice (Ozgene, Perth, Australia). The targeting vector consisted of two loxP sites flanking the murine endogenous exons 8 and 9 with a poly A sequence, a PGK-neomycin selection cassette (flanked by Flp recombinase target FRT sequences) and a cDNA sequence of murine exon 8 ligated with the human exon 9 del52 with the endogenous murine 3' UTR. Embryonic stem cells were transfected with the targeting construct, selected with neomycin and the cells with homologous recombination were identified by Southern blotting. The targeted stem cells were then microinjected into C57BL/6 blastocysts to obtain chimeras that were bred with C57BL/6 mice to generate germline transmitting CALR-del52 conditional knock-in mice. The PGK-neo cassette was deleted using the OzFlp: Ubic-Flpe mice ( Figure 1A). The conditional transgenic mice were bred with OZCre mice (PGK-cre in ROSA26 locus) to generate germline heterozygous CALR-del52 mice. The germline knock-in mice did not survive past embryogenesis in the homozygous state. The mice were bred with Vav-iCre transgenic mice 11 for hematopoietic specific expression of CALR-del52.

| Animal experiments
The mice were housed and maintained under standard conditions and all experiments were performed in accordance with Austrian law under the animal experimentation license number BMWFW-66.015/0004-WF/V/3b/2016. The genotyping primers are listed in Table S1. Blood was obtained from Vena facialis and at end point by heart puncture. Blood parameters were measured using the animal blood counter scil Vet abc. The sternum and a part of the spleen were processed for histopathology. Single cell suspensions were made from the rest of the spleen and the bone marrow from the femurs and tibia and analyzed by flow cytometry.

| Enzyme-linked immunosorbent assay (ELISA) for quantification of secreted mutCALR
To quantify secreted mutCALR (sCALR) in murine and human sera, two antibodies against mutCALR were usedpolyclonal antibodies against a 22-mer peptide derived from the mutCALR C-terminus in rabbits (capture antibody) and polyclonal antibodies against the recombinant CALR-ins5 protein in chicken (detection antibody). The rabbit anti-mutCALR specific polyclonal antibody (anti-mutCALRab -SAT601) was diluted in PBS at a concentration of 2 μg/mL and 50 μL was used to coat each well of a 96-well plate. The assay plates were incubated overnight at 4°C and later washed with wash buffer (0.05% Tween20 in PBS).
Plasma samples from vavCre, CALRdel52 fl/+ ;vavCre (mutCALR het ), CALRdel52 fl/fl ;vavCre (mutCALR hom ), CALR mutated patients and healthy control were diluted 1:4 in the blocking buffer (2% BSA, 0.05% Tween20 in 1xPBS) and were added to the plates. The samples were incubated for 2 h at 30°C and later washed with wash buffer. Chicken anti-mutCALR polyclonal antibody was added to all the wells at a concentration of 0.5 μg/mL for 1 h at 30°C. Following washing of the plates, secondary antibody anti-chicken IgY-HRP was used in the assay (1:50 000) for 1 h at room temperature. The plates were then washed and developed with TMB for 5 min at room temperature. The absorbance was read at 450 and 620 nm using Spectra-Max i3 (Molecular Devices). Recombinant CALRdel52 and CALRins5 proteins (MyeloPro) were used as a standard to calculate the concentration of the samples.

| Anti-mutCALR antibody treatment of mice
The mice (3n) were injected intraperitoneally with 3 mg/kg of mouse IgG2a anti-mutCALR monoclonal antibody twice a day, for 2.5 days.
The control group received an equal volume of the vehicle (PBS).
Blood was obtained from Vena facialis. The end point analysis was performed as described above. In the presence of Cre recombinase, the endogenous murine exon 9 of Calr is replaced by human del52 mutated exon 9 ( Figure 1A). This (murine Calr with human mutant C-terminal end), under the control of the endogenous murine Calr promoter. We generated the conditional CALR-del52 mice by breeding with vavCre transgenic animals to restrict the expression of CALR-del52 to the hematopoietic system.
To confirm the expression of the mutant CALR protein, Western blot analysis was performed with the whole cell extracts of splenocytes and probed using antibodies against total CALR and mutant CALR.
The expression level of mutant CALR were higher in mutCALR homozygous (mutCALR hom ) mice for the CALR-del52 transgene compared to mutCALR heterozygous (mutCALR het ) mice. The expression level of total CALR is significantly lower in the mutCALR hom mice compared to that in the wild type mice due to the increased degradation and secretion of the mutCALR protein ( Figure 1B). 9 We followed the peripheral blood values of a cohort of mutCALR het and mutCALR hom mice and vav-Cre control mice ( Figure 1C). Compared to wild type controls, mutCALR het mice had elevated platelet counts from 6 weeks of age without any significant change in total white blood count (WBC) and red blood count (RBC). However, in the mutCALR hom mice, the platelet counts were significantly higher. In older mice, leukocytosis and erythropenia were evident along with thrombocytosis. We analyzed the mice in more detail at 6 months and 1 year of age and made similar observations in the peripheral blood ( Figure S1A).
The mutCALR hom mice manifested splenomegaly at 6 months of age with high spleen to body weight ratio (2.25-fold) that increased at 1 year of age (3.3-fold). The mutCALR het mice developed mild splenomegaly only at 1 year (1.3-fold) ( Figure 1D,E). The total number of cells in the bone marrow (BM) was significantly reduced in 6 months (4-fold) and 1 year (6-fold) old mutCALR hom mice ( Figure S1C). We saw a significant increase in the percentage of megakaryocytes (Mks) in the bone marrow of mutCALR het (1.7-fold) and mutCALR hom (6-fold) mice at 6 months. The percentage of Mks increased further in mutCALR hom mice at 1 year (7-fold) ( Figure 1F). The percentage of hematopoietic stem cells (HSCs, LSK, lineage − sca1 + kit + ) was also significantly increased in the mutCALR hom mice at 6 months (2-fold), which further increased at 1 year (6-fold) of age. However, the mutCALR het mice showed increased LSK cells only at 1 year (1.7-fold) of age ( Figure 1G). In mutCALR hom mice this increase in LSK cells was Histopathological analysis of the bone marrow confirmed a myeloproliferative phenotype with prominent megakaryocytosis and megakaryocytic dyspoiesis in the bone marrow mutCALR hom mice ( Figure 1K, Figure S1B). These features were milder in the mutCALR het mice and in 6 months old mice. The one-year-old mutCALR hom mice had mild to moderate trabecular osteosclerosis and minimal to mild increase in reticulin positive fibers. Overt myelofibrosis was not a prominent feature in these mice ( Figure 1K).
In the spleen, the splenic red pulp was significantly expanded by the myeloproliferative process with marked megakaryocytosis and megakaryocytic dyspoeisis in 6 month-and 1 year-old mutCALR hom mice.
Lymphoid follicles were significantly attenuated in the spleens of these mice. In the mutCALR het mice, these features were much less prominent at 1 year of age, and insignificant to marginal at 6 months ( Figure S1H).
The conditional CALR-del52 mice were bred with Oz-Cre transgenic animals (which express Cre in germ line cells) to generate mice with germ line expression of CALR-del52 (CALR del52/+ ). Germline homozygous mice are not present in these litters suggesting embryonic lethality as previously reported. 12 Germline heterozygous mice are viable and develop a phenotype similar to the mutCALR het mice ( Figure S2). Taken together, the CALR-del52 transgenic animals develop a disease reminiscent of human essential thrombocythemia.

| CALR mutant stem cells have increased proliferation capacity in a competitive bone marrow transplant
We performed a competitive bone marrow transplant assay by trans-

| Treatment with anti-mutCALR monoclonal antibody reduces platelet and LSK counts
The mutant CALR protein represents a bona fide tumor antigen that is a target for immunotherapeutic strategies. We were able to detect the presence of the mutCALR protein on the surface of the total cells, lineage negative cells, LSK cells and megakaryocytes in the bone marrow of mutCALR hom mice, by FACS ( Figure 4A). Since, the expression of the protein was higher in the homozygous mice and the homozygous mice best represent the human MPN phenotype, we chose to use these mice for immunotherapy experiments. We treated the mice with a monoclonal antibody targeting the human mutant C-terminus.
The mutCALR hom mice were injected with anti-mutCALR mAb, intraperitoneally 3 mg/kg, twice a day for a period of 2.5 days ( Figure 4B). After treatment, only platelet counts were significantly tion persisted on day 2. We did not observe any effects of the treatment on blood parameters in wild type mice (Figures 4C, S6A).
At terminal workup, the antibody treated mice had a significant reduction in the percentage of LSKs and MPPs in the spleen of mutCALR hom mice. There was no change in the levels of ST-HSCs, and LT-HSCs and megakaryocytes ( Figure S6B). The mutCALR LSK cells in the bone-marrow were also significantly reduced in the antibody treated mice, while the other cell populations remained unchanged ( Figure 4D). In conclusion, anti-mutCALR antibody treatment of mice resulted in immunodepletion of platelets and hematopoietic stem cells in vivo in mutCALR-dependent manner.
Since shedding and secretion of antigen constitutes a potential antibody sink, we examined the soluble mutCALR (sCALR) levels in the plasma of transgenic mice. The mutCALR hom mice had a significantly increased amount of sCALR compared to the wild type and mutCALR het mice ( Figure 4E, ELISA setup in Figure S6C,D). This is consistent with the levels of sCALR detected in MPN patients with CALR mutations, using the same ELISA assay ( Figure 4F). Notably, the mutCALR hom mice treated with anti-mutCALR antibody have significantly reduced levels of sCALR in the plasma in comparison to PBS treated mice ( Figure 4G). We ruled out any competition between the two antibodies by performing a competitive ELISA (data not shown).
Therefore, immunological depletion of the antigen (by formation of antigen-antibody complexes) could be a possible reason for the reduction of ELISA signal. Accordingly, we were able to detect antigen-antibody complexes of the monoclonal antibody bound to sCALR in the plasma of the treated mice ( Figures 4H, S6D). These data indicate that sCALR was reduced in the plasma due to immunodepletion. The use of vavCre transgenic mice restricted the expression of the mutCALR protein to the hematopoietic tissue. We were able to detect the expression of the mutCALR protein in the splenocytes of transgenic mice. Notably, the mutCALR hom mice expressed higher levels of the CALR-del52 protein than the mutCALR het mice. Transgenic mice expressing human CALR-del52 under MHC-I promoter 13 or murine mutant CALR protein 12 have a relatively mild increase in platelets. However, transgenic mice expressing mouse-human hybrid CALR mutant proteins generated by" knock-in" strategy similar to ours, 14,15 developed a myeloproliferative disease which was more severe in homozygous mice. Similarly, the phenotype in the mutCALR hom mice was characterized by increased platelets and WBC and reduced RBC in the peripheral blood. The mutCALR hom mice also developed splenomegaly and increased percentage of megakaryocytes and stem cells in the bone marrow ( Figure 1). The LSK cells from mutCALR hom mice also showed proliferative advantage in a competitive bone marrow transplantation assay, as previously reported by Benlabiod et al. 15 Using transgenic mice with the expression of Cre in germ line tissue, we generated mice with systemic CALR-del52 expression (Calr del52/+ ). The mice do not survive past embryogenesis when they are homozygous for CALR-del52, similar to the Calr knockout mice and CRISPR/Cas9 knock-in mice. 12,16 To dissect the signature pattern associated with the mutant CALR, we performed transcriptome profiling of the LSK cells sorted from the mutCALR hom mice. The pathways revealed by GSEA, fall into four major categories: oncogenic signaling (K-ras signaling, mTORC1, P53 and JAK-STAT signaling pathways), cell cycle control (E2F targets, G2M checkpoint), inflammation (TNFα, IFNα and IFNγ signaling) and metabolism (bile acid, cholesterol homeostasis). Constitutive JAK-STAT signaling is one of the hallmarks of classical MPNs. We see an enrichment of genes that are known to be bona fide STAT5 targets. Major signaling networks are also known to feed into the JAK-STAT signaling. 17 The prominent ones that have emerged in the transcriptome data are categorized into oncogenic signaling list. One of the hallmarks of cancer is deregulation of cell cycle. Interestingly, we see that the components of G2M checkpoints and E2F targets are downregulated in mutCALR LSKs. The crucial role of CDK6 in stem cell quiescence, cytokine secretion and cell proliferation has already been demonstrated, in the context of JAKV617F. 18 It would be interesting to investigate the role of cell cycle control associated genes in CALR-positive MPN progression. A chronic state of inflammation is another hallmark of MPN. 19 NFKB is known to be the master regulator in promoting inflammation. We see that the components of TNFα signaling via NFKB, IFNα and IFNγ are upregulated in mutCALR LSKs.

| DISCUSSION
We also see that cholesterol homeostasis is significantly upregulated in mutCALR LSKs. As the ER luminal calcium plays an important role in sensing cholesterol, 20 the perturbation of cholesterol homeostasis may be due to the loss of calcium binding capacity in the del52-homozygous cells. Compared to the report of Prins et al. 21 we identified several other pathways in the GSEA analysis. This is likely because we performed transcriptome analysis from 1000 LSK cells rather than single cells. While single cell RNA-seq has the advantages of identifying transcriptional signature of individual cells, information of low-expressed genes can be lost in the process, which can be accessed in conventional transcriptome analysis.
MutCALR associates with the thrombopoietin receptor and is trafficked onto the cell surface as a complex. 4 We were able detect the presence of the mutCALR protein on the surface of the target cells in the bone marrow of mutCALR mice. The cell-surface residing protein can act as a neoantigen and, therefore, be targeted by antibody-based immunotherapy. To test this hypothesis, we administered a murine anti-mutCALR IgG2a raised against the human oncoprotein to the mutCALR hom mice. During antibody treatment, we observed a short-term platelet immunodepletion in the mutCALR hom mice. At the end of treatment, we detected a significant reduction in the percentage of LSK cells in both the spleen and the bone marrow, in the monoclonal antibody treated mutCALR mice. The control vavCre mice were not affected by the antibody treatment, demonstrating the specificity of the antibody against mutCALR. Interestingly, the platelet values were replenished within the 18-h gap between two injections. The amount of antibody used here may not be sufficient to combat the very high rate at which the mutCALR megakaryocytes shed platelets.
Secretion of mutCALR has been shown in cell culture supernatants. 4,9 However, the biologic effects of the secreted neoantigen has not yet been investigated in mice or patients. 22 To test if the transgenic mice could be used to model the pathophysiologic role of secreted mutCALR, we measured the soluble mutCALR (sCALR) in the transgenic mice and in MPN patients. The plasma levels in patients with CALR mutations and in mutCALR mice were comparable suggesting that the mutCALR mice also recapitulate this aspect of the MPN phenotype.
The sCALR in plasma may act as a "sink" for the anti-mutCALR monoclonal antibody and lead to dampening of the antibody therapeutic effects. Although we detected the presence of immunocomplexes of the anti-mutCALR antibody and the sCALR in the plasma of treated mice, this did not prevent the immunodepletion of LSK cells in both spleen and bone marrow ( Figure 4). These data suggest that the secreted mutCALR and the abundant platelets in the peripheral blood do not constitute a significant antibody sink. The mechanisms of how the antibody treatment leads to platelet and LSK cell depletion is unclear at the moment and further studies are needed to clarify the mode of action.
In summary, we have generated transgenic mice that recapitulate the MPN phenotype and provide proof of concept that an antibody raised against the mutant C-terminus can be effectively used to target the platelets and mutant LSK cells. The mutCALR mice serve as a modeling platform to study the molecular mechanisms and immunotherapeutic strategies to target the disease.