Accumulation of the GSK3 target protein β-catenin is lethal for B cell precursors and malignant B cells

Glycogen synthase kinase 3 (GSK3) is a ubiquitously expressed kinase involved in a myriad of biological processes. Although GSK3 mediated phosphorylation has been shown to induce the degradation of many pro-survival and pro-proliferation factors, cancer cells of different origin show reduced proliferation or survival after GSK3 inhibition. Our current understanding of the role GSK3 plays in normal mature B cells, B cell precursors and transformed B cells is incomplete and does not allow to assess whether GSK3 inhibitors can be used to treat B cell derived malignancies. Here we identify β-catenin as the major factor driving GSK3-inhibition induced changes in B cells. We show that β-catenin accumulation has opposing effects on cell metabolism and survival in mature B cells and B cell precursors. Moreover, we demonstrate that β-catenin destabilizes the commitment to the B cell lineage. In summary, our study identifies β-catenin induced signaling as a factor that can be exploited to limit the survival of malignant B cells.


44
GSK3 is a serine/threonine kinase expressed in α and β isoforms in most 45 mammalian tissues. More than 100 substrates are known to be phosphorylated by 46 GSK3, many of which play important roles in regulating cell metabolism, 47 proliferation and survival (Sutherland, 2011). GSK3 is an atypical kinase in that it is 48 constitutively active in resting cells and disabled upon mitogenic stimulation. In 49 many cell types GSK3 inhibits proliferation by targeting pro--survival and pro--50 proliferation factors such as cMyc (Gregory et al., 2003) or β--catenin (Doble et al.,51 9 β--catenin can fulfill two separate functions in cells, it can act as a transcriptional co--245 activator and as a coordinator of cell--cell adhesion (Valenta et al., 2012). To confirm 246 that β--catenin induces changes in the transcriptional profile and to assess which 247 gene clusters are affected by β--catenin stabilization in B cell precursors, we 248 performed transcriptome analysis of B220--positive control and bCat LoxEx3  expected considering that GSK3 has many other targets beyond β--catenin. However, 266 when we analyzed the set of genes that were significantly upregulated in bCat LoxEx3 B 267 cell precursors when compared to control cells and the set of genes significantly 268 upregulated in LY2090314 treated B cell precursors when compared to untreated 269 cells, we found an overlap of 433 genes (Supplementary table 1 (Cobaleda et al., 2007) and Blimp1 is a transcription factor known to drive plasma 278 cell differentiation in mature B cells, but cell death in B cell precursors (Setz et al., 279 2018 cells contained only few B220--positive cells (Fig.6B). These cultures also contained 286 a second population of B220--low cells that was not seen in the WT sample. Both 287 B220--low and B220--positive cells from the bCat LoxEx3 culture contained only slightly 288 higher Pax5 levels than B220--negative cells (Fig.6C). These results thus confirm that 289 Pax5 expression is reduced in response to β--catenin accumulation. Moreover both 290 B220--positive and B220--low bCat LoxEx3 cells expressed higher Blimp1 protein levels 291 than B220--positive wildtype cells (Fig.6D). To test whether β--catenin accumulation 292 can disrupt the gene expression profile defining B cell identity in already established 293 B cell precursors, we treated wildtype CD19+ B220+ B cell precursors overnight 294 with LY2090314 and analyzed Pax5 and Blimp1 expression. Strikingly, we found 295 Pax5 expression to be reduced and Blimp1 expression to be increased after GSK3 296 inhibition (Fig.6E,F). Similar to normal B cell precursors we found Blimp1 297 expression to increase upon GSK3 inhibition in transformed B cell precursors 298 ( Fig.6G) Blimp1 has been reported to induce cell death of B cell precursors (Setz et 299 al., 2018). To test whether Blimp1 is primarily responsible for inducing apoptosis 300 after GSK3 inhibition, we treated control and Prdm1--deficient B cell precursors with 301 LY2090314. GSK3 inhibition induced cell death in both WT and Prdm1--deficient B 302 cell precursors suggesting that β--catenin has other detrimental effects on cell 303 survival in addition to driving Blimp1 expression (Fig.6H). In our search of 304 additional factors deregulated downstream of β--catenin we found the protein levels 305 of the transcriptional factor Foxo1 to be reduced after GSK3 inhibition in both 306 11 normal and transformed B cell precursors (Fig.6I, J). Foxo1 is a transcription factor 307 that not only governs essential steps in B cell development but has also been shown 308 to be crucial for the survival and cell cycle progression of acute B cell 309 leukemia (Alkhatib et al., 2012;Dengler et al., 2008;Wang et al., 2018). In conclusion 310 these results demonstrate that β--catenin stabilization not only alters the metabolic 311 profile of B cell precursors but also induces a transcription profile that reverts 312 lineage fate decisions. 313 314

316
B cell derived lymphomas are the most common malignant lymphoid neoplasms 317 (Küppers, 2005;Young and Staudt, 2013) and new strategies are required to treat 318 refractory disease. GSK3 has emerged as a potential new target for medical 319 intervention in different types of cancer. Owing to the complexity of the signaling 320 processes governed by GSK3 it remains however difficult to predict the biological 321 outcome of GSK3 inhibition in B cell derived malignancies. Inactivation of genes 322 encoding both of the GSK3 isoforms results in enhanced oxygen consumption, cell 323 mass accumulation and proliferation in mature stimulated B cells (Jellusova et al., 324 2017). This phenotype is consistent with the accumulation of the transcription 325 factors cMyc and β--catenin in these cells. Both of these transcription regulators have 326 been shown before to support a transcription profile favoring increased metabolic 327 activity (Dang, 2013;Sherwood, 2015). Similar to mature stimulated cells, B cell 328 lymphoma cells have been shown to benefit from GSK3 inhibition (Varano et al., 329 2017). Signals originating from the B cell receptor have been suggested to result in 330 GSK3 inhibition, which in turn increases the competitive fitness of the cells (Varano 331 et al., 2017). These findings would thus discourage the use of GSK3 inhibitors for B 332 lymphoma treatment. However, other studies have found GSK3 inhibition to 333 successfully inhibit lymphoma B cell proliferation (Wu et al., 2019). In our study we 334 have confirmed that GSK3 inhibition reduces proliferation and survival in 335 lymphoma cells. The role of GSK3 in transformed B cell precursors has not been 336 studied before and our results demonstrate for the first time that GSK3 inhibition 337 12 reduces metabolic activity and proliferation of transformed B cell precursors. Thus 338 we have identified a possible new target that could be exploited for therapy in of B 339 cell precursor acute lymphoblastic leukemia. 340 Moreover, we show that while GSK3 inhibition induces increased mitochondrial 341 activity and faster proliferation in mature stimulated B cells, this treatment exerts 342 the opposing effect on B cell precursors arguing for a context dependent role of 343 GSK3 in B cells. GSK3 possesses a plethora of potential targets and teasing apart the 344 role of these various proteins is essential in order to be able to identify patients that 345 could benefit from GSK3 inhibition. GSK3 has been shown to bind to centrosomes 346 and suggested to play an important role in cell cycle progression (Wu et al., 2019). 347 While it is possible that GSK3 plays a role in mitosis progression, this function is not 348 absolutely necessary for B cells to proliferate as both normal and lymphoma B cells 349 have been shown to be able to undergo cell division in the presence of GSK3 350 inhibitors or in the absence of the GSK3α/β proteins (Jellusova et al., 2017;Varano et 351 al., 2017). In this study we propose that the outcome of β--catenin activity may be the 352 determining factor of whether GSK3 inhibition results in accelerated proliferation or 353 cell death. Despite the wide array of proteins being dysregulated after GSK3 354 inhibition, we found that β--catenin accumulation alone is sufficient to replicate the 355 phenotype of GSK3--inhibited cells. Similar to B cells treated with a GSK3 inhibitor, 356 mature B cells with hyper--stabilized β--catenin show increased oxygen consumption, 357 ROS production and proliferation. Equally, both GSK3--inhibited B cell precursors 358 and B cell precursors with hyper--stabilized β--catenin display reduced production of 359 ROS and increased cell death. Of note, we found cMyc protein levels and the 360 phosphorylation of ribosomal protein S6 to be differentially regulated in mature B 361 cells and B cell precursors upon GSK3 inhibition. GSK3 has been previously reported 362 to negatively impact on both cMyc levels (Gregory et al., 2003) and mTORC1 activity 363 (Inoki et al., 2006). In other cell types, GSK3 has been shown to phosphorylate and 364 activate the TSC complex, which in turn inhibits mTORC1. However in B cells, this 365 signaling pathway does not seem to play a dominant role in regulating S6 366 phosphorylation, since both GSK3--deficient and GSK3--inhibited mature B cells show 367 normal S6 phosphorylation (Jellusova et al., 2017). cMyc is targeted for degradation 368 13 via GSK3 induced phosphorylation (Gregory et al., 2003) and GSK3  Mice 420 Mice bearing the Catnb lox(ex3) locus have been described previously (Harada et al., 421 1999). Catnb lox(ex3) mice were crossed to mb1--cre ERT2 mice (Hobeika et al., 2015;422 Hug et al., 2014). Mice were i.p. injected with 1mg tamoxifen (Sigma) + 10% ethanol 423 (Roth) in olive oil on three consecutive days. Control animals were injected with 424 tamoxifen the same way as experimental animals. To induce deletion of catnb exon 425 3 in B cell precursors Catnb lox(ex3) mice were crossed to Mb1--cre mice. For both 426 lines, mice homozygous or heterozygous for the Catnb lox(ex3) locus were used as 427 experimental animals. Since no significant differences were observed between 428 homozygous and heterozygous mice, the exact genotype is not indicated when 429 presenting data. As controls, both Cre--positive and negative animals were used. No 430 15 significant differences were observed between these two types of control animals. 431 Both male and female mice were used for experiments. Animals were maintained in 432 a specific pathogen free environment. Experiments were approved by the regional 433 council in Freiburg, Germany and carried out in accordance with the German Animal 434 Welfare Act. For experiments with Blimp1--deficient B cell precursors, bone marrow 435 was obtained from Mb1--cre x Prdm1 lox mice (Setz et al., 2018). were fixed and permeabilized using the BD Cytofix/Cytoperm buffer system (BD) 506 and permeabilization buffer (eBioscience), incubated with 0.3mg/ml DNAse 507 (eBioscience) for 1h at 37°C, washed with Perm buffer (BD) and treated with anti--508 BrDU (eBioscience) in Perm buffer for 30min at room temperature. Cells were 509 washed, 0.5µg/ml DAPI or 7AAD was added to detect DNA, and live cells were 510 identified gating on forward side scatter. Cell cycle analysis was performed on day 511 5--6 of cell cuture, cells were SCF+IL7 dependent at this point. 512 Cells were analyzed by flow cytometry. The following cytometers were used for 513 acquisition of flow cytometry data: LSR II (BD), CyAn (Beckman Coulter), Attune 514 (Thermo Fisher Scientific). FlowJo software (TreeStar) was used for analysis. 515 516

Analysis of metabolic parameters and proliferation 517
Oxygen consumption was assessed using a Seahorse XFe96 metabolite analyzer 518 (Agilent). 10 5 Ramos cells, 1x10 5 transformed or 3x10 5 normal B cell precursors or 519 10 6 mature stimulated B cells were plated on Cell--Tak (Corning) coated Seahorse 520 cell culture plates. The cells were first incubated in Seahorse base medium + 1mM 521 sodium pyruvate (Thermo Fisher Scientific) + 2mM L--glutamine (Thermo Fisher 522 Scientifc) + 10mM glucose (Sigma) in a volume of 50µl for 30min, 130µl medium 523 18 were added in a second step and the cells were incubated for an additional 1h. 524 Oligomycin, FCCP and rotenone+antimycin were sequentially injected during the 525 measurement to a final concentration of 1µM each to assess different parameters of 526 respiration. To measure the production of reactive oxygen species, cells were 527 stained with 10µM carboxy--H2DCFDA (Thermo Fisher Scientific) for 20min at 37°C 528 and washed before measurement. To assess glucose uptake cells were incubated 529 with 30µM 2NBDG (Cayman Chemical) in PBS (Invitrogen) for 30min at 37°C and 530 analyzed by flow cytometry. To analyze proliferation, cells were loaded with 5µM 531 proliferation dye eFluor670 (eBioscience) and cultured for up to 3 days. Dilution of 532 the proliferation dye was measured by flow cytometry. 533 534

Data availability 586
Raw data obtained from transcriptome analysis were deposited with GEO and will 587 be made publicly accessible upon manuscript acceptance. Original, uncropped 588 pictures of shown western blots are included in source data. Additional repeats of 589 the experiments are also included in the source data. Numerical data used for all the 590 shown graphs are included in the source data. 591 592

Conflict of interest 593
The authors declare no conflict of interest. 594  The sources data include numerical values underlying graphs shown in Fig.1, Fig.1S,