Targeting DYRK1A/B kinases to modulate p21‐cyclin D1‐p27 signalling and induce anti‐tumour activity in a model of human glioblastoma

Abstract The dual‐specificity tyrosine‐regulated kinases DYRK1A and DYRK1B play a key role in controlling the quiescence‐proliferation switch in cancer cells. Serum reduction of U87MG 2D cultures or multi‐cellular tumour spheroids induced a quiescent like state characterized by increased DYRK1B and p27, and decreased pRb and cyclin D1. VER‐239353 is a potent, selective inhibitor of the DYRK1A and DYRK1B kinases identified through fragment and structure‐guided drug discovery. Inhibition of DYRK1A/B by VER‐239353 in quiescent U87MG cells increased pRb, DYRK1B and cyclin D1 but also increased the cell cycle inhibitors p21 and p27. This resulted in exit from G0 but subsequent arrest in G1. DYRK1A/B inhibition reduced the proliferation of U87MG cells in 2D and 3D culture with greater effects observed under reduced serum conditions. Paradoxically, the induced re‐expression of cell cycle proteins by DYRK1A/B inhibition further inhibited cell proliferation. Cell growth arrest induced in quiescent cells by DYRK1A/B inhibition was reversible through the addition of growth‐promoting factors. DYRK inhibition‐induced DNA damage and synergized with a CHK1 inhibitor in the U87MG spheroids. In vivo, DYRK1A/B inhibition‐induced tumour stasis in a U87MG tumour xenograft model. These results suggest that further evaluation of VER‐239353 as a treatment for glioblastoma is therefore warranted.

quiescence. 1,10,11 Additionally, DYRK1A and 1B can phosphorylate cyclin D1 (resulting in its destabilization) [12][13][14] and p27Kip1 (increasing its stabilization) [15][16][17] thereby controlling the S-phase checkpoint. DYRK1B expression is increased in quiescent cells and inhibition of DYRK1B can sensitize quiescent pancreatic and ovarian carcinoma cell lines to DNA damaging agents 18,19 and quiescent GIST cells to imatinib. 11 We have previously described the use of fragment and structurebased lead discovery methods to identify potent and selective ATP competitive inhibitors of DYRK1A and DYRK1B. 20 The structure and key data of VER-239353 (compound 34 in 20 ) are reproduced in Figure S1. Here, we further evaluate the anti-tumour activity of this compound in a model of human glioblastoma.

| Cell lines and cell culture
All cell lines were purchased from the American Type Culture Collection (ATCC), established as a low passage cell bank and then routinely passaged in our laboratory for less than 3 months after resuscitation. A375, PANC-1, PC-3 and U87MG cells were routinely cultured in DMEM or NCI-H1299 in RPMI-1640 containing 10% foetal calf serum (FCS) and 1% penicillin / streptomycin at 37°C in a normal humidified atmosphere supplemented with 5% CO 2 . Cells were authenticated by STR profiling (LGC Standards, Teddington UK).
For quiescence induction, cells were trypsinized and resuspended in media with 10% FCS, centrifuged and washed twice with FCS-free media and then resuspended in media containing 0.2% FCS and counted. Cells were subsequently plated in media containing 0.2% FCS and incubated for 72 h before analysis. Multi-cellular tumour spheroids were generated as previously described. 21

| In vitro combination synergy
Pre-formed U87MG spheroids were treated with a combination of VER-239353 and the CHK1 inhibitor VER-157932 22 for 7 days at fixed ratio concentrations. Synergy was calculated using the Median Dose Effect in CalcuSyn software (Biosoft, UK) according to the method of Chou and Talalay. 23

| High content cell cycle analysis
Determination of cell cycle fractions was conducted using high content imaging as previously described. 24 For multiparametric cell cycle analysis, cells were labelled with 10 μM EdU for 15 min immediately prior to fixation with 3.7% paraformaldehyde in PBS at room temperature for 15 min. Cells were washed twice in PBS then twice in 3% BSA in PBS before permeabilization with 0.5% Triton X100 in PBS for 20 min at room temperature. Cells were washed twice with 3% BSA in PBS before incorporated EdU was labelled with an Alexa Click-iT EdU labelling kit (Life Technologies). Following blocking for 30 min with 5% normal goat serum in PBS, cells were incubated with an anti-pHH3 (S10) primary antibody diluted in antibody dilution buffer (1% BSA, 0.3% Triton X100 in PBS) at 4°C for 16 h. Cells were washed with PBS then incubated with an Alexa-labelled secondary antibody (1:500, Life Technologies) and Hoechst 33342 (1 μg/ml) in antibody dilution buffer at room temperature for 60 min. Following washing with PBS, cells were imaged with an Operetta high content imaging system (Perkin Elmer) at 10× magnification and analysed using Harmony software (Perkin Elmer). The antibodies used are listed in Table S1.

| Immunoblotting
Cells were washed once with PBS and lysed in RIPA buffer containing protease and phosphatase inhibitor cocktail (Roche). Protein concentration was determined using a BCA kit (Pierce). Equal amounts of lysate were separated by SDS-PAGE and Western blot analysis conducted using the antibodies indicated in Table S1. Primary antibodies were detected with HRP-conjugated secondary antibody (Santa Cruz Biotechnology) and detected with Western Lightning (Perkin Elmer) or Immobilon (Millipore) chemiluminescent HRP substrate.
Densitometry was determined using Image J software (NIH).

| High content immunofluorescent imaging
Following compound treatment, cells were fixed in 3.7% paraformaldehyde in PBS at room temperature for 15 min, washed with PBS, blocked with 5% normal goat serum in 0.3% Triton X100 in PBS for 1 h at room temperature then incubated with primary antibody diluted in antibody dilution buffer (1% BSA, 0.3% Triton X100 in PBS) at 4°C for 16 h. Cells were washed with PBS then incubated with an Alexalabelled secondary antibody (1:500, Life Technologies) and Hoechst 33342 (1 μg/ml) in antibody dilution buffer at room temperature for 60 min. Following washing with PBS, cells were imaged with an Operetta high content imaging system (Perkin Elmer) at 10× or 20× magnification and analysed using Harmony software (Perkin Elmer).

| In vivo efficacy in U87MG xenografts
Female nude mice (Crl:NU(NCr)-Foxn1nu) were purchased from Charles River Laboratories. Xenograft studies were undertaken by Charles River Laboratories Discovery Services, North Carolina and accredited by the Association for Assessment and Accreditation of Laboratory Animal Care International. The U87MG tumour line was maintained by serial SC transplantation in female athymic nude mice.
Tumour fragments, approximately 1 mm 3 each, were implanted SC into the right flank of each animal and allowed to grow towards a target size of 100-150 mm 3 . On day 1 of the study, tumours were randomized into treatment groups before compound administration.
VER-00239353 was formulated in 60 mM HCl and 40% hydroxypropylβ-cyclodextrin and administered by oral gavage once daily for 21 days. Tumour size was measured twice weekly with electronic callipers and tumour volume calculated according to the formula ((width × width) × length)/2. Body weight was measured daily for the first 5 days then twice weekly thereafter. The study endpoint was defined as a tumour volume of 2000 mm 3 or D28, whichever came first.
Each animal was euthanized when its tumour reached the endpoint. TGI (%) = 1 − ((T 11 /T 0 )/(C 11 /C 0 ))/1 − (C 0 /C 11 ) × 100 where T 11 and T 0 equals mean tumour volume of treated group at day 11 and day 1, respectively, and C 11 and C 0 equals median tumour volume of control group at day 11 and day 1, respectively. Significance was determined using a one-way ANOVA in GraphPad PRISM 8.4.3.
The time to endpoint (TTE) for each mouse was calculated from the following equation: TTE = log10 (endpoint volume) -b/m where b is the intercept and m is the slope of the line obtained by linear regression of a log-transformed tumour growth data set. The data set is comprised of the first observation that exceeded the study endpoint volume and the three consecutive observations that immediately preceded the attainment of the endpoint volume. Animals that did not reach endpoint were euthanized at the end of the study and assigned a TTE value equal to the last day of the study (D28).
Animals determined to have died from treatment-related (TR) causes were assigned a TTE value equal to the day of death. If the animal died from non-treatment-related (NTR) causes, it was excluded from the TTE calculations.

| DYRK1/A inhibition of serum-starved cancer cells induces re-expression of cell cycle proteins as well as cell cycle inhibitors
The 2D culture of U87MG cells in low serum (0.5% FCS or less) resulted in reduced cell proliferation ( Figure 1A), a reduction in cycling cells ( Figure 1B) and an increase in the fraction of cells staining negative for Ki67 (indicative of G0 phase) from 23.0 to 55.4% ( Figure 1C). This increase in quiescence was also evident following Western blot analysis of U87MG cells grown in reduced serum. A reduction in phosphorylation of Rb at S807/811, a decrease in total Rb and cyclin D1 protein, and an increase in DYRK1B and p27 expression were observed ( Figure 1D). Likewise, formation and growth of U87MG multi-cellular tumour spheroids in lower (2% compared to 10%) FCS resulted in a similar decrease in Rb S807/811 phosphorylation, total Rb and cyclin D1 protein and an increase in DYRK1B and p27 expression ( Figure 1D).   (Table S2). These changes happened rapidly within 24 h of VER-239353 addition and were sustained for at least 96 h ( Figure S3B,C). These changes to cell cycle protein phosphorylation and expression were not just restricted to U87MG cells grown as 2D monolayers. Treatment of U87MG cells grown as multi-cellular spheroids in either 2 or 10% FCS with VER-239353 resulted in a dose-dependent increase in pRb and total DYRK1B, cyclin D1, p21 and p27 ( Figure S4A,B). The effects were more marked in the spheroids growing in 2% FCS compared to those growing in 10% FCS with EC 50 values for cyclin D1, p21 and DYRK1B of 132, 94 and 103 nM, respectively ( Figure S4C and Table S2).
These observations were confirmed by high content immunofluorescent imaging. Treatment of U87MG cells grown in low FCS with VER-239353 dramatically increased the percentage of cells staining positive for p21, p27 or cyclin D1 ( Figure 3A). In comparison, switching the cells from growth in low (0.2%) to high (10%) serum did not.
In cells growing in high serum, VER-239353 treatment also increased the fraction of cells staining positive for p21 or p27 but the magnitude of effect was much lower than that in low serum. An analysis of the cell cycle phase (using DNA and EdU content) by high content imaging revealed that the p21 and p27 positive cells following growth in low serum were in the G0/G1 phase. VER-239353 treatment did not significantly alter the cell cycle distribution of the p21 or p27 positive cells with the majority still in G0/G1 ( Figure 3B). Switching U87MG cells grown in low serum (0.2%) to high serum (10%) resulted in an obvious re-entry into S-phase after 24 h ( Figure 3C).

| DYRK1A/B inhibition induces DNA damage in U87MG multi-cellular spheroids and synergizes with CHK1 inhibition
Previous studies have demonstrated that DYRK1A/B inhibition can increase the sensitivity of quiescent cancer cells to genotoxic drugs such as cisplatin and gemcitabine. 18   against human glioblastoma models through EGFR destabilization. 27,28 This is reported to occur through the activation of p53-MDM2. 29 The reason for the lack of EGFR downregulation at target-hitting concentrations in our studies is not clear, although the discrepancy may be related to the fact that these published studies used either shRNA knockdown of DYRK1A, rather than inhibition of its kinase activity, or the less-selective