CRISPR/Cas9 screening indicates loss of ARRDC3 provides a competitive advantage to Eµ-Myc lymphoma cells after TRP53 activation
Using Eµ-Myc lymphoma-derived cell lines, we performed a CRISPR/Cas9 screen with the mouse whole-genome “Yusa” sgRNA library (Koike-Yusa et al. 2014), and utilised the MDM2 inhibitor nutlin-3a to activate TRP53 in a non-genotoxic manner. Specifically, Eµ-Myc lymphoma cells stably transduced with a Cas9 expression vector were further transduced with the “Yusa” sgRNA library and expanded for nine days before being separated into two streams – 24 h treatment with DMSO (vehicle control) or nutlin-3a (Fig. 1A). Surviving cells were sorted by FACS, their genomic DNA extracted, and next generation sequencing (NGS) undertaken. Bioinformatic analyses were then performed to identify the enriched sgRNAs in each of the three streams of our experiment. When comparing the nutlin-3a-treated cells to the untreated control cells, loss of Trp53 was, as expected, the top hit (Fig. 1B). This, along with loss of the apoptosis mediator Bbc3/PUMA also being an expected strong hit (Valente et al. 2016), provided strong validation of our screening approach. This same comparison also returned Arrdc3 as the 5th top hit (Fig. 1B), suggesting that loss of Arrdc3 confers a survival/growth advantage in Eµ-Myc lymphoma cells treated with the TP53/TRP53-activating drug nutlin-3a. Interestingly, when comparing the nutlin-3a-treated samples to the DMSO-treated samples, Arrdc3 dropped to the 21st top hit. This indicates that there was also some low-level selection for loss of Arrdc3 in the DMSO-treated samples (which had also undergone the additional process of cell sorting) compared to the untreated samples. This likely indicates that Arrdc3 loss may generally enhance the survival/growth of Eµ-Myc lymphoma cells, which are highly apoptosis-prone, under normal (or slightly stressful; i.e. DMSO treatment) conditions (Figures S1A,B). Since Arrdc3 was more strongly enriched in the nutlin-3a-treated samples than the DMSO-treated samples, we chose to pursue the validation of Arrdc3 as a potential factor in TRP53-mediated tumour growth suppressing responses.
Validation that Arrdc3 is a TRP53-target gene, and loss ofArrdc3provides a competitive advantage toEµ-Myclymphoma cells treated with TRP53-activating drugs
To validate Arrdc3 as a hit from our screen, we used CRISPR/Cas9 and an sgRNA targeting Arrdc3 to generate Arrdc3 knockout (Arrdc3KO) cells in three well-characterised Eµ-Myc mouse lymphoma cell lines – AH15A, AF47A, and 560 (Thijssen et al. 2021). The efficacy of Arrdc3 disruption was confirmed by NGS (Figure S2).
We first hypothesised that Arrdc3 might be a transcriptional target (direct or indirect) of the master regulator TRP53. As such, we examined the expression of Arrdc3, and as positive controls the well-known TRP53 targets Pmaip1 (encodes NOXA), Bbc3 (encodes PUMA), and Cdkn1a (encodes p21), in isogenic NTsgRNA and Trp53KO (previously validated (Thijssen et al. 2021)) AF47A Eµ-Myc lymphoma cells after 6 and 24 h of treatment with nutlin-3a or etoposide (Figs. 2A, S3). We found that while baseline Arrdc3 expression levels were similar between untreated control and Trp53KOEµ-Myc lymphoma cells, there was a marked increase in Arrdc3 expression after treatment with nutlin-3a in the NTsgRNA Eµ-Myc lymphoma cells (~ 7-17-fold induction over 6–24 h) and treatment with etoposide (~ 5-10-fold induction over 6–24 h), but no such increase was seen in the Trp53KOEµ-Myc lymphoma cells (Figs. 2A, S3). As expected, we observed marked increases in the expression of the positive control TRP53 target genes in the NTsgRNA cells, but not in the Trp53KO cells, after both treatments. These data demonstrate that Arrdc3 can be (directly or indirectly) transcriptionally regulated by TRP53 in Eµ-Myc lymphoma cells.
We next examined whether Arrdc3 loss affected the proliferation rate/cycling of Eµ-Myc lymphoma cells after TRP53 activation. We treated the isogenic NTsgRNA control and Arrdc3KOEµ-Myc lymphoma cell lines with 5 µM nutlin-3a (~ IC85) for 6 h and assessed cell cycle stages by staining the DNA with DAPI. The observed reductions in the numbers of cells in S-phase and increases in the numbers of cells in the G1-phase, comparing nutlin-3a treated cells with DMSO (vehicle control) treated cells (Fig. 2B), is expected after TRP53 activation (Adams et al. 1985; Valente et al. 2016; Vassilev et al. 2004). However, we could discern no differences between the control cells or the Arrdc3KO cells in all cell backgrounds (Figs. 2B, S3A). This demonstrates that ARRDC3 does not play a major role in TRP53-mediated cell cycle arrest.
Next, viability assays were undertaken on these Arrdc3KOEµ-Myc lymphoma cell lines, employing Eµ-Myc lymphoma cells with an sgRNA targeting human BIM (hereafter referred to as non-targeting sgRNA (NTsgRNA)) as a control (Aubrey et al. 2015). The lymphoma cells were treated for 24 h with increasing concentrations of nutlin-3a, etoposide (a DNA damaging agent that causes activation of TP53/TRP53), and thapsigargin (induces apoptosis in a TP53/TRP53-independent manner by causing endoplasmic reticulum stress (Jaskulska, Janecka, and Gach-Janczak 2020)) (Figs. 2C, S4B,C). For both nutlin-3a and etoposide, we observed a slight but not statistically significant increase in the viability of Arrdc3KO lymphoma cells compared to the NTsgRNA control lymphoma cells, while thapsigargin killed lymphoma cell lines of all genotypes to a similar extent. These data suggest that while Arrdc3 might have a small role in TRP53-mediated apoptosis, it is likely not its predominant function. As a positive control, we also validated our hit of Bbc3 (encoding PUMA) via 24 h viability assay, and observed resistance to nutlin-3a- and etoposide-mediated killing (Figure S4D).
Finally, to mimic an in vivo scenario more closely, where only some cells possess a particular mutation, we examined whether Arrdc3KOEµ-Myc lymphoma cells had a competitive advantage over control Eµ-Myc lymphoma cells when grown in sub-lethal doses of these drugs over a longer period. To this end, we set our Arrdc3KOEµ-Myc lymphoma cells against their concomitant NTsgRNA Eµ-Myc controls in competition assays, and also in parallel against isogenic Trp53KOEµ-Myc lymphoma cells. These mixed lymphoma cell populations were then treated with either nutlin-3a (1.5 µM) or thapsigargin (1 nM), at doses chosen to kill significant proportions of the cells. In the Arrdc3KOvs control lymphoma cell competition, we observed outgrowth of the Arrdc3KO population over control cells even without any treatment, and this competitive advantage was enhanced in the presence of nutlin-3a (Figs. 2D, S4E). Interestingly, when treated with thapsigargin, Arrdc3KO lymphoma cells did not exhibit a competitive advantage vs control lymphoma cells beyond that observed after DMSO treatment. This suggests that while ARRDC3 likely plays a role in TRP53-mediated suppression of lymphoma cell expansion, it may have only limited involvement in TRP53-independent suppression of lymphoma cell growth or survival. By contrast, Trp53KO lymphoma cells outcompeted control and Arrdc3KO lymphoma cells when treated with either nutlin-3a or thapsigargin in both the AH15A and 560 cell lines, though the Arrdc3KO AF47A cells proved slightly more resilient (Figs. 2D, S4E). Overall, these data indicate that Arrdc3 is a TRP53 target, and its loss gives cells a competitive advantage in the face of TRP53 activation, possibly via low level apoptosis protection that can be selected for over time.
Arrdc3 is essential for normal mouse development
To explore the role of Arrdc3 in vivo, we generated Arrdc3 knockout mice by deleting 7 of 8 Arrdc3 exons (Fig. 3A). After obtaining a stable colony of Arrdc3+/− mice, we found that inter-crossing these mice did not yield viable Arrdc3−/− adult offspring, as has been previously observed (Patwari et al. 2011; Shea et al. 2012). Inter-crosses between Arrdc3+/− animals revealed a statistically significant difference between the observed and expected numbers of Arrdc3−/− animals at weaning (Fig. 3B). To determine the developmental stage when Arrdc3−/− animals die, genotypes were assessed at embryonic day 14.5 (E14.5) after timed inter-crosses of Arrdc3+/− mice, which revealed expected Mendelian ratios (Fig. 3B). We next assessed the foetal genotypes at E18.5/19.5, after timed inter-crosses of Arrdc3+/− mice, by administering progesterone to pregnant females on E17.5 and E18.5, preventing labour. This allowed for a Caesarean section to be carried out to deliver the pups at E18.5/19.5. Genotyping of the E18.5/19.5 pups across 20 separate litters showed all genotypes were in line with Mendelian ratios (Fig. 3B). This reveals the Arrdc3 loss-induced lethality likely occurs during or soon after birth.
To identify abnormalities that might be contributing to this lethality, we carried out a full assessment of E19.5 pups (n = 21 animals examined across 3 litters) (File S1). There was a slight, but non-significant, trend towards Arrdc3+/− and Arrdc3−/− animals weighing more than their wild-type littermates (data not shown). Notably, we observed a number of incompletely penetrant developmental abnormalities in the Arrdc3−/− pups (Fig. 3C), including: underdeveloped eyes (also seen in some Arrdc3+/− animals) (Figs. 3C,D), external and internal haemorrhaging, manifesting, for example, as small areas of haemorrhage on the surface of the thymus (Figs. 3C,F), liver discolouration (Fig. 3C), and one Arrdc3−/− animal presented with an omphalocele (Figs. 3C,E). Some Arrdc3−/− pups had breathing difficulties, with one of these also displaying subcutaneous oedema composed of serous fluid and blood (Figs. 3C,G). We hypothesised that heart defects might contribute to the mortality of Arrdc3−/− pups, given the lack of consistently lethal external morbidities. Histological sections of the hearts revealed some Arrdc3−/− pups (n = 2/7) exhibited ventricular-septal defects. These defects varied in severity, with one Arrdc3−/− animal missing gross internal ventricle structure (compare Figures S5A + B, demonstrating Arrdc+/+ and Arrdc+/− hearts, to S5C, demonstrating an Arrdc3−/− heart), while another Arrdc3−/− animal had a very small ventricular septal defect (Figure S5D). These findings indicate that a range of developmental defects may contribute to the perinatal lethality of Arrdc3−/− mice.
Arrdc3 has no role in TRP53-mediated cell cycle arrest or apoptosis in primary non-transformed cells
Considering the requirement of ARRDC3 in development, we next investigated whether ARRDC3 plays a role in the survival and growth of primary tissues. We first examined murine embryonic fibroblasts (MEFs) derived from Arrdc3+/+ and Arrdc3−/− E14.5 foetuses. Gene expression analysis via qRT-PCR was used to assess Arrdc3 expression in MEFs after treatment with nutlin-3a or etoposide. As expected, Arrdc3−/− MEFs had entirely lost Arrdc3 expression (Figure S6A), but in wild-type MEFs we observed that Arrdc3 expression was relatively weakly induced (~ 2-fold induction) in response to both drugs (Figure S6B). While considerably smaller than the responses observed in Eµ-Myc lymphoma cells (Figs. 2A,S3), the similarly reduced levels of induction of known TRP53-target control genes (Pmaip1, Bbc3, Cdkn1a) suggests that MEFs are overall less sensitive to TRP53-activating stimuli than Eµ-Myc lymphoma cells. We next examined the cell cycle behaviour of MEFs after treatment with nutlin-3a, which revealed the expected reduction in numbers of cells in S phase and increased numbers of cells in G1 phase, but no significant differences were evident between the two genotypes (Figure S6C). Finally, we assessed MEF viability after treatment with different concentrations of nutlin-3a or etoposide. Each treatment resulted in a noticeable reduction in MEF viability, but no significant differences were observed between the two genotypes (Figure S6D).
We next examined the role of Arrdc3 in the development of primary haematopoietic cells. To enable this, we used the foetal liver cells of E14.5 Arrdc3+/+ and Arrdc3−/− foetuses to perform haematopoietic reconstitutions, injecting these cells into lethally irradiated recipient wild-type congenic mice. After 10 weeks we harvested the bone marrow, spleen, and thymus and assessed the proportions and numbers of different haematopoietic cell types in these tissues by flow cytometry. Examining B cell development in the bone marrow did not reveal any differences between the Arrdc3+/+ (i.e. wt) vs Arrdc3−/− reconstituted mice (Figure S7A). Similarly, in both genotypes, follicular and marginal zone B cells from the spleen were roughly equal in number (Figure S7B), as were T lymphoid cells of the major stages of differentiation (as defined by expression of CD4 and CD8) in the thymus (Figure S7C). We also examined the responses over time of cultured bone marrow-derived B cells and thymocytes from Arrdc3+/+ and Arrdc3−/− reconstituted mice to treatment with different doses of nutlin-3a (Figure S7D). Like in MEFs, we found no differences in the viability of these cells at any timepoint between the two genotypes. Lastly, qRT-PCR was used to evaluate the expression of Arrdc3 and Cdkn1a (p21, as a TRP53 target control) in splenic B cells and thymocytes from Arrdc3+/+ and Arrdc3−/− reconstituted mice. In both B and T cells, Arrdc3 expression was increased in the nutlin-3a treated Arrdc3+/+ cells but, as expected, absent in the Arrdc3−/− samples, whereas Cdkn1a expression was strongly induced after treatment with nutlin-3a in cells from both genotypes (Figure S7E). The extent of Arrdc3 induction in B cells was less pronounced than in T cells after 24 h of treatment with nutlin-3a (~ 2-fold vs ~ 10-fold) (Figures S7D,E).
These data demonstrate that Arrdc3 is required for normal embryonic development but does not have a major role in haematopoiesis or the response of lymphoid cells or MEFs to anti-cancer agents that activate TRP53.
The absence of Arrdc3 markedly accelerates lymphoma development in Eµ-Myc transgenic mice
Having found that loss of Arrdc3 confers a competitive advantage in malignant Eµ-Myc lymphoma cells, and having generated an Arrdc3 knockout mouse model, we next investigated whether Arrdc3 might impact MYC-driven lymphoma development in vivo. To this end, we inter-crossed Eµ-MycT/+;Arrdc3+/− male mice with Arrdc3+/− female mice. Genotyping the offspring of these crosses revealed that, for mice with or without an Eµ-Myc transgene, Arrdc3−/− mice were significantly underrepresented at the adult stage (Fig. 4A). We then monitored those mice possessing an Eµ-Myc transgene to determine the impact of Arrdc3 loss on MYC-driven lymphoma development. Surprisingly, one Eµ-MycT/+;Arrdc3−/− animal survived post-weaning but had to be sacrificed due to lymphoma at 56 days (Fig. 4B). Assessing the tumour-free survival of the other genotypes, we observed a slight but non-significant decrease in tumour latency in Eµ-MycT/+;Arrdc3+/− animals (median survival = 77 days) compared to the Eµ-MycT/+;Arrdc3+/+ control mice (median survival = 91 days) (Mantel-Cox test, df = 1, X2 = 2.981, p > 0.05 (p = 0.0842)) (Fig. 4B). Stratifying the mice by gender did not reveal any additional variation between genotypes.
As we were unable to obtain more than one Eµ-MycT/+;Arrdc3−/− adult, we turned to the process of haematopoietic reconstitution. At E14.5 we observed the genotypes fell into expected Mendelian ratios (Fig. 4A), and therefore we were able to use both Eµ-MycT/+;Arrdc3+/+ and Eµ-MycT/+;Arrdc3−/− foetal liver cells to reconstitute lethally irradiated recipient mice. These recipients were then monitored for lymphoma development. We observed a remarkable (statistically significant) decrease in tumour latency in mice reconstituted with Eµ-MycT/+;Arrdc3−/− foetal liver cells (median survival = 67 days) compared to recipients reconstituted with Eµ-MycT/+;Arrdc3+/+ foetal liver cells (median survival = 210 days) (Mantel-Cox test, df = 1, X2 = 13.22, p < 0.001 (p = 0.000276)) (Fig. 4C). Examining the peripheral blood content of the lymphoma burdened mice at sacrifice revealed no clear differences between the two genotypes in their cellular makeup (Figure S8A). Similarly, organ weights did not reveal any clear differences between the two genotypes and hence severity of lymphomatous disease (Figure S8B). Interestingly, immunophenotyping of the malignant cells derived from the spleens of the reconstituted mice illustrated some differences between the genotypes. The Eµ-MycT/+;Arrdc3−/− lymphomas were more immature in origin (> 60% tumours were majority B220+/IgD-/IgM- pro-B/pre-B) compared to the Eµ-MycT/+;Arrdc+/+ lymphomas (> 60% tumours were majority B220+/IgD-/IgM + immature B) (Figure S8C).
Collectively, these findings demonstrate Arrdc3 loss leads to a marked acceleration of MYC-driven lymphoma development.