Introduction

The intrinsic apoptosis pathway is a tightly regulated process necessary for normal development and tissue homeostasis, but can also be activated by external cellular stress, such as radiation or nutrient deprivation [1]. The first step in intrinsic apoptosis signalling is the transcriptional and/or post-transcriptional upregulation of pro-apoptotic BH3-only proteins (e.g. PUMA, NOXA, BIM) in response to, for example, activation of the tumour suppressor TRP53 [2,3,4,5]. BH3-only proteins can bind and inhibit the pro-survival BCL-2 proteins (e.g. BCL-2, BCL-XL, MCL-1), resulting in the release and activation of the pro-apoptotic effectors BAK and BAX, though some BH3-only proteins have been reported to also activate BAK/BAX directly [6]. Activated BAK/BAX oligomerise and form pores in the mitochondrial outer membrane (MOMP), allowing release of apoptogenic factors from the space between the inner and outer mitochondrial membranes. This instigates apoptosome formation and activation of the caspase cascade, effecting cell destruction.

Mutated in more than 50% of all human cancers, the transcription factor TP53 (called TRP53 in mice) is a critical tumour suppressor [7]. Moreover, the efficacy of many chemotherapeutic drugs, particularly those that cause DNA damage (e.g., etoposide, cisplatin), depends upon intact TP53/TRP53 functionality, consequently meaning that mutant TP53 cancers often respond poorly to cancer therapy [8]. In stressed cells, TP53/TRP53 controls the expression of genes that regulate cellular responses that cooperate to suppress tumour development, such as apoptotic cell death, cell cycle arrest and senescence, coordination of DNA repair, and adaptation of cellular metabolism [7]. Several TP53/TRP53 target genes critical for some of these processes have been identified, such as those encoding the pro-apoptotic BH3-only proteins PUMA and NOXA (BBC3 and PMAIP1, respectively) [2, 3, 5], as well as the cyclin-dependent kinase inhibitor P21/CDKN1A [9]. Notably, however, mice deficient for all three of these genes—Bbc3, Pmaip1, and Cdkn1a—do not spontaneously develop tumours [10], in striking contrast to the highly tumour prone Trp53 knockout mice [11]. This indicates that additional target genes, and the processes they operate in, must play critical roles in TP53/TRP53-mediated tumour suppression. For example, in vivo shRNA screens have identified DNA repair genes (e.g. MLH1) as important effectors of TP53/TRP53-mediated tumour suppression [12].

In this study, we aimed to identify novel negative regulators of lymphoma cell expansion/survival that function downstream of TP53/TRP53 activation. Such regulators could be potential therapeutic targets that, when activated, bypass drug resistance caused by mutation or loss of TP53/TRP53. To this end, we conducted CRISPR/Cas9 whole-genome screens to identify novel genes that mediate TRP53-tumour suppressive responses in murine Eμ-Myc lymphoma cells, a well-established model of aggressive B cell lymphoma [13]. We identified and validated arrestin domain containing 3 (Arrdc3), the loss of which provided a competitive growth/survival advantage to lymphoma cells after TRP53 activation. ARRDC3 (also known as TBP-2-like inducible membrane protein (TLIMP)) is one of six α-arrestins, with 2 visual- and 2 β-arrestins rounding out the wider arrestin family [14]. ARRDC3 has most commonly been identified in roles also ascribed to other, better-studied arrestins, such as in the suppression of G-protein coupled receptor (GPCR) signalling, where it mediates receptor ubiquitination and lysosomal degradation [15,16,17,18,19,20,21]. Several publications have associated abnormalities in ARRDC3 with a wide variety of cancers, particularly in epithelial to mesenchymal transition (EMT) and invasiveness, where loss/downregulation of ARRDC3 contributes to more severe phenotypes [22,23,24,25,26,27,28]. However, to our knowledge, ARRDC3 has not previously been described in the context of TP53-mediated tumour suppression. Extending our discovery, we generated Arrdc3 knockout mice, which revealed that Arrdc3 is an essential gene, as Arrdc3−/− mice died perinatally. Interestingly, transplanting lethally irradiated mice with foetal liver cells from E14.5 Eμ-MycT/+;Arrdc3−/− or control Eµ-MycT/+ foetuses demonstrated that loss of ARRDC3 greatly accelerated MYC-driven lymphoma development. Altogether, our research demonstrates that Arrdc3 is an essential gene, and plays an important role in TRP53-mediated suppression of MYC-driven lymphoma development.

Results

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 [29], 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 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 [30], 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; e.g. DMSO treatment/cell sorting) 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.

Fig. 1: Arrdc3 identified as a hit in a CRISPR/Cas9 screen for regulators of TRP53-mediated lymphoma growth suppression.
figure 1

A Diagram of the CRISPR screen methodology using the MDM2 inhibitor nutlin-3a, which activates TP53/TRP53 in a non-genotoxic manner, as a selection pressure. In quadruplicate, AF47A Eμ-Myc lymphoma cells were transduced with the whole-genome “Yusa” sgRNA library, expanded, and either cultured without treatment or treated with either DMSO (vehicle control) or nutlin-3a (10 μM) for 24 h. Live cells were then sorted by FACS, DNA extracted from each sample (marked by asterisks), indexing PCR performed, and NGS combined with bioinformatic analyses used to identify sgRNAs that conferred a survival/competitive advantage after TRP53 activation. B Top hits from the CRISPR screen, comparing the untreated control cells (yellow asterisk) with the cells treated with nutlin-3a (red asterisk). Arrdc3 was identified as one of the top hits, with Arrdc3 targeting sgRNAs highly enriched in the surviving nutlin-3a-treated lymphoma cells.

Full size image

Validation that Arrdc3 is a TRP53-target gene, and loss of ARRDC3 provides a competitive advantage to Eμ-Myc lymphoma 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 [31]. 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 AF47A Eμ-Myc lymphoma cells with a non-targeting sgRNA (NTsgRNA) [32] or made Trp53KO by CRISPR/Cas9 (previously validated [31]) after 6 and 24 h of treatment with nutlin-3a or etoposide (Figs. 2A, S3). While baseline Arrdc3 expression levels were similar between untreated control and Trp53KO Eμ-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 Trp53KO Eμ-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 transcriptionally regulated by TRP53 in Eμ-Myc lymphoma cells. Whether this transcriptional regulation is direct or indirect remains unclear.

Fig. 2: Arrdc3 was validated as a TRP53 regulated gene and loss of Arrdc3 provided a competitive advantage to Eμ-Myc lymphoma cells after TRP53 activation.
figure 2

A qRT-PCR analysis to assess the levels of Arrdc3 expression in AF47A Eμ-Myc lymphoma cells after treatment for 6 or 24 h with nutlin-3a (10 μM) or etoposide (40 ng/mL), relative to expression in DMSO-treated control cells. Additionally, expression of Pmaip1, Bbc3, and Cdkn1a were assessed as controls, and the expression of each gene was also examined in Trp53KO cells to assess the level of reliance of Arrdc3 expression on TRP53 activity. Gapdh was used as a housekeeping gene. Each treatment was performed three times, and the qRT-PCR was undertaken with 3 technical replicates. Error bars represent standard error of the mean. Statistical tests were one-way ANOVAs with Šídák’s multiple comparisons tests, performed to compare the indicated samples. Note that due to variation between biological replicates, some differences are not statistically significant despite obvious trends. B Cell cycle assay using the AF47A Eμ-Myc lymphoma cell line with CRISPR/Cas9-mediated knockout of Arrdc3 compared to the NTsgRNA transduced control lymphoma line, treated with 5 μM nutlin-3a for 6 h. Each treatment was performed 3 times. Data are presented as means +/− standard deviation. Statistical analyses are given in Table S1. C Cell death assays using the AF47A Eμ-Myc lymphoma cell line edited using CRISPR/Cas9 to ablate Arrdc3, treated for 24 h with one of three apoptosis-inducing drugs: nutlin-3a and etoposide that act via TP53/TRP53, or thapsigargin that functions in a TP53/TRP53-independent manner. Each cell line treatment was performed 3 times, with 2 technical replicates each time. Data are presented as means +/− standard deviation. D Cell competition assays using the 560 Eμ-Myc lymphoma cell line with CRISPR/Cas9-mediated knockout derivative lines of either Arrdc3 or Trp53, or a non-targeting sgRNA (NTsgRNA). Mixed cell populations (1:1) were treated with sub-optimal doses of either nutlin-3a (1.5 µM, ~IC20) or thapsigargin (1 nM, ~IC90) over a period of 14 days. Each competition assay was performed twice for each cell line, with 2 technical replicates each time, with a representative example being shown. Data are presented as means +/− standard deviation.

Full size image

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 Arrdc3KO Eμ-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 [13, 30, 33]. However, we could discern no consistent differences between the control cells or the Arrdc3KO cells in all cell backgrounds (Figs. 2B, S4A; statistical analyses in Table S1). This demonstrates that ARRDC3 does not play a major role in TRP53-mediated cell cycle arrest.

Next, viability assays were undertaken on these Arrdc3KO and NTsgRNA Eμ-Myc lymphoma cell lines. 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 [34]) (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 (Fig. S4D).

Finally, to mimic an in vivo scenario more closely, where only some cells possess a particular mutation, we examined whether Arrdc3KO Eμ-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 Arrdc3KO Eμ-Myc lymphoma cells against their concomitant NTsgRNA Eµ-Myc controls in competition assays, and also in parallel against isogenic Trp53KO Eμ-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 Arrdc3KO vs 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 [20, 35]. 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 (perinatally).

Fig. 3: Arrdc3 knockout mice are not viable, with a range of abnormalities evident at embryonic day E19.5.
figure 3

A The coding region of the Arrdc3 gene spans the region 81,031,503…81,044,161 on chromosome 13 in mice. Arrdc3 knockout mice were generated using CRISPR/Cas9 gene editing, with sgRNAs (red) targeting the 3’ UTR upstream of the gene and between exons (purple) 7 and 8, resulting in the deletion of the majority of the Arrdc3 gene locus. B The genotypes of offspring of inter-crosses between Arrdc3+/− mice do not obey Mendelian ratios post-birth (X2 = 82.87, df = 2, p < 0.05 (p < 1 × 10−15)), with only 8 Arrdc3−/− offspring being observed among 302 animals, none of which survived past weaning. By contrast, E14.5 foetuses generated from inter-crosses of Arrdc3+/− mice were seen to obey Mendelian ratios (X2 = 0.53, df = 2, p > 0.05 (p = 0.766)), with 31 of 137 foetuses genotyped as Arrdc3−/−. E18.5/19.5 pups also obeyed Mendelian ratios for each Arrdc3 genotype (X2 = 1.511, df = 2, p > 0.05 (p = 0.470)). Expected numbers are marked with a red E. C E19.5 Arrdc3−/− pups (n = 7 animals examined across 3 litters) displayed numerous gross morphological defects, though none were completely penetrant. D Underdeveloped eyes were observed in some Arrdc3−/− (and Arrdc3+/−) animals. In this stronger example, the right eye of mouse 470.3 was not externally visible. E An omphalocele observed in Arrdc3−/− mouse 468.3. F Haemorrhaging was observed on the exterior of some Arrdc3−/− animals and in some internal organs, including the surface of the thymus in Arrdc3−/− mouse 472.5. G Many Arrdc3−/− pups had breathing difficulties, including mouse 468.2, which also displayed additional morphological defects, including subcutaneous oedema which included serous fluid and blood.

Full size image

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) (Fig. 3C, D), external and internal haemorrhaging, manifesting, for example, as small areas of haemorrhage on the surface of the thymus (Fig. 3C, F), liver discolouration (Fig. 3C), and one Arrdc3−/− animal presented with an omphalocele (Fig. 3C, E). Some Arrdc3−/− pups had breathing difficulties, with one of these (pup 468.2) also displaying subcutaneous oedema composed of serous fluid and blood (Fig. 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 Arrdc3+/+ and Arrdc3+/− hearts, to S5C, demonstrating a disrupted 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 (wt) 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) suggest 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) (Figure S7E).

While Arrdc3 is required for normal embryonic development, it does not have a major role in haematopoiesis, or in 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.

Fig. 4: Loss of Arrdc3 markedly accelerated MYC-driven lymphomagenesis in mice.
figure 4

A Offspring of inter-crosses between Arrdc3+/− and Eμ-MycT/+;Arrdc3+/− mice do not obey Mendelian ratios post-birth (X2 = 122.11, df = 5, p < 0.05 (p < 1 × 10−15)). By contrast, E14.5 foetuses generated from such inter-crosses were seen to obey Mendelian ratios (X2 = 4.95, df = 5, p > 0.05 (p = 0.422)). Expected numbers are marked with a red E. B Kaplan–Meier survival curve of Eμ-MycT/+;Arrdc3+/+ mice (median survival = 91 days), Eμ-MycT/+;Arrdc3+/− mice (median survival = 77 days), and only one Eμ-MycT/+;Arrdc3−/− mouse that survived the developmental perinatal lethality. The absence of one allele of Arrdc3 slightly, but not significantly, accelerated lymphoma development in Eμ-Myc mice (Mantel-Cox test, df = 1, p > 0.05 (p = 0.0686)). C Kaplan–Meier lymphoma-free survival curve of lethally irradiated recipient mice that had been transplanted with Eμ-MycT/+;Arrdc3+/+ or Eμ-MycT/+;Arrdc3−/− foetal liver cells. We observed a statistically significant acceleration of lymphoma development in mice that had been transplanted with Eμ-MycT/+;Arrdc3−/− foetal liver cells (median survival = 67 days) compared to control mice that had been transplanted 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)).

Full size image

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.

Discussion

In this study, we have identified a novel role for ARRDC3 as a mediator of TRP53-mediated tumour suppression in Eμ-Myc-driven lymphomagenesis. ARRDC3 is one of 10 arrestin family members in mammals, with research showing there is at least partial overlap between the functionality of these proteins, and some level of physical association between them too, though the significance of the association is not completely understood [18, 20]. Likely the best characterised function for ARRDC3, shared with the β-arrestins, is as a regulator of receptor-mediated signalling. In this role, ARRDC3 wears multiple hats, as while it has been shown to promote G-protein coupled receptor (GPCR) signalling via the beta-2 adrenergic receptor [21], it can also negatively regulate various signalling pathways. One such example is insulin signalling, where ARRDC3 suppresses phosphorylation of FOXO1 by AKT [36]. Another GPCR example is PAR1, where ARRDC3 suppresses Hippo signalling by binding and inhibiting pathway effectors YAP and TAZ, a function that also appears to be conserved in tissues of the vinegar fly, Drosophila melanogaster [16, 26, 37]. A number of these roles occur at the endosomal level, with α-arrestins 1 and 4 also having been shown to act at the level of extracellular vesicles [38, 39], suggesting a general importance for α-arrestins in cellular trafficking processes. Mechanistically, clues regarding ARRDC3 functions can be found in its structure—specifically, two PPxY domains (shared by most α-arrestins) facilitate binding to WW domain-containing proteins, such as various NEDD4-family E3 ubiquitin ligases, including ITCH, NEDD4, NEDD4L, WWP1, and WWP2, some of which have been shown to contribute to ARRDC3-mediated regulation of molecular signalling [17,18,19,20, 25, 40]. Despite this impressive body of work, and its discovery over a decade ago [41], ARRDC3 function has still only begun to be characterised.

No role for ARRDC3 as a mediator of the TP53 tumour suppressive pathway has previously been identified. However, interestingly, better characterised arrestin family members, such as the β-arrestins (arrestin beta 1 and arrestin beta 2 (ARRB1/2)), have been identified as either promoters or suppressors of apoptosis [42]. Due to the nature of our identification of Arrdc3 in screens involving activation of TRP53, the most intriguing connection is the ability of ARRB2 to bind MDM2, the E3 ubiquitin ligase which is primarily responsible for the degradation of TP53 [43]. Indeed, in this role, ARRB2 activity appears to be partially reliant on GPCR signalling, an increase of which correlated with ARRB2-MDM2 binding strength [43]. While a similar relationship between MDM2 and ARRDC3 would line up well with our observations of a role for ARRDC3 in TP53-mediated tumour suppression, no evidence exists for ARRDC3 regulating MDM2. However, it is noteworthy that NEDD4, WWP1, and WWP2, some of the HECT E3 ubiquitin ligases thought to interact physically with ARRDC3, are also thought to act as regulators of MDM2 and/or TP53 [44,45,46]. As such, exploration of the proteome surrounding ARRDC3, particularly in malignant cells, would help inform our understanding of the role of ARRDC3.

At least two other groups have previously generated Arrdc3 knockout mice via independent gene trap insertions, as opposed to our CRISPR/Cas9 approach [20, 35]. Interestingly, while we observed perinatal lethality in Arrdc3−/− animals, Patwari et al. observed ~7% homozygous knockout animals surviving to weaning [35], and while Shea et al. observed no surviving homozygous knockout animals, they were able to partially rescue homozygous knockout lethality by providing the breeding mice with a high-fat diet [20]. While varying diet is not something we tested, we did observe significant morphological defects in E19.5 animals (Fig. 3), something other groups did not examine. Particularly noteworthy are the heart defects we observed which, while incompletely penetrant (Figure S5), generate some intriguing connections with other literature. For example, in humans, ARRDC3 copy number variation has been linked to congenital heart defects [47]. In Drosophila melanogaster, an unnamed ARRDC3 orthologue (CG1105) was the 10th strongest hit in a 7000+ gene RNAi screen for regulators of heart function [48]. ARRDC3 variants have also been associated with congestive heart failure in cattle [49]. In the context of human malignancies, analysis of 5989 patient samples of haematopoietic cancers using the cBioPortal database showed no correlation between ARRDC3 mutation (loss/amplification) with either MYC amplification or TP53 loss [50,51,52]. However, expanding this search to include all types of cancers, across 69223 patient samples, the database indicated ARRDC3 mutation (loss/amplification) co-occurred significantly (two-sided Fischer exact tests, each p < 0.001) with both MYC amplificiation and TP53 loss. These results suggest there may be an unappreciated role for ARRDC3 in human cancers and, coupled with our phenotypic observations and the literature studying ARRDC3 in a highly diverse set of contexts, there is some level of universal importance for ARRDC3.

In summary, we have characterised a role for ARRDC3 in regulating the process of TP53-mediated suppression of Eμ-Myc-driven lymphomagenesis. This is a novel role for this relatively understudied member of the arrestin gene family. Further investigations into ARRDC3 and related arrestin family members, and in particular their relationship to TRP53, are certainly warranted, and may uncover new therapeutic avenues or key regulatory avenues that could bypass defects in TP53 for cancer therapy.

Materials and methods

CRISPR/Cas9 screening with nutlin-3a

Once stably transfected AF47A-Cas9-Yusa Eμ-Myc lymphoma cells were generated (see Supplementary Data), input samples were collected for DNA. 16 × 106 AF47A-Cas9-Yusa lymphoma cells were then plated into T125 flasks in quadruplicate, and each replicate treated with nutlin-3a (10 μM, Cayman Chemical #18585) or DMSO (vehicle control; Sigma-Aldrich #D4540) for 24 h. From the treated samples, the remaining live cells (propidium iodide (PI)-negative) were sorted via FACS, and samples taken for DNA extraction.

CRISPR screen sequencing and analysis

DNA was extracted from all samples using the DNeasy Blood and Tissue kit (QIAGEN #69506). Vector indexing was performed by single-step PCR using established primers [29] which contain overhangs to enable preparation of an Illumina sequencing library. Indexed samples were then pooled, DNA clean up performed using Ampure XP Beads (Beckman Coulter #A63880), and pools prepared for sequencing on a MiSeq (Illumina) machine according to the manufacturer’s instructions. MAGeCK v0.4 and v0.5 [53] was used to rank sgRNA enrichment within the different treatment samples.

Mouse husbandry and generation of Arrdc3 knockout mice

Eµ-Myc transgenic mice (which are always kept as heterozygotes) have been reported [13] and were maintained on a C57BL/6 background. See supplementary materials for additional information.

To generate the Arrdc3 knockout mouse strain, exons 1-7 of Arrdc3 (mouse chromosome 13) were deleted in C57BL/6J fertilised oocytes by CRISPR editing using two sgRNAs with the sequences 5′-GAGACTACTAGGTGACGGGAAGG-3′ and 5′-AGCCATCCTCATCGACTACAGGG-3′ following established protocols [32]. Genotyping primers and expected sizes are listed in Table S2. To minimise CRISPR off-targets, mice were backcrossed twice to C57BL/6 mice, and frequently outcrossed to Eµ-Myc mice. Deletion of the targeted region in the Arrdc3 gene was confirmed by NGS.