Abstract
Increased protein synthesis supports the rapid proliferation associated with cancer. The Rpl24Bst mutant mouse reduces the expression of the ribosomal protein RPL24 and has been used to suppress translation and limit tumorigenesis in multiple mouse models of cancer. Here we show that Rpl24Bst also suppresses tumorigenesis and proliferation in a model of colorectal cancer with two common patient mutations, Apc and Kras. In contrast to previous reports, Rpl24Bst mutation has no effect on ribosomal subunit abundance but suppresses translation elongation through phosphorylation of eEF2, reducing protein synthesis by 40% in tumour cells. Ablating eEF2 phosphorylation in Rpl24Bst mutant mice by inactivating its kinase, eEF2K, completely restores the rates of elongation and protein synthesis. Furthermore, eEF2K activity is required for the Rpl24Bst mutant to suppress tumorigenesis. This work demonstrates that elevation of eEF2 phosphorylation is an effective means to suppress colorectal tumorigenesis with two driver mutations. This positions translation elongation as a therapeutic target in colorectal cancer, as well as other cancers where the Rpl24Bst mutation has a tumour suppressive effect in mouse models.
Introduction
Tumour cells require rapid protein synthesis to acquire sufficient biomass in order to divide and, as such, protein synthesis is directly regulated by many oncogenic signalling pathways (Proud 2019; Robichaud et al. 2019; Smith et al. 2021). As well as exploiting protein synthesis to drive cell division, cancers use translation to selectively synthesise a proteome geared towards proliferation, survival and immune evasion. For example, in colorectal cancer (CRC) translation of the mRNA encoding the proto-oncogene c-MYC is selectively upregulated by eIF4E and mTORC1 signalling (Knight et al. 2020a). Likewise, independent reports have shown that the expression of the immune suppressive ligand PD-L1 is maintained on tumour cells by the activity of the translation factors eIF4A and eIF5B as well as phosphorylation of eIF4E and eIF2 (Suresh et al. 2020; Xu et al. 2019; Cerezo et al. 2018).
APC is the most commonly mutated gene in CRC, followed by TP53 and then KRAS (Guinney et al. 2015). We have previously shown that Apc-deficient mouse models of CRC are dependent on fast translation elongation, a process which can be suppressed by rapamycin leading to near complete reversal of tumorigenesis (Faller et al. 2015). This approach has had clinical success, where rapamycin (sirolimus) regressed APC-deficient polyps of familial adenomatous polyposis patients in two independent clinical trials (Yuksekkaya et al. 2016; Roos et al. 2020). Clinical data also suggest that CRCs increase translation elongation to potentiate proliferation, exemplified by the lower expression of the elongation inhibiting kinase eEF2K correlating with worse patient survival (Ng et al. 2019). However, the regulation of translation elongation in CRC is complex, notably being influenced by specific cancer-associated mutations. We have shown that mutation of Kras drives resistance to rapamycin in Apc-deficient models both in terms of its effect on elongation and on proliferation (Knight et al. 2020a). This is consistent with KRAS-mutant CRCs being resistant to rapalogues, and other therapeutics (Ng et al. 2013; Spindler et al. 2013; DeStefanis et al. 2019) and highlights the unmet need for effective therapies against KRAS-mutant cancers. Although a recently developed compound covalently targeting KRASG12C mutation has shown remarkable potency against this specific mutation (Hong et al. 2020).
Evidence suggests that targeting translation in KRAS-mutant colorectal cancer can still be effective. As well as our recent study re-sensitizing Kras-mutant CRCs to rapamycin by targeting translation initiation (Knight et al. 2020a), we have demonstrated that Kras-mutant models of CRC depend upon the transporter SLC7A5 to maintain protein synthesis by facilitating the influx of amino acids (Najumudeen et al. 2021). These data support protein synthesis as a tractable target in CRC, with the discovery of additional factors regulating these pathways only improving the potential to target protein synthesis in the clinic (Knight and Sansom 2021).
In this study we analyse the previously characterised Rpl24Bst mutation in models of CRC with Apc-deletion and Kras mutations. This spontaneously arising 4 nucleotide deletion in the Rpl24 gene, which encodes RPL24 (a component of the 60S ribosomal subunit also called large ribosomal protein subunit eL24), disrupts splicing of its mRNA, effectively resulting in a Rpl24 heterozygous animal (Oliver et al. 2004). Animals present with impaired dorsal pigmentation and malformed tails, among other defects, leading to the designation of a belly spot and tail (Bst) phenotype and the Rpl24Bst designation. This tool has been used to suppress overall protein synthesis in genetically engineered mouse models of c-MYC driven B-cell lymphoma (BCL), Pten-deficient T-cell acute lymphoblastic leukaemia (T-ALL) and T-cell specific Akt2 activation (Barna et al. 2008; Signer et al. 2014; Hsieh et al. 2010). In these studies, tumorigenesis increased total protein synthesis, which was rescued by combination with the Rpl24Bst mutation. Suppression of protein synthesis was sufficient to slow tumorigenesis, with some Rpl24Bst/+ mice surviving over 3 times longer than the median survival of tumour model mice wild-type for Rpl24. However, the means by which the Rpl24Bst mutation suppresses protein synthesis was not addressed in these studies, instead deferring to the original observation that there is likely a defect in ribosome production (Oliver et al. 2004).
Here we show that decreased expression of RPL24 suppresses proliferation and extends survival in Apc-deficient Kras-mutant pre-clinical mouse model of CRC. Importantly, we find that reduced RPL24 does not alter the available pool of ribosomal subunits, as previously suggested, but instead alters signalling that regulates a translation factor. Specifically, we observe increased phosphorylation of eEF2, an event that negatively regulates translation elongation. We directly measure translation elongation to show that the Rpl24Bst mutation suppresses protein synthesis at the elongation step, consistent with increased phosphorylation of eEF2. Relieving P-eEF2 by inactivating its inhibitory kinase, eEF2K, completely restores translation elongation and protein synthesis rates as well as reversing the beneficial effect of Rpl24Bst mutation in our tumour models. Interestingly, we find that the Rpl24Bst mutation has no effect in Kras-wild-type models. We attribute this to a specific requirement for physiological RPL24 in Kras-mutant cells, which may provide additional mechanisms to target these cells clinically.
Finally, we provide evidence from transcriptomic and proteomic analyses of patient tissue that supports the signalling pathways uncovered in our pre-clinical models being altered in the human disease. Altogether this work demonstrates that the Rpl24Bst mutation is tumour suppressive in a solid tumour model, and elucidates an unexpected mode of action underlying its impact on protein synthesis. This has implications for targeting translation elongation in cancer and provides mechanistic insight to supplement the previously published efficacy of the Rpl24Bst mouse in models of cancer and other diseases.
Results
RPL24Bst mutation does not alter intestinal homeostasis but suppresses the rate of translation
Prior to addressing the role of RPL24 in intestinal tumorigenesis we first analysed whether the Rpl24Bst mutation had any effect on normal intestinal homeostasis (Figure 1A). We observed a reduction in RPL24 expression (Figure 1B), but no dramatic differences in intestinal architecture, crypt proliferation shown by BrdU incorporation (Figure 1A) or abundance of stem cells (Olfm4), Paneth cells (lysozyme) or goblet cells (AB/PAS) (Figure S1A). Similarly, homeostasis in the colons of Rpl24Bst mutant mice was unaffected, exemplified by no change in proliferation scored by BrdU incorporation (Figure S1C). In accordance with these in vivo observations, ex vivo organoids made from the small intestines of Rpl24Bst/+ mice established in culture and grew comparably to wild-type controls (Figures 1C and S1D). Surprisingly, we measured a >40% reduction in total protein synthesis, by 35S-methionine labelling, in Rpl24Bst/+ organoids compared to wild-type counterparts (Figure 1D). Therefore, despite no change in homeostasis, Rpl24Bst mutation had a dramatic effect on protein synthesis. This indicates a resistance to reduced protein synthesis in the wild-type intestine, which appear to function normally despite a dramatic reduction in protein output.
We then investigated how Rpl24Bst mutation suppresses translation. Performing sucrose density gradients to quantify the number of ribosomes engaged in active translation we observed an increase in ribosomes bound to mRNAs in polysomes in Rpl24Bst/+ organoids, particularly the heavy polysomes (Figure 1E). This appears to contradict the reduction in global protein synthesis observed in Figure 1D. However, ourselves and others have previously observed increased polysomes in conjunction with reduced protein synthesis in model systems where translation elongation is reduced (Knight et al. 2015; Faller et al. 2015). In these instances slowed translation elongation increased the abundance of polysomes via changes in signalling to the elongation factor eEF2. We therefore assayed the regulatory phosphorylation of eEF2 (threonine 56 / T56) in wild-type and Rpl24Bst mutant samples. This phosphorylation event excludes eEF2 from the ribosome thereby impairing the translocation step of translation elongation, reducing protein synthesis (Ryazanov and Davydova 1989; Carlberg et al. 1990).
In small intestinal tissue assayed by IHC we observed an increase in the P-eEF2 T56, specifically in the proliferating crypt and transit amplifying zone of Rpl24Bst/+ mouse intestines (Figure 1B). Likewise we observed a 75% increase in P-eEF2 T56 in lysates generated from Rpl24Bst/+ organoids compared to wild-type organoids (Figure 1D). These organoids also showed a 50% reduction in RPL24 expression by western blot (Figure 1D). Thus, in these two proliferative settings (intestinal crypts in situ and ex vivo organoids) Rpl24Bst mutation increases the phosphorylation of eEF2, which is known to suppress translation. Surprisingly, we observed that in the differentiated villus, Rpl24Bst mutation suppressed P-eEF2 T56 (Figure 1B). The reasons for this are unclear, but may relate to different cell functions in the two compartments. This effect on P-eEF2 T56 is specific, as we observed no effect on the phosphorylation of other translation related proteins, 4E-BP1 or RPS6 (at S240/S244), readouts for modulation of signalling downstream of mTORC1 (Figure S1A-B).
Altogether, this analysis of wild-type tissue indicates that physiological RPL24 expression is not required for proliferation or function, but reduction of RPL24 expression reduces the rate of translation. Interestingly, this involves regulation of translation elongation and correlates with increased P-eEF2 (Figure 1G). Previously the Rpl24Bst mutation had been suggested to suppress ribosome biogenesis, resulting in uneven 40S and 60S ratios. However, we observed no alteration in the relative levels of the 40S and 60S subunits in sucrose density gradients (Figure S1E), consistent with previous reports that RPL24 deletion has little effect on ribosome biogenesis (Barkić et al. 2009).
The wild-type mouse intestine regenerates following γ-irradiation, dependent on Wnt and MAPK signalling pathways. These pathways are often deregulated in colorectal tumours, such that this intestinal regeneration acts as a surrogate for oncogenic potential, with reduced regenerative capacity indicative of reduced tumorigenic proliferation (Faller et al. 2015). We observe that mutation of Rpl24 restricts regeneration of the small intestine (Figure S1F). Thus, RPL24 expression enables regeneration, which may correlate with effects in tumorigenesis.
RPL24 is required for proliferation in Apc-deficient Kras-mutant intestinal tumours
Next, we analysed the effect of the Rpl24Bst mutation on a model of CRC driven by tamoxifen-inducible VillinCreER mediated deletion of Apc and activation of KRAS with a G12D mutation. This model can be used with homozygous deletion of Apc (Apcfl/fl) where intestinal hyperproliferation generates a short term (3-4 days) hyperproliferation model or with heterozygous deletion of Apc (Apcfl/+) where intestinal adenomas form following spontaneous loss of the second copy of Apc. The recombination of a lox-STOP-lox allele at the endogenous Kras locus expresses a constitutively active G12D mutant form of the protein. Mice were generated with the Apcfl/fl and KrasG12D alleles with and without the Rpl24Bst mutation (Figure 2A). Hyperproliferation in the Apcfl/fl KrasG12D/+ Rpl24Bst/+ mutant mice was significantly suppressed compared to Apcfl/fl KrasG12D/+ control mice (Figure 2B and C). Reduced RPL24 expression was confirmed by IHC (Figure 2C) and coincided with increased P-eEF2 throughout the proliferative crypt area of the Apcfl/fl KrasG12D/+ Rpl24Bst/+ intestine (Figure 2C and S2A). Hyper-proliferation in the colon mirrored that of the small intestine, with reduced proliferation in mice mutant for Rpl24 compared to those expressing wild-types levels of RPL24 (Figure S2B). In parallel, organoids were derived from the small intestines of the same genotypes (Apcfl/fl KrasG12D/+ and Apcfl/fl KrasG12D/+ Rpl24Bst/+) and their growth compared. Rpl24Bst mutation resulted in significantly less proliferation ex vivo (Figure 2D), consistent with the in vivo experiment shown in Figure 2B.
In the tumour model, where a single copy of Apc is deleted, Apcfll+ KrasG12D/+ Rpl24Bst/+ mice lived on average 32 days longer than Apcfll+ KrasG12D/+ controls, an extension of survival of 45% (Figure 2E). Apcfll+ KrasG12D/+ Rpl24Bst/+ organoids derived from the adenomas in this tumour model grew more slowly than controls (Figure S2C). There was no significant difference in the number of tumours at experimental endpoint (Figure S2D), indicating that adenomas can form but take longer to reach a clinically significant burden. Therefore, RPL24 enables proliferation in Apc-deficient, KRAS-activated cells within the intestinal epithelium of the mouse.
RPL24 maintains translation elongation in Apc-deficient Kras-mutant intestinal tumour models
The suppression of tumorigenesis in the Apc-deficient Kras-mutant model correlated with increased phosphorylation of eEF2 (Figure 2C and S2A). To investigate this further we used 3 methods to measure the rate of translation; polysome profiling, 35S-methionine labelling and harringtonine run-off assays (Figure 3A). Polysome profiling from extracted crypts from Apcfl/fl KrasG12D/+ and Apcfl/fl KrasG12D/+ Rpl24Bst/+ showed an increase in polysomes with the Rpl24 mutation (Figure 3B), and notably a significant increase in the quantity of heavy polysomes (Figure 3C). Intestinal organoids of the same genotype showed a 35% reduction in 35S-methionine incorporation (Figure 3D). These same organoids had a greater than 40% decrease in elongation rate measured by harringtonine run-off (Figures 3D and S3A). Together these data provide strong evidence that normal RPL24 expression is required to maintain translation elongation in this CRC model. Polysome profiles and protein synthesis rate measurements from Apcfl/+ KrasG12D/+ Rpl24Bst/+ adenoma cultures also showed more polysomes and lower protein synthesis compared to control adenoma cultures (Figures S3B and C). Furthermore, the reduced protein synthesis rate does not correlate with differences in free ribosomal subunit availability as the ratio of 40S to 60S subunits is unchanged by the Rpl24 mutation in Apc-deficient Kras-mutant cells (Figure S3D).
Interpreting this molecular analysis in conjunction with the effects on tumorigenesis leads to the conclusion that RPL24 expression maintains translation elongation and protein synthesis rates, which in turn maintain tumour-related proliferation (Figure 3F). Furthermore, suppressing RPL24 expression increases P-eEF2, which decreases the rate of elongation and overall protein synthesis and correlates with suppressed tumorigenesis and proliferation.
Rpl24 mutation has no effect in CRC models expressing wild-type Kras
In parallel to analysing the effect of Rpl24 mutation in Apc-deficient Kras-mutant intestinal tumours, we also assessed its role in Apc-deficient models wild-type for Kras. We have previously shown that these are dependent on signalling from mTORC1 to maintain low levels of P-eEF2, and that this can be suppressed by rapamycin treatment to great therapeutic benefit (Faller et al. 2015). We observed that the hyperproliferation in the small intestine or colon of Apcfl/fl was not reduced by the Rpl24Bst mutation (Figure 4A and B, and S4A). Furthermore, in both germline ApcMinl+ and inducible Apcfll+ models of Apc-deficiency we see no benefit of the Rpl24Bst mutation, with no difference in survival or tumour development (Figures 4C and S4B). In agreement, Apcfl/fl organoids with the Rpl24Bst mutation grew at an identical rate to those wild-type for Rpl24 in culture (Figure 4D). Together these data demonstrate that reduced RPL24 expression does not limit tumorigenesis in Apc-deficient CRC models with wild-type Kras.
Despite no effect on proliferation, we observed an increase in P-eEF2 in Apcfl/fl Rpl24Bst/+ intestines compared to Apcfl/fl (Figure 4B), showing that P-eEF2 is consistently increased in the intestinal crypts of Rpl24Bst/+ mice. However, we see no change in the ratio of polysome to sub-polysomes in Apcfl/fl Rpl24Bst/+ intestines compared to Apcfl/fl (Figure S4C-D) and no change in protein synthesis rate between organoids of these same genotypes (Figure 4D). In contrast, rapamycin treatment significantly reduces protein synthesis in Apcfl/fl organoids treated in parallel (Figure 4D). From these data we conclude that the change in P-eEF2 does not limit the rate of protein synthesis which likely allows efficient tumorigenesis in these Kras-wild-type models of CRC. Importantly, neither P-4E-BP1 or P-RPS6 S240/244 are reduced by Rpl24Bst mutation in the Apcfl/fl model (Figure S4E), indicating that translation-promoting mTORC1 signalling remains high.
We hypothesised that the reason for the KRAS-specificity seen with the Rpl24Bst mutation may relate to expression levels between the different genotypes analysed. Using unbiased RNA sequencing data from wild-type, Apcfl/fl and Apcfl/+ KrasG12D/+ small intestinal tissue we observed a consistent increase in ribosomal protein expression following Apc deletion, then again following KRAS activation (Figure S4F). This is consistent with previous reports (Smit et al. 2020), and a requirement to increase protein synthesis as a direct consequence of KRAS activation. Indeed, the mRNA for all ribosomal proteins with sufficient reads were increased on average nearly 1.5-fold by KRAS activation in the small intestine (Figure 4F). In contrast, the Rpl24 mRNA was only increased by 1.14-fold following Kras mutation, despite a nearly two-fold increase following deletion of Apc (Figure 4F). This manifests as a significant difference in the RNA expression of Rpl24 compared to the other ribosomal proteins. Therefore, RPL24 expression may be sufficient in Rpl24Bst mice in Apc-deleted models, but then becomes limiting following Kras-mutation due to the relatively limited upregulation of Rpl24 expression accompanying KRAS activation.
Genetic inactivation of eEF2K completely reverses the anti-proliferative benefit of Rpl24 mutation
Thus far we have demonstrated a correlation between the increase in P-eEF2 and the slowing of translation elongation following Rpl24Bst mutation. To test whether the slowing of elongation caused by Rpl24Bst mutation was dependent on P-eEF2, we used a whole-body point mutant of eEF2K, the kinase that phosphorylates eEF2, which almost completely inactivates its kinase activity (Gildish et al. 2012). We crossed this Eef2kD273A/D273A allele to the Apcfl/fl KrasG12D/+ Rpl24Bst/+ mice to generate Apcfl/fl KrasG12D/+ Rpl24Bst/+ Eef2kD273A/D273A mice. In the short term hyperproliferation model, the inactivation of Eef2k completely reversed the suppression of proliferation seen in Apcfl/fl KrasG12D/+ Rpl24Bst/+ small intestines (Figure 5A-C), and the medial colon (Figure S5A). The kinase inactive Eef2k allele resulted in no detectable P-eEF2 (Figures 5C and S2A), and we previously reported no difference in hyper-proliferation between Apcfl/fl KrasG12D/+ and Apcfl/fl KrasG12D/+ Eef2kD273A/D273A models (Knight et al. 2020a). The reversal in proliferation rate with Eef2k and Rpl24Bst mutations was also seen in intestinal organoid growth after 3 days (Figure 4D). Furthermore, inactivation of eEF2K reverted the survival benefit of the Rpl24Bst mutation in the Apcfll+ KrasG12D/+ tumour model (Figure 4E). This experiment also shows the lack of effect of the Eef2kD273A/D273A mutation on tumorigenesis. Indeed, Eef2kD273A/D273A had no impact on tumorigenesis in inducible Apcfll+ and germline ApcMinl+ models of Apc-deficient intestinal tumorigenesis (i.e. expressing wild-type Kras) (Figures S5B-C). Heterozygous inactivation of one copy of Eef2k resulted in a slight reversal of the extension of survival associated with Rpl24Bst mutation in the Apcfll+ KrasG12D/+ tumour model (Figure S5D).
Therefore, mutation of Rpl24 requires functional eEF2K to suppress proliferation and extend survival in this model of CRC. There was no alteration in the number of tumours in the ageing model at endpoint (Figure S4E), again identifying tumour cell proliferation, rather than tumour initiation, as the important factor regulated by RPL24 and eEF2K. These data also identify the effect of Rpl24Bst mutation on the tumour phenotype was entirely dependent upon eEF2K activity.
Rpl24Bst mutation suppresses translation exclusively via eEF2K/P-eEF2
Next, we addressed the molecular consequences of inactivation of eEF2K downstream of Rpl24Bst mutation. The reduction in protein synthesis that results from Rpl24Bst mutation is completely reversed by eEF2K inactivation (Figure 6A), with Apcfl/fl KrasG12D/+ Rpl24Bst/+ Eef2kD273A/D273A organoids having an almost identical translation capacity as Apcfl/fl KrasG12D/+ controls. Similarly, the rate of ribosome run-off was also reverted in Apcfl/fl KrasG12D/+ Rpl24Bst/+ Eef2kD273A/D273A compared to Apcfl/fl KrasG12D/+ Rpl24Bst/+ organoids, again to the same rate as controls with wild-type Rpl24 (Figures 6B and S6A). In agreement, crypt fractions from Apcfl/fl KrasG12D/+ Rpl24Bst/+ Eef2kD273A/D273A mice have a reduced number of heavy polysomes compared to Apcfl/fl KrasG12D/+ Rpl24Bst/+ crypt cells (Figure S6B), indicating faster translation elongation following eEF2K inactivation.
This has important implications for the function of RPL24, showing that ribosomes in Rpl24Bst mutant cells can elongate efficiently despite the reduction in RPL24 expression. However, reduced RPL24 increases P-eEF2 which inhibits elongation, with an absolute requirement for eEF2K for this (Figure 6C). Importantly, the combination of the Rpl24 and eEF2K mutants shows that when P-eEF2 is abolished elongation occurs at normal speed, despite reduced expression of RPL24.
The expression of RPL24, EEF2K and EEF2 is indicative of fast elongation in human CRC
Using pre-clinical mouse models we have demonstrated that physiological RPL24 expression maintains low eEF2K-mediated phosphorylation of eEF2 (Figure 7A). Hypo-phosphorylated eEF2 then ensures rapid protein synthesis enabling tumour proliferation in vivo. This requirement for RPL24 is specific to mutant-Kras models, indicating a likely relationship between KRAS and RPL24 that is beyond the scope of this work. We next sought to position this pre-clinical work in the context of clinical studies of the human disease. Using publicly available datasets for RNA expression in normal and colon cancer tissue we observe increased RPL24 and EEF2 expression in conjunction with reduced EEF2K expression (Figure 7B). This mirrors the signalling pathways in our mouse models which focus on ensuring high expression of active eEF2. Elevated EEF2 message and reduced EEF2K message in these clinical samples is consistent with conservation of this signalling in the clinic. Similar results are seen for the protein expression of RPL24, eEF2K and eEF2 from colon adenocarcinoma samples (Figure S7A), and for the three mRNAs in rectal adenocarcinoma (Figure S7B). These expression analyses highlight the conservation of the proliferative tumour-associated signalling pathways characterised in our pre-clinical mouse models and patient samples from the clinic.
Discussion
The original characterisation of the Rpl24Bst mutation identified a defect in ribosome biogenesis affecting the synthesis of 60S subunits (Oliver et al. 2004). The evidence to support this was limited to analyses of fasted/refed mouse livers for nascent rRNAs and by polysome profiles purporting to show reductions in 28S rRNA precursors and mature 60S subunits respectively. In contrast, RNAi depletion of RPL24 in human cell lines had no effect on ribosome biogenesis, or on the relative abundance of 40S and 60S subunits (Barkić et al. 2009; Wilson-Edell et al. 2014a). Furthermore, two reports have identified RPL24 as an exclusively cytoplasmic protein, leading to the hypothesis that it is assembled into mature ribosomes in the cytoplasm after rRNA synthesis and processing has already occurred (Barkić et al. 2009; Saveanu et al. 2003). In agreement with this, in the hindbrain of E9.5 embryos, the Rpl24Bst mutation had no effect on nucleolar architecture, indicating that it is not required for nucleolar function (Herrlinger et al. 2019). Our data agree with these later examples, as we fail to see any effect of the Rpl24Bst mutation on 60S:40S ratio in wild-type and transformed mouse intestines. Of further importance to the field, we also demonstrate in our tumour model that the translation defect in Rpl24Bst mutant mice is restored to normal levels following inactivation of eEF2K in Rpl24Bst mutant mice showing that the defect is dependent on eEF2K. From this, we conclude that ribosomes produced in Rpl24Bst/+ mice would allow protein synthesis to proceed at near physiological levels, however this process is restricted via signalling through eEF2K/P-eEF2.
RPL24 depletion suppresses tumorigenesis in Apc-deficient Kras-mutant (APC KRAS) mouse models of CRC, but not in Apc-deficient Kras-wild-type (APC) models. In line with this, protein synthesis is reduced following depletion of RPL24 in our APC KRAS models, but not in APC models, despite induction of P-eEF2 in both cases. We previously showed that suppression of translation elongation via mTORC1/eEF2K/P-eEF2, using rapamycin, in the same APC models dramatically suppressed proliferation and extended survival (Faller et al. 2015). Thus, while RPL24 depletion or rapamycin treatment of Apc-deficient intestines each induce P-eEF2, only rapamycin treatment suppresses protein synthesis. The effect on protein synthesis appears to be the differential driving this divergence, with proliferation only impaired when protein synthesis is reduced. Crucially, RPL24 deficiency does not suppress mTORC1 activity since phosphorylation of the mTORC1 effectors 4E-BP1 and RPS6 (at S240/S244) are not reduced in Apc-deficient Rpl24Bst, providing an explanation of how proliferation and protein synthesis are maintained in Apc-deficient Rpl24Bst mutant animals. We also provide evidence that Rpl24 mRNA is not elevated in line with other ribosomal protein mRNAs following Kras-mutation, which could explain why physiological expression levels of RPL24 are required for proliferation in Kras-mutant models.
It is interesting to reflect on previous work with the Rpl24Bst mutant mouse considering the mechanism we have described here. The Rpl24Bst mutation suppresses tumorigenesis in three mouse models of different types of blood cancer (Signer et al. 2014; Barna et al. 2008; Hsieh et al. 2010). In each case the Rpl24Bst mutation was found to decrease translation, although how the mechanism of translational suppression was not fully determined. Barna et al demonstrated a reduction in the cap-dependent translation using a luciferase reporter in Rpl24Bst MEFs (Barna et al. 2008). Furthermore, Rpl24Bst mutation dramatically suppressed cap-dependent translation following MYC activation, consistent with Rpl24Bst slowing tumorigenesis via reduced translation. We have shown that in mouse models of CRC the molecular mechanism by which normal levels of RPL24 maintains translation is via eEF2K and P-eEF2, therefore implicating this pathway in these previously studied blood cancer models.
The Rpl24Bst mutation has also been used to analyse brain development from neural progenitor cells, and explored as a model for retinal degenerative disease (Riazifar et al. 2015; Herrlinger et al. 2019). These studies found a revertant phenotype in neural progenitor cells overexpressing LIN28A and the defect in subretinal angiogenesis in Rpl24Bst/+ mice. The role of translation elongation with respect to these phenotypes should now be analysed. eEF2K, and the many pathways it integrates, are potential targets for intervention in these in vivo models. It will also be of interest to determine how the inactivation of eEF2K affects the whole-body phenotypes of the Rpl24Bst mouse, such as the coat pigmentation and tail defects.
eEF2K has a pleiotropic effect in tumorigenesis, acting akin to a tumour suppressor or promoter dependent on the context (Knight et al. 2020b). In some cancers eEF2K promotes tumorigenesis. For example, under nutrient deprivation eEF2K acts as a pro-survival factor in transformed fibroblasts and tumour cell lines by suppressing protein synthesis to ensure survival (Leprivier et al. 2013). Here we show that eEF2K inactivation does not modify intestinal tumorigenesis. However, eEF2K is required for the suppression of tumorigenesis and protein synthesis following rapamycin/mTORC1 inhibition in APC cells (Faller et al. 2015) or Rpl24Bst mutation in APC KRAS cells (this work). Therefore, although eEF2K does not directly drive tumorigenesis, low eEF2K activity ensures that there is no blockade of translation or proliferation in tumour cells. Furthermore, eEF2K expression is required for drug or signalling responses that suppress tumorigenesis, giving it tumour suppressive activity. In accordance, CRC patients with low eEF2K protein expression suffer a significantly worse prognosis (Ng et al. 2019). Furthermore, we present mRNA and protein expression data showing that eEF2K is reduced in clinical CRC samples compared to normal tissue. This agrees with a model where low eEF2K allows rapid translation elongation to promote proliferation.
Using the same clinical data sets we demonstrate that both RPL24 and eEF2 are elevated in CRC, consistent with their roles in promoting translation and proliferation. This presents the possibility of direct targeting of either RPL24 or eEF2 for anti-cancer benefit. RNAi against RPL24 in human breast cancer cell lines dramatically reduced proliferation (Wilson-Edell et al. 2014b). Similarly, targeting eEF2 using a compound reported to slow its exit from the ribosome and thus the rate of protein synthesis, reduces cell line and patient derived organoid growth (Stickel et al. 2015; Keysar et al. 2020). These reports agree with the data presented here suggesting that targeting of RPL24 or eEF2 would be beneficial in CRC.
This work uncovers an unexpected role for the ribosomal protein RPL24 in the regulation of translation elongation, acting via eEF2K/P-eEF2 signalling. We demonstrate that depletion of RPL24 suppresses tumorigenesis in a pre-clinical mouse model of a solid tumour, supplementing the three reports of the mutation suppressing blood tumour models. We provide genetic evidence supported by molecular assays of translation elongation to demonstrate that RPL24 depletion activates eEF2K to elicit tumour suppression in our models. We also speculate as to the role of eEF2K in the previously published blood cancer models where RPL24 depletion was beneficial. This work provides additional evidence for the anti-tumorigenic role of the eEF2K in CRC, highlighting the potential for targeting translation elongation for this disease.
Materials and methods
Materials availability
The mouse strains used will be made available on request. However, this may require a Materials Transfer Agreement and/or a payment if there is potential for commercial application. We ourselves are limited by the terms of Materials Transfer Agreements agreed to by the suppliers of the mouse strains.
Mouse studies
Experiments with mice were performed under a licence from the UK Home Office (licence numbers 60/4183 and 70/8646). All mice used were inbred C57BL/6J (Generation ≥8) and were housed in conventional cages with a 12-hour light/dark cycle and ad libitum access to diet and water. Mice were genotyped by Transnetyx in Cordova, Tennessee. Experiments were performed on mice between the ages of 6 and 12 weeks, both male and female mice were used. Sample sizes for all experiments were calculated using the G*Power software (Faul et al. 2009) and are shown in the figures or legends. Researchers were not blinded during experiments. The VillinCreER allele (El Marjou et al. 2004) was used for intestinal recombination by intraperitoneal (IP) injection of tamoxifen in corn oil at a final in vivo concentration of 80mg/kg. The Apc flox allele, ApcMin, KrasG12D lox-STOP-lox allele, Eef2kD273A and Rpl24Bst alleles were previously described (Jackson et al. 2001; Shibata et al. 1997; Gildish et al. 2012; Oliver et al. 2004; Moser et al. 1990). Tumour model experiments began with a single dose of tamoxifen after which mice were monitored until they showed signs of intestinal disease – paling feet from anaemia, weight loss and hunching behaviour. Tumours were scored macroscopically by counting the number visible after fixation of intestinal tissue. For short-term experiments, mice wild-type for Kras were induced on consecutive days (days 0 and 1) and sampled 4 days after the first induction (day 4). Mice bearing the KrasG12D allele were induced once and sampled on day 3 post-induction. Where indicated, 250μL of BrdU cell proliferation labelling reagent (Amersham Bioscience RPN201) was given by IP injection 2 hours prior to sampling. For the regeneration experiments, mice were exposed to 10 Gy of γ-irradiation at a rate of 0.423 Gy per minute from a caesium 137 source. They were then sampled 72 hours after irradiation.
Histology and immunohistochemistry
Tissue was fixed in formalin and embedded in paraffin. Immunohistochemistry (IHC) staining was carried out as previously (Faller et al. 2015), using the following antibodies; BrdU (BD Biosciences #347580), P-eEF2 T56 (Cell Signaling Technology (CST) #2331), P-4E-BP1 T37/46 (CST #2855), P-RPS6 S235/6 (CST #4858), P-RPS6 S240/4 (CST #5364), RPL24 (Sigma Aldrich HPA051653) and Lysozyme (Dako A0099). The IHC protocol followed the Vector ABC kit (mouse #PK-6102, rabbit #PK-6101). RNAScope analysis was carried out according to the manufacturers guidelines (ACD) using probes to murine Olfm4 (#311838). For all staining, a minimum of 3 biological replicates were stained and representative images used throughout. For BrdU scoring in short-term model experiments tissue was fixed in methanol:choloform:acetic acid at a ratio 4:2:1 then transferred to formalin and embedded paraffin.
Intestinal organoid culture
Crypt cultures were isolated and then maintained as previously described (Knight et al, in press). In all crypt culture experiments, multiple biologically independent cultures were generated and analysed from different animals of the shown genotypes. DMEM/F12 medium (Life Technologies #12634-028) was supplemented with 5 mM HEPES (Life Technologies #15630-080), 100 U/mL penicillin/streptomycin (Life Technologies #1540-122), 2 mM L-glutamine (Life Technologies #25030-024), 1x N2, 1x B27 (Invitrogen #17502-048 and #12587-010), 100 ng/mL noggin (Peprotech #250-38) and 50 ng/mL EGF (Peprotech #AF-100-15). Wild-type cultures were also supplemented with 500 ng/mL R-spondin (R&D Systems #3474-RS). For ex vivo growth assays, cells were plated in technical triplicate in 96 well plates, and proliferation measured using Cell-Titer Blue (Promega #G8080) added to previously untreated cells each day for up to 4 days. Rapamycin (LC Laboratories R-5000) was dissolved in DMSO and administered for 24 hours, comparing to just DMSO vehicle treated cells.
Western blotting
Samples were lysed (10 mM Tris (pH 7.5), 50 mM NaCl, 0.5% NP40, 0.5% SDS supplemented with protease inhibitor cocktail (Roche #11836153001) PhosSTOP (Roche #04906837001) and benzonase (Sigma Aldrich # E1014)) on ice and then the protein content estimated by BCA assay (Thermo Fisher Scientific #23225). 20μg of protein were denatured in loading dye containing SDS then resolved by 4-12% SDS-PAGE (Invitrogen #NP0336BOX). Protein was transferred to nitrocellulose membranes, blocked with excess protein and immunoblotted overnight at 4°C using the following antibodies; P-eEF2 T56 (CST #2331), eEF2 (CST #2332), RPL24 (Sigma Aldrich HPA051653) and β-actin (Sigma Aldrich #A2228) as a sample control. 1 hour incubation at room temperature with secondary antibodies (HRP-conjugated anti-mouse secondary (Dako #P0447); HRP-conjugated anti-rabbit secondary (Dako #P0448)) was followed by exposed to autoradiography films with ECL reagent (Thermo Fisher Scientific #32106). Quantification was performed using Image J (Rueden et al. 2017).
Sucrose density gradients
Cells were replenished with fresh medium for the 6 hours before harvesting. This medium was then spiked with 200 µg/mL cycloheximide (Sigma Aldrich #C7695) 3 minutes prior to harvesting on ice. Crypt fractions from mice were isolated by extraction of the epithelial from 10cm of proximal small intestine. Each data point plotted represents an individual animal. PBS-flushed linearly-opened small intestines were incubated in RPMI 1640 medium (Thermo Fisher Scientific #21875059) supplemented with 10mM EDTA and 200µg/mL cycloheximide for 7 minutes at 37°C with regular agitation to extract villi. Crypts were isolated by transferring remaining tissue to ice-cold PBS containing 10mM EDTA and 200 µg/mL cycloheximide for a further 7 minutes, again with agitation. The remaining tissue was discarded. Samples were lysed (300 mM NaCl, 15 mM MgCl2, 15 mM Tris pH 7.5, 100 µg/mL cycloheximide, 0.1% Triton X-100, 2 mM DTT and 5 U/mL SUPERase.In RNase Inhibitor(Thermo Fisher Scientific #AM2696)) on ice and post-nuclear extracts placed on top of 10-50% weight/volume sucrose gradients containing the same buffer (apart from no Triton X-100, DTT, or SUPERase.In). These were then spun in an SW40Ti rotor at 255,000rcf for 2 hours at 4°C under a vacuum. Samples were then separated through a live 254 nM optical density reader (ISCO). Polysome to subpolysome (P:S), heavy to light polysome (H:L) or 60S to 40S (60S:40S) ratios were calculated using the manually defined trapezoid method. For harringtonine run-off assays, cultures were prepared in duplicate for each genotype. One was pretreated with 2µg/mL harringtonine (Sant Cruz #sc-204771) for 5 minutes (300 seconds) prior to cycloheximide addition. This was then processed as above. Run-off rates were calculated as previously (Knight et al. 2015).
35S methionine incorporation assay
Organoids were replenished with medium 6 hours prior to analysis while in the optimal growth phase post-splitting. Technical triplicates were used from 3 different animals per experiments. 35S-methionine (Perkin Elmer #NEG772002MC) was used at 30 µCi/mL for 30 minutes. Samples were prepared in technical triplicate then lysed using the same buffer described for western blotting. Protein was precipitated in 12.5% (w/v) trichloroacetic acid onto glass microfiber paper (Whatmann #1827-024) by use of a vacuum manifold. Precipitates were washed with 70% ethanol and acetone. Scintillation was read on a Wallac MicroBeta TriLux 1450 scintillation counter using Ecoscint scintillation fluid (SLS Ltd #LS271) from these microfiber papers. In parallel the total protein content was determined by the BCA assay using un-precipitated sample. Protein synthesis rate was expressed as the scintillation normalised to the total protein content (CPM / µg/mL protein), which was then changed to a relative value compared to relevant controls for each experiment.
Publicly available clinical data analysis
The TNMplot and UALCAN web portals were used to analysis publicly available dataset from The Cancer Genome Atlas and Clinical Proteomic Tumor Analysis Consortium. Details of these portals are available in these publications (Bartha and Győrffy 2021; Chandrashekar et al. 2017).
Statistical analyses
All statistical analyses are detailed in the relevant figure legends. In all cases, calculated P values less than or equal to 0.05 were considered significant. N numbers for each experiment are detailed within each figure, as individual points on graphs or within figure legends.
Author contributions
Study design: JRPK, OJS. Data acquisition: JRPK, NV, DMG, RAR, WJF. Data analysis: JRPK, NV. Data interpretation: JRPK, NV, DMG, CGP, GM, TvdH, CMS, AEW, OJS. Writing the manuscript: JRPK, AEW, OJS.
Acknowledgements
The Sansom laboratory was funded by CRUK (A17196, A24388, A21139), The European Research Council ColonCan (311301). This work was also funded by a Wellcome Trust Collaborative Award in Science (201487) to GM, CMS, TvdH, AEW and OS. We are grateful to the Advanced Technologies and Core Services at the Beatson Institute (funded by CRUK C596/A17196 and A31287), particularly the Biological Services Unit, Histology Services and Transgenic Technology Laboratory. CGP is supported by funding from the National Health and Medical Research Council (Australia). We thank Fiona Warrander for critical reading of the manuscript.