SMG1 heterozygosity exacerbates haematopoietic cancer development in Atm null mice by increasing persistent DNA damage and oxidative stress

Abstract Suppressor of morphogenesis in genitalia 1 (SMG1) and ataxia telangiectasia mutated (ATM) are members of the PI3‐kinase like–kinase (PIKK) family of proteins. ATM is a well‐established tumour suppressor. Loss of one or both alleles of ATM results in an increased risk of cancer development, particularly haematopoietic cancer and breast cancer in both humans and mouse models. In mice, total loss of SMG1 is embryonic lethal and loss of a single allele results in an increased rate of cancer development, particularly haematopoietic cancers and lung cancer. In this study, we generated mice deficient in Atm and lacking one allele of Smg1, Atm−/−Smg1gt/+ mice. These mice developed cancers more rapidly than either of the parental genotypes, and all cancers were haematopoietic in origin. The combined loss of Smg1 and Atm resulted in a higher level of basal DNA damage and oxidative stress in tissues than loss of either gene alone. Furthermore, Atm−/−Smg1gt/+ mice displayed increased cytokine levels in haematopoietic tissues compared with wild‐type animals indicating the development of low‐level inflammation and a pro‐tumour microenvironment. Overall, our data demonstrated that combined loss of Atm expression and decreased Smg1 expression increases haematopoietic cancer development.

decay (NMD), the pathway used by cells to detect and degrade mRNA with premature termination codons which may code for truncated proteins. 7 SMG1 has previously been implicated in regulation of DNA damage responses, telomere maintenance and oxidative and hypoxic stress responses and stress granule formation. [7][8][9][10][11] Complete loss of SMG1 expression in mice is lethal early during embryogenesis but we have demonstrated that loss of a single SMG1 allele increased cancer development, particularly lung adenocarcinomas and lymphomas. 12 SMG1 haploinsufficiency in these mice did not result in sufficient protein loss to affect its roles in nonsense-mediated decay, DNA damage responses or apoptosis induction. However, SMG1-deficient animals showed elevated levels of basal inflammation and oxidative damage to tissues prior to development of cancers indicating a potential role for these pathways in enhancing tumourigenesis in this model. SMG1 and ATM have previously been shown to co-regulate DNA damage responses and p53 signalling. 13,14 Brumbaugh et al (2004) demonstrated that both enzymes contribute to the phosphorylation of Upf1 and p53 in response to ionizing radiation (IR). Further SMG1 and ATM are both required for maximal activation of the G1/S checkpoint following exposure to ionising radiation or during oxidative stress. 13,14 SMG1 can also regulate alternative splicing of p53 in response to DNA damage. 15 To further examine the interplay between SMG1 and ATM in cancer development, we crossed mice heterozygous for both the Smg1 genetrap allele (Smg1 gt/+ ) and the Atm null allele to generate Atm knockout mice that were also heterozygous for Smg1 (Atm −/− Smg1 gt/+ ). These mice were viable and developed lymphomas at a more rapid rate than mice carrying only the Atm −/− or Smg1 gt/+ alleles.

| Generation of Atm −/− Smg1 gt/+ mice and animal husbandry
Parental strain Atm −/− and Smg1 gt/+ mice have been described previously. 12,16 Animals heterozygous for both alleles were bred to generate Atm −/− Smg1 gt/+ mice, and sex of parents carrying each genotype was alternated during breeding. All animal ex- F I G U R E 1 Combined Atm loss and Smg1 heterozygosity decreases mouse lifespan. A, Smg1 +/gt mice were bred with Atm +/− mice to generate Smg1 +/gt Atm −/− animals. Kaplan-Meier survival curve shows that combined loss of Atm and Smg1 significantly decreases lifespan. B, Smg1 +/gt mice were also crossed to generate Smg1 +/gt p53 −/− animals. SMG1 heterozygosity had no additional effect on lifespan in these animals as demonstrated by Kaplan-Meier survival curve. C, Median lifespan for each of the mouse lines examined as determined using Kaplan-Meier analysis in GraphPad Prism. ***P < .001, ****P < .0001 Genomic DNA was extracted from ear clips for genotyping as described previously. 12

| Isolation and culture of mouse embryonic fibroblasts (MEFs)
Mating of mice was assumed at midnight and timed from 0.5.

| Histology
Tissues were isolated and fixed in 10% formalin. Samples were processed by the QIMR Berghofer histology facility. Serial sections of 4μm thickness were used for haematoxylin and eosin staining or immunofluorescence for 8-oxo-dG or 4HNE as described previously. 17 Images were scanned using Aperio Turbo or Aperio FL (Leica Biosystems) and analysed using ImageScope software (Leica Biosystems). Histology slides were examined by an experienced pathologist.

| Irradiation and DNA damage analysis
Cells were irradiated at 6Gy with a GammaCell40Exactor (Best Theratronics Ltd.) with a Cobalt60 source of 960c Gγ/min. Cells were fixed with 4% paraformaldehyde in phosphate-buffered saline (PBS) at indicated time-points. γH2AX analysis by immunofluorescence was performed as described previously. 18

| Flow Cytometry-cell death assays, cytokine bead assays and cell surface markers
Flow cytometry was performed with a FACSCanto or Fortessa (BD Biosciences). Annexin V (BD Pharmingen) apoptotic analysis was performed according to manufacturer's instruction. Propidium iodide (Sigma) cell cycle analysis was performed as described previously. 19 Measurement of serum cytokines levels was performed with cytokine bead array, 20 and staining for cell surface markers was performed as described previously. 21,22

| Quantitative PCR
Real-time PCR reactions were performed as described previously 23 and analysed using an Applied Biosystems QuantStudio (Thermo Fisher). Primer sequences are shown in the table below.

| Generation of Atm −/− Smg1 gt/+ mice
The Smg1 Genetrap line (Smg1 gt/+ ) and Atm knockout mice (Atm −/− ) have been described previously. 12,16 Animals heterozygous for both alleles were bred to generate Atm −/− Smg1 gt/+ mice. Animals were generated at approximately the expected frequency (1/7, when accounting for lethality of Smg1 knockout animals).   Figure S1. This cancer profile is more similar to the Atm −/− animals which also developed lymphomas in the majority of animals (80%, Table 1), in contrast Smg1 gt/+ mice who developed a combination of lymphomas and papillary lung adenocarcinomas as we have described previously (Table 1). 12 We performed a similar experiment crossing Smg1 gt/+ mice to p53 −/− mice. In contrast to the results with Atm −/− crosses, SMG1 haploinsufficiency had no effect on the rate of tumour development in  (Table 1). IHC was also performed with examples in Figure S2. These data indicate that decreased SMG1 expression exacerbates the pro-oncogenic effect of ATM loss, potentially by increasing the dysregulation of a pathway in which both SMG1

| More rapid lymphoma development in
and ATM provide regulatory feedback or by a combined effect on independently regulated pathways.

| Effect of combined loss of SMG1 and ATM on haematopoietic cell composition
Given the predominance of lymphoma formation in Atm −/− Smg1 gt/+ mice, we examined whether the addition of Smg1 heterozygosity resulted in increased defects in the immune system compared with the known decrease in T cells caused by loss of ATM expression. 2,3 We showed previously that SMG1 heterozygosity alone did not significantly alter the composition of immune system prior to disease onset. 12 Here we analysed the composition of circulating blood cells and lymphocytes in the spleen, thymus and lymph nodes. There were no significant differences in the major circulating cell populations between any of the genotypes ( Figure 2A)

| Effect of Smg1 haploinsufficiency on DNA damage responses and oxidative stress
In both Atm −/− and Smg1 gt/+ mice, high levels of oxidative damage to tissues have been detected. 12, 16 We examined tissues from  Figure S3). ATM has a well-established role in the DNA damage response. 26 We previously showed that the level of SMG1 protein expressed in Smg1 gt/+ mice was sufficient for its roles in DNA damage repair. 12  We also examined cell death in response to irradiation, thymocytes were isolated from mice and exposed to 5Gy irradiation, and apoptosis was determined by Annexin V/ PI staining and flow cytometry analysis. As expected from previous literature, Atm −

| Effect of Smg1 haploinsufficiency on systemic inflammation
Our previous investigations showed that Smg1 gt/+ mice had ele- and we could observe a small difference in cytokine levels between wild-type and Smg1 gt/+ mice, though a smaller magnitude difference than we had observed previously ( Figure 4A). This is likely due to the earlier sacrifice time of the animals in this study compared with 6-9 months in our original study. 12 The number of animals with higher levels of cytokines increased in Atm −/− mice and Atm −/− Smg1 gt/+ mice although the difference in level was not significantly different due to the variability between individuals ( Figure 4A). However, it appears that mice lacking Atm and especially when combined with Smg1 heterozygosity are more likely to express higher levels of these cytokines. Also of interest was the pattern of expression of IL-13. IL-13 was undetectable in wild-type and expressed at low levels in Smg1 gt/+ mice, two Atm −/− mice expressed high levels of IL-13, and half of the Atm −/− Smg1 gt/+ mice expressed detectable IL-13 levels ( Figure 4B). As serum levels provide an overview of systemic inflammation, we also measured IL-1β, IL-6 and colony-stimulating factor 1 (CSF-1) levels by quantitative PCR ( Figure 4C). The only significant differences in cytokine expression were IL-1β expression in the heart with Smg1 gt/+ , Atm −/− and Atm −/− Smg1 gt/+ mice all having significantly elevated levels compared with wild-type. IL-6 in the heart and lung appeared to F I G U R E 3 Smg1 heterozygosity combined with Atm loss increases basal oxidative stress and DNA damage burden. A, Smg1 heterozygosity combined with Atm loss increases oxidative damage to tissues. Spleens were harvested from pre-disease animals, and immunofluorescence was performed for the marker of oxidative damage to DNA, 8-oxo-dG (green) and DAPI to highlight nuclei (blue). B, C, Murine embryonic fibroblasts were generated and exposed to 5 Gy irradiation (IR). Cells were fixed at the indicated time-points post-IR and stained for the presence of γH2AX as a marker of unrepaired DNA damage. The number of γH2AX foci in each cell nucleus was counted, and quantification is shown in panel A and example 0hr images in panel B γH2AX (green) and DAPI (blue). Statistical significance was determined using a t test with Welch's correction for unequal variance. Bars indicate the mean and error bars the standard error of the mean. *P < .05, **P < .01, ***P < .001. D, Thymocytes were isolated from mice and exposed to 5Gy irradiation (IR). At baseline and 24 h post-IR, apoptosis was measured using AnnexinV/PI staining and flow cytometry. Bars indicate the mean and error bars the standard error of the mean be increased in Atm −/− and Atm −/− Smg1 gt/+ mice but this did not reach statistical significance. We also measured the levels of IFNβ in splenic tissues but this was undetectable in nearly all samples ( Figure 4D).

| D ISCUSS I ON
Both SMG1 and ATM have been shown previously to act as tumour suppressors. 12,[27][28][29] Here we demonstrated that loss of one of the alleles of Smg1 in addition to Atm loss resulted in more rapid cancer development. Also all cancers arose from the haematopoietic system in contrast to either Smg1 +/gt or Atm −/− mice where there were a large proportion of haematopoietic cancers but solid cancers, particularly lung adenocarcinomas, were also prevalent 12,16 (Table 1) to be blocked in their passage through the cell cycle at the G1/S phase transition, often referred to as radioresistant DNA synthesis. 13,30 This is due to defective p53 phosphorylation, in the absence of ATM at the G1/S checkpoint. SMG1 can also control p53 activation by regulating the expression of alternatively spliced p53 isoforms. 15 Further loss of either Smg1 or Atm alone results in increased oxidative stress and the resultant damage to DNA. 2,12 As such when we combined loss of Atm and Smg1 in Atm −/− Smg1 gt/+ mice we saw exacerbated oxidative damage in tissues and increased basal DNA damage load. Given that increased damage burden did not result in greater cell death in Atm −/− Smg1 gt/+ cells ( Figure 3C), the combined data suggest that cells are surviving but are more likely to develop an oncogenic mutation, due to low level DNA damage, which could contribute to the development of cancers in these animals.
ATM and SMG1 have also both been implicated in the regulation of inflammation with loss increasing basal cytokine levels (reviewed in 31 ). A low-level, continuous inflammatory response or 'smouldering' inflammation is a pro-tumourigenic microenvironment. 32 This is particularly true for haematopoietic cancers where the cytokines can also act directly as growth factors; for example, increased IL-6 drives B cell growth and proliferation via activation of STAT3. 33 Along with STAT3, NF-κB is a key transcription factor which can drive tumourigenesis and support cancer cell survival. 34 In Atm −/− Smg1 gt/+ mice, we saw a trend of increasing levels of NF-κB-dependent cytokine expression indicating that there was ongoing NF-κB activation in these animals ( Figure 4A and 4).  Figure 1A). Some of these animals would have been much closer in time to developing cancers than others, and as such, only a proportion of mice had detectable tumour promoting cytokines at this time-point. Interestingly, we also saw a significant increase in IL-13 levels in the serum in Atm −/− Smg1 gt/+ mice compared with wild-type littermates ( Figure 4B). IL-13 also has a role in creating a pro-tumour environment via the activation of tumour-associated macrophages and myeloid-derived suppressor cells. 35 Together these data suggest that as they age Atm −/− Smg1 gt/+ mice may more quickly develop a pro-tumour microenvironment compared with control animals.
What was not increased in Atm −/− Smg1 gt/+ mice was type I interferon production. Long-term unrepaired DNA damage results in accumulation of DNA in the cytoplasm of cells which activates the cGAS/STING pathway to induce interferon (IFN) production. 36 This has previously been demonstrated in bone marrow-derived cells from Atm −/− mice and in isolated cells and tissues from rats lacking ATM expression. 5,6,17 However, in tissues analysed here IFNβ was barely detected in any Atm −/− Smg1 gt/+ mice ( Figure 4D). At this same time-point, increased oxidative DNA damage was evident in spleen ( Figure 3A) but it is possible that this had not persisted for long enough to lead to IFN induction. Alternatively, loss of SMG1 in addition to ATM loss may limit IFNβ production by an unknown mechanism.
Overall, our data demonstrate that combined loss of expression of ATM and heterozygosity of SMG1 results in more rapid cancer development particularly haematopoietic cancers. Pre-disease animals showed increased unrepaired DNA damage and oxidative stress and propensity to develop "smouldering" inflammation most clearly in haematopoietic tissues. The combination of persistent DNA damage and a permissive tumour microenvironment represents a likely mechanism resulting in increased and more rapid development of blood cancers. In total, this work further highlights the extensive crosstalk and dual regulation of pathways by members of the PIKK family.

ACK N OWLED G EM ENTS
The authors would like to thank the Ingham Institute Biological YCL is supported by a Danish Cancer Society Fellowship.

CO N FLI C T O F I NTE R E S T
The authors declare no conflicts of interest.

AUTH O R CO NTR I B UTI O N
UH, JL, AJ, CSL, HQ, HCL, SA, YCL and TLR performed experimentation and analysed data; MFL, UH and TLR designed experiments; and UH and TLR wrote the manuscript with feedback from all authors.