Two Pathways of p27Kip1 Degradation Are Required for Murine Lymphoma Driven by Myc and EBV Latent Membrane Protein 2A.

Epstein-Barr virus (EBV) latent membrane protein 2A (LMP2A), expressed in EBV latency, contributes to Burkitt lymphoma (BL) development in a murine model by acting as a constitutively active B cell receptor (BCR) mimic. Mice expressing both LMP2A and MYC transgenes (LMP2A/λ-MYC) develop tumors significantly faster than mice only expressing MYC (λ-MYC). Previously, we demonstrated the cell cycle inhibitor p27Kip1 is present at significantly lower levels in LMP2A/λ-MYC mice due to increased posttranslational degradation. P27Kip1 degradation can occur in the cytoplasm following phosphorylation on serine 10 (S10) or in the nucleus via the SCFSkp2 complex, which depends on Cks1. We previously demonstrated an S10A knock-in of p27Kip1 (p27S10A/S10A) that prevented S10 phosphorylation failed to significantly delay tumor onset in LMP2A/λ-MYC mice. We also previously demonstrated that a Cks1 knockout partially delayed tumor onset in LMP2A/λ-MYC mice, but onset was still significantly faster than that in λ-MYC mice. Here, we have combined both genetic manipulations in what we call p27Super mice. LMP2A/λ-MYC/p27Super mice and λ-MYC/p27Super mice both displayed dramatic delays in tumor onset. Strikingly, tumor development in LMP2A/λ-MYC/p27Super mice was later than that in λ-MYC mice and not significantly different from that in λ-MYC/p27Super mice. The p27Super genotype also normalized G1-S-phase cell cycle progression, spleen size, and splenic architecture in LMP2A/λ-MYC mice. Our results reveal both major pathways of p27Kip1 degradation are required for the accelerated BL development driven by LMP2A in our BL model and that blocking both degradation pathways is sufficient to delay Myc-driven tumor development with or without LMP2A.IMPORTANCE BL is a cancer that primarily affects children. The side effects of chemotherapy highlight the need for better BL treatments. Many BL tumors contain EBV, and our goal is to determine what makes EBV-positive BL different from EBV-negative BL. This may lead to more specific treatments for both types. All cases of BL require overexpression of MYC Mice engineered to express EBV LMP2A along with MYC (LMP2A/λ-MYC mice) develop tumors much more quickly than mice only expressing MYC (λ-MYC mice). Blocking degradation of the cell cycle inhibitor protein p27Kip1 in LMP2A/λ-MYC mice causes tumors to develop later than in λ-MYC mice, showing that p27Kip1 degradation may play a larger role in EBV-positive BL than EBV-negative BL. Furthermore, our studies suggest the cell cycle is an attractive target as a treatment option for LMP2A-positive cancers in humans.

mainly in equatorial Africa, sporadic (sBL), found throughout the rest of the world, and AIDS-associated (AIDS-BL), found in HIV-positive patients. Nearly all eBL tumors, and smaller proportions of sBL and AIDS-BL tumors, are associated with EBV (1). In vitro, EBV can transform resting B cells into growing lymphoblastoid cell lines by expressing the growth program, which drives cell proliferation and survival (3)(4)(5). In vivo, however, EBV exploits B cell biology to achieve latency.
In order for naive B cells to become memory B cells, they undergo clonal expansion following antigen exposure and form germinal centers (GC) in the spleen and lymph nodes (5,6). While in the GC, antigen binding to the B cell receptor (BCR) leads to phosphorylation of the tyrosine kinase Syk, which drives prosurvival signaling through proteins such as phosphatidylinositol 3-kinase, Akt, and extracellular signal-regulated kinase (7). The clones that survive begin to differentiate and exit the GC, becoming memory B cells (7). Taking advantage of this process, EBV infects naive B cells, driving clonal expansion of EBV-infected B cells (5). Once in the GC, EBV transitions to a latency program in which a small number of genes, including those encoding latent membrane proteins 1 (LMP1) and 2A (LMP2A), as well as EBNA1, are expressed (5). In this environment, LMP2A acts as a constitutively active mimic of the BCR, driving survival and differentiation (8,9). As the infected B cells differentiate into memory B cells, EBV switches to another latency program, expressing EBNA1 and other non-protein-coding genes (5,10). B cells isolated from BL tumors display a GC phenotype and while EBV-positive BL tumors were first observed expressing only EBNA1, LMP2A transcripts were subsequently detected in eBL cell lines and snap-frozen eBL tumors (11)(12)(13)(14)(15)(16)(17). It is therefore likely that EBV-infected B cells transform into BL tumor cells when EBV drives clonal proliferation early after EBV infection.
A key feature of all BL tumors is a translocation of the proto-oncogene c-MYC (MYC) and an immunoglobulin (Ig) gene locus, leading to Myc overexpression under an Ig promoter (18). A C57BL/6 murine model of BL has been developed in which a MYC transgene is expressed under the Ig locus (-MYC) (19). Myc is a transcription factor that drives many types of cancer. It promotes the expression of genes that promote cell cycle progression from G 1 -to S-phase and is required for normal B cell activation and proliferation (20). In addition, Myc induces expression of several tumor suppressor genes that promote cell cycle arrest and apoptosis, including TP53 and ARF (21). For this reason, inactivation of tumor suppressor pathways is required for the development of Myc-driven tumors. When -MYC mice are crossed with LMP2A transgenic mice, the resulting LMP2A/-MYC mice develop tumors significantly faster than -MYC mice (22,23). We previously found that spleens of LMP2A/-MYC mice displayed a greater percentage of S-phase B cells than -MYC mice, and the cell cycle inhibitor p27 Kip1 was rapidly degraded and expressed at lower levels in LMP2A/-MYC splenic B cells (24).
Degradation of p27 Kip1 occurs through two major pathways. Phosphorylation on serine 10 (S10) results in the export from the nucleus and subsequent degradation of p27 Kip1 in the cytoplasm (25). Alternatively, phosphorylation can occur on threonine 187 (T187), leading to degradation of p27 Kip1 by the SCF Skp2 complex in the nucleus (26). Previously, we attempted to delay lymphomagenesis in LMP2A/-MYC mice by using a p27 Kip1 S10A mutant homozygous knock-in (p27 S10A/S10A ) mouse, which proved unsuccessful, although we did observe modest effects on G 1 -S cell cycle progression in -MYC mice (24). Subsequent research determined that preventing p27 Kip1 degradation by SCF Skp2 via the homozygous knockout of Cks1 (Cks1 Ϫ/Ϫ ), a member of the SCF Skp2 complex that is essential for the recognition of phosphorylated T187, resulted in a significant delay in lymphomagenesis for LMP2A/-MYC mice (27). These mice, however, still developed tumors much faster than the -MYC mice (27).
To investigate the two pathways of p27 Kip1 degradation in LMP2A-mediated lymphomagenesis, we crossed Cks1 knockout mice (8) with p27 S10A/S10A mice (28) and identified offspring that were homozygous for the S10A knock-in and Cks1 knockout, which we termed p27 Super mice. Both LMP2A/-MYC/p27 Super and -MYC/p27 Super mice displayed dramatically delayed tumor onset. Strikingly, tumor onset in the LMP2A/-MYC/p27 Super mice was later than that in the -MYC mice and not significantly different from that in -MYC/p27 Super mice. These data show that preventing p27 Kip1 degradation by using the p27 Super genotype is the most effective genetic manipulation yet for preventing tumor growth in our model of BL.

RESULTS
The p27 Super genotype delays tumor onset in both LMP2A/-MYC and -MYC mice. To observe the effect of degradation-resistant p27 Kip1 on Myc-induced tumorigenesis, tumor-free survival in LMP2A/-MYC, LMP2A/-MYC/p27 Super , -MYC, and -MYC/p27 Super mice was examined (Fig. 1). All tumors appeared in lymph nodes in the cervical, abdominal, or thoracic area. Tumor onset was delayed in both LMP2A/-MYC/ p27 Super and -MYC/p27 Super mice ( Fig. 1). Median tumor-free survival time was 378 days in LMP2A/-MYC/p27 Super mice compared to 63 in LMP2A/-MYC mice, a delay of 315 days. In our previous study, median tumor-free survival in LMP2A/-MYC/ Cks1 Ϫ/Ϫ mice was delayed only 61.5 days compared to that in LMP2A/-MYC mice (27). The S10A knock-in alone was previously shown to have no effect on tumor onset (36 days for both the LMP2A/-MYC/p27 S10A/S10A and LMP2A/-MYC mice) (24). Strikingly, tumor onset in the LMP2A/-MYC/p27 Super mice was 172 days later than that in the -MYC mice, and there was no significant difference in tumor onset between the LMP2A/-MYC/p27 Super and the -MYC/p27 Super mice, which had a median tumor-free survival time of 400 days. In our previous study, tumor onset in LMP2A/-MYC/Cks1 Ϫ/Ϫ mice was 25 days earlier than that in -MYC mice (27). Although the p27 S10A/S10A knock-in by itself was previously shown to have no effect on tumor-free survival (24), our data show that it improved tumor-free survival in the presence of the Cks1 knockout. From these data, we conclude that blocking both major pathways of p27 Kip1 degradation completely blocked accelerated LMP2A-driven lymphomagenesis in our BL model.
The p27 Super genotype does not alter splenic cellular architecture or B cell numbers in the spleen compared to those of the WT. We next investigated whether the p27 Super genotype affects normal B cell development in 4-week-old mice. The p27 Super mice developed a similar number of mature B cells in the periphery ( Fig. 2A). B cell numbers were lower in the bone marrow of p27 Super mice, whereas the splenic In previous studies, tumor onset was delayed 61.5 days in LMP2A/-MYC/Cks1 Ϫ/Ϫ and not delayed in the LMP2A/-MYC/p27 S10A/S10A mice (24,27). Tumor onset in LMP2A/-MYC/p27 Super mice is 172 days later than that in -MYC mice. Previously, tumor onset in LMP2A/-MYC/Cks1 Ϫ/Ϫ mice was 25 days earlier than that in -MYC mice (27). Sample size (n) for each genotype is indicated below the curve, as well as P values determined by log-rank (Mantel-Cox) test. *, P Ͻ 0.05; ****, P Ͻ 0.0001; ns, not significant. P27 Kip1 Degradation Required in Murine Lymphomagenesis ® B cell number in p27 Super mice was similar to that of wild-type (WT) mice ( Fig. 2A). We next analyzed splenic architecture in p27 Super mice and compared it to that of WT mice by staining spleens for B220 and p27 Kip1 . B cell development is a highly regulated process that results in the formation of follicles of developing B cells that can be readily observed in the spleen by immunohistochemistry (IHC). B cell follicle formation and p27 Kip1 staining were similar in the p27 Super spleens and WT spleens (Fig. 2B), demonstrating that the p27 Super genotype does not significantly alter splenic B cell number or follicle formation.
The p27 Super genotype normalizes B cell development in LMP2A/-MYC and -MYC spleen. Our previous studies showed pretumor LMP2A/-MYC and -MYC mice have a significantly higher percentage of splenic B cells in S-phase than WT mice of the same age (24,27). By analyzing the cell cycle, we found the percentage of S-phase cells in LMP2A/-MYC/p27 Super spleens was 27.7% lower than that in LMP2A/-MYC mice and was similar to what is observed in WT mice (Fig. 3A). In our previous studies, the percentage of pretumor splenic B cells in S-phase was 0.59% lower in LMP2A/-MYC/ p27 S10A/S10A mice and 19.8% lower in LMP2A/-MYC/Cks1 Ϫ/Ϫ mice than in LMP2A/-MYC mice (24,27). The percentage of splenic S-phase B cells was 10.5% lower in -MYC/p27 Super spleens than in -MYC spleens (Fig. 3A), while in previous studies the ) and LMP2A/-MYC/p27 S10A/S10A (0.59%) mice (24,27). (B) The ratios of spleen weight to total body weight were calculated for 4-to 5-week-old mice of the indicated genotypes. (C) IHC was performed on spleens of 6-to 8-week-old pretumor mice as described in the legend to Fig. 1. Spleens were sectioned and stained for B220 and p27 Kip1 . Three mouse samples are shown for each genotype. (D) Immunoblots were performed on protein isolated from splenic B cells of pretumor mice. Membranes were probed for p27 Kip1 and GAPDH. Densitometry was performed to calculate relative p27 Kip1 levels. A representative blot (left) and relative p27 Kip1 levels (right) are shown. Data represent the means Ϯ SD. P values were determined by two-tailed t test. *, P Ͻ 0.05; ***, P Ͻ 0.001; ****, P Ͻ 0.0001. P27 Kip1 Degradation Required in Murine Lymphomagenesis ® percentage was 9.13% lower in -MYC/p27 S10A/S10A and 6.5% lower in -MYC/Cks1 Ϫ/Ϫ mice (24,27) than in -MYC mice. Finally, the percentages of S-phase cells in LMP2A/ -MYC/p27 Super and -MYC/p27 Super mice were not significantly different (18% and 20%, respectively) (Fig. 3A), compatible with the similar tumor onset observed in these two genotypes. These data show that blocking both p27 Kip1 degradation pathways together has a synergistic effect on halting G 1 -S-phase cell cycle transition, and that this synergy is specific to LMP2A/-MYC mice.
The p27 Super genotype also prevented the splenomegaly observed in LMP2A/-MYC mice. As previously shown (24,27), LMP2A/-MYC mice at 4 to 5 weeks old displayed significantly higher spleen weight relative to body weight than WT mice (Fig. 3B). The LMP2A/-MYC/p27 Super mice, however, had spleen-to-body weight ratios that were not significantly different from those of WT mice (Fig. 3B).
After exiting the bone marrow, normal B cells form follicles in the spleen and other secondary lymphoid organs. This B cell organization is completely lost in the spleens of LMP2A/-MYC mice (Fig. 3C). This suggests that the LMP2A/-MYC genotype leads to a defect of normal B cell development or follicular homing. The p27 Super genotype, however, restored B cell follicle formation in the LMP2A/-MYC/p27 Super mice (Fig. 3C). While the B cell follicles in all four -MYC genotype-containing spleens appeared to be less defined than those in the WT, B cell organization in LMP2A/-MYC/p27 Super , -MYC, and -MYC/p27 Super spleens closely resembled that of the WT rather than that of the disorganized LMP2A/-MYC spleens (Fig. 3C).
P27 Kip1 protein degradation is blocked in p27 Super mice. We next analyzed p27 Kip1 levels in splenic B cells. Immunohistostaining of pretumor spleens showed an observable increase in the intensity of the p27 Kip1 signal in both the LMP2A/-MYC/ p27 Super and -MYC/p27 Super mice compared to that in LMP2A/-MYC and -MYC mice (Fig. 3C). To confirm the IHC data, we performed Western blot analysis to quantify p27 Kip1 levels in pretumor splenic B cells (Fig. 3D). Spleens were dissected from mice prior to tumor onset, and B cells were isolated and purified. As expected, p27 Kip1 levels were significantly elevated in LMP2A/-MYC/p27 Super and -MYC/p27 Super mice compared to those in LMP2A/-MYC and -MYC mice, respectively. In our previous studies, the Cks1 knockout alone increased p27 Kip1 levels in LMP2A/-MYC mice nearly to wild-type levels but resulted in only a partial delay in tumor development (27). In our current study, the LMP2A/-MYC/p27 Super and -MYC/p27 Super mice have higher p27 Kip1 levels than WT mice, although the difference is not statistically significant. This greater increase in p27 Kip1 levels may account for the greater delay in tumor onset caused by the p27 Super genotype than by the Cks1 knockout alone.
The p27 Super genotype significantly increases p27 Kip1 levels in both LMP2A/-MYC and -MYC tumors. In order to determine whether p27 Kip1 levels remain elevated in LMP2A/-MYC/p27 Super and -MYC/p27 Super tumors, mice were sacrificed once tumors became apparent. Lymph node tumors were dissected from either the cervical or abdominal area. IHC was performed to assess p27 Kip1 levels (Fig. 4A), as was done with pretumor spleens. Western blot analyses were also performed to quantify the levels of p27 Kip1 in B cells isolated from the tumors (Fig. 4B). Both the IHC and Western blot analysis showed that p27 Kip1 levels are significantly elevated in LMP2A/-MYC/p27 Super and -MYC/p27 Super tumors compared to levels in LMP2A/-MYC and -MYC mice, respectively. These data suggest that while tumors still form in LMP2A/-MYC/p27 Super and -MYC/p27 Super mice, it is likely due to factors other than p27 Kip1 degradation.

DISCUSSION
The mechanisms driving BL tumors can differ drastically depending on the presence or absence of EBV. EBV-positive BL carries fewer, and distinct, driver mutations compared to EBV-negative BL (29). Previous studies, including those by our group, have shown that LMP2A accelerates tumor development in combination with dysregulated Myc, obviating the need for mutations in the ARF-Mdm2-p53 pathway (22,23). We have also demonstrated that LMP2A cooperates with Myc to increase G 1 -S-phase cell cycle transition prior to tumorigenesis (24,27). Determining the effect of EBV latent proteins like LMP2A on the cell cycle can uncover important distinctions between EBV-positive and EBV-negative BL, which may lead to more specific therapies for BL patients based on disease subtype.
Gene expression profiling has previously shown elevated expression of genes involved in cell cycle regulation in eBL compared to those genes in sBL (30). Additionally, transcriptome sequencing found mutations in CCND3, which encodes cyclin D3, a critical regulator of G 1 -S transition, were frequent in sBL but not eBL (31). Furthermore, expression of LMP2A in eBL correlates with increased expression of genes promoting G 1 -S-phase cell cycle transition (32). In our previous studies, posttranslational degradation of p27 Kip1 correlated with earlier tumor development in LMP2A/-MYC mice, which was partially attenuated by blocking Cks1-dependent p27 Kip1 degradation (27). As noted in the introduction, our previous attempts to delay tumor onset in LMP2A/ -MYC and -MYC mice by blocking S10 phosphorylation alone were unsuccessful. In the current study, we delayed tumor onset by blocking both Cks1-dependent and S10  phosphorylation-dependent p27 Kip1 degradation with the p27 Super genotype. We found that p27 Super mice had a dramatic delay in tumor onset compared to the Cks1 knockout alone in both LMP2A/-MYC/p27 Super mice and -MYC/p27 Super mice.
Of twelve proteins known to regulate G 1 -S-phase transition (p27 Kip1 ; p21 Cip1 ; p15 Ink4b ; p16 Ink4a ; cyclins D1, D2, D3, E, and A; CDK2; Rb; and E2f1), p27 Kip1 was the only one expressed at significantly lower levels in LMP2A/-MYC than in -MYC pretumor B cells (24). For this reason, we predicted that blocking p27 Kip1 degradation in LMP2A/ -MYC mice would prevent tumors from developing more quickly than in -MYC mice. Our results show that tumor onset in LMP2A/-MYC/p27 Super mice was significantly later than that in -MYC mice and was not significantly different from that in -MYC/ p27 Super mice. With the p27 Super genotype, mice expressing LMP2A and Myc do not develop tumors more quickly than mice only expressing Myc, demonstrating that both pathways of p27 Kip1 degradation are required for the accelerated tumorigenesis driven by LMP2A in the presence of dysregulated Myc.
The effect of the p27 Super genotype on the pretumor spleen and splenic B cells mirrored its effect on tumor onset. The percentage of splenic B cells in S-phase was significantly decreased in LMP2A/-MYC/p27 Super and -MYC/p27 Super mice (Fig. 3A), and the splenomegaly observed in LMP2A/-MYC mice was normalized in LMP2A/-MYC/p27 Super mice (Fig. 3B). Additionally, B cell follicle formation is partially restored in LMP2A/-MYC/p27 Super pretumor spleens (Fig. 3C). The increased p27 Kip1 expression observed in LMP2A/-MYC/p27 Super and -MYC/p27 Super splenic B cells (Fig. 3D) is maintained in tumors from both genotypes (Fig. 4B), suggesting these tumors develop independently of p27 Kip1 degradation. Interestingly, in the absence of -MYC, p27 Super does not affect splenic B cell numbers (Fig. 1A) and does not decrease the percentage of S-phase B cells (Fig. 3A) or the spleen-to-body ratio (Fig. 3B). This suggests the effects of p27 Super on the cell cycle and B cell development are specific to LMP2A/-MYC and -MYC mice.
While there is no significant difference in tumor onset between the LMP2A/-MYC/ p27 Super and -MYC/p27 Super mice, the pathways that drive tumorigenesis in each genotype are likely different. Increased expression of Myc leads to apoptosis through the ARF-Mdm2-p53 tumor suppressor pathway. As a result, MYC-driven tumors often involve mutations in p19 ARF or p53, which render them inactive (33,34). We have previously shown that tumors isolated from -MYC mice frequently displayed such inactivating mutations, while those from LMP2A/-MYC mice did not (22). Because LMP2A combines with Myc to increase degradation of p27 Kip1 in our model, tumor development can occur without the necessity for inactivation of the ARF-Mdm2-p53 pathway, as we have previously observed (24). We wanted to determine whether p19 ARF and/or p53 was frequently inactivated in tumors from p27 Super mice. We found that 43% (3 of 7) of -MYC/p27 Super tumors had p19 ARF and/or p53 abnormalities, while only 14% (1 of 7) of LMP2A/-MYC/p27 Super tumors did (data not shown). This indicates that inactivation of the ARF-Mdm2-p53 pathway can contribute to tumor development in LMP2A/-MYC/p27 Super mice, but it does so less frequently than in -MYC/p27 Super mice. Future studies will explore additional pathways that may differ between these two models, as well as between the LMP2A/-MYC and -MYC models in general.
Our current studies indicate the requirement that both pathways of p27 Kip1 degradation be blocked to fully prevent the accelerated tumorigenesis driven by LMP2A. Overall, our results show that normalizing G 1 -S-phase cell cycle progression by elevating levels of p27 Kip1 both delays Myc-driven lymphoma and offsets the contribution of LMP2A in accelerating tumor development. Our study points to both the nuclear and cytoplasmic pathways of p27 Kip1 degradation, as well as G 1 -S-phase regulation, as potential targets for developing more specific treatments of BL. Preclinical experiments testing the effectiveness of drugs targeting these pathways may uncover more effective therapies with fewer long-term side effects than the chemotherapies that are currently used.

MATERIALS AND METHODS
Mice. The Tg6 line of E-LMP2A transgenic mice expresses LMP2A under immunoglobulin (Ig) heavy-chain promoter and intronic enhancer (E), while -MYC mice overexpress human MYC. Both lines are in the C57BL/6 background and have been described previously (8,19). Cks1 null (Cks1 Ϫ/Ϫ ) mice (28) were obtained from the Steven Reed laboratory at The Scripps Research Institute in La Jolla, CA. Mice expressing the p27 Kip1 S10A knock-in (Cdkn1b tm2Jro ) (35) were obtained from The Jackson Laboratory. Tumor mice were sacrificed when lymph node tumors could be observed externally or when mice were moribund. Animals were maintained at Northwestern University's Center for Comparative Medicine in accordance with the university's animal welfare guidelines.
Tumor, spleen, and bone marrow cell isolation. Pretumor splenic B cells were purified using the mouse pan-B cell isolation kit by StemCell Technologies. Bone marrow cells were flushed from femurs and tibia. Tumor-bearing lymph node cells were prepared as previously described (22,24,36).
Flow cytometry and cell cycle analysis. To measure B cell numbers, purified B cells from spleens and bone marrow of 4-to 5-week-old mice were stained with IgM, IgD, and B220 antibodies (BD Biosciences). For cell cycle analysis, purified splenic B cells from 4-to 5-week-old mice were fixed in 70% ethanol and stained with propidium iodide/RNase staining buffer according to the manufacturer's instructions (BD Biosciences). All flow cytometry was performed with a FACS-CantoII flow cytometer (BD Biosciences), and all results were analyzed with FlowJo software (FlowJo, LLC).
Immunohistochemistry. Spleens from 4-to 8-week-old mice and tumor-bearing lymph nodes were fixed in 10% buffered formalin phosphate, stored in 70% ethanol, and embedded in paraffin. Samples were sectioned and stained with hematoxylin and eosin (H&E) or anti-p27 Kip1 (Invitrogen) or anti-B220 (BD Biosciences) antibody. Stained tissue slides were imaged using an EVOS XL Core microscope.
Immunoblots. Purified pretumor B cells or tumor cells were lysed in radioimmunoprecipitation assay lysis buffer with protease and phosphatase inhibitor cocktails. Lysates were separated by SDS-PAGE (Bio-Rad). Protein was transferred from the gel to a nitrocellulose membrane (Bio-Rad). Membranes were probed with anti-p27 Kip1 (Santa Cruz) or glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (Abcam) primary antibody and then incubated with IRDye secondary antibodies (LI-COR Biosciences). Protein bands were visualized with an Odyssey Imager and analyzed with Image Studio (LI-COR Biosciences).
Statistical analysis. Two-tailed t test, survival analysis, and log-rank (Mantel-Cox) test were performed using Prism 7 (GraphPad Software). A P value of Ͻ0.05 was considered statistically significant.

ACKNOWLEDGMENTS
We thank Nanette Susmarski for timely and excellent technical assistance and members of the Longnecker laboratory for help in the completion of this study, particularly Kamonwan Fish for initiating these studies and reading the manuscript prior to submission. We thank the Steven Reed laboratory for the Cks1 Ϫ/Ϫ mice. The Northwestern University Mouse Histology and Phenotyping Laboratory provided histology services and is supported by NCI P30-CA060553, awarded to the Robert H. Lurie Comprehensive Cancer Center. Flow cytometry work was supported by the Northwestern University Interdepartmental Immunobiology Flow Cytometry Core Facility. This work was supported by National Institutes of Health, National Cancer Institute grant R01 CA073507 (R.L.), and Carcinogenesis Training Program grant T32CA009560 (R.P.S.). R.L. is a Dan and Bertha Spear Research Professor in Microbiology-Immunology. R.L., R.P.S., and M.I. designed the research and wrote the manuscript. R.P.S. and M.I. performed the research and analyzed the data.
We have no competing financial interests to declare.