Ninjurin 1 has two opposing functions in tumorigenesis in a p53-dependent manner

Significance

Nerve injury-induced protein 1 (Ninj1), a p53 target, forms a feedback loop with p53 by repressing p53 translation. Here, we show that cell growth was enhanced by Ninj1 deficiency in cells carrying a mutant p53, but inhibited in cells carrying a WT p53. We also show that in WT p53 background, Ninj1-deficient mice were prone to systemic inflammation, but not to spontaneous tumors. Importantly, loss of Ninj1 altered lifespan and tumor susceptibility in mutant p53 and p53-null mice. Taken together, our results reveal a critical role of Ninj1 in p53-dependent tumor suppression and the Ninj1–p53 loop may be explored as a potential therapeutic target for the treatment of inflammatory diseases and cancers.

Abstract

WT p53 is critical for tumor suppression, whereas mutant p53 promotes tumor progression. Nerve injury-induced protein 1 (Ninj1) is a target of p53 and forms a feedback loop with p53 by repressing p53 mRNA translation. Here, we show that loss of Ninj1 increased mutant p53 expression and, subsequently, enhanced cell growth and migration in cells carrying a mutant p53. In contrast, loss of Ninj1 inhibited cell growth and migration in cells carrying a WT p53. To explore the biological significance of Ninj1, we generated a cohort of Ninj1-deficient mice and found that Ninj1^+/− mice were prone to systemic inflammation and insulitis, but not to spontaneous tumors. We also found that loss of Ninj1 altered the tumor susceptibility in both mutant p53 and p53-null background. Specifically, in a mutant p53(R270H) background, Ninj1 deficiency shortened the lifespan, altered the tumor spectrum, and increased tumor burden, likely via enhanced expression of mutant p53. In a p53-null background, Ninj1 deficiency significantly increased the incidence of T-lymphoblastic lymphoma. Taken together, our data suggest that depending on p53 genetic status, Ninj1 has two opposing functions in tumorigenesis and that the Ninj1–p53 loop may be targeted to manage inflammatory diseases and cancer.

The tumor suppressor p53 is often referred to as the “guardian of the genome” and loss of p53 function is known to be central for tumor development (1, 2). The importance of p53 in tumor suppression is underscored by its ability as a transcription factor. In response to various stress signals, p53 is activated and then induces a plethora of target genes that mediate cell cycle arrest, apoptosis, and senescence (3). p53 is the most commonly mutated gene in human cancers (4). Unlike most tumor-suppressor genes that are mainly inactivated by deletion, the vast majority of tumor-derived p53 mutations are missense mutations and clustered within the central DNA-binding domain (5). Consequently, these mutations make p53 protein defective in DNA binding and loss of tumor suppression (6, 7). However, many p53 mutants acquire new and distinct oncogenic properties, generally referred to as gain-of-function, such as the ability to promote cell proliferation, metastasis, and chemoresistance (8, 9).

Cell adhesion molecules (CAMs) are typically transmembrane proteins and composed of an intracellular domain that mediates the intracellular signal pathways, a transmembrane domain, and an extracellular domain that interacts with the same CAMs (homophilic binding) or with different CAMs or extracellular matrix (heterophilic binding) (10). In addition to cell adhesion, CAMs serve as a biosensor to the changing surrounding microenvironment (10). As a result, CAMs are critical for many fundamental processes, such as cell and tissue homeostasis, tissue repair, and wound healing (11). In addition, altered function of CAMs have been implicated in tumor progression and metastasis (12).

Nerve injury-induced protein 1 (Ninjurin1, Ninj1) is a homophilic CAM and originally identified as a gene to be highly induced by nerve injury in Schwann cells and dorsal root ganglion (13). Ninj1 is ubiquitously expressed in epithelial cells (14) and plays a role in the progression of multiple sclerosis (15) and migration of T cells to the intraluminal surface of blood vessels in the central nervous system of rodents with experimental autoimmune encephalomyelitis (EAE) (16). Conversely, Ninj1 deficiency attenuates the severity of EAE in mice (17). Ninj1 also plays a role in angiogenesis by modulating neovessel formation in vitro (18) and the hyaloid vascular system in vivo (19). Additionally, inhibition of Ninj1 by a neutralizing Ninj1 antibody enhances penile angiogenesis and restores erectile function in streptozotocin-induced diabetic mice (20). Moreover, Ninj1 is capable of suppressing cancer cell invasion and migration in Ninj1-expressing cells and transgenic mice (21, 22). Previously, we showed that Ninj1 is a target of p53 and in turn regulates p53 mRNA translation (23). Here, we found that Ninj1 is capable of regulating mutant p53 expression. Most importantly, we found that Ninj1 exerts two opposing effects on cell growth, migration, and tumorigenesis via WT and mutant p53, respectively.

Results

The Effect of Ninj1 on Cell Growth and Migration Is Dependent on the Genetic Status of the p53 Gene.

Previously, we showed that Ninj1 modulates WT p53 expression via mRNA translation (23). Thus, we explored whether mutant p53 is regulated by Ninj1 in SW480 and MIA-PaCa2 cells. SW480 cells contain one allele of the p53 gene with two mutations (R273H and P309S), whereas MIA-PaCa2 cells contain a mutant p53(R248W). The level of WT p53 in MCF7 cells was measured as a control and found to be decreased upon ectopic expression of Ninj1 (Fig. 1A), consistent with our previous study (23). Similarly, upon ectopic expression of Ninj1, the level of mutant p53 was decreased in SW480 and MIA-PaCa2 cells (Fig. 1 B and C). Moreover, Egr-1, a target of mutant p53 (24), was decreased by Ninj1 (Fig. 1 B and C). Next, we examined the effect of endogenous Ninj1 on p53 expression in multiple cell lines in that the NINJ1 gene was knocked out by CRISPR-Cas9 (SI Appendix, Fig. S1A). Sequencing analysis showed that Ninj1-KO MCF7 and MIA-PaCa2 cells had a 42-nt deletion in exon 1 of the NINJ1 gene (SI Appendix, Fig. S1 B and C). In contrast, Ninj1-KO SW480 cells had a 60-nt deletion in one allele and a 92-nt deletion in other allele (SI Appendix, Fig. S1 B and C). Due to nonsense-mediated mRNA decay, Ninj1 protein was not detectable in these Ninj1-KO cells (Fig. 1 D–F). In Ninj1-KO MCF7 cells, the level of p53 was increased (Fig. 1D). To verify that Ninj1 represses p53 translation, we showed that the binding of eIF4E to the p53 mRNA was increased by Ninj1 deficiency (SI Appendix, Fig. S2, compare lane 5 with 6), consistent with our previous observation (23). Moreover, we found that the levels of mutant p53 were increased by Ninj1-KO along with increased expression of Egr-1 in SW480 and MIA-PaCa2 cells (Fig. 1 E and F).

Fig. 1.

Ninj1 regulates WT and mutant p53 expression. (A–C) Ninj1, p53, Egr-1, and actin proteins were examined by Western blotting in MCF7 (A), SW480 (B), and MIA-PaCa2 (C) cells transiently transfected with a control or Ninj1-expressing vector. (D–F) Ninj1, p53, Egr-1, and actin proteins were examined by Western blotting in isogenic controls and Ninj1-KO MCF7 (D), SW480 (E), and MIA-PaCa2 (F) cells.

WT p53 is a tumor suppressor, whereas mutant p53 acts as an oncogene (25). Thus, we speculated that the regulation of WT vs. mutant p53 by Ninj1 would result in two distinct biological responses. To test this, a colony formation assay was performed with isogenic controls and Ninj1-KO cells. We found that loss of Ninj1 reduced the number of colonies in MCF7 cells (Fig. 2A), consistent with our previous report (23). In contrast, loss of Ninj1 markedly increased the number of colonies in SW480 (Fig. 2B) and MIA-PaCa2 (Fig. 2C) cells. Next, a wound-healing assay was performed and showed that loss of Ninj1 inhibited cell migration in MCF7 cells (Fig. 2D). Conversely, cell migration was enhanced by loss of Ninj1 in both SW480 and MIA-PaCa2 cells (Fig. 2 E and F). These results suggest that Ninj1 exerts two opposing functions in cell growth and migration via WT and mutant p53, respectively.

Fig. 2.

The effect of Ninj1 on cell growth and migration is p53-dependent. (A–C) Colony formation assay was performed with isogenic controls, Ninj1-KO MCF7 (A), SW480 (B), and MIA-PaCa2 (C) cells. (D–F) Wound-healing assay was performed with isogenic controls and Ninj1-KO MCF7 (D), SW480 (E), and MIA-PaCa2 (F) cells. The images were proportionally scaled with an original magnification of 100×.

A 12-aa Ninj1-Blocking Peptide Elicits a Strong Effect on p53 Expression and Cell Growth.

Pharmacological modulation of p53 expression is actively explored as a strategy for cancer therapy. Thus, we synthesized a 12-aa peptide corresponding to the N-terminal extracellular region of Ninj1 (SI Appendix, Fig. S3A), which is known to block homophilic interaction of Ninj1 (16). We found that upon treatment with the 12-aa peptide, the levels of WT and mutant p53 proteins were increased in isogenic controls, but not in Ninj1-KO MCF7 and MIA-PaCa2 cells (SI Appendix, Fig. S3 B and D, compare lanes 1 and 3 with 2 and 4, respectively). We note that the basal levels of WT and mutant p53 were elevated by Ninj1-KO compared with that in isogenic controls (SI Appendix, Fig. S3 B and D, compare lane 1 with 3). Additionally, we found that the number of colonies was decreased by the Ninj1-blocking peptide in MCF7 cells (SI Appendix, Fig. S3C), but increased in MIA-PaCa2 cells (SI Appendix, Fig. S3E). These data suggest that the effect of Ninj1-KO on p53 expression is recapitulated by the 12-aa Ninj1-blocking peptide.

Ninj1-Heterozygous Mice Are Prone to Systemic Inflammation.

Previously, we showed that Ninj1-null mice succumb to hydrocephalus (23), and thus are not suitable for long-term study. In this regard, a cohort of WT (n = 24) and Ninj1^+/− (n = 21) mice was generated and monitored throughout their lifespan (SI Appendix, Tables S1 and S2). We note that 17 WT mice have been analyzed for another tumor study (26). We found that many Ninj1^+/− mice developed skin lesions (severe ulcerative psoriasiform dermatitis) and had to be killed. As a result, the median lifespan for Ninj1^+/− mice (98.71 wk) was shorter than—but still insignificant from—that for WT mice (120 wk; P = 0.08 by log-rank test) (Fig. 3A). If mice with skin lesions were excluded, the median survival for Ninj1^+/− mice (n = 11, 115 wk) was similar to that for WT mice (n = 21, 121 wk; P = 0.695 by log-rank test) (SI Appendix, Fig. S4A). In contrast, the incidence of ulcerative dermatitis was significantly higher in Ninj1^+/− mice than that in WT mice (P = 0.0192 by Fisher’s exact test) (Fig. 3B). Next, histopathological analysis was performed and showed that some WT (5 of 24) and Ninj1^+/− (8 of 21) mice developed spontaneous tumors (SI Appendix, Tables S1 and S2). However, the tumor incidence between WT and Ninj1^+/− mice was not statistically significant (P = 0.3234 by Fisher’s exact test).

Fig. 3.

Mice deficient in Ninj1 are prone to systemic inflammation. (A) Kaplan–Meier survival curve for WT vs. Ninj1^+/− mice (P = 0.080 by log-rank test). (B) The percentage of mice with dermatitis in WT vs. Ninj1^+/− mice (P = 0.0192 by Fisher’s exact test). (C) Representative images of H&E-stained skin sections from a pair of gender-matched WT and Ninj1^+/− mice. (D) Representative images of H&E- and IHC-stained liver and kidney sections from a pair of gender-matched WT and Ninj1^+/− mice. The red arrows indicate the positive-staining cells. The images were proportionally scaled with an original magnification of 200×.(E and F) IL-6 and actin transcripts were examined in the liver and kidney of two pairs of gender-matched WT and Ninj1^+/− mice. WT#1: F/56w/ID#5-19-3. WT#2: F/50w/ID#9-3-2. Ninj1^+/− #1: F/51w/ID#11-8-5. Ninj1^+/− #2: F/44w/ID#12-2-4.

Histopathological analysis indicates that the skin lesion in Ninj1^+/− mice exhibited marked psoriasiform epidermal hyperplasia and infiltrates of intrademeral perivascular neutrophils and lymphohistiocytes along with multifocal ulceration (Fig. 3C). Moreover, we found that compared with WT mice, Ninj1^+/− mice were prone to systemic inflammation in the liver, kidney, salivary gland, pancreas, and lung (Fig. 3D, H&E staining panel, and SI Appendix, Fig. 4B and Tables S1 and S2). Immunohistochemistry (IHC) indicated that the chronic inflammatory infiltrates were composed of mixed CD3⁺ T and B220⁺ B cells along with a few macrophages (F4/80⁺) (Fig. 3D). Next, the levels of IL-6, a key cytokine known to stimulate inflammatory and autoimmune responses (27), were examined and found to be increased in the liver, kidney, and spleen of Ninj1^+/− mice compared with that in WT mice (Fig. 3 E and F and SI Appendix, Fig. S4C). Similarly, the levels of several proinflammatory cytokines (TNF-α, IFN-γ, IL-8, and IL-1β) were also increased by Ninj1 deficiency (SI Appendix, Fig. S4D). These results suggest that Ninj1 deficiency promotes infiltration of immune cells and enhances production of proinflammatory cytokines, resulting in systemic chronic inflammation.

Mice Deficient in Ninj1 Are Prone to Pancreatic Insulitis.

When examining various organs of WT and Ninj1^+/− mice for inflammatory responses, we found that Ninj1^+/− mice had chronic pancreatic insulitis with expanded lymphohistiocytic infiltrates surrounding the islets (Fig. 4A). The incidence of insulitis was significantly higher in Ninj1^+/− mice than in WT mice (P = 0.0088 by Fisher’s exact test) (Fig. 4B and SI Appendix, Tables S1 and S2). We would like to mention that insulitis is a hallmark of type I diabetes, a T cell-mediated autoimmune disease characterized by selective destruction of pancreatic insulin-secreting β-cells (28). Because T cells are not a constitutive component of normal islets, the infiltration of T cells to islets represents a prerequisite for development of autoimmune type I diabetes (28). Thus, IHC analysis was performed to confirm that normal islets from a WT mouse were essentially negative for inflammatory cells (Fig. 4C). However, a mixed T and B cell population was detected surrounding the islets from a Ninj1^+/− mouse with histological features typical of insulitis (Fig. 4D). Interestingly, the immune cells within the islets were predominantly T cells (Fig. 4D).

Fig. 4.

Mice deficient in Ninj1 are prone to insulitis. (A) Representative images of H&E-stained pancreas sections from two pairs of gender-matched WT and Ninj1^+/− mice. The images were proportionally scaled with an original magnification of 200×. (B) The incidence of insulitis in WT vs. Ninj1^+/− mice (P = 0.0088 by Fisher’s exact test). (C and D) Pancreas sections from WT (C) and Ninj1^+/− (D) mouse were stained with antibodies against CD3, B220, or F4/80. The box plots were used to show the staining area in the Right. Red arrow: T cells at the inflammation site. Blue arrow: T cells inside the islet. Yellow arrow: B cells at the inflammation site.

Ninj1 Deficiency Shortens the Lifespan, Increases the Tumor Burden, and Alters the Tumor Spectrum in p53^R270H/- Mice.

To examine whether Ninj1 has an effect on mutant p53-mediated tumorigenesis, we generated a cohort of p53^R270H/− (n = 31) and Ninj1^+/−;p53^R270H/− (n = 28) mice (SI Appendix, Tables S3 and S4). Murine p53(R270H) corresponds to human p53(R273H), a gain-of-function mutant (7, 29). All mice succumbed to tumors, except for two p53^R270H/− mice who died because of an unknown reason (SI Appendix, Tables S3 and S4). We found that the median survival for p53^R270H/− mice was 26.71 wk, which was shortened to 23.14 wk in Ninj1^+/−;p53^R270H/− mice (P = 0.014 by log-rank test) (Fig. 5A). In addition, histopathological analysis indicated that Ninj1 heterozygosity altered the tumor spectrum in p53^R270H/− mice (Fig. 5B and SI Appendix, Tables S3 and S4). The penetrance of T-lymphoblastic lymphoma (T-LBL) was significantly higher in Ninj1^+/−;p53^R270H/− mice (24 of 28) than that in the p53^R270H/− mice (15 of 31; P = 0.0029 by Fisher’s exact test) (Fig. 5C). Additionally, we found that T-LBLs developed in Ninj1^+/−;p53^R270H/− mice were disseminated to multiple organs, including the heart, kidney, skin, liver, and lung (20 of 24 for Ninj1^+/−;p53^R270H/− mice vs. 4 of 15 for p53^R270H/− mice; P = 0.0007 by Fisher’s exact test) (Fig. 5 D and E). Moreover, the number of mice with two or more tumors derived from two different cell lineages was significantly higher in Ninj1^+/−;p53^R270H/− (14 of 28) mice than that in p53^R270H/− (4 of 31) mice (P = 0.0039 by Fisher’s exact test) (Fig. 5F and SI Appendix, Fig. S5A and Tables S3 and S4). We would like to note that several types of tumors, such as angiosarcoma, liposarcoma, osteosarcoma, and adenocarcinoma, were only detected in Ninj1^+/−;p53^R270H/− but not in p53^R270H/− mice (Fig. 5B and SI Appendix, Fig. S5B and Tables S3 and S4). Similarly, fibrosarcoma and squamous cell carcinoma developed in p53^R270H/− mice were not found in Ninj1^+/−;p53^R270H/− mice (Fig. 5B and SI Appendix, Tables S3 and S4).

Fig. 5.

Ninj1 deficiency shortens the lifespan, increases the tumor burden, and alters the tumor spectrum in p53^R270H/− mice. (A) Kaplan–Meier curve for p53^R270H/− vs. Ninj1^+/−;p53^R270H/− mice (P = 0.014 by log-rank test). (B) Tumor spectrum in p53^R270H/− vs. Ninj1^+/−;p53^R270H/− mice. (C) The incidence of T-LBL in p53^R270H/− vs. Ninj1^+/−;p53^R270H/− mice (P = 0.0029 by Fisher’s exact test). (D) Representative images of H&E-stained T-LBL disseminated to peripheral organs in a Ninj1^+/−;p53^R270H/− mouse. (E) The incidence of disseminated T-LBL in p53^R270H/− vs. Ninj1^+/−;p53^R270H/− mice (P = 0.0007 by Fisher’s exact test). (F) The tumor burden in p53^R270H/− vs. Ninj1^+/−;p53^R270H/− mice (P = 0.0039 by Fisher’s exact test). (G) Representative images of H&E- and IHC- (p53) stained kidney sections from normal and sarcoma areas in Ninj1^+/−;p53^R270H/− mouse. (H) p53 and actin protiens were examined in T-LBLs from two pairs of p53^R270H/− and Ninj1^+/−;p53^R270H/− mice. T-LBL#1: p53^R270H/−/M/16w/ID#1-15-4. T-LBL#2: Ninj1^+/−;p53^R270H/−/M/15w/ID#2-28-2. T-LBL#3: p53^R270H/−/M/18w/ID#11-26-18. T-LBL#4: Ninj1^+/−;p53^R270H/−/M/18w/ID#11-5-7.

Next, IHC analysis was performed to examine mutant p53 expression in normal and tumor tissues from Ninj1^+/−;p53^R270H/− mice. We found that mutant p53 protein was detected in the tumor tissues but not in the normal tissues from Ninj1^+/−;p53^R270H/− mice (Fig. 5G and SI Appendix, Fig. S6). In addition, in the tumor tissues, only tumor cells, but not intratumoral stromal cells, were positive in p53 staining (Fig. 5G and SI Appendix, Fig. S6). These observations are consistent with a previous report that mutant p53 protein was highly expressed in tumor cells from p53^R172H-knockin mice (30). Moreover, to determine whether the aggressive tumor phenotypes observed in Ninj1^+/−;p53^R270H/− mice are due to increased expression of mutant p53 by Ninj1 deficiency, we measured the levels of mutant p53 protein in the T-LBLs derived from two pairs of gender-matched p53^R270H/− and Ninj1^+/−;p53^R270H/− mice. Indeed, we found that the levels of mutant p53 were much higher in the T-LBLs from Ninj1^+/−;p53^R270H/− mice than that from p53^R270H/− mice (Fig. 5H). To verify this, mouse embryo fibroblasts (MEFs) were isolated and showed that the level of mutant p53(R270H) protein was much higher in Ninj1^−/−;p53^R270H/− MEFs than that in p53^R270H/− MEFs (SI Appendix, Fig. S5C). These results were consistent with the above observations that Ninj1 deficiency increased mutant p53 expression in SW480 and MIA-PaCa2 cells (Fig. 1 E and F). Since mutant p53 is known to create a permissive tumor microenvironment via enhancing protumor inflammation (31), we examined the levels of proinflammatory cytokines, including TNF-α, IFN-γ, IL-8, IL-6, and IL-1β, in the T-LBLs from p53^R270H/− and Ninj1^+/−;p53^R270H/− mice. We found that the levels of these proinflammatory cytokines were all increased in Ninj1^+/−;p53^R270H/− T-LBLs compared with that in p53^R270H/− T-LBLs (SI Appendix, Fig. S7).

Ninj1 Deficiency Increases the Penetrance of T-LBL in p53-Null Mice.

While Ninj1^+/− mice were susceptible to systemic inflammation, these mice were still not prone to spontaneous tumors (SI Appendix, Table S2), suggesting that increased expression of WT p53 by Ninj1 deficiency may have tempered inflammation-induced tumorigenesis. Thus, we sought to determine whether Ninj1 deficiency plays a role in tumorigenesis independent of p53. To test this, we generated a cohort of p53^−/− and Ninj1^+/−;p53^−/− mice. We would like to mention that except for one animal, 24 p53^−/− mice used as control were analyzed in a previous study (26). All mice succumbed to tumors, with the exception of two Ninj1^+/−;p53^−/− mice that died from hydrocephalus (SI Appendix, Tables S5 and S6). The median survival for Ninj1^+/−;p53^−/− mice (19.43 wk) was shorter than—but still insignificant from—that for p53^−/− mice (24 wk; P = 0.669 by log-rank test) (Fig. 6A). Histopathological analysis showed that Ninj1^+/−;p53^−/− mice were prone to T-LBL (Fig. 6 B and C and SI Appendix, Table S6). Statistical analysis indicated that the penetrance of T-LBL was significantly higher in Ninj1^+/−;p53^−/− mice (21 of 24) compared with p53^−/− mice (12 of 25; P = 0.0054 by Fisher’s exact test) (Fig. 6B). Additionally, Ninj1^+/−;p53^−/− mice did not develop hibernoma, hemangioma, granulocytic sarcoma, histiocytic sarcoma, osteosarcoma, and rhabdomyosarcoma (Fig. 6C and SI Appendix, Table S6). Instead, Ninj1^+/−;p53^−/− mice developed high-grade pleomorphic sarcoma, which was not observed in p53^−/− mice (Fig. 6C and SI Appendix, Tables S5 and S6).

Fig. 6.

Ninj1 deficiency increases the penetrance of T-LBL in p53^−/− mice. (A) Kaplan–Meier curve for p53^−/− vs. Ninj1^+/−;p53^−/− mice (P = 0.669 by log-rank test). (B) The incidence of T-LBL in p53^−/− vs. Ninj1^+/−;p53^−/− mice (P = 0.0054 by Fisher’s exact test). (C) Tumor spectrum in p53^−/− vs. Ninj1^+/−;p53^−/− mice.

Discussion

Ninj1 is a target of p53 and found to repress WT p53 expression via mRNA translation (23). In the present study, we extended our work to investigate the role of the Ninj1–p53 loop in vivo. We showed that Ninj1 is capable of modulating mutant p53 expression (Fig. 1). Interestingly, depending on the genetic status of the p53 gene, cell proliferation and migration can be inhibited by the loss of Ninj1 in cells carrying a WT p53, but increased in cells carrying a mutant p53 (Fig. 2). We also showed that in the WT p53 background, mice deficient in Ninj1 are not prone to spontaneous tumors (SI Appendix, Tables S1 and S2). However, in a mutant p53^R270H/− background, mice deficient in Ninj1 have a shorter lifespan and are prone to aggressive tumors (Fig. 5 and SI Appendix, Tables S3 and S4). Because Ninj1 deficiency leads to increased expression of WT and mutant p53, we conclude that Ninj1 has two opposing functions in tumorigenesis via WT and mutant p53, respectively.

Here, we found that Ninj1-dificient mice are prone to systemic chronic inflammation in the skin, liver, kidney, and pancreas (Figs. 3 and 4 and SI Appendix, Fig. S4 B–D and Tables S1 and S2). We also showed that the immune cells at the inflammation sites were mainly T and B cells (Figs. 3D and 4D). In addition, we showed that the levels of proinflammatory cytokines were increased in the Ninj1^+/− mice tissues compared with that in the WT mice tissues (Fig. 3 E and F and SI Appendix, Fig. S4 C and D). We speculate that loss of Ninj1 would attenuate the cell-adhesion activities and, subsequently, promote migration of T and B cells. Thus, additional studies are warranted to address the role of Ninj1 in modulating the immune responses.

Recent advances have highlighted a role for WT p53 in the inflammatory response as a part of its tumor-suppressor activity (32). In contrast, mutant p53 fuels protumor inflammation and promotes inflammation-induced tumor progression through enhanced secretion of chemokines by tumor cells carrying mutant p53 (31). We showed that Ninj1-het mice with an intact p53 remain near normal in terms of tumor formation. However, in a mutant p53 or p53-null background, Ninj1 deficiency creates a permissive microenvironment and synergizes with loss of p53 or mutant p53 to promote tumor progression and metastasis. In support of this notion, we showed that the incidence of T-LBL is markedly increased by Ninj1 deficiency in Ninj1^+/−;p53^R270H/− and Ninj1^+/−;p53^−/− mice (Figs. 5C and 6B, and SI Appendix, Tables S3–S6). Consistent with this, the levels of proinflammatory cytokines (TNF-α, IFN-γ, IL-8, IL-6, and IL-1β) were increased by Ninj1 deficiency in Ninj1^+/−;p53^R270H/− T-LBLs (SI Appendix, Fig. S7).

Insulitis is an inflammatory response surrounding or within pancreatic islets and often found in patients with autoimmune type I and II diabetes (33). Here, we found that mice deficient in Ninj1 are prone to insulitis along with predominantly T cell infiltration within the islet and mixed accumulation of T and B cells surrounding the islet (Fig. 4). Because Ninj1^+/− mice are prone to systemic inflammation, it is likely that the systemic inflammation serves as a contributing factor to the increased incidence of insulitis. However, it remains unclear whether the insulitis-prone Ninj1-deficient mice will develop type I diabetes. Additionally, as p53 plays a role in pancreatic β-cell apoptosis via p53 up-regulated modulator of apoptosis (34), future studies are warranted to determine whether increased expression of p53 by Ninj1 deficiency is involved in the insulitis and pancreatic β-cell death.

We showed that a 12-aa Ninj1-blocking peptide has an opposing effect on cell proliferation: through up-regulation of WT p53 to inhibit cell growth in MCF7 cells (SI Appendix, Fig. S3 B and C) and through up-regulation of mutant p53 to promote cell growth in MIA-PaCa2 cells (SI Appendix, Fig. S3 D and E). Consistent with this finding, a previous report showed that the Ninj1-blocking peptide inhibits T cell migration in the vessels in a rat EAE model (16). However, the Ninj1-blocking peptide is also found to alleviate inflammatory response in a septic model (35), which seems to contradict the observations in this study and by an earlier report (16). One possible explanation is simply due to the mouse models used: Ninj1-deficient mouse model for this study and rat EAE model for one early report (16) vs. a septic mouse model for the other study (35). The other possibility is that Ninj1 deficiency in our mouse model may have a long-term and systemic effect on the immune systems, whereas the Ninj1-blocking peptide may exert a transient effect on the immune systems in septic mice. Thus, additional studies are needed to determine the role of the Ninj1-blocking peptide in inflammatory responses and in p53-dependent tumor suppression in vitro and in vivo.

Materials and Methods

Cell Lines and Primary Mouse MEFs.

MCF7, SW480, and MIA-PaCa2 cells were cultured in DMEM (Invitrogen) supplemented with 10% FBS (HyClone). To generate Ninj1-KO cell lines using the CRISPR/Cas9 system, we followed the previously described protocol (36). Individual single-guide RNA constructs targeting Ninj1 (SI Appendix, Fig. S1A) were cloned into the pSpCas9(BB)-2A-Puro (PX459) vector (Addgene). Upon transfection of the targeting vector into cells, Ninj1-KO clones were selected by puromycin and confirmed by sequencing. p53^R270H/− and Ninj1^−/−;p53^R270H/− MEFs were isolated from 13.5-d embryos as described previously (37).

Mice Deficient in Ninj1, p53, or Carrying a p53^R270H Knockin Allele.

Ninj1-deficient mice were generated as described previously (23). p53^+/− and p53^R270H/+ knockin mice (38) were purchased from the Jackson Laboratory. Ninj1^+/−;p53^+/− mice were crossed with p53^+/− or p53^R270H/+ mice to generate Ninj1^+/−;p53^−/− or Ninj1^+/−;p53^R270H/− mice. All animals and experimental protocols were approved by the University of California, Davis Institutional Animal Care and Use Committee. To genotype Ninj1 and p53-deficent mice, the primers were used as previously described (23, 39). To test p53^R270H knockin allele, the primers are 5′-AGC CTG CCT AGC TTC CTC AGG-3′ and 5′-CTT GGA GAC ATA GCC ACA CTG-3′.

Histological Analysis.

Tissue processing, H&E staining, and IHC analysis were performed as previously described (26).

RNA Isolation and RT-PCR Analysis.

Total RNAs were purified with TRIzol (Invitrogen) and then subjected to cDNA synthesis using M-MLV reverse transcriptase (Promega). The program used for amplification was: (i) 94 °C for 5 min, (ii) 94 °C for 45 s, (iii) 58 °C for 45 s, (iv) 72 °C for 1 min, and (v) 72 °C for 10 min. The primers to detect murine IL-6 are 5′-GAG GAT ACC ACT CCC AAC AGA CC-3′ and 5′-AAG TGC ATC ATC GTT GTT CAT ACA-3′; and for murine actin, 5′-CCC ATC TAC GAG GGC TAT-3′ and 5′-AGA AGG AAG GCT GGA AAA-3′.

Western Blot Analysis and Colony Formation Assay.

Western blotting was performed as described previously (23) with anti-human p53 (DO-1; monoclonal), anti-p53 (Santa Cruz), anti-Egr-1 (Santa Cruz), anti-Ninj1 (Abcam), and antiactin (Sigma). Colony formation assay was also carried out as described previously (23).

Wound-Healing Assay.

Cells were scratched with a 200-µL pipette tip at 70% confluency, incubated with fresh medium for 48 h and then captured with Nikon microscope (Nikon Corporation).

Statistical Analysis.

Fisher’s exact test was used for comparison between two genotypes. Log-rank test was used to determine the difference in median survival between two genotypes.