Activation-induced cytidine deaminase is dispensable for virus-mediated liver and skin tumor development in mouse models

Right This dissertation is author version of following the journal article. Tung Nguyen, Jianliang Xu, Shunsuke Chikuma, Hiroshi Hiai, Kazuo Kinoshita, Kyoji Moriya, Kazuhiko Koike, Gian Paolo Marcuzzi, Herbert Pfister, Tasuku Honjo, and Maki Kobayashi. Activation-induced cytidine deaminase is dispensable for virus-mediated liver and skin tumor development in mouse models. Int. Immunol. (2014) 26 (7): 397-406 first published online February 25, 2014 doi:10.1093/intimm/dxu040

(2). Transgenic, ubiquitous over-expression of AID causes T cell lymphoma and micro-adenoma in the lung (3) along with mutations in the T cell receptor and c-myc genes. Chronic infections with micro-organisms such as helicobacter pylori (4), hepatitis C virus (HCV) (5-7), and human T cell leukemia virus type 1 (8) induce the aberrant AID expression, which has been proposed to cause tumors by introducing translocations and somatic mutations into proto-oncogenes. In addition, AID expression is associated with chronic infections of these pathogens in human cases, in which it has also been proposed to contribute to tumor formation at least in part (4,9). However, it has not been directly determined if viral or bacterial-induced oncogenesis requires the action of AID.
Hepatocellular carcinoma (HCC) is the fifth most frequent cancer, and hepatitis B virus (HBV) and HCV infections are the major risk factors for developing this cancer worldwide (10). In fact the risk of developing HCC is increased 11.5-to 17-fold in HCV-infected patients; however, antiviral therapies have limited effectiveness in only a small fraction of patients. Thus, elucidation of the mechanism(s) involved in promoting liver tumorigenesis is urgently required for developing a prevention strategy. As natural infection of HCV is restricted to humans and chimpanzees, several transgenic mouse models harboring parts of the HCV polyprotein have been generated to recapitulate HCC development (11). HCV, a small RNA virus, belongs to the Flaviviridae family and contains a 9.6-kb single-stranded RNA genome. The polyprotein encoded by the HCV genome is processed into the structural proteins including (core, E1 and E2) and the nonstructural proteins (NS2-NS5) required for RNA genome replication by host and viral proteases (11). Among them, the HCV core protein has unique, multifunctional roles in apoptosis, signal transduction, reactive oxygen species formation, transformation, and immune modulation (such as the up-regulation of TGF-β) (10), by interacting with many cellular proteins. Out of the 14 lines of HCV-transgenic mice developed, 5 HCV core protein-containing transgenic lines and one nonstructural (NS) 5A transgenic line can give rise to HCC, after the development of severe steatosis, a characteristic pathology associated with HCV infection (11-13).
Especially, a transgenic mice model expressing HCV core protein (HCV-Tg) driven HBV regulatory elements (12) which limits HCV core To determine whether AID is aberrantly induced by HCV or HPV8 and required for virally induced tumorigenesis, we crossed HCV-Tg or HPV8-Tg mice with AID -/mice and compared the tumorigenesis frequencies in the AID +/+ and AID -/mice. We also examined the AID expression levels in the affected tissues. We found that HCV-Tg mice exhibited enhanced AID expression in the B cells infiltrating the liver, and that the steatosis and lymphocytic follicle formation were more severe in the HCV-Tg/AID +/+ than in the HCV-Tg/AID -/mice. However, the HCC prevalence at 20 months of age was not remarkably different between the two groups. Similarly, the time course of papilloma development was indistinguishable between the HPV8-Tg/AID +/+ and HPV8-Tg/AID -/mice. Furthermore, AID expression was not induced in the skin papillomatous tissues of the HPV8-Tg/AID +/+ mice. We conclude that AID is not necessary for the viral protein-induced oncogenesis in these two mouse models.

Mice maintenance and genotyping
All the mice used in this study were maintained at the Institute of Laboratory Animals in accordance with the guidelines of the Animal Research Committee, Graduate School of Medicine, Kyoto University.
AID -/mice (16) back-crossed to C57/B6 (B6) were crossed with HCV core protein transgenic mouse line (HCV-Tg) (12, 13) and HPV8-Tg mice (on an FVB/N background) (15). AID-Cre and Rosa-RFP compound mice (17) were crossed with HCV-Tg mice to enable the detection of previous and current AID expression. The genotyping primers are described in Supplementary Table 1.

Western blotting
Western blotting was performed by conventional methods. Mouse organs were dissected and homogenized in RIPA buffer. The primary antibodies used were the rat monoclonal anti-mouse AID antibody 2 (MAID-2) (eBioscience, San Diego, USA), anti-Tubulin antibody (Calbiochem, MERCK, Darmstadt, Germany) and anti-HCV core protein antibody (clone B2, Yes Biotech Lab, Anogen, Ontario, Canada).

RT-PCR and qRT-PCR
Mouse organs were excised and homogenized in Sepasol RNA I Super (Nacalai Tesque, Kyoto, Japan) following the manufacturer's instructions.

Histology and immunohistochemistry
For AID immunohistochemical staining, freshly excised livers were fixed in 4% paraformaldehyde and processed for frozen section as previously described (19). AID protein was detected by MAID-2 and peroxidase-labeled donkey F(ab')2 anti-rat IgG (Jackson ImmunoResearch, West Grove, USA), and stained with Diaminobenzidine. Images were captured with a DM5000B microscope (Leica; Wetzlar, Germany).
Hematoxylin and eosine (HE)-stained samples were fixed with Mildform 10N (Wako Pure Chemical Industries, Osaka, Japan), embedded in paraffin, and stained by standard methods.

Liver cell fractionation and FACS analysis
The isolation of intra-hepatic immune cells (IHICs) from the liver of HCV-Tg mice was performed as previously described, with some minor

ELISA
TNF-α, IL-1β, and TGF-β were detected using ELISA kits specific for each cytokine (BioSource, Life Technologies), according to the manufacturer's instructions.

Statistical analysis
The Mann-Whitney U test was used to calculate the statistical differences in AID expression (Fig. 1B). Fisher's exact test was used to determine significant differences in tumor incidence (Table 3). Student's t-test was used to determine significant differences in cytokine expression ( Fig. 3B and 3C), and Welch's t-test was used for pathological severity validation (Table 2). p values < 0.05 were considered statistically significant.  Figure 1). Because HCV-Tg mice develop severe steato-hepatitis within 9 months after birth (12), and this chronic inflammation is supposed to reproduce the similar cytokine environment to the TNF-α-stimulated hepatocyte cell lines that express AID (9), we examined AID expression in the liver. AID transcripts were detected in the liver from 16-month-old HCV(+)AID +/+ mice, but not from wild-type B6 mice ( Fig. 1A, Supplementary Figure 2). However, the AID expression detected in the HCV(+)AID +/+ liver was comparable to the low levels observed in primary un-stimulated spleen cells, which included B and T lymphocytes. Transcripts for CD19, a specific marker for B lymphocytes, were also higher in the HCV(+) compared to the B6 liver, suggesting that the HCV(+) liver may contain a considerable number of B lymphocytes, which could contribute to the increased AID expression. Real-time PCR analysis revealed that the AID transcript level in the liver from 16 month-old HCV(+)AID +/+ females was 15-fold greater than that from the liver of similarly aged B6 mice and of 12-month-old HCV(+)AID +/+ males ( Fig. 1B).
The AID protein levels were measured in the same liver samples by Western blotting (Fig. 1C). However, using the anti-mouse AID antibody 2 (MAID2), no protein signal could be detected in the same samples that contained AID transcripts (Fig. 1A). To quantify the limitation of AID protein detection by MAID2, we used spleen cells as a control (Supplementary Figure 3). We assigned one arbitrary unit of AID mRNA to the q-PCR signal detected from 500 ng of naïve spleen cell RNA. Extracts prepared from the same number of spleen cells contained 8.6 µg protein, which did not elicit a detectable AID signal in the Western blot. Since the AID transcript level from the HCV(+)AID +/+ samples in Supplementary   Figure 2 was lower than that in naive spleen cells, we conclude that the AID protein signal in the HCV(+)AID +/+ liver was below the level detected by MAID2.
We next explored the possibility that the AID protein expression was limited to a specific location in the liver, such as the immune cells in the hepatic blood vessels. We therefore performed an

AID transcripts are detected in intra-hepatic immune cells (IHICs), but not in hepatocytes from HCV-Tg mice.
We next explored the possibility that the low level of AID transcripts was contributed by B cells infiltrating the HCV-Tg liver. Liver cells from three HCV(+)AID -/or HCV(+)AID +/+ mice at 16 months of age were fractionated to separate the IHICs from the hepatocytes ( Fig. 2A-B). RNA was purified from both fractions of each genotype, and the AID, CD19, and albumin transcripts were analyzed to confirm the purity of these fractions and to identify the cellular origin of the AID mRNA (Table 1) Table 3).
To detect both current and past AID expression, transgenic reporter mice expressing tdRFP under the control of BAC-AID-Cre (17) were crossed to HCV-Tg mice ( Fig. 2B, C, Supplementary Table 3). This mouse reveals tdRFP fluorescence in any cells that have (or had) expressed AID.
Using this approach, tdRFP + cells were found to represent 3 to 4% of the total IHICs in both HCV-Tg and no HCV-Tg (HCV(-)) mice at 3 months of age ( Fig. 2B). tdRFP was not detected in hepatocytes from 3-month-old HCV(+) mice that had already developed mild steatohepatitis (12).

AID deficiency reduces the severity of histopathological phenotypes and cytokine expression profiles in the liver of HCV-Tg mice
It is reported that 14-30% of HCV-Tg male mice develop HCC between 16 and 19 months of age (13). We therefore analyzed the histopathological phenotypes of HE-stained liver sections from HCV(+)AID -/and HCV(+)AID +/+ mice at 12 (male) and 16 (female) and 20 months (male) of age ( Fig. 3A, 4A and Table 2). The severities of steatosis and lymphocyte infiltration were graded from 0 to 3, and the average scores for each group were calculated and tested by Welch's t-test (Table 2).
This decrease in steatosis severity may have been due to the female composition of the mice, because females of this HCV-Tg did not show tumor development (13) and human HCV infected cirrhotic females develop HCC less frequently (21). In addition, lymphoid follicle formation was apparent at 16 months of age and was more frequent in the HCV(+)AID +/+ than the HCV(+)AID -/mice ( Fig. 3A, Table 2). Consistent with the previous report (13), the fibrotic or regenerative nodular changes were very mild at 16 months of age.
However, the liver samples from both HCV(+)AID -/and HCV(+)AID +/+ mice at 20 months of age revealed marked progressive changes, with nuclear atypia detected in 10 out of 21 of each genotype, and liver cell degeneration or regenerative changes detected in 10 and 11 out of 21 HCV(+)AID -/and HCV(+)AID +/+ mice, respectively (Fig. 4A).
Interestingly, both the steatosis and lymphoid follicle severity scores appeared to be higher in the HCV(+)AID +/+ than the HCV(+)AID -/-20-month-old mice (1.33 versus 1.00 for steatosis and 1.48 versus 1.14 for lymphoid follicles, respectively), although the observed differences were not statistically significant (Table 2).
We next examined whether the more advanced inflammatory histological phenotypes observed in the HCV(+)AID +/+ mice were associated with increases in cytokine expression. Since cytokine production levels are altered in chronic HCV hepatitis (22) and in HCV-Tg mouse (23), we used q-PCR to measure representative pro-inflammatory Th1 and Th2 cytokine expression levels in the livers of 16-month-old female mice (Fig.   3B). The presence of the HCV transgene was associated with significantly higher levels of IL-1β, TNFα, and TGFβ mRNA, and HCV(+)AID +/+ mice exhibited higher TNFα levels than HCV(+)AID -/mice. Consistent with the q-PCR results, the protein levels of IL-1β and TGFβ were also elevated by the presence of the HCV transgene, and TNFα protein production level was dependent on the presence of AID, suggesting that TNFα production may have led to the aggressive pathological findings observed in HCV(+)AID +/+ mice (Fig. 3C).

panels).
These results suggest that AID is not essential for HCV-induced carcinogenesis. Consistent with these findings, the AID transcript level in the tumor region of an AID +/+ mouse (#241) was equivalent to the level detected in a non-tumor area (Fig. 4B). Comparison of the AID expression in the tumor and non-tumor areas from the two tumor-bearing mice of the HCV(+)AID -/and the two of HCV(+)AID +/+ indicated that the tumor tissues did not contain elevated AID expression levels (Fig. 4C). The AID protein levels were not detectable by Western blotting in either the tumor or non-tumor areas (Fig. 4D).

Dispensability of AID for the development of skin tumors in HPV8-Tg mice
To examine AID's involvement in the development of HPV8-induced skin tumors, HPV8-Tg (HPV(+)) mice were crossed with AID -/mice. Then  The lower papilloma prevalence (~30%) compared to the original report describing the HPV8(+) mice (15) may be due to the mixed genetic background of FVB/N and B6 in both groups (the HPV8(+)AID -/and HPV8(+)AID +/+ mice), since mice with an FVB/N genetic background are reported to have more severe papilloma progression than those with a B6 background (15). Although the malignant progression to SCC in these skin tumors was not examined, AID expression was absent, and tumor development was equivalent in the AID -/and AID +/+ mice. We thus conclude that AID is not involved in HPV8-induced skin tumorigenesis.

Discussion
In the current study, we investigated the requirement of AID for virally induced tumorigenesis by using compound mice that were generated by crossing mice transgenic for either HCV core proteins or HPV8 early proteins with either AID wild-type or knockout mice. Our results indicated that AID was expressed in neither hepatocytes of HCV-Tg nor skin tissue of HPV8-Tg. Thus AID was not shown to be required for the development of both HCV-and HPV8-promoted tumorigenesis. We could not conclude that the frequency of the liver malignancy is statistically different between HCV(+)AID +/+ (4 out of 21 mice) and HCV(+)AID -/-(2 out of 21 mice), partly because the frequency of the liver malignancy was unexpectedly lower than that in the original report (13). Studies on 5 times number of mice may allow us to obtain statistically significant conclusion in the frequency of the liver malignancy between HCV(+)AID +/+ and HCV(+)AID -/-. We note, however, that the HCV(+)AID +/+ mice exhibited higher levels of TNF-α production with more severe histological phenotypes than the HCV(+)AID -/mice.
AID is reported to be expressed in HCC and in the surrounding non-cancerous liver tissues (6, 7). In addition, the hepatoma-derived cell lines HepG2, Hep3B, and Huh-7 have all been shown to express AID in response to HCV core protein-induced NF-κB signaling (9). It therefore has been assumed that AID is induced in HCV-Tg mice and contributes to liver carcinogenesis. However, AID expression was not detected in HCV core protein-positive hepatocytes, and the low levels of AID transcripts in liver tissues were attributable to infiltrating B cells in the present study.
The different AID expression levels observed in the present mouse model and HCV-infected patients could be due in part to pathogenic differences between the two systems. HCV-Tg mice lack cirrhotic changes Although the difference in the AID promoter between human and mice is not known, we can not totally exclude this possibility to explain the different AID expression response.
Although HCV-Tg had weaker inflammatory responses than natural HCV infection, the inflammation observed in HCV(+)AID +/+ mice tended to be more severe than that in HCV(+)AID -/mice. The HCV core protein expression induced infiltration of IHICs including T, B and probably NK cells, and AID +/+ B cells enhanced TNF-α production more than AID -/-B cells. The mechanism of TNF-α upregulation by AID is unknown, however, higher levels of TNF-α production is likely to affects the pathogenesis or prognosis. TNF-α and IL-1β were increased in the whole liver lysates of 16-month-old HCV-Tg mice, as previously reported (23). Clinically, increased TNF-α production from liver-infiltrating monocytes in Non-A, Non-B hepatitis has also been reported (34).
The HCC development without severe "cirrhotic" findings may be due to carcinogenic properties associated with the HCV core protein, including suppression of apoptosis (by interacting with p53 and pRb),    The severity of steatosis and lymphoid follicle formation is classified as: 0 (none), 1 (mild), 2 (moderate), or 3 (severe
Trans-IT LT1 transfection reagent (Mirus bio corporation, Madison, WI, USA) was used for pEF1α-HCVcore-iresZsGreen1 transfection. Briefly, one day before transfection, 0.5 × 10 6 cells were plated and transfected following manufacturer's instruction. 24 hours after transfection, the cells were treated with or without TNFα (100 ng/ml, Sigma-Aldrich, St. Louis, USA) and finally used for RNA extraction.

BL2-AIDER cells
Human B cell lymphoma-derived BL2 cells was stably transfected by AID-ER (estrogen receptor) fusion protein plasmid.

Naïve or stimulated spleen cells
Spleen of 12 months-aged HCV-Tg was gently grilled and pushed through