**4. Discussion**

30 Liver Tumors

indicate that the tumorigenic effect of *Hcs7* does not involve acute changes in the rates of

 **Num. mice Mitoses / field Apoptoses / field Age Treatment B6 2R8 B6 2R8 B6 2R8**  13 days No DEN 7 11 1.50 1.25 0.50 0.50 14 days No DEN 9 9 2.00 1.5 0.25 0.50 15 days No DEN 7 8 1.40 1.00 0.30 0.25 13 days DEN 7 7 0.5 0.5 0.30 0.25 14 days DEN 9 11 2.25 2.50 0.25 0.50 15 days DEN 7 12 2.00 1.85 0.50 0.55 Table 3. Mitosis and apoptosis do not differ significantly between line 2R8 and B6 during

**3.8 The** *Hcs7***-interval genes most differentially expressed between C3H and B6 are** 

To identify potential candidate genes and pathways regulated by the *Hcs7* region, we collected livers from DEN-treated B6 and 2R8 mice at 13 days, 28 hours after DEN injection. RNA from these livers was reverse-transcribed and hybridized to Agilent whole-genome arrays, in competition with a B6 mixed-sex control, to identify differentially expressed genes. These arrays carry 43,379 probes representing most genes in the genome. Of the 44 genes in the 175.35 to 178.64 Mb region, 35 are represented by probes. The remaining nine include genes for three spliceosomal RNAs, two olfactory receptors, one protein of unknown function, and three pyrin-domain-containing genes. Genome-wide, *Ifi202b* is the only significantly differentially expressed transcript based on FDR correction (*q* < 10-2). The

To identify potential candidates that may not be expressed at 13 days, RNA from three B6 and three 2R8 tumors was reverse transcribed and hybridized individually in competition with the mixed-sex control DNA. Genome-wide, *Ifi202b* was again the only significantly differentially expressed transcript, with levels 48-fold higher in 2R8 tumors than in B6 tumors (genome-wide FDR *q* < 0.05). When limiting analysis of the expression array data set to only the 35 candidate genes in the 175.35 to 178.64 Mb minimal region, both *Ifi202b* and the closely related *Aim2* (*Absent in melanoma 2*) were expressed at significantly different

Recent resequencing of the C3H genome has revealed many SNPs in the *Hcs7* region. Data compiled by the Mouse Genome Database (MGD) at the Mouse Genome Informatics website (Blake et al., 2011; URL: http://www.informatics.jax.org; 8/2011) show that the C3H alleles of eight genes in the *Hcs7* region encode amino acids that differ from B6. *Ifi202b* carries a Thr/Ser polymorphism. The most dramatic changes lie in *Mndal*, which carries Arg/Gly and Asp/Tyr polymorphisms. Other genes in the interval with non-synonomous

apoptosis or mitosis in response to DEN injection.

expression of *Ifi202b* is 57-fold higher in line 2R8 than in B6.

SNPs include *Olfr433, Olfr419, Olfr220, Fmn2, Chml,* and *Exo1*.

levels (*q* < 10-4 and *q* < 10-2, respectively).

tumor initiation.

**immune genes** 

The Chromosome 1 liver cancer modifier *Hcs7* has been localized to a 3.3 Mb region on distal Chromosome 1. In DEN-treated mice, the C3H allele of *Hcs7* bred onto a resistant B6 background confers 3- to 7-fold greater tumor multiplicity, dominantly. Consistent with these results, the B6 allele of *Hcs7* bred onto a susceptible C3H background confers 4-fold resistance, recessively. Importantly, C3H alleles on Chromosome 1 including *Hcs7* confer susceptibility to spontaneous tumorigenesis that is comparable to the susceptibility of inbred C3H mice. The Chr 1 hepatocarcinogenesis susceptibility allele in C57BR/cdJ mice, *Hcif2*, does not appear to correspond to *Hcs7*. B6.BR mice congenic for BR alleles distal of 170 Mb are not more susceptible to liver tumorigenesis than B6 mice.

The initiation of tumors after DEN injection is thought to involve the proliferation of DENmutated hepatocytes in response to DEN-induced cell death, within 72 hours of injection (Maeda et al., 2005). We found that *Hcs7* affects neither apoptosis nor mitosis in this time period. We have previously shown that preneoplastic lesions in inbred C3H males grow faster than B6 lesions, and that this differential growth can be seen from prior to 16 weeks until at least 28 weeks of age (Hanigan et al., 1988). Here, we show that *Hcs7* accounts for part of this early growth advantage: the volume of preneoplastic lesions in the B6.C3H congenic lines 1R33 and 2R8 is greater than for B6 lesions by 16 weeks and remains so at 24 weeks. Our results indicate that *Hcs7* begins to affect cell proliferation after the acute reaction to DEN injection and likely tumor initiation, but before 16 weeks, to promote lesion growth.

The 175.35 to 178.64 Mb minimal interval carries 44 genes, of which 12 are related pyrinand/or p200-domain-containing interferon-gamma response genes such as *Ifi202b*, *Aim2*, and *Mnda*. Two additional genes encode proteins that are known or likely to mediate immune responses (*Exo1* and the *Crp*-like gene 1810030J14Rik). This region also contains the cytoskeletal/cell polarity gene *Fmn2* that is overexpressed in B-cell leukemias, a renal cell cancer tumor suppressor (*Fh1*), four known or likely signaling protein genes, and 14 olfactory receptor genes.

*Ifi202b* is a strong candidate for the HCC modifier gene in the *Hcs7* locus. It was the only gene significantly differentially expressed on a genome-wide basis between B6.C3H*Hcs7* congenics and B6 mice at 13 days of age (in the presence and absence of DEN) and in tumors from 32-week-old males. Importantly, CGH analysis indicates that the 5' end of *Ifi202b* is altered in the resistant B6.C3H congenic line 1R5, though the effect of this alteration on the protein or its expression still needs to be established. Finally, *Ifi202b* lies in the minimal susceptibility region determined by congenic lines, between 175.35 and 178.64 Mb. An important test of its candidacy will be to determine whether the CBA inbred strain, closely related to C3H and similarly susceptible to liver tumors, also expresses high levels of *Ifi202b*.

When only genes in the minimal susceptibility region were evaluated for differential expression in tumors, *Aim2* was the only other significantly differentially expressed gene. Aim2 is an interferon-regulated pro-apoptotic protein that is closely related to, can heterodimerize with, and affects the expression of *Ifi202b* (Panchanathan et al., 2010; Choubey et al., 2010). Aim2 is a component of inflammasomes that sense cytoplasmic double-stranded DNA and activate caspase 1, which in turn activates IL1-β and induces inflammation (Fernandes-Alnemri et al., 2009; Bürckstümmer et al., 2009).

In transfected macrophage cell lines, *Ifi202b* overexpression reduces *Aim2* expression, and *Aim2* knock-down increases *Ifi202b* expression (Panchanathan et al., 2010). Similarly, *Aim2* null mice overexpress *Ifi202b* in spleen tissue. Absence of *Aim2* expression correlates with cell growth *in vitro* and with cell immortalization *in vitro*, and *Aim2* is inactivated in approximately half of colon cancers with microsatellite instability (Woerner et al., 2007; Choubey et al., 2010). Therefore, high levels of *Ifi202b* may prevent *Aim2* expression and induce growth. Both *Aim2* and *Ifi202b* were overexpressed in line 2R8 tumors relative to B6 tumors. This result may reflect a tumor-specific response, as 13-day-old line 2R8 mice did not express more *Aim2* than B6 mice.

The high expression of *Ifi202b* in line 2R8 mice relative to B6 might reflect promoter and/or enhancer polymorphisms in both *Aim2* and *Ifi202b*. The NZB allele of *Ifi202b* has been shown to be more active than the B6 allele, most likely due to a TATA-box-creating polymorphism in the NZB *Ifi202b* promoter (Choubey et al., 2010). Analysis of congenic lines suggests that B6-specific polymorphisms, conversely, enhance the expression of *Aim2*. Autoimmunesensitive congenics carrying NZB alleles from 154.7 to ~194 Mb express high levels of *Ifi202b* mRNA and protein relative to B6. However, mice from a line that carries a smaller NZB congenic region between approximately 174.5 to 194 Mb – including the entire *Aim2* gene at 175.2 Mb and the entire *Ifi202b* gene at 175.9 Mb – are not predisposed to autoimmunity. These resistant congenic mice express more *Aim2* mRNA and protein and less *Ifi202b* mRNA and protein than the larger congenic. Indeed, levels of Ifi202b protein are similar to those of B6 mice (Panchanathan et al., 2010). The breakpoint between B6 and NZB alleles in this smaller congenic lies less than one megabase proximal of *Aim2*. The effect of B6 polymorphisms near the 5' end of the *Aim2* gene on *Aim2* expression and, indirectly, *Ifi202b* expression may also explain the resistance to hepatocarcinogenesis of line 2R9. Assessing *Aim2* and *Ifi202b* mRNA and protein levels in lines 2R8, 2R16, 2R9, and 2R7 would address this hypothesis.

Interferon-inducible p200 family members have been shown to interact functionally with p53, MyoD, and Rb and appear to regulate proliferation, differentiation, apoptosis, and senescence (Gariglio et al., 2011). Overexpression of *Ifi202b* in particular has been shown to promote cell survival, while lower levels of expression increase cell death (Roberts et al., 2009; Choubey et al., 2010). The expression of *Ifi202b* is upregulated by IL-6, which has been shown to be virtually required for hepatocarcinogenesis in DEN-treated mice, and *Ifi202b* can stimulate or inhibit the transcription of NFκB target genes depending on cell type (Naugler et al., 2007; Choubey et al., 2010). Given these many cell regulatory functions and its effect on *Aim2* expression, *Ifi202b* could promote lesion growth in multiple ways. Determining whether preneoplastic cells in lines carrying C3H alleles undergo less cell death, an increase in mitosis, or both, will be an important first step in assessing the oncogenicity of *Hcs7*.

Based on genome-wide analysis, no genes other than *Ifi202b* were significantly differentially expressed. However, nine genes were not represented on the expression array. These include several unlikely candidates (olfactory receptors and splicing RNAs that are found throughout the genome), as well as the pyrin/p200 proteins *BC094916*, *Pydc3*, and *Mndal*. The expression of these genes and the effect of amino acid changes in genes in the region will need to be evaluated. Finally, the test of *Ifi202b* as a candidate will require its overexpression (*e.g.*, from a transgene) in B6, or its disruption in C3H.

In transfected macrophage cell lines, *Ifi202b* overexpression reduces *Aim2* expression, and *Aim2* knock-down increases *Ifi202b* expression (Panchanathan et al., 2010). Similarly, *Aim2* null mice overexpress *Ifi202b* in spleen tissue. Absence of *Aim2* expression correlates with cell growth *in vitro* and with cell immortalization *in vitro*, and *Aim2* is inactivated in approximately half of colon cancers with microsatellite instability (Woerner et al., 2007; Choubey et al., 2010). Therefore, high levels of *Ifi202b* may prevent *Aim2* expression and induce growth. Both *Aim2* and *Ifi202b* were overexpressed in line 2R8 tumors relative to B6 tumors. This result may reflect a tumor-specific response, as 13-day-old line 2R8 mice did

The high expression of *Ifi202b* in line 2R8 mice relative to B6 might reflect promoter and/or enhancer polymorphisms in both *Aim2* and *Ifi202b*. The NZB allele of *Ifi202b* has been shown to be more active than the B6 allele, most likely due to a TATA-box-creating polymorphism in the NZB *Ifi202b* promoter (Choubey et al., 2010). Analysis of congenic lines suggests that B6-specific polymorphisms, conversely, enhance the expression of *Aim2*. Autoimmunesensitive congenics carrying NZB alleles from 154.7 to ~194 Mb express high levels of *Ifi202b* mRNA and protein relative to B6. However, mice from a line that carries a smaller NZB congenic region between approximately 174.5 to 194 Mb – including the entire *Aim2* gene at 175.2 Mb and the entire *Ifi202b* gene at 175.9 Mb – are not predisposed to autoimmunity. These resistant congenic mice express more *Aim2* mRNA and protein and less *Ifi202b* mRNA and protein than the larger congenic. Indeed, levels of Ifi202b protein are similar to those of B6 mice (Panchanathan et al., 2010). The breakpoint between B6 and NZB alleles in this smaller congenic lies less than one megabase proximal of *Aim2*. The effect of B6 polymorphisms near the 5' end of the *Aim2* gene on *Aim2* expression and, indirectly, *Ifi202b* expression may also explain the resistance to hepatocarcinogenesis of line 2R9. Assessing *Aim2* and *Ifi202b* mRNA and protein levels in lines 2R8, 2R16, 2R9, and 2R7 would address

Interferon-inducible p200 family members have been shown to interact functionally with p53, MyoD, and Rb and appear to regulate proliferation, differentiation, apoptosis, and senescence (Gariglio et al., 2011). Overexpression of *Ifi202b* in particular has been shown to promote cell survival, while lower levels of expression increase cell death (Roberts et al., 2009; Choubey et al., 2010). The expression of *Ifi202b* is upregulated by IL-6, which has been shown to be virtually required for hepatocarcinogenesis in DEN-treated mice, and *Ifi202b* can stimulate or inhibit the transcription of NFκB target genes depending on cell type (Naugler et al., 2007; Choubey et al., 2010). Given these many cell regulatory functions and its effect on *Aim2* expression, *Ifi202b* could promote lesion growth in multiple ways. Determining whether preneoplastic cells in lines carrying C3H alleles undergo less cell death, an increase in mitosis, or both, will be an important first step in assessing the

Based on genome-wide analysis, no genes other than *Ifi202b* were significantly differentially expressed. However, nine genes were not represented on the expression array. These include several unlikely candidates (olfactory receptors and splicing RNAs that are found throughout the genome), as well as the pyrin/p200 proteins *BC094916*, *Pydc3*, and *Mndal*. The expression of these genes and the effect of amino acid changes in genes in the region will need to be evaluated. Finally, the test of *Ifi202b* as a candidate will require its over-

expression (*e.g.*, from a transgene) in B6, or its disruption in C3H.

not express more *Aim2* than B6 mice.

this hypothesis.

oncogenicity of *Hcs7*.

If Ifi202b is the molecule responsible for promoting hepatocarcinogenesis in mice carrying C3H *Hcs7* alleles, it would make a tempting target for pharmacological intervention. The *Hcs7* region corresponds to parts of human chromosome 1q22-23.1 and 1q43. Chromosome 1q, including these regions, is amplified in 58% to 86% of human hepatocellular carcinomas, independent of stage (Thorgeirsson and Grisham, 2002; Lau and Guan, 2005; Chochi et al., 2009; Zhang et al., 2010). When only parts of 1q are amplified, 1q22-23 is frequently affected. These observations imply that chromosome 1q alterations are an early event in hepatocarcinogenesis. Chromosome 1q is also frequently amplified in other liver cancers such as hepatoblastoma, cholangiocarcinoma, and fibrolamellar carcinoma, as well as cancers of other tissues such as the pancreas and the breast (Buendia, 2002; Climent et al., 2002; Birnbaum et al., 2011; Ward and Waxman, 2011). Humans have four homologs of *Ifi202b* in the 1q22-23.1 region, including an ortholog of *Aim2*, an ortholog of *Mnda*, *Ifi16* and *IfiX*/*Pyhin1* (www.ensembl.org; 9/2011; Gariglio et al., 2011). All of these genes have been implicated in growth suppression in cell lines and ectopic tumor models; however, *Ifi16* in particular is also associated *in vivo* with rapidly dividing epithelium and with undifferentiated tissue including fetal hepatocytes and hepatocytes following liver transplantation (Gariglio et al., 2011). In an analysis of gene expression in a human tumor carrying a 1q amplification, four chromosome 1 genes were overexpressed. Only two of these correspond to mouse Chromosome 1: *Mnda* and "interferon-gamma inducible gene" at 1q22 (*Ifi16* or *IfiX*; Niketeghad et al., 2001). All four *Ifi* p200 human genes express transcripts that encode both the pyrin cell death and p200 protein interaction domains (www.ensembl.org; 9/2011). Each also expresses transcripts that carry only one of the domains, and *Ifi16* in particular expresses transcripts that carry only the pyrin domain, only the p200 domain, or both. These many alternative transcripts suggest the possibility that one of them is orthologous to *Ifi202b*, which expresses only the p200 domain. An analysis of human tumors should reveal which p200 family transcripts they express. Transgenesis could then be used to test the oncogenicity of these selected transcripts.

C3H alleles from several regions of mouse Chromosome 1 conferred some significant susceptibility to B6 mice. Similarly, several regions on human chromosome 1 are amplified, sometimes simultaneously but separately, in human tumors (Lau and Guan, 2005; Chochi et al., 2009). It seems likely that multiple genes on Chromosome 1 in both mice and humans affect hepatocarcinogenesis.
