Cancer Cell
`
`Article
`
`Inhibition of De Novo NAD+ Synthesis
`by Oncogenic URI Causes Liver
`Tumorigenesis through DNA Damage
`
`Krishna S. Tummala,1 Ana L. Gomes,1 Mahmut Yilmaz,1 Osvaldo Gran˜ a,2 Latifa Bakiri,3 Isabel Ruppen,4
`Pilar Xime´ nez-Embu´ n,4 Vinayata Sheshappanavar,5 Manuel Rodriguez-Justo,6 David G. Pisano,2 Erwin F. Wagner,3
`and Nabil Djouder1,*
`1Growth Factors, Nutrients and Cancer Group, BBVA Foundation-Cancer Cell Biology Programme, Spanish National Cancer Research
`Centre, CNIO, 28029 Madrid, Spain
`2Bioinformatics Unit, Structural Biology and Biocomputing Programme, Spanish National Cancer Research Centre, CNIO, 28029 Madrid,
`Spain
`3Genes, Development, and Disease Group, BBVA Foundation-Cancer Cell Biology Programme, Spanish National Cancer Research Centre,
`CNIO, 28029 Madrid, Spain
`4Proteomics Core Unit, ProteoRed ISCIII, Biotechnology Programme, Spanish National Cancer Research Centre, CNIO, 28029 Madrid, Spain
`5Department of Pathology, Royal London Hospital, London E1 1BB, UK
`6Department of Cellular Pathology, University College London NHS Trust, London NW1 2BU, UK
`*Correspondence: ndjouder@cnio.es
`http://dx.doi.org/10.1016/j.ccell.2014.10.002
`
`SUMMARY
`
`Molecular mechanisms responsible for hepatocellular carcinoma (HCC) remain largely unknown. Using
`genetically engineered mouse models, we show that hepatocyte-specific expression of unconventional pre-
`foldin RPB5 interactor (URI) leads to a multistep process of HCC development, whereas its genetic reduction
`in hepatocytes protects against diethylnitrosamine (DEN)-induced HCC. URI inhibits aryl hydrocarbon (AhR)-
`and estrogen receptor (ER)-mediated transcription of enzymes implicated in L-tryptophan/kynurenine/
`nicotinamide adenine dinucleotide (NAD+) metabolism, thereby causing DNA damage at early stages of
`tumorigenesis. Restoring NAD+ pools with nicotinamide riboside (NR) prevents DNA damage and tumor
`formation. Consistently, URI expression in human HCC is associated with poor survival and correlates nega-
`tively with L-tryptophan catabolism pathway. Our results suggest that boosting NAD+ can be prophylactic or
`therapeutic in HCC.
`
`INTRODUCTION
`
`Hepatocellular carcinoma (HCC) is the commonest, usually lethal,
`human primary liver neoplasm (GLOBOCAN v2.0, 2008). The early
`stage is characterized by low- to high-grade dysplastic nodules,
`‘‘preneoplastic lesions’’ (Kudo, 2009). These frequently develop
`in chronic inflammatory liver disease or hepatitis, which can pro-
`mote fibrosis, cirrhosis, and progression to HCC. Thus, precan-
`cerous lesions have clinical value for HCC prediction (Libbrecht
`et al., 2001), but therapeutic options are limited (El-Serag, 2011).
`
`In early stages of many cancers, including HCC, oncogene
`activation induces replicative stress, resulting in DNA damage
`leading to chromosomal
`instability (CIN), which accelerates
`tumor development (Teoh et al., 2008). DNA damage elicits a
`key repair mechanism, the DNA damage response (DDR), initi-
`ated by phosphorylation of checkpoint proteins Chk1, Chk2,
`and p53 (Reinhardt and Schumacher, 2012). p53-dependent
`responses, including cell cycle arrest and/or senescence, are
`induced, limiting preneoplastic lesions’ growth. When DNA dam-
`age is too pronounced, p53 engages an apoptotic program by
`
`Significance
`
`HCC is the third leading cause of cancer death worldwide with limited therapeutic options. Here we demonstrate that NAD+
`deficit-induced genotoxic stress is critical to initiate liver tumorigenesis and unravel a critical link between nutrient meta-
`bolism and genome integrity. Because our findings are relevant in human HCC, we propose that nutritional supplementation
`of NR, a vitamin B3 derivative, or other NAD+ boosters can be used as preventive and curative therapies in oncogene-
`induced NAD+ depletion-mediated DNA damage and carcinogenesis, especially in patients with precancerous lesions.
`Therapeutic intervention on metabolic alterations prior to genomic instability should be further considered to prevent
`tumorigenesis.
`
`826 Cancer Cell 26, 826–839, December 8, 2014 ª2014 Elsevier Inc.
`
`Elysium Health Exhibit 1017
`Page 1 of 14
`
`

`

`Cancer Cell
`URI-Induced NAD+ Depletion Causes HCC Development
`
`upregulation of Bcl-2 family proteins (Noxa, Puma, Bid, and/or
`Bax). p53 dysfunctions allow tumor cells to escape apoptosis,
`and thus, mutations inactivating p53 are the most common alter-
`ations observed in HCC (Reinhardt and Schumacher, 2012).
`In pathophysiological situations, the balance between cell pro-
`liferation and apoptosis can be altered, perturbing tissue homeo-
`stasis. Apoptotic dysregulations are important in liver disease.
`Insufficient apoptosis, eliminating mutated cells, combined
`with inflammation-mediated proliferation can promote liver can-
`cer development. Excessive or sustained apoptosis causes liver
`injuries, increased hepatocyte regeneration, which enhances ge-
`netic errors and predisposes to HCC (Malhi and Gores, 2008).
`Still, the initiating hepatocarcinogenesis events remain unclear.
`Developing experimental models mimicking distinct stages of
`HCC development would help to explore molecular mechanisms
`linking histopathological changes to hepatocarcinogenesis.
`Unconventional prefoldin RBP5 interactor (URI), a member
`of the R2TP/URI-prefoldin (PFD)-like complex containing the
`heat shock protein 90 (HSP90) (Boulon et al., 2010), is an onco-
`gene amplified in human ovarian carcinomas and downstream
`effector of the growth factor and nutrient-regulated mTOR/
`S6K1 signaling cascade (Theurillat et al., 2011). URI inhibits
`phosphatase PP1g, thereby increasing S6K1 activity-dependent
`survival signaling. Thus, URI/PP1g complexes maintain the
`mitochondrial threshold for apoptosis in accordance to nutrient
`availability. URI overexpression promotes survival, while its dele-
`tion enhances cancer cell death (Djouder et al., 2007; Theurillat
`et al., 2011). Prompted by these observations, and the fact
`that HCC occurs on the basis of mitochondrial dysfunction-
`mediated hepatocyte death and liver injury (Luedde et al.,
`2014; Malhi and Gores, 2008), we investigate the role of URI in
`hepatocarcinogenesis.
`
`RESULTS
`
`URI Expression in Mouse Hepatocytes Induces
`Spontaneous Liver Tumors
`We generated a Col1a1 knockin mouse (Figures S1A and S1B
`available online), expressing human URI (hURI) via a tetracy-
`cline-dependent transactivator controlled by the hepatocyte-
`specific liver activated protein promoter (Figures S1C–S1E).
`These mice, designated hURI-tetOFFhep, and littermates lacking
`hURI expression are referred to hereafter as ‘‘mutants’’ and
`‘‘controls,’’ respectively. Without doxycycline, hURI was ex-
`pressed specifically in hepatocytes from one allele from E10.5
`(Carpenter et al., 2005), roughly twice as much as mouse
`URI (Figures 1A, S1F, and S1G), similar to the increase of URI
`expression in human HCC (see below).
`We observed no pathological signs in 3-week-old mutants. In
`8-week-old mutants, hematoxylin and eosin (H&E) staining re-
`vealed anisokaryotic clusters (Figure 1B) resembling low-grade
`dysplastic nodules observed in human hepatitis (Libbrecht
`et al., 2001). At 12 weeks the clusters developed into high-grade
`dysplastic nodules (Figures 1B, S1H, and S1I), similar to human
`large liver cell dysplasia (LLCD) (Libbrecht et al., 2001). Fibrosis
`was detected at 8 weeks and increased over time until 24 weeks,
`as assessed by Sirius Red (SR), Masson Trichrome, alpha
`smooth muscle actin, type I collagen (COL1A1), and reticulin
`staining (Figure S1J). Quantification showed that about 1% to
`
`3% of livers were SR positive in mutants, representing 100%
`to 300% increase over littermates (Figure S1K). Increases in
`fibrotic markers were measured by quantitative RT-PCR (Figures
`S1L–S1N), but serum alanine aminotransferase (ALT) values
`remained unchanged (Figure S1O).
`Between 24–54 weeks macroscopic lesions including ade-
`noma and early HCC emerged. Recent reports described malig-
`nant transformation of human adenomas, but tumors in our
`model developed simultaneously (Pilati et al., 2014). Between
`54–65 weeks low-grade and differentiated HCC were fully
`apparent, and between 65–75 weeks, 40% of mutants devel-
`oped macroscopic high-grade tumors occupying 20%–60%
`of the liver (Figures 1B, 1C, and S1P). There were 25%–50% of
`hepatocytes that were Ki67-positive, suggesting aggressive
`tumors (data not shown). According to World Health Organi-
`zation criteria (WHO, 2008), all tumors were well/moderately
`differentiated: 20% glandular/acinar, indicative of telangiectatic
`variants, and 80% trabecular. No cholangiocarcinoma were
`detected (Figure 1C). Serum glucose, ALT, and total bile
`acids were affected (Figure S1Q). Surprisingly, serum albumin
`was increased, suggesting that liver function might not be fully
`compromised (Figure S1Q).
`All mutants died at 85 weeks, with a median survival
`of 76 weeks before complete liver failure (Figure 1D). Immuno-
`histochemistry (IHC) and pathological analyses revealed hURI-
`positive hepatocyte-like cells in 30% of mutant lungs with
`HCC, indicating aggressive metastases (Figures S1R and S1S).
`Histopathological characterization confirmed the presence of
`heterogeneous tumors with collapsed reticulin fibers, suggesting
`increased hepatocyte death and proliferation, as indicated by
`Ki67 (Figure 1E). Increases in alpha fetoprotein (AFP) levels, a
`clinical marker for human HCC varied, but all tumors displayed
`dramatic increases in p53 abundance and phosphorylation (Fig-
`ures 1E and 1F), suggesting that p53 may either carry mutations
`or may be improperly folded (Trinidad et al., 2013), thus possibly
`inactive.
`Fully developed HCC appeared at 30 weeks in hepatocarcino-
`gen diethylnitrosamine (DEN)-treated hURI-tetOFFhep mice (Ves-
`selinovitch and Mihailovich, 1983) (Figure S1T). When hURI was
`expressed from two alleles, increasing its expression to 6-fold
`compared to heterozygous hURI-tetOFFhep mice, HCC were
`detected at 10 weeks (Figure S1U), highlighting the importance
`of URI dosage. Embryonic development was not
`involved
`because liver tumors were also detected in mice kept on doxy-
`cycline until 8 weeks (expressing hURI from 8 weeks) then trans-
`ferred to normal (chow) diet (Figure S1V). Thus, hURI expression
`in mouse hepatocytes induces spontaneous HCC.
`
`Continuous URI Expression Is Essential for
`Hepatocarcinogenesis
`Ceasing hURI expression in 8-week-old mutants for 24 weeks
`reduced fibrosis and abolished dysplastic foci and prevented
`early tumors, without affecting liver-to-body weight ratios (Fig-
`ures 2A–2D and S2A). S6K1 activity was increased in 24-week-
`old mice, but remained constant when hURI expression ceased,
`indicating that mTOR/S6K1 activation was hURI-independent
`(Figure 2B). Switching hURI expression off until 60 weeks pre-
`vented tumor development and normalized ALT levels (Figures
`S2B–S2E). Similarly, when hURI was expressed for 24 weeks,
`
`Cancer Cell 26, 826–839, December 8, 2014 ª2014 Elsevier Inc. 827
`
`Elysium Health Exhibit 1017
`Page 2 of 14
`
`

`

`Cancer Cell
`URI-Induced NAD+ Depletion Causes HCC Development
`
`Figure 1. URI Expression in Mouse Hepatocytes Induces Spontaneous Liver Tumors
`(A) Representative images of IHC stained liver sections from 3-week-old hURI-tetOFFhep mice using hURI and FLAG antibodies. Insets represent the periportal
`area, showing hepatocyte specific hURI expression. (n > 10).
`(B) Representative images of H&E stained liver sections from 3- (n > 6), 8- (n > 19), 12- (n > 11), and 32-week-old (n > 7) hURI-tetOFFhep mice. Bottom two rows are
`representative images of whole livers from hURI-tetOFFhep mice at 32 and 75 weeks of age. Black dotted circles mark LLCD-like lesions and black arrows point
`anisokaryotic clusters in mutant hURI-tetOFFhep mice. Yellow dotted circles depict adenoma and HCC at 32 and 75 weeks of age, respectively.
`(C) Percentage of control and mutant hURI-tetOFFhep mice bearing liver abnormalities in 60- to 75-week-old-mice.
`(D) Kaplan Meier curve of control (n = 17) and mutant (n = 17) hURI-tetOFFhep mice. Log rank test p = 0.0036; Hazard ratio = 0.1603.
`(E) Representative images of H&E, IHC, and reticulin stained liver sections from control and four tumors derived from one mutant hURI-tetOFFhep. NT, PT, and
`T denote nontumoral, peritumoral, and tumoral tissues, respectively.
`(F) WB analysis of control and mutant hURI-tetOFFhep livers. ‘‘T’’ denotes tumor.
`See also Figure S1.
`
`until high-grade dysplastic nodules/early HCC and adenomas
`were apparent, then, switched-off for 28 weeks, only residual
`anisokaryotic clusters were detected, but no adenomas or
`HCCs (Figure 2E). However, ultrasound analysis demonstrated
`that well/moderately differentiated HCC (above 60 weeks) did
`not regress when hURI expression was ceased for 5 weeks (Fig-
`ures S2F–S2H). Thus, continuous hURI expression is required for
`the maintenance of preneoplastic lesions and early tumors.
`Aggressive HCCs with sufficient genetic mutations become
`URI independent, even though ceasing URI expression for a
`longer time remains to be tested.
`
`We genetically inactivated URI specifically in hepatocytes by
`crossing URI(lox/lox) and serum albumin (SA)-CreERT2 mice
`(Schuler et al., 2004). URI deletion in hepatocytes after tamoxifen
`treatment to obtain URI(+/D)hep or URI(D/D)hep mice, was confirmed
`by IHC and western blotting (WB) (Figures 2F and 2G). Homozy-
`gous deletion of URI led to death of URI (D/D)hep mice around
`10 days (Figure S2I). Disruption of tissue architecture, presence
`of atypia, dilated veins with intrahepatic bleeding, signs of
`necrosis, and inflammatory cell
`infiltration were observed by
`H&E staining. Additionally, SR staining, collapsed reticulin
`fibers, and increased ALT indicated that hepatocytes underwent
`
`828 Cancer Cell 26, 826–839, December 8, 2014 ª2014 Elsevier Inc.
`
`Elysium Health Exhibit 1017
`Page 3 of 14
`
`

`

`Cancer Cell
`URI-Induced NAD+ Depletion Causes HCC Development
`
`Figure 2. Continuous URI Expression Is Essential for Hepatocarcinogenesis
`(A) Representative images of H&E stained liver sections from 32-week-old hURI-tetOFFhep mice fed with (+) or without () doxycycline (Dox) after dysplatic lesion
`formation at 8 weeks. Dotted black circle represents premalignant lesions. (n R 5).
`(B) WB analysis of hURI-tetOFFhep livers as described in (A).
`(C) Representative images of Sirius Red stained liver sections from mice described in (A). (n R 5).
`(D) Quantification of Sirius Red stained liver sections from mice described in (C). (n R 5).
`(E) Representative images of full livers and H&E stained liver sections from hURI-tetOFFhep mice treated with Dox for 28 weeks. Treatment started at 24 weeks of
`age, after the appearance of dysplastic lesions, adenomas, and early HCC. Dotted black circles denote reminiscent anisokaryotic areas. (n R 5).
`(F) Representative images of IHC stained liver sections for endogenous URI in URI(+/+)hep, URI(+/D)hep, and URI(D/D)hep mice. (n R 3).
`(G) WB liver analysis for endogenous URI in URI(+/+)hep, URI(+/D)hep, and URI(D/D)hep livers.
`(H) Representative images of whole livers from URI(+/+)hep and URI(+/D)hep mice treated with diethylnitrosamine (DEN) and sacrificed at 24 weeks of age. Dotted
`yellow circles depict liver tumors. (n R 5). Bottom pictures represent reticulin stained livers, black circle depicts the dysplastic area.
`(I) Tumor burden of mice described in (H).
`(J) Serum ALT levels from mice described in (H).
`Data represented as mean ± SEM (p % 0.05 = *). See also Figure S2.
`
`massive apoptotic program leading to liver injury, suggesting
`that these mice die from fulminant liver failure (Figures S2J and
`S2K). However, URI(+/D)hep mice, in which URI expression was
`approximately halved (Figures 2F and 2G), supplied a liver dam-
`age-inducing 3,5-diethoxycarbonyl-1,4-dihydrocollidine (DDC)-
`supplemented diet, presented significantly less liver damage
`and fibrosis as shown by ALT, SR, and reticulin stainings (Figures
`S2L–S2N). Thus URI reduction protects from hepatocyte injury.
`DEN-treatment increased URI levels in 3-week-old URI(+/+)hep
`mouse livers and in wild-type murine HCC (Figures S2O
`and S2P). Moreover, DEN induced tumor development at
`24 weeks in 60% of URI(+/+)hep mice and enhanced ALT levels,
`but URI(+/D)hep mice did not show any tumors at this age and dis-
`played normal ALT values (Figures 2H–2J). Levels of cytochrome
`P450 2E1 and 1A1 catalyzing DEN were unchanged in 3-week-old
`
`livers of DEN-treated URI(+/+)hep and URI(+/D)hep mice (Figure S2O),
`suggesting that halving URI levels does not affect DEN bio-
`activation (Kang et al., 2007). Thus, while complete deletion of
`URI induces liver injury, halving URI is beneficial to maintain liver
`homeostasis and prevents liver injury and HCC development.
`
`URI-Induced DNA Damage Initiates Liver Tumorigenesis
`To further elucidate the mechanisms of hepatocyte death and
`their contribution to HCC, we checked for phosphorylation of
`histone H2AX (gH2AX), a DNA damage marker that triggers
`apoptosis via the p53-DDR pathway. gH2AX and p53 phosphor-
`ylation and abundance did not differ in 1-week-old livers (Fig-
`ure S3A). At 3 weeks, a nonpathological stage with no dysplastic
`lesions, gH2AX, phosphorylation of the 32 kDa subunit of repli-
`cation protein A (RPA32) at Ser-4 and Ser-8, and p53 abundance
`
`Cancer Cell 26, 826–839, December 8, 2014 ª2014 Elsevier Inc. 829
`
`Elysium Health Exhibit 1017
`Page 4 of 14
`
`

`

`Cancer Cell
`URI-Induced NAD+ Depletion Causes HCC Development
`
`Figure 3. URI-Induced DNA Damage Initiates Liver Tumorigenesis
`(A) Representative images of gH2AX IHC stained liver sections from 3-week-old hURI-tetOFFhep mice. Insets denote gH2AX positive nuclei. (n = 5).
`(B) Quantification of (A).
`(C) WB analysis of 3-week-old hURI-tetOFFhep livers.
`(D) Representative images of gH2AX IHC stained liver sections from 8-week- and 12-week-old hURI-tetOFFhep mice. Dotted black shapes depict anisokaryotic
`clusters positive for gH2AX. (n = 6).
`(E) Quantification of (D).
`(F) WB analysis of 8-week-old hURI-tetOFFhep livers
`(G) WB analysis of 8-week-old hURI-tetOFFhep livers, with or without p53 inactivation.
`(H) Reticulin and Sirius Red stained livers described in (G).
`(I) Serum ALT levels of mutant mice with or without p53 inactivation. (n > 4).
`(J) Kaplan-Meier survival curve of control and mutant hURI-tetOFFhep mice with and without p53 inactivation. (Log rank test p % 0.001.)
`(K) Percentage of tumor incidence in hURI-tetOFFhep mice with or without p53 inactivation.
`Data represented as mean ± SEM (*p % 0.05 and **p % 0.01). See also Figure S3.
`
`and phosphorylation were higher in mutants (Figures 3A–3C),
`suggesting that DDR precedes precancerous lesions. While
`p53-dependent apoptosis occurred in cells that unsuccessfully
`repair DNA (cleaved caspase 3; Figure 3C), hepatocyte prolifer-
`ation rate was reduced (Figures S3B and S3C), suggesting that
`high proliferation is not the initial hepatocarcinogenic event.
`gH2AX-positive nuclei were more abundant in older mutant
`livers (8- to 12-week-old) with obvious dysplastic lesions.
`Furthermore, Ser-345 phosphorylation (and hence activation)
`of Chk1 was enhanced in mutants, but not Thr-68 phosphoryla-
`tion of Chk2, suggesting single strand break (Reinhardt and
`Schumacher, 2012) (Figures 3D–3F). Enhanced p53 phosphory-
`lation at Ser-18 (Chk1 target), acetylation at Ac-lys379 (Ito et al.,
`2001) (Figure 3F), and expression of p19ARF (Reinhardt and
`Schumacher, 2012) (Figure S3D) indicate p53 stabilization.
`
`Senescence-associated b-galactosidase activity, expression
`of several p53 target genes, BAX, and p21 protein abundance
`were increased in 8- and 12-week-old mutants, while expression
`of Xiap was decreased (Figures 3F, S3E, and S3F) (Reinhardt and
`Schumacher, 2012), suggesting increased hepatocyte death as
`also shown by collapsed reticulin fibers (Figure S3G). Increased
`abundance of proliferating cell nuclear antigen, cyclin D1, and
`Ki67-positive nuclei suggested compensatory proliferation (Fig-
`ures 3F and S3H). Finally, MAD2, a CIN marker and downstream
`effector of cyclin D1, was increased (Figure 3F). Thus, hURI
`expression in hepatocytes induces genotoxic stress, apoptosis,
`compensatory proliferation, and CIN.
`To assess genotoxic stress-induced apoptosis in HCC devel-
`opment, p53 was inactivated in hURI-tetOFFhep mice. In p53-in-
`activated 8-week-old mutants (hURI (+/Ki); p53ERTAM (+/Ki)),
`
`830 Cancer Cell 26, 826–839, December 8, 2014 ª2014 Elsevier Inc.
`
`Elysium Health Exhibit 1017
`Page 5 of 14
`
`

`

`Cancer Cell
`URI-Induced NAD+ Depletion Causes HCC Development
`
`cleaved caspase 3, Bax expression, and collapsed fibers were
`decreased (Figures 3G and 3H). Furthermore, SR staining and
`ALT levels were reduced (Figures 3H and 3I), indicating that
`DNA damage-activated p53 is required for hepatocyte death
`and liver injury. While apoptosis was drastically suppressed,
`inactivation of p53 significantly reduced survival and accelerated
`liver tumorigenesis (Figure 3J): 80% of mice displayed aggres-
`sive HCC (Figure 3K). Deletion of Cdkn2a did not modify mouse
`survival or tumor burden (data not shown). Thus, genotoxic
`stress, rather than excessive apoptosis, is the critical initiating
`event in liver carcinogenesis.
`
`URI Causes DNA Damage and Liver Tumorigenesis by
`Inhibiting De Novo NAD+ Synthesis
`To identify URI-mediated hepatocarcinogenetic events, we first
`examined mTOR activation, which had been implicated in HCC
`development via DNA damage (Menon et al., 2012). No increases
`in S6K1 activity were detected at 1 week (Figure S4A). In sequen-
`tial immunoprecipitation experiments, using 1-week-old liver ex-
`tracts, free hURI molecules were revealed by WB after complete
`depletion of PP1g (Figures S4B and S4C), and vice versa (data
`not shown). When 3-week-old mice were supplied a rapamy-
`cin-containing diet, progression to preneoplastic abnormalities
`continued, if not further pronounced (data not shown). Thus,
`although a fraction of hURI binds PP1g, hURI apparently has a
`PP1g-independent role in DNA damage and liver tumorigenesis.
`Additionally, no differences in reactive oxygen species (ROS)
`were observed in 1- and 8-week-old livers (Figures S4D and
`S4E), suggesting that DNA damage is ROS-independent.
`Global transcriptomic and proteomic profiling were performed
`in a very early nonpathological stage and early premalignant
`state (1- and 8-week-old livers). Transcripts’ sequencing re-
`vealed small fractions of genes differentially expressed upon
`hURI expression: 303 out of 12,295 genes at 1 week, and 740
`out of 11,133 (false discovery rate [FDR] < 0.05) at 8 weeks (Fig-
`ures 4A and S4F). Similarly, isobaric tags for relative and abso-
`lute quantification (iTRAQ) identified 2,394 proteins: 122 and
`597 of which were differentially expressed in 1- and 8-week
`livers, respectively (Figures 4B and S4G; Table S1).
`Heatmapping revealed that most differentially expressed pro-
`teins were downregulated (Figure S4H). Significant overlaps in
`the differentially expressed transcripts and proteins at 1 and
`8 weeks (Figures S4I and S4J), indicated hURI-dependent tran-
`scriptional repression mechanisms. Ingenuity pathway analysis
`(IPA) revealed that among canonical metabolic pathways, the
`L-tryptophan/kynurenine catabolism leading to de novo nicotin-
`amide adenine dinucleotide (NAD+) synthesis was one of the
`most significant downregulated pathways (Figures 4C and S4K).
`Enzymes implicated in the L-tryptophan/kynurenine degrada-
`tion, including tryptophan 2,3-dioxygenase (TDO2) and arylfor-
`mamidase (AFMID) catalyzing the initial rate-limiting step and
`kynurenine 3-monooxygenase (KMO), kynureninase (KNYU), and
`3-hydroxyanthranilate 3,4-dioxygenase (HAAO) were all downre-
`gulated (Figure 4D). Gene set enrichment analysis (GSEA) (Subra-
`manian et al., 2005), using the RNA sequencing data and Kyoto
`Encyclopedia of Genes and Genomes database, corroborated
`these defects (data not shown). WB confirmed that TDO2 and
`AFMID expression was reduced >50% in these livers (Figures 4E
`and S4L) and in adult livers expressing hURI (Figure S4M).
`
`NAD+ concentrations were reduced in 3- and 6-week mutant
`livers (Figures 4F and S4N), while increases in TDO2, AFMID,
`and NAD+ levels were detected in URI(+/D)hep livers (Figures 4G
`and 4H). Consistent with previous observations (Konishi et al.,
`1986), liver NAD+ levels were depleted in DEN-treated mice,
`and URI reduction enhanced NAD+ levels (Figure S4O). Thus,
`URI reduction enhances NAD+ de novo synthesis, potentially ex-
`plaining the protective effect of URI deletion in HCC. Further-
`more, NAD+ concentrations inversely correlated with URI levels
`in four human HCC cell
`lines (Huh-7, HepG2, SNU-398, and
`SNU-449). While URI depletion significantly increased NAD+
`levels, URI overexpression reduced NAD+ values (Figure S4P).
`URI overexpression in SNU-449 cells, which had high NAD+
`values and low endogenous URI levels, increased their growth,
`whereas URI depletion in Huh-7 and HepG2 cells displaying
`high endogenous URI, significantly reduced their growth (Fig-
`ure S4Q). URI-regulating NAD+ levels may therefore be relevant
`for human liver tumorigenesis.
`line SNU-449
`Depleting TDO2 and AFMID in HCC cell
`significantly reduced NAD+ levels (Figure S4R). Importantly, 14C-
`labeled NAD+ levels in four human HCC cell lines incubated with
`14C-tryptophan reduced significantly following AFMID depletion
`(Figure S4S), indicating that L-tryptophan degradation accounts
`for de novo NAD+ synthesis. Furthermore, expression of key en-
`zymes of three other pathways implicated in oncogenesis was
`unaffected by hURI expression: SHMT1, G6PD, and GOT1 of
`the glycine/serine/threonine, pentose phosphate and glutamine/
`aspartate pathways, respectively (Figure S4T). Finally, expression
`of nicotinamide phosphoribosyltransferase (NAMPT, implicated
`in NAD+ biosynthesis through salvage reactions, Figure S4U)
`and activity of poly (ADP-ribose) polymerase (PARP), the main
`NAD+-consuming enzyme (Figure S4U) were not affected at early
`stages (1 week), and levels of NADH and several dehydrogenases
`that reduce NAD+ to NADH were decreased (Figure S4U;
`Table S2). Reduction of NAD+ is thus mainly due to downregula-
`tion of L-tryptophan/kynurenine catabolism.
`We next induced liver injury and hepatocyte proliferation in
`C57BL/6 mice with DDC-supplemented diet for 4 days, treated
`them with DMSO or Ro-61-8048, a KMO inhibitor, for the next
`3 days, and sacrificed mice on day 8 (Figures S4V and S4W).
`NAD+ concentrations were reduced and DNA damage foci
`significantly elevated in Ro-61-8048-treated livers (Figures 4I–
`4K). Thus, L-tryptophan/kynurenine pathway inhibition in vivo
`leads to reduced NAD+ concentrations and DNA damage, reca-
`pitulating effects of hURI expression. Finally, nontumorigenic
`mouse liver cells AML-12, stably depleted of TDO2 and AFMID
`and transplanted into immunodeficient mice, formed aggressive
`tumors (Figure S4X), suggesting that inhibition of L-tryptophan
`pathway leads to transformation and tumorigenesis.
`We assessed whether DNA damage was a consequence
`of inactivation of NAD+-consuming enzymes, such as SIRT1 or
`PARP (Durkacz et al., 1980; Herranz et al., 2010). In SNU-449
`cells, SIRT1 inhibition by EX-527, which may enhance NAD+,
`reduced RPA32 phosphorylation, whereas activating SIRT1
`with resveratrol, which may lower NAD+, increased phosphoryla-
`tion of RPA32 (Figure S4Y). Because in our model NAD+ deficits
`increased replicative stress, DNA damage is unlikely due
`to SIRT1 inhibition alone. Additionally, in URI-overexpressing
`SNU-449 cells, in which NAD+ levels were lowered (Figure S4P),
`
`Cancer Cell 26, 826–839, December 8, 2014 ª2014 Elsevier Inc. 831
`
`Elysium Health Exhibit 1017
`Page 6 of 14
`
`

`

`Cancer Cell
`URI-Induced NAD+ Depletion Causes HCC Development
`
`Figure 4. URI Causes DNA Damage and Liver Tumorigenesis by Inhibiting De Novo NAD+ Synthesis
`(A) Volcano plots from RNA sequencing representing differentially expressed significant (blue) and unchanged (red) mRNA species in livers from 1- and 8-week-
`old hURI-tetOFFhep mice. (n > 3).
`(B) Histogram of differentially expressed proteins analyzed by iTRAQ in livers from 1- and 8-week-old hURI-tetOFFhep mice. Numbers of proteins significantly
`downregulated (green) and upregulated (red) are shown. (n = 5).
`(C) Top downregulated canonical metabolic pathways based on iTRAQ data from 8-week-old mice, analyzed by using IPA software.
`(D) Scheme of de novo NAD+ synthesis. Fold change of protein expression detected in iTRAQ are represented within the brackets. Ro-61-8048 is an inhibitor
`for KMO.
`(E) WB analysis (left) and quantification of reduction (mutant over control, right) of TDO2 and AFMID of 8-week-old hURI-tetOFFhep livers.
`(F) Liver NAD+ concentrations in 3-week-old hURI-tetOFFhep mice. (n R 10).
`(G) WB analysis of URI(+/+)hep and URI(+/D)hep livers.
`(H) NAD+ levels in livers from URI(+/+)hep and URI(+/D)hep mice. (n = 5).
`(I) Liver NAD+ levels in C57BL/6 mice previously fed with DDC and treated with either DMSO (1%) or Ro-61-8048 (25 mg/Kg) compound. (n R 5).
`(J) Representative images of gH2AX IHC stained liver sections from C57BL/6 mice described in (I). (n = 5).
`(K) Quantification of (J).
`Data represented as mean ± SEM (*p % 0.05 and ***p % 0.001). See also Figure S4 and Tables S1 and S2.
`
`increased RPA32 phosphorylation.
`SIRT1 activation further
`Thus, modulating SIRT1 activity may affect PARP activity either
`via modulation of NAD+ levels or through regulation of acet-
`ylation-dependent PARP1 activity (Rajamohan et al., 2009).
`Notably, URI overexpression increased RPA32 phosphorylation,
`which was not further enhanced when PARP was inhibited (Fig-
`ure S4Y). Finally, PARP activity was reduced in 3-week-old
`mutants, while NAMPT expression remained unchanged (Fig-
`ure S4Z). Thus, hURI-mediated NAD+ depletion may induce
`DNA damage via PARP inhibition.
`
`Restoring NAD+ Pools Protects from DNA Damage and
`Prevents Tumor Formation
`To investigate whether restoring NAD+ pools would prevent
`dysplastic nodules and tumor formation, 3-week-old hURI-
`
`832 Cancer Cell 26, 826–839, December 8, 2014 ª2014 Elsevier Inc.
`
`tetOFFhep mice were supplied with a nicotinamide riboside
`(NR) diet. NR significantly increased hepatic NAD+ concentra-
`tions (Figure S5A) without affecting liver-to-body weight ratio
`(Figure S5B). We detected dysplastic lesions and DNA damage
`in all mutants on chow, but not in those on NR, which also had
`reduced fibrosis, p53 abundance, and Ser-18 phosphorylation
`(Figures 5A–5D, S5C, and S5D). Prolonged NR treatment pre-
`vented tumor development and reduced ALT levels (Figures
`5E–5G). Similarly liver tumors were prevented in 30-week-old
`homozygous mutants with higher URI
`levels (Figure S5E).
`Thus, restoring NAD+ pools protects from hURI-induced DNA
`damage, preneoplastic lesions, and tumor development. Sur-
`prisingly, 12-week-old homozygous mutants with full blown tu-
`mors then on 48 weeks of NR regimen showed significant tumor
`regression (Figures S5F and S5G), and their livers had high levels
`
`Elysium Health Exhibit 1017
`Page 7 of 14
`
`

`

`Cancer Cell
`URI-Induced NAD+ Depletion Causes HCC Development
`
`Figure 5. Restoring NAD+ Pools Protects from DNA Damage and Prevents Tumor Formation
`(A) Representative images of H&E and gH2AX IHC stained liver sections from 12-week-old hURI-tetOFFhep mice fed with either chow (n R 15) or NR diets started
`at 3 weeks of age (n R 15). Dotted black lines indicate anisokaryotic clusters present in mutant hURI-tetOFFhep mice under chow diet.
`(B) Quantification of dysplastic lesions in the hURI-tetOFFhep mice described in (A).
`(C) Quantification of gH2AX positive nuclei in the hURI-tetOFFhep mice described in (A).
`(D) WB analysis of mutant hURI-tetOFFhep livers as described in (A).
`(E) Representative images of whole livers and H&E stained liver sections from 30- or 60-week-old hURI-tetOFFhep mice supplemented with NR diet from 3 weeks
`of age until mice were sacrificed. (n R 10 for chow fed or NR fed.) Yellow dotted circles depict early tumors and black arrows point mitotic bodies.
`(F) Tumor burden of 60-week-old mice described in (E).
`(G) Serum ALT levels of 60-week-old mice described in (E).
`(H) WB analysis of hURI-tetOFFhep mice expressing hURI for 8 weeks and switched OFF for 24 weeks.
`(I) gH2AX IHC stained liver sections from 32-week-old mutant hURI-tetOFFhep mice fed with either chow or Dox diets. (n = 5). Red arrows point to DNA damage foci.
`Data represented as mean ± SEM (*p % 0.05 and ***p % 0.001). See also Figure S5.
`
`of cleaved caspase 3 (Figure S5H), suggesting that boosting
`NAD+ levels may be cytotoxic for tumor cells.
`Furthermore, ceasing hURI expression in 8-week-old mice for
`24 weeks restored AFMID levels, suppressed DNA damage,
`abolished the DDR, and reduced acetylation of p53 at Lys-379,
`possibly due to activated NAD+-dependent SIRT1 (Luo et al.,
`2001) (