NF[kappa]B1 is a suppressor of neutrophil-driven hepatocellular carcinoma

Abstract

Hepatocellular carcinoma (HCC) develops on the background of chronic hepatitis. Leukocytes found within the HCC microenvironment are implicated as regulators of tumour growth. We show that diethylnitrosamine (DEN)-induced murine HCC is attenuated by antibody-mediated depletion of hepatic neutrophils, the latter stimulating hepatocellular ROS and telomere DNA damage. We additionally report a previously unappreciated tumour suppressor function for hepatocellular nfkb1 operating via p50:p50 dimers and the co-repressor HDAC1. These anti-inflammatory proteins combine to transcriptionally repress hepatic expression of a S100A8/9, CXCL1 and CXCL2 neutrophil chemokine network. Loss of nfkb1 promotes ageing-associated chronic liver disease (CLD), characterized by steatosis, neutrophillia, fibrosis, hepatocyte telomere damage and HCC. Nfkb1^S340A/S340A mice carrying a mutation designed to selectively disrupt p50:p50:HDAC1 complexes are more susceptible to HCC; by contrast, mice lacking S100A9 express reduced neutrophil chemokines and are protected from HCC. Inhibiting neutrophil accumulation in CLD or targeting their tumour-promoting activities may offer therapeutic opportunities in HCC.

ARTICLE

Received 24 Oct 2014 | Accepted 2 Mar 2015 | Published 16 Apr 2015

C.L. Wilson1, D. Jurk2, N. Fullard1, P. Banks1, A. Page1, S. Luli1, A.M. Elsharkawy3, R.G. Gieling4,J. Bagchi Chakraborty5, C. Fox1, C. Richardson6, K. Callaghan1, G.E. Blair7, N. Fox7, A. Lagnado2, J.F. Passos2, A.J. Moore8, G.R. Smith1, D.G. Tiniakos1, J. Mann1, F. Oakley1,* & D.A. Mann1,*

Hepatocellular carcinoma (HCC) develops on the background of chronic hepatitis. Leukocytes found within the HCC microenvironment are implicated as regulators of tumour growth. We show that diethylnitrosamine (DEN)-induced murine HCC is attenuated by antibody-mediated depletion of hepatic neutrophils, the latter stimulating hepatocellular ROS and telomere DNA damage. We additionally report a previously unappreciated tumour suppressor function for hepatocellular nfkb1 operating via p50:p50 dimers and the co-repressor HDAC1. These anti-inammatory proteins combine to transcriptionally repress hepatic expression of a S100A8/9, CXCL1 and CXCL2 neutrophil chemokine network. Loss of nfkb1 promotes ageing-associated chronic liver disease (CLD), characterized by steatosis, neutrophillia, brosis, hepatocyte telomere damage and HCC. Nfkb1S340A/S340Amice carrying a mutation designed

to selectively disrupt p50:p50:HDAC1 complexes are more susceptible to HCC; by contrast, mice lacking S100A9 express reduced neutrophil chemokines and are protected from HCC. Inhibiting neutrophil accumulation in CLD or targeting their tumour-promoting activities may offer therapeutic opportunities in HCC.

1 Fibrosis Research Group, Institute of Cellular Medicine, Newcastle University, Newcastle upon Tyne NE2 4HH, UK. 2 Newcastle University Institute for Ageing and Institute for Cell and Molecular Biosciences, Newcastle University, Newcastle Upon Tyne NE4 5PL, UK. 3 Liver Unit, University Hospitals Birmingham, Birmingham B15 2TH, UK. 4 Hypoxia and Therapeutics Group, Manchester Pharmacy School, University of Manchester, Manchester M13 9PT, UK. 5 Department of Medicine, Immunology and Inammation, Imperial College of Science, Technology and Medicine, Hammersmith Hospital, London W12 0NN, UK. 6 Centre for Behaviour and Evolution/Institute of Neuroscience, Medical School, Newcastle University, Newcastle Upon Tyne NE2 4HH, UK.

7 Faculty of Biological Sciences, School of Molecular and Cellular Biology, University of Leeds, Garstang Building, Leeds LS2 9JT, UK. 8 Institute for Cell and Molecular Biosciences, Newcastle University, Catherine Cookson Building, Framlington Place, Newcastle Upon Tyne NE2 4HH, UK. * These authors contributed equally to this work. Correspondence and requests for materials should be addressed to F.O. (email: mailto:[email protected]

Web End [email protected] ) or to D.A.M. (email: mailto:[email protected]

Web End [email protected] ).

NATURE COMMUNICATIONS | 6:6818 | DOI: 10.1038/ncomms7818 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 1

& 2015 Macmillan Publishers Limited. All rights reserved.

DOI: 10.1038/ncomms7818 OPEN

NFkB1 is a suppressor of neutrophil-driven hepatocellular carcinoma

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms7818

The mechanisms driving the progression of chronic inammatory liver diseases such as alcoholic steatohepatitis and non-alcoholic steatohepatitis (ASH and NASH) to

hepatocellular carcinoma (HCC) are poorly understood. The majority of HCC develops on the background of cirrhosis1, which is the end-stage result of brogenesis, a process of excessive wound repair. This has led to the concept that the brotic extracellular matrix and/or pro-brogenic myobroblasts may be stimulators of HCC2. However, the increasing awareness that non-cirrhotic HCC represents a high proportion of liver cancers argues for additional mechanisms inherent in the inammatory process. Inammation underpins roughly 20% of all solid tumours and experimental studies have provided mechanistic links between inammatory mediators such as interleukin (IL)-6 and carcinogenesis3. The histopathological changes associated with chronic ASH and NASH include persistent tissue inltration by neutrophils and lymphocytes4. Appearance of neutrophils in the hepatic parenchyma is normally a transient process triggered by damage-induced expression of soluble and insoluble neutrophil guidance cues and is subsequently terminated by neutrophil death and/or clearance by Kupffer cells5,6. This transitory response is important for resolution of inammation and for protecting epithelial cells from bystander damage resulting from exposure to neutrophil-derived reactive oxygen species (ROS) and proteases. Persistence of neutrophils in chronic liver disease (CLD) is therefore a pathological feature; however, it remains unclear as to how neutrophils contribute to disease progression. Experimental work in rodent models argues against a role for neutrophils in wound repair or brogenesis7,8. As neutrophils are present in human HCC and its surrounding tissue (Supplementary Fig. 1a), we were interested to determine whether they functionally contribute to tumour development. Here we demonstrate a requirement of hepatic neutrophils for the development of carcinogen-induced HCC and dene a protumour mechanism in which neutrophils cause telomere DNA damage in bystander hepatocytes. We additionally dene a tumour suppressor role for the nfkb1 gene, which through the expression of anti-inammatory p50:p50 nuclear factor-kB (NF-kB) dimers is able to transcriptionally repress a neutrophil chemokine network.

ResultsNeutrophils are required for experimentally induced HCC. Diethylnitrosamine (DEN) acts as a complete liver carcinogen when administered to mice 15 days after birth9. However, a role for inammatory mechanisms in this model is demonstrated by the requirement for NF-kB activation in myeloid cells and for the production of IL-6 by Kupffer cells, for injury to progress to HCC10,11. To begin to investigate a role for neutrophils in this process, we examined hepatic neutrophils across a 5- to 40-week time course of DEN-induced disease in male C57BL/6 mice (Fig. 1a). Elevated numbers of hepatic neutrophils were evident throughout the disease process and underwent a steady increase until 40 weeks where they were accompanied by enlarged livers, substantial tumour frequency (Fig. 1b), large tumours (Supplementary Fig. 1b) and high numbers of PCNA

proliferating hepatocytes (PCNA ) (Fig. 1c). To determine

whether neutrophils contribute to DEN-induced HCC, mice were treated between weeks 32 and 40 with an antibody specic for the mouse neutrophil antigen Ly6G, known to effectively suppress neutrophil recruitment to tissues12. We conrmed that anti-Ly6G suppressed DEN-induced hepatic neutrophil accumulation with little effect on F4/80 macrophages or CD3 lymphocytes

(Supplementary Fig. 1c). Selective depletion of neutrophils was associated with a 3.5-fold reduction in tumour burden at 40 weeks and a slight reduction in liver/body weight ratio (Fig. 1d).

Mice lacking nfkb1 (encoding the NF-kB p50 subunit) display exaggerated neutrophil inammatory responses to injury in multiple organs, including the liver1315. Early (5 weeks) and persistent elevation of hepatic neutrophils was observed in DEN-injured nfkb1 / mice (Fig. 1a). Furthermore, these animals exhibited rapid progression to aggressive HCC (Fig. 1b). Of note, Cyp2E1 expression (Supplementary Fig. 1d) and levels of liver injury and apoptosis (Supplementary Table 1 and Supplementary Fig. 1e) were similar between the two genotypes, ruling out the possibility that differences in DEN-induced liver injury caused the increased tumour burden. Dysplastic cells and small tumours were evident at 20 weeks (Supplementary Fig. 2a) in nfkb1 / mice, and by 30 weeks there was an average of 32 visible surface liver tumour foci compared with just 3 for wild type (WT; Fig. 1b). By week 40, an average of 73 and 12 surface growths were recorded for nfkb1 / and WT, respectively. Accelerated disease in nfkb1 / liver was conrmed by increased cell proliferation (Fig. 1c), elevated hepatic expression of cyclin D1 (Supplementary

Fig. 2b) and loss of phospho-p38a (Supplementary Fig. 2c), the latter being a feature of advanced human HCCs16. C57BL/6 mice have a characteristically low rate (o4%) for the development of spontaneous liver cancer. We were therefore surprised to observe spontaneous CLD, a nodular appearance of the liver surface, emergence of dysplastic nodules, an increase in liver/body weight ratio and tumour frequency at 20 months in nfkb1 / males (Fig. 1e). Histological assessment of the livers revealed that nfkb1 / mice develop a spectrum of features of CLD, including steatohepatitis, similar to that in humans, lobular mixed inammation and hepatocyte ballooning occasionally with MalloryDenk inclusions or megamitochondria, sinusoidal brosis and bridging brosis in 50% of the mice (Fig. 1e and Supplementary Fig. 2d). In ve out of ten mice, hepatocellular adenomas and/or well-to-moderately differentiated HCC were observed (Supplementary Fig. 2e,f) in non-cirrhotic liver. In two cases, portal biliary lesions resembling biliary microhamartomas with no evidence of epithelial dysplasia were noted (Supplementary Fig. 2f). Conversely, old WT mice develop mild steatosis and some focal inammation (Supplementary Tables 2 and 3). The spontaneous hepatic lesions in aged nfkb1 / mice were associated with highly elevated numbers of hepatic neutrophils (Supplementary Fig. 2g). To determine a functional contribution of neutrophils in the nfkb1 / background, DEN-injured mice were treated with either the neutrophil-depleting antibody anti-Ly6G or control IgG between weeks 22 and 30 (this being the period of progression from dysplasia to HCC). Anti-Ly6G treatment effectively reduced numbers of hepatic neutrophils, normalized the liver/body weight ratio (Supplementary Fig. 2h and Fig 1f) and blunted tumour development (Fig. 1f) without affecting macrophage or T-lymphocyte recruitment (Supplementary Fig. 2h). Taken together, these data implicate neutrophils as important functional contributors to inammation-driven HCC.

Accelerated HCC in the nfkb1 / mouse. Cancer models in extra-hepatic organs have suggested pro-tumour functions for neutrophils, and that inhibition of neutrophil recruitment or activities may be therapeutic options in the context of chronic inammatory disease1721. Interrogating the mechanisms responsible for enhanced neutrophil inammation and aggressive HCC phenotype of the nfkb1 / mouse will therefore provide valuable mechanistic insights. Initial investigations ruled out intrinsic differences in neutrophil functions such as apoptosis or production of ROS (Supplementary Fig. 3a,b). We therefore reasoned that either nfkb1 / hepatocytes have a deciency in

2 NATURE COMMUNICATIONS | 6:6818 | DOI: 10.1038/ncomms7818 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2015 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms7818 ARTICLE

WT

nfkb1/

WT

*** ***

Ave NIMP1+cells 15HPF

150 90 0.120.10

0.08

0.06

0.04

0.02

75 60 45 30 15

WT nfkb1/

WT

nfkb1/

Ave number of tumors

Liver/body weight ratio

120

90

*** *** ***

***

***

***

nfkb1/

**

***

5 wk 10 wk 15 wk 20 wk 30 wk 40 wk

5 wk

20-month-old livers Steatohepatitis Sinusoidal fibrosis Sirius red/fast green

HCC

0 30 wk 40 wk

0 30 wk

wk 40 harvest

WT DEN +/ Ly6G Ab plan

Ave PCNA+ cells 15 HPF

60

30

0

350 300 250 200 150 100

50

0

Day 15DEN IP wk 32 wk 36

Liver/body weight ratio

0.060.050.040.030.020.01 0

Birth

**

IgG control/Ly6G Ab mini-pump

0.07

**

10 wk 15 wk 20 wk 30 wk 40 wk

40 wk

Ly6G

Control lgG21.09 +/ 4.31

Ly6G**6.60 +/ 2.37

IgG WT

WT

H&E

*

0.07 **

Liver/body weight ratio

0.060.050.040.030.020.01 0

Dysplastic focus

H&E

0.08 1.8

1.61.41.21.00.80.60.40.2 0

Ave tumor frequency

per biopsy

nfkb1/

H&E

WT nfkb1/

WT nfkb1/

nfkb1/ DEN +/ Ly6G Ab plan

Day 15DEN IP wk 22 wk 26

wk 30 harvest

Liver/body weight ratio

0.070.060.050.040.030.020.01 0

Birth

IgG control/Ly6G Ab mini-pump

Control lgG

41.0 +/ 9.3

Ly6G*

15.3 +/ 3.6

IgG Ly6G

nfkb1/

Figure 1 | Neutrophils are required for experimentally induced HCC. (a) Graph shows average total number of hepatic neutrophils ( NIMP-R14 ) in DEN-injured WT or nfkb1 / mice. Red line denotes basal neutrophil levels in normal WT liver. (b) Average tumour counts and liver/body weight ratio in 30- and 40-week DEN-injured WT and nfkb1 / mice. Representative pictures of livers from 40-week DEN-injured WT and nfkb1 / mice.

(c) Average total number of PCNA hepatocytes in DEN-injured WT and nfkb1 / mice. (d) Diagram showing the experimental plan to deplete

neutrophils (Ly6G antibody) in WT mice from week 3240 after DEN injury. Average tumour counts, representative pictures of livers and average liver/body weight ratio graph in 40-week DEN-injured WT mice treated with control IgG or Ly6G antibody (Ab) for 8 weeks. n 10 (e) Representative

pictures of liver from 20 month (aged under normal conditions) WT and nfkb1 / mice. Representative photomicrographs of haematoxylin and eosin (H&E)- and Sirius red/fast green-stained liver sections from 20-month nfkb1 / mice (upper panel) revealed steatohepatitis (fat, blue arrows;

inammation, yellow arrows; ballooned hepatocytes with MalloryDenk bodies, green arrows) and brosis (black arrows). Focal dysplasia (yellow dotted line) with focal inammation (black arrow) and HCC in H&E-stained nfkb1 / aged livers (lower panel). Graphs show average liver/body weight ratio and tumour frequency identied histologically using H&E sections from 20-month, aged WT and nfkb1 / mice. n 7 WT and ten nfkb1 / mice.

(f) Diagram showing the neutrophil depletion experimental protocol in nfkb1 / mice from week 2230 after DEN injury. Average tumour counts, representative pictures of livers and graph showing average liver/body weight ratio from 30-week DEN-injured nfkb1 / mice treated with control IgG or

Ly6G for 8 weeks (n 10). All data are meanss.e.m; scale bars, 200 mm. For a and c, n 6, 4, 6, 7, 15 and 9 (nfkb1 / ), and 4, 4, 4, 5, 11 and 14 (WT)

for the 5- to 40-week time points). Statistical signicance was determined using one-way analysis of variance with Tukeys post-hoc test (a,c) or an unpaired t-test (b,e), *Po0.05, **Po0.01 or ***Po0.001 compared with control.

NATURE COMMUNICATIONS | 6:6818 | DOI: 10.1038/ncomms7818 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 3

& 2015 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms7818

their control of neutrophil recruitment, or/and nfkb1 / neutrophils have an enhanced ability to trafc to the liver. To test the former idea, WT neutrophils were labelled ex-vivo with Cellvue NR185 tracker before transfer into the circulation of DEN-injured WT and nfkb1 / mice (Fig. 2a). Mice were then imaged 4-h post transfusion by in vivo imaging system (IVIS) for hepatic accumulation of labelled neutrophils, which were discovered to be more abundant in nfkb1 / compared with

WT liver (Fig. 2b). This difference was also observed ex-vivo in isolated organs (Fig. 2c) and, of note, labelled neutrophils were restricted to the liver and the spleen. Carrying out the reverse experiment, labelled WT or nfkb1 / neutrophils

were transferred into DEN-injured WT animals and hepatic accumulation determined by IVIS. No differences were observed between the genotypes, suggesting that the absence of nfkb1 /

within the neutrophil does not have an intrinsic effect on its migratory ability (Supplementary Fig. 3c). To conrm these data, we produced chimeric mice in which either nfkb1 / bone marrow was reconstituted into an irradiated WT background (nfkb1 / BM4WT) or WT bone marrow was reconstituted in the nfkb1 / background (WTBM4nfkb1 / ). In response to acute DEN injury, neutrophils were recruited to the liver in both backgrounds, but of note higher numbers of hepatic neutrophils were present in WTBM4nfkb1 / mice (Supplementary

WT

IVIS image animals

IVIS image dissected organs

WT>WT

WT>nfkb1/

t=4 h

t=4 h

IV inject t=0 h

Isolate neutophils

t= 18 h

Far red dye labelled cells

Liver Kidney Kidney

Spleen Spleen

Liver

14

E+06 average radians

in animal imaging

*

IV inject t=0 h

12 10

8 6 4 2 0

nfkb1/

WT nfkb1/

WT

nfkb1/

E+06 average radians

ex vivoimaging

20

15

10

5

25 3 5

4

3

2

1

0

7

** ***

**

*

RLTD fold WT

acute DEN livers

2

1

350

6

3 2 1 0

4

3

2

1

0

5

4

0 WT nfkb1/

0 CXCL1 CXCL2 S100A9 TNF

RLTD fold nfkb/ hepatocytes

40 20

16

12

8

4

0

**

1.801.60

1.40 **

***

**

Ave NIMP1+ cells

15 HPF, acute DEN

WT

nfkb1/

1.201.000.800.600.400.20 0

RLTD fold WT

in hepatocytes

30

20

10

0

300 250 200 150 100

50

0 Control IgG

Heps CXCL1 CXCL2

CXCL1+2 IgG

BMM

Day 15 DEN IP

Birth

WT

17.2 +/ 3.2

wk 40 Harvest

0.12

1.5

1.5

CXCL1 CXCL2

Liver/body weight ratio

0.100.080.060.040.02 0

RLTD fold WT

40 wk DEN livers

*

*

1.0

1.0

0.5

0.0 0.0

0.5

s100a9 /6.7 +/ 0.8

WT

s100a9 /

Figure 2 | Dysregulation of hepatic chemokine expression and accelerated HCC in the nfkb1 / mouse. (a) Diagram of the experimental plan using whole-body imaging to track WT neutrophils to the liver of acute DEN-injured WT or nfkb1 / mice. (b) Representative IVIS pictures of mice given

NIR815-labelled WT neutrophils intravenously, showing neutrophils tracking to livers of acute DEN-injured WT or nfkb1 / mice. Graph shows average radians from IVIS-imaged WT or nfkb1 / mice. (c) Representative ex vivo images of the liver (left), kidney (middle) and the spleen (right), and graph showing average radians from WTor nfkb1 / livers imaged ex vivo. n 3 (d) Hepatic CXCL1, CXCL2, S100A9 and tumour necrosis factor-a (TNFa) mRNA

levels expressed as relative level of transcription difference (RLTD) compared with WT 48 h post acute DEN in WT and nfkb1 / mice, n 6. (e) Graph

shows RLTD of S100A9 in bone marrow macrophages (BMM) compared with hepatocytes isolated from WT and nfkb1 / mice, n 3. (f) Hepatocyte

CXCL1 and CXCL2 mRNA levels expressed as RLTD in nfkb1 / mice compared with WT, n 3. (g) Graph shows average total number of NIMP-R14 cells

in liver sections from WTor nfkb1 / mice after acute DEN injury-treatedIgG or CXCL1 and two neutralizing antibody, n 5. (h) Average tumour counts,

representative pictures of livers and graph showing average liver/body weight ratio from 40-week DEN-injured WT and s100a9 / mice, n 19. Hepatic

CXCL1 and CXCL2 mRNA levels expressed as RLTD in 40-week, chronic, DEN-injured WT and s100a9 / mice, n 6. Data are meanss.e.m. Statistical

signicance was determined using an unpaired t-test, *Po0.05, **Po0.01 or ***Po0.001 compared with control.

4 NATURE COMMUNICATIONS | 6:6818 | DOI: 10.1038/ncomms7818 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2015 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms7818 ARTICLE

Fig. 3d). Of note, both the nfkb1 / BM4WT and WTBM4nfkb1 / animals were considered too sick for use in longer-term studies that would have conrmed the hepatocytespecic actions of nfkb1 / in HCC; this phenotype may be related to our recent report that nfkb1 / cells are highly susceptible to radiation-induced senescence22. We conclude that enhanced neutrophil recruitment in nfkb1 / mice most probably reects a defect in the hepatic regulation of neutrophil trafcking. We have recently reported that neutrophil migration to the injured liver is critically dependent on a hepatic neutrophil chemokine network comprised of calprotectin (S100A8/S100A9), CXCL1 (Groa/KC) and CXCL2 (Grob/MIP2), and as the expression of these proteins is under the transcriptional control of NF-kB, it was relevant to determine their expression in nfkb1 / liver7. CXCL2, S100A9 and tumour necrosis factor-a were all signicantly overexpressed in DEN-injured nfkb1 /

liver (Fig. 2d and Supplementary Table 4). Modest increases for CXCL1, S100A8 and IL-6 were also observed but failed to reach statistical signicance (Fig. 2d and Supplementary Fig. 4a). CXCL1, CXCL2, S100A9, IL-6 and tumour necrosis factor-a were also highly upregulated in the diseased livers of 3-, 9-, 12-, 15-and 20-month nfkb1 / males (Supplementary Fig. 4b and

Supplementary Table 4). At the protein level, S100A9 overexpression was located to mononuclear cells and hepatocytes, and particularly strong S100A9 staining was found within nfkb1 / tumour tissue (Supplementary Fig. 4c).

Analysis of S100A9 messenger RNA expression in isolated macrophages demonstrated no difference between genotypes; by contrast, S100A9 was not expressed in WT hepatocytes but was detected in nfkb1 / hepatocytes (Fig. 2e). Expression of the hepatocyte markers albumin and HNF4a, and macrophage markers F4/80 and CD68 veried the purity of the cell isolations (Supplementary Fig. 4d). CXCL1 and CXCL2 transcripts were highly overexpressed in cultured primary nfkb1 / versus WT hepatocytes (Fig. 2f), this being conrmed by enzyme-linked immunosorbent assay (ELISA) for

CXCL2 (Supplementary Fig. 4e). Furthermore, number of S100A9 cells and hepatic expression of S100A9 mRNA were

higher in acute DEN-injured WTBM4nfkb1 / compared with nfkb1 / BM4WT chimeric mice (Supplementary Fig. 3d).

These observations conrm a previous report that calprotectin expression is induced in damaged epithelia and HCC cells23. To formally establish a requirement of CXCL1 and CXCL2 for migration of neutrophils to DEN-injured liver, we neutralized the

8

8

8

7

***

***

Fold enrichment lgG

NFB1 ChlP

Fold enrichment lgG

HDAC1 ChlP

*** ***

7

7

50 WT

nfkb1/

**

CXCL1, CXCL2 and S100A9 promoter

Repressed transcription

6

6

6

5

5

5

40

30

**

4

4

4

3

3

3

in a and b

2

2

2

HDAC1

NFB1 NFB1

NFB1 binding site

1

1

20

10

1 0 CXCL1 promoter

Strongylocentrotus purpuratus

Xenopus laevis

Mus musculus

Gallus gallus

Homo sapiens

S340 S340

CXCL2 promoter

0 S100A9 promoter

0 CXCL1 promoter

CXCL2 promoter

0 S100A9 promoter

NFB1

NFB2

RelA

cRel

ReIB

+RelAeGFP

Fp50S337A

Fp50S342A

Control

Fp50

HAp50

Fp50

Fp50T145A

+HAp50

Fp50S210A

Fp50S315A

Fp50S337A

Fp50S342A

Fp50

Fp50T145A

Fp50S210A

Fp50S315A

IP flag Blot RelA

Fp50

IP flag Blot HA

IP flag Blot flag

Total HA

IP flag Blot flag

Total RelAeGFP Endogen. RelA

Figure 3 | p50 homodimers complexed with HDAC1 regulate the chemokines CXCL1 and 2. (a,b) ChIP assay analysis of p50 (a) and HDAC1 (b) recruitment to the CXCL1, CXCL2 and S100A9 promoters in acute DEN-injured WT or nfkb1 / livers. (c) Model showing repression of the CXCL1,

CXCL2 and S100A9 promoters by p50:p50 homodimers complexed with HDAC1. (d) Sequence alignment of nfkb1 gene shows conservation of amino acid serine 340 (in mouse) throughout evolution from purple sea urchin through to human. The conserved serine is shown in bold and red. (e) Sequence alignment of the mouse NF-kB subunits, show conservation of serine 340 (in mouse) throughout all subunits (shown in bold and red) and conservation of serine 337 in NF-kB1, RelA and cRel (shown in bold and blue). (f) X-ray crystal structure of the mouse p50:p50 homodimer (cyan) bound to DNA (deep blue), PDB entry 1NFK. White arrows show serine 340, which is rendered in space-lling representation. (g) Representative western blots of anti-Flag or anti-RelA after Flag IP on lysates from Cos-7 cells transfected with RelA-eGFPFlag-p50 or Flag-p50 mutant shows that p50 mutants retain interaction with RelA. (h) Western blots for anti-HA or anti-Flag after Flag IP on lysates from Cos-7 cells transfected with Flag-p50 or Flag-p50 mutantHA-p50 reveals that p50 mutants bind p50, except p50 S342A (human equivalent of mouse 340). Data are meanss.e.m. and representative of ve mice per group. Statistical signicance was determined using an unpaired t-test, **Po0.01 or ***Po0.001 compared with control.

NATURE COMMUNICATIONS | 6:6818 | DOI: 10.1038/ncomms7818 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 5

& 2015 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms7818

chemokines by an antibody approach in the context of acute hepatic injury. Antibody-mediated combined inhibition of CXCL1 and CXCL2 abrogated DEN-induced accumulation of neutrophils in nfkb1 / mice (Fig. 2g). Unfortunately, this approach was not practical for determining the role of CXCL1 and CXCL2 in HCC due to the need for longer-term administration of high cost antibodies. However, calprotectin regulates the hepatic expression of both CXCL1 and CXCL2. s100a9 / mice express reduced levels of both chemokines upon liver injury7 and transgenic expression of S100A8 and S100A9 stimulates overproduction of CXCL1 and mobilization of neutrophils from the bone marrow24. Furthermore, neutrophil recruitment is reduced in s100a9 / mice in acute DEN injury (Supplementary Fig. 4f). We therefore determined the susceptibility of s100a9 / to DEN-induced HCC and discovered that these animals are protected compared with WT.

Reduced tumour burden in these animals was associated with reduced hepatic expression of CXCL1 and CXCL2 (Fig. 2h). We conclude that hepatocellular nfkb1 negatively regulates a neutrophil chemokine network comprising calprotectin, CXCL1 and CXCL2. In support of this network having a pro-tumour function, inhibition of CXCL1 and CXCL2 receptor CXCR2 is sufcient to suppress inammation-driven skin and intestinal cancers19. Furthermore, s100a9 / mice are protected against cancers of inammatory origin, including breast, pancreatic and colitis-associated tumours2527, although loss of S100A9 is associated with susceptibility to experimental skin cancer28. To further dene the relative contributions of these chemokines to the pathobiology of HCC, it will be necessary to generate conditional genetic deletions enabling temporal control of their expression targeted selectively to hepatocytes.

Disruption of p50 homodimers increases susceptibility to HCC. Previous studies propose a model in which the anti-inammatory properties of nfkb1 are mediated by p50:p50 homodimers2933. Importantly, p50 lacks inherent transcriptional activity and in homodimer form associates with co-repressors such as HDAC1 or Bcl3, to actively repress inammatory gene transcription29,32,34,35. Of note, the nuclei of unstimulated cells contains a signicant number of p50:p50 dimers that are assumed to repress transcription36. Chromatin immunoprecipitation (ChIP) assays conrmed that p50 and HDAC1 are recruited to the cxcl1, cxcl2 and s100a9 promoters (Fig. 3a,b). In contrast, HDAC1 did not associate with these promoters in nfkb1 /

liver (Fig. 3b), supporting a model in which p50:p50 dimers recruit repressive HDAC1 (or/and Bcl3) to suppress hepatic expression of neutrophil chemokines (Fig. 3c). The mechanisms that control NF-kB dimer assembly and, in particular, the choice of heterodimers versus homodimers are very poorly understood, yet will dictate the formation of anti-inammatory p50:p50:HDAC1 versus pro-inammatory RelA:p50 complexes. Sequence analysis of p50 protein reveals a region spanning residues 325 to 354 (human p50) displaying high conservation across evolutionarily distant species separated by 743 million years (Fig. 3d). Aligning this region with equivalent sequences in the mammalian NF-kB subunits revealed perfect conservation of two serine residues, Ser337 and Ser342 (Fig. 3e). We noted in the structure of mouse p50:p50 dimers bound to DNA that the equivalent residue to human Ser342 (mouse Ser340) is located on the opposing face from the dimerization interface where it might exert conformational inuences on dimer assembly (Fig. 3f). Alanine mutations were introduced at Ser337, Ser342 and at other evolutionary conserved residues (Thr145, Ser210 and Ser315) in the context of human FLAG-p50. WT and mutant Flag-p50 proteins were co-expressed together with HA-p50 or EGFP-RelA,

to determine their relative abilities to assemble homodimers or heterodimers, respectively. Pull-down assays demonstrated that all mutant p50 constructs interacted with RelA to a similar degree as WT-p50. Of note, the Ala342 mutant protein was consistently expressed at reduced levels compared with WT-p50 and other mutants (Fig. 3g). Homodimer interactions were similar between WT-p50 and the 145, 210, 315 and 337 alanine mutants (Fig. 3h). In contrast, the Ala342 mutant disabled p50:p50 assembly (Fig. 3h). These mutagenesis experiments therefore reveal the Ser342 residue to be a specic and critical determinant of p50 homodimer assembly. In further experiments, FLAG-Supershift EMSA experiments conrmed that FLAG-p50-Ala342 fails to bind to kB motifs as would be expected, as p50 monomers are unable to bind DNA (Supplementary Fig. 5a). The effects of the mutation on chemokine gene expression were examined by adenoviral overexpression of WT and mutant p50 proteins in mouse hepatocytes (Supplementary Fig. 5b). As anticipated, elevated levels of WT-p50 suppressed expression of CXCL1, but this effect was not observed in cells expressing the mutant homodimerization-defective p50 (Supplementary Fig. 5c). This result conrms that p50:p50 dimers are repressors of neutrophil chemokine expression.

We next generated a genetically modied mouse carrying a single alanine (Ala) mutation at Ser340 (Supplementary Fig. 6). Homozygote nfkb1S340A/S340A mice were viable and developed to adulthood without obvious defects or gross pathologies. Mutant p105S340A and p50S340A proteins were expressed in all major organs but at reduced levels relative to WT (Supplementary Fig. 7a), although transcript levels were similar between WT and nfkb1S340A/S340A liver tissues, indicating that loss of the serine residue may reduce p105/p50 translation or stability (Supplementary Fig. 7b). Of note, we did not detect signicantly altered expression of other NF-kB subunits. Furthermore, although total tissue levels of p50S340A were reduced, p50 and RelA DNA binding were induced to a similar degree by DEN injury at kB target sequences when comparing WT and nfkb1S340A/S340A (Supplementary Fig. 7c). To assess the impact of the mutation on neutrophil recruitment, adult nfkb1S340A/S340A mice were injured acutely with DEN. Highly augmented acute neutrophil inammation was observed in the nfkb1S340A/S340A background relative to WT controls (Fig. 4a). Furthermore, hepatic CXCL1, CXCL2 and S100A9 were overexpressed when compared with WT (Fig. 4b and Supplementary Fig. 7d). To determine the susceptibility of nfkb1S340A/S340A mice to DEN-induced HCC, a 40-week disease model was carried out alongside WT controls. nfkb1S340A/S340A males developed signicantly more tumours than WT (Fig. 4c). This accelerated disease phenotype was conrmed by increased liver weight, elevated numbers of PCNA proliferating hepatocytes

(Fig. 4c,d) and substantially higher numbers of hepatic neutrophils (Fig. 4e). We have therefore identied a single amino acid (Ser342 human or Ser340 mouse) that on mutation enhances injury-induced expression of neutrophil chemokines and increases susceptibility to HCC.

Neutrophils promote hepatocellular telomere damage. A remaining question was the nature of the neutrophil-mediated pro-tumour mechanism that is opposed by hepatocellular nfkb1. This is likely to be multifactorial given the vast array of cytotoxic molecules, inammatory mediators and mitogenic factors released by neutrophils37. However, when examining human HCC tissue we were intrigued by the juxtaposition of neutrophil-rich inammatory foci with hepatocytes positive for 8-hydroxyguanosine, the latter being a biomarker for oxidative DNA damage (Fig. 5a). In addition, immuno-uorescent in situ

6 NATURE COMMUNICATIONS | 6:6818 | DOI: 10.1038/ncomms7818 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2015 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms7818 ARTICLE

Ave number of NIMP1+ cells per 15 HPF, acute DEN

400 350 300 250 200 150 100

50

0 NIMP-R14 liver cells

WT

*** *

wk 40 harvest

RLTD fold WT

acute DEN livers

4

2

1

RLTD fold WT

acute DEN livers

3

2

1

0

3

RLTD fold WT

acute DEN livers

3

2

1

WT

nfkb1S340A

nfkb1S340A

In all panels

0 CXCL1 CXCL2

0 S100A9

Day 15 DEN IP

Birth *

0.12

0.10

0.08

0.06

0.04

0.02

0 0

WT

WT

Liver/body weight ratio

Average number of

PCNA + cells per 15HPF

120

100

80

60

40

20

***

Average number of

NIMP1+ cells per 15HPF

250

200

250

200

50

nfkb1S340A

nfkb1S340A

WT21.1 +/ 5.93

nfkb1S340A***

58.0 +/ 8.89

WT

nfkb1S340A

PCNA+ liver cells

0 NIMP-R14 liver cells

Figure 4 | nfkb1S340A knock-in mice have increased neutrophils and tumour burden in experimentally induced HCC. (a) Representative photo-micrographs of NIMP-R14 staining and graph showing average total number of neutrophils in liver sections from WT or nfkb1S340A mice following acute

DEN injury. Black arrows denote NIMP-R14 stained cells, n 5 (b) Hepatic CXCL1, CXCL2 and S100A9 mRNA 48 h post acute DEN in WT and nfkb1S340A

mice. n 5 (c) Representative pictures of livers, average tumour counts and liver/body weight ratio in 40-week DEN-injured WT and nfkb1S340A mice.

(d) Graphs show average total number of PCNA stained cells and representative photomicrographs at 200 magnication of PCNA staining in

WT and nfkb1S340A 40-week DEN-injured livers, black arrows denote PCNA stained cells, n 6 (e) Graph shows NIMP-R14 cells in liver sections from

40-week DEN-injured WTand nfkb1S340A mice, n 6. Representative photomicrographs at 200 magnication of NIMP-R14 staining in WTand nfkb1S340A

40-week DEN-injured livers, black arrows denote NIMP-R14 stained cells. Data are meanss.e.m.; all scale bars, 100 mm. Statistical signicance was determined using an unpaired t-test, *Po0.05 or ***Po0.001 compared with control.

hybridization (FISH) using antibody against DNA damage marker gH2A.X and FISH for the telomere-specic (C3TA2)3

PNA probe revealed the presence of tight clusters of hepatocytes with telomere-associated DNA damage in human alcoholic liver disease (Fig. 5b and Supplementary Fig. 8a). By contrast, we detected negligible telomere-associated foci (TAF ) hepatocytes

in normal human liver (Fig. 5b). DNA damage in telomeres is slow to repair38,39 and these lesions contribute to loss of telomere integrity, chromosome instability and cancer40,41. This latter concept was examined experimentally in the context of HCC by Begus-Nahrmann et al.42 who reported that transient telomere dysfunction in the liver is sufcient alone to promote chromosome instability and hepatocarcinogenesis. We therefore employed immuno-FISH to examine telomere DNA damage in DEN-injured mouse livers. A higher percentage of TAF

hepatocytes were found at 30 weeks in nfkb1 / livers compared with WT (Fig. 5c). Comparison of TAF between WT and nfkb1 / livers at early time points revealed no differences in numbers of TAF hepatocytes at 5 weeks post-DEN injury

(Supplementary Fig. 8b); however, elevated numbers of TAF

cells were detected in nfkb1 / livers before tumour growth at 15 and 20 weeks post DEN (Supplementary Fig. 8c).

Furthermore, higher numbers of TAF cells were found in the

livers of 3-month-old nfkb1 / mice compared with age-matched WT mice, again indicating TAFs precede the development of HCC (Supplementary Fig. 8d). Notably, in vivo treatment of DEN-injured nfkb1 / mice with anti-Ly6G resulted in a signicant reduction in TAF hepatocytes

(Fig. 5d) coincident with reduced tumour burden (Fig. 1f). From these data we conclude that the aggressive inammatory reaction that develops between 5 and 15 weeks post DEN in nfkb1 / liver is accompanied by enhanced levels of hepatocellular telomere DNA damage, and that the neutrophilic

component of the inammatory reaction is the cellular mediator of these telomere lesions.

Neutrophils induce oxidative DNA damage in bystander cells by intercellular transfer of ROS43. Intracellular ROS causes lipid peroxidation and generation of trans-4-hydroxy-2-nonenol (4HNE), which following conversion to 2,3-epoxy-4HNE reacts with deoxyadenosine and deoxycytidine resulting in DNA lesions44. 4HNE-positive hepatocytes were consistently found in greater numbers in nfkb1 / livers compared with WT across the 5- to 40-week DEN time course (Fig. 6a). This correlated at most time points with higher levels of hepatocellular DNA damage (gH2A.X cells) (Fig. 6b). In vitro experiments were

designed to test our hypothesis that neutrophils induce bystander ROS and DNA damage in hepatocytes. Primary hepatocytes were co-cultured either in direct contact with neutrophils or were exposed to neutrophil-derived diffusible factors by means of transwell culture (Fig. 6c). Measurement of intracellular ROS within co-cultured hepatocytes revealed that either direct or indirect contact with neutrophils was sufcient to elevate hepatocellular ROS (Fig. 6d). Representative 4,6-diamidino-2-phenylindole (DAPI)/53BP1-stained hepatocytes from these co-cultures were used to quantify the percentage of hepatocytes with DNA damage foci. 53BP1 foci were more abundant in

co-cultures compared with monoculture controls (Fig. 6e and Supplementary Fig. 9a). As in vivo conrmation that neutrophils are contributors to ROS-induced lipid peroxidation, treatment with anti-Ly6G reduced the numbers of 4HNE hepatocytes

(Supplementary Fig. 9b). Further in vivo support for ROS as a mediator of neutrophil-stimulated HCC was demonstrated by therapeutic application of the dietary anti-oxidant butylated hydroxyanisole (BHA), which protected against progression of the aggressive DEN-induced HCC disease in the nfkb1 /

background (Fig. 6f). This protective effect of BHA was

NATURE COMMUNICATIONS | 6:6818 | DOI: 10.1038/ncomms7818 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 7

& 2015 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms7818

Cirrhotic liver

Max projection

Single plane

nfkb1/ 30 wk DEN ALD

FISH H2A.X

Merge

i

ii

80 70 60 50 40 30 20 10

140 i ii

i ii

250

**

120

8-OHG stain

Fluorescence

intensity (AU)

100

200

150

Telomeres H2A.X

80

HCC

60

100

40

20

50

0

0.0 0.4 0.8 1.2

Distance (M)

0 0.0 0.4 0.8 1.2

Distance (M)

0 Normal liver

ALD

i

ii

Merge

Percent TAF + cells

80

70

60

50

40

30

20

10

0 0.0 0.4 0.8 1.2 Distance (M)

Distance (M)

FISH H2A.X

Percent TAF + cells

300 **

0 WT nfkb1/

250

***

250

Fluorescence

intensity (AU)

Percent TAF + cells

80 70 60 50 40 30 20 10

200 150

200 150

Telomeres H2A.X

100

100

50

50

IgG

nfkb1/

Ly6G

0 0.0 0.4 0.8 1.2 1.6 0

1.6

Figure 5 | Neutrophils promote hepatocellular telomere DNA damage. (a) Representative photomicrographs at 200 magnication of 8-OHG staining

in liver sections from cirrhotic livers and HCC, black arrows denote brown positively stained areas of damage in hepatocytes and red arrows show inammatory cells. Scale bars, 100 mm . (b) Representative deconvolved maximum intensity projections of telomere FISH and phospho-H2A.X (gH2A.X)

staining in alcoholic liver disease (ALD) liver sections (n 2 normal human liver and n 4 ALD). Graph shows per cent TAF hepatocytes in ALD liver

compared with normal control liver. (b,c) Graphs (iii) show the uorescence intensity and co-localization of telomere FISH and gH2A.X staining of the corresponding single plane images. Scale bars, 3 mm . (c) Representative immuno-FISH images (deconvolved maximum intensity projections) from 30-week DEN-injured nfkb1 / livers, n 5. Graph per cent TAF TAF hepatocytes in WT versus nfkb1 / 30-week DEN livers. (d) Graph shows

per cent TAF hepatocytes in 30-week DEN-injured nfkb1 / mice treated with control IgG or Ly6G antibody for 8 weeks, n 10. All data are

meanss.e.m. TAF analysis, a minimum of 50 cells per liver counted for human sections and a minimum of 80 cells counted per mouse section). Statistical signicance was determined using an unpaired t-test, **Po0.01 or ***Po0.001 compared with control.

accompanied by reduced 4HNE (Fig. 6g) and gH2A.X-stained

(Fig. 6h) hepatocytes, providing a clear in vivo correlation between neutrophil-derived ROS, oxidative DNA damage and development of HCC.

DiscussionNeutrophil inltration is associated with poor prognosis in a variety of human cancers including colorectal carcinoma, head and neck squamous carcinoma, bronchioloalveolar carcinoma and HCC45. Here we have described how inammatory signals released from hepatocytes in the damaged liver orchestrate the recruitment of circulatory neutrophils into the hepatic parenchyma where they stimulate genotoxic damage in bystander hepatocytes inclusive of telomere lesions that are

sufcient to stimulate HCC40,42. This contrasts with an apparent lack of contribution of neutrophils to regenerative or brogenic responses in the injured liver7,8 and indicates that the presence of neutrophils in human ASH and NASH liver may be a risk factor for progression to HCC, making these cells potential therapeutic targets. Signicant challenges will be to conrm the role of neutrophils in human CLD and to then translate this knowledge to strategies for manipulating neutrophils in the diseased liver. For the latter, opportunities include in-development small molecular inhibitors that block activities of the IL-8 family chemokines or their neutrophil receptor CXCR2 (ref. 46). Alternatively, the tumour suppressor activities of the NFkB1 gene may be exploitable. We have presented data in support of an anti-tumour mechanism for NFkB1 via p50:p50-mediated transcriptional repression of neutrophil-recruiting chemokines.

8 NATURE COMMUNICATIONS | 6:6818 | DOI: 10.1038/ncomms7818 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2015 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms7818 ARTICLE

WT

Ave 4HNE+ cells 15 HPF

Ave H2A.X+ cells 15 HPF

1,000

WT

250

WT

nfkb1/

***

800

600

400

200

nfkb1/

** ***

** **

***

nfkb1/

*

***

0

5wk 10wk 15wk 20wk 30wk 40wk 5wk 10wk 15wk 20wk 30wk 40wk

Mean fluorescence intensity ROX

Mean fluorescence intensity DHE

10

8

6

4

2

Percent 53bp1 +ve hepatocytes

Direct co-culture

Indirect co-culture

6

60

***

***

Control

**

*

4

2

50

40

30

20

10

*

*

Direct

0 Control

Direct

Indirect

Neutrophil/ hepatocytes

0 Control

Direct

Indirect

0 Control

Direct

Indirect

Neutrophil/ hepatocytes

200

150

100

50

0

Indirect

Neutrophil/ hepatocytes

Day 15 DEN IP Birth

wk 15 BHA diet

wk 30 Harvest

Control diet

Control diet

Ave 4HNE+ cells 15 HPF

40 35 30 25 20 15 10

5 0

Ave H2A X+ cells 15 HPF

100

80

60

40

20

BHA diet

BHA diet

***

***

nfkb1/ 30 wk DEN

Control diet

24.52 +/ 1.02 1.62 +/ 0.44

BHA diet***

Control

BHA

diet

nfkb1/

0 Control

BHA

diet

nfkb1/

Figure 6 | Neutrophil-dependent bystander ROS and hepatocellular cancer is limited by anti-oxidants. (a,b) Graph shows average total number of HNE- (a) and gH2A.X- (b) positive cells in liver sections from WTor nfkb1 / mice 540 weeks post DEN injury (n 6, 4, 6, 7, 15, 9 (nfkb1 / ) and 4, 4,

4, 5, 11, 14 (WT) for the 5- to 40-week time points, respectively). (a) Representative photomicrographs at 200 magnication show hepatic 4HNE

staining in 40-week DEN-injured WT or nfkb1 / mice, black arrows denote 4HNE hepatocytes; scale bars, 100 mm. (c) Diagram showing either direct

(top) or indirect (bottom) transwell co-culture of WT hepatocytes and WT neutrophils. (d) Graphs show mean uorescence intensity of ROX and dihydroehidium (DHE; ROS production) in WT hepatocytes culturedWT neutrophils n 3. (e) Representative immunocytochemistry images and cell

counts of DAPI/53BP1-stained WT hepatocytes only or WT hepatocytes in either direct or indirect (transwell) co-culture with WT neutrophils n 3;

scale bars, 10 mm. (f) Diagram of BHA therapy in 30-week DEN-injured nfkb1 / mice. Representative liver pictures and average tumour counts in 30-week DEN-injured nfkb1 / mice15 weeks dietary supplementation with BHA (n 5 control diet and 8 BHA diet). (g,h) Graphs show average HNE

(g) and gH2A.X (h) counts and representative photomicrographs of liver sections from 30-week DEN-injured nfkb1 / miceBHA. Black arrows denote 4HNE hepatocytes (g) or gH2A.X hepatocytes (h); scale bars, 100 mm. All data are meanss.e.m. Statistical signicance was determined

using an unpaired t-test, *Po0.05, **Po0.01 or ***Po0.001 compared with control.

From our observations, we predict that enhancing the anti-inammatory properties of p50:p50 and its co-repressor HDAC1 in hepatocytes would have suppressive effects on the hepatic expression of multiple neutrophil chemokines, thereby limiting hepatic neutrophil accumulation and reducing the incidence of genotoxic and telomeric damage in bystander hepatocytes. Possible approaches would be to exploit the ability of co-factor Bcl3 to enhance stability of p50:p50 dimers at promoters34 or to

manipulate the assembly of p50 subunits into homodimers over heterodimers. Towards the latter aim, we have begun to investigate structural features of p50 that dictate the choice of dimer partner and have identied human S342/mouse S340 as non-essential for heterodimer assembly but essential for homodimers. A mouse carrying a mutation at S340 displayed similar disease phenotypes to the nfkb1 / mouse including increased susceptibility to HCC. We have therefore demonstrated

NATURE COMMUNICATIONS | 6:6818 | DOI: 10.1038/ncomms7818 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 9

& 2015 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms7818

that p50 contains discrete structural components that can be manipulated to modify its selection of dimer partner and which have an impact on inammation-associated cancer. Therefore, more detailed structurefunction studies on p50 are warranted with the aim of discovering how to experimentally promote the assembly and stabilization of anti-inammatory p50:p50 dimers.

A caveat with the concept of inhibiting hepatic neutrophil recruitment in CLD is the potential dual role for neutrophils in cancer and the risk of removing neutrophil-mediated cytotoxic activities directed against tumour cells47. Disarming the harmful bystander effects of neutrophils on hepatocytes while retaining their anti-tumorigenic activities may be an optimal approach. Our discovery that neutrophils induce ROS-mediated telomere damage on bystander hepatocytes is therefore signicant given the associations in the literature between telomere lesions, telomerase reactivation and cancer42. Telomere DNA damage foci were detected in vivo in diseased livers and found to be exacerbated in nfkb1 / mice, but were ameliorated in animals treated with anti-Ly6G. In vitro co-culture experiments conrmed the ability of neutrophils to induce telomere DNA lesions in hepatocytes and indicated a diffusible mediator is responsible for this bystander damage. In this regard, long-lived neutrophil-derived ROS species (for example, HOCl and N-chloramines) and/or proteolytic enzymes are worthy of future consideration and, in particular, as they are also naturally occurring DNA repair inhibitors48.

It is pertinent to consider the relevance of our ndings for inammation-driven cancer in humans. The degree to which currently available rodent models of liver disease recapitulate the pathologies underpinning the development of HCC in humans is questionable. It is worth noting the absence of the ageing process in these models, which is intimately linked with CLD and HCC in humans. Ageing is associated with a physiological increase in hepatic lipid accumulation, elevated systemic expression of inammatory mediators, loss of replicative potential for hepatocytes and a substantially increased risk for the development of brosis and HCC49. We have recently reported that the nfkb1 / mouse has a reduced lifespan that is associated with increased levels of cellular senescence in regenerative organs including the liver, which can be ameliorated by treatment with the anti-inammatory drug ibuprofen22. Here we now document in these mice the spontaneous age-dependent development of numerous lesions that are reminiscent of those found in age-associated human liver disease including steatosis, hepatitis, brosis, ductular proliferation, dysplasia and HCC. Further detailed investigations of the interplay between mechanisms of accelerated liver ageing and cancer in the nfkb1 / background may be highly illuminating. Finally, although there is no obvious human model for loss of p50 homodimers, there are common polymorphisms in the human NFKB1 gene that have been genetically linked with inammatory disease. In particular, the 94 del/ins (rs28362491)-promoted polymorphism, which is predicted to result in reduced expression of p50, is associated with increased risk for HCC and for other cancers including breast, prostate, gastric, colorectal and oral5053. However, as yet there are no investigative studies in the literature examining the impact of the 94 del/ins polymorphism on dimer assembly and occupancy at NF-kB-regulated genes.

Methods

Mice and models of liver injury. All experiments on male C57BL/6 (WT), nfkb1 / , s100a9 / and nfkb1S340A/S340A mice were performed under approval from the Newcastle Ethical Review Committee and a UK Home Ofce licence. s100a9 / mice were kindly provided by Professor Nancy Hogg (UK), nfkb1 /

mice were a gift from Professor J. Caamano (UK). nfkb1 / and s100a9 / mice were bred in-house as homozygous lines and compared with in-house C57BL/6 mice. nfkb1S340A/S340A homozygous knock-in mice and WT controls were derived

from nfkb1S340A/WT heterozygous knock-in mice custom made at Taconic. Day 15 mice were given 30 mg kg 1 N-Nitrosodiethylamine (DEN) in 0.9% saline by intraperitoneal (i.p.) injection to induce liver cancer. Acute DEN, 8-week-old mice were given 100 mg kg 1 by i.p. injection to induce liver DNA damage. Either power calculations or previous studies, which achieved statistical signicance, were used to determine all group sizes.

Genotyping of nfkb1S340A/S340A mice. Genotyping was performed by PCR using genomic DNA isolated from ear clips. Tissue sample was the digested tissue in a buffer of 0.5% SDS, 20 mM EDTA, 200 mM NaCl, 40 mM Tris pH8.0, with50 mg ml 1 proteinase K at 55 C overnight, followed by phenol/chloroform extraction and DNA precipitation. The WT or nfkb1S340A/S340A alleles were amplied using NeoDel forward 50-GTCTTCAAAACGCCAAAGTATAAGGAT

GTC-30 and NeoDel reverse 50-CCCCTCCTGGTGGAGGACCAC-30 specic primers for 30 cycles; denaturation: 98 C, 10 s; annealing: 58 C, 15 s; extension: 68 C, 1 min). Amplication of a single product of 490 or 620 bp corresponds to WT or 340 / allele, respectively. Amplication of both products corresponds to a heterozygous ( / ) genotype (Supplementary Fig. 6ac).

Therapies. Mini-pumps (Alzet, model 2004) were implanted subcutaneously into 22-week DEN-injured (nfkb1 / ) or 32-week DEN-injured (WT) mice to deliver28.5 mg per day of Ly6G (clone 1A8) neutrophil depleting antibody or Rat IgG2a (clone 2A3) (BioXCell) for 8 weeks. nfkb1 / mice were fed BHA (0.7% w/w)or normal chow 15 weeks post DEN for 15 weeks. CXCL1 and CXCL2 were neutralized by administering i.p. 25 mg of anti-CXCL1 (AF-453-NA) and 25 mg of anti-CXCL2 (AF-452-NA) or 50 mg goat IgG (AB-108-C) 1 h before and 18 h post acute DEN. At the end of the experiment, animals were culled, and the liver and serum harvested for analysis.

In vivo imaging. WT neutrophils were labelled using the Cell Vue NIR815 uorescent kit (Li-Cor), according to manufacturers instructions, then 1 107 cells

were injected intravenously via the tail vein into 18 h acute DEN-injured WT or nfkb1 / mice. The IVIS220 series imaging system (Ex max 786 nm; em max 814 nm lters) was used to longitudinally track neutrophil migration; after the nal scan, animals were culled and the liver, kidney and the spleen were removed and imaged ex vivo. Data were analysed using Living Image 4.2 software, regionsof interest were drawn and Average Radiant Efciency (p s 1cm 2 sr 1)/

(mW cm 2) was calculated.

Neutrophil isolation. Briey, bone marrow was extracted from the femur and tibia of WT, nfkb1 / and nfkb1S340A/S340A mice by ushing with HBBSCa2with 5% serum. Neutrophils were isolated by percoll gradient (62%) and purity was established by Ly6G and CD11b (BD Biosciences) ow cytometry (BD FACScantoII). We used the 7AAD and Annexin V apoptosis detection kit according to manufacturers instruction (BD Biosciences) and ow cytometry to measure neutrophil apoptosis.

Generation of bone marrow chimera. Eight- to 12-week-old WT and nfkb1 /

mice were used as donors and recipients. Bone marrow cells were isolated from WT and nfkb1 / mice as described above. The bone marrow cell suspension was washed, centrifuged at 400g for 5 min, then re-suspended and counted. Recipient mice underwent whole body irradiation (NDT 320KV 3.2KW irradiator) at 10 Gy and then received 107 bone marrow cells intravenously.

Hepatocyte isolation. Hepatocytes were isolated from the livers of WT,nfkb1 / and nfkb1S340A/S340A mice by digestion with collagenase from Clostridium histolyticum (Sigma), then ltered through a 70-mm cell strainer. Cells were collected by centrifugation (500 r.p.m. for 3 min), washed three times in KrebsRinger buffer (Sigma) and then re-suspended in Williams medium E with 10% serum (WME Gibco), and plated onto collagen-coated plates (type I collagen, BD Biosciences). After 4 h, medium was removed and cells were cultured in fresh 10% or 0.5% Williams medium E.

Hepatocyte and neutrophil co-culture and ROS production. WT and nfkb1 /

hepatocytes were plated on collagen-coated transwell plates, then co-cultured with WT or nfkb1 / neutrophilsa 3-mm transwell insert (Thincert Griener) at a 1:1 ratio for 18 h. Intracellular ROS of hepatocytes and isolated neutrophils was measured following incubation with 10 mM diuorouorescein diacetate (FITC 488 nm), 5 mM dihydroehidium (Rhodamine, 594 nm) or 5 mM Cell Rox Orange for 30 min at 37 C. Median uorescence intensity was measured by ow cytometry. Up to 10,000 events were analysed on FACScan/FACS Canto II (BD, Oxford, UK) using Flowjo software (FlowJo, Inc.).

Bone marrow macrophages. Bone marrow cells from the femurs of WT or nfkb1 / mice were differentiated into macrophages (7 days with RPMI media

10 NATURE COMMUNICATIONS | 6:6818 | DOI: 10.1038/ncomms7818 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2015 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms7818 ARTICLE

supplementedwith 5% horse serum (Sigma-Aldrich) and 10% L929 supernatant (refreshed on days 3 and 6)).

Cell culture. Cos-7 cells were cultured in DMEM supplemented with 100 U ml 1 penicillin, 100 mg ml 1 streptomycin, 2 mM L-glutamine and 16% fetal bovine serum at 37 C at an atmosphere of 5% CO2.

Adenoviral expression of WT and S340A p50. Hepatocytes were transduced with one green uorescent forming unit per cell of control adenovirus or adeno-virus expressing either WT murine p50 or S340A p50 for 24 h. Adenovirus was kindly provided by Professor G Eric Blair (UK).

Human tissue. HCC and alcoholic liver disease liver samples for histology were collected under full ethical approval with patient consent (REC references 10/H0906/41).

ChIP assay. Cross-linked chromatin was prepared from acute DEN-injured WT, nfkb1 / and nfkb1S340A/S340A mouse livers, using the protocol outlined in the Upstate Biotechnology Immuno-precipitation ChIP assay kit. ChIP was performed using 25 mg of cross-linked chromatin per reaction and 5 mg of antibody to HDAC1 (05-100 Millipore), p50 (ab7971 Abcam) or IgG control (Abcam) for immuno-precipitation. Mouse S100A9, CXCL1 and two promoters were amplied by quantitative reverse transcriptasePCR using specic primers (Supplementary Methods Table 1).

Electromobility shift assay. Cos-7 cells were co-transfected with 1 mg Flag-tagged WT-p50 or mutant-p50 using the effectene kit (Qiagen). After 48 h, cells were lysed with Dignam A buffer and incubated on ice for 15 min with vortexing, then centrifuged at 900g at 4 C for 30 s. The cell pellet was re-suspended in Dignam C buffer and incubated on ice for 15 min. After centrifugation at 900g (4 C) for 5 min nuclear extracts were removed and stored at 80 C. p50 For EMSA, briey

810 mg of nuclear extract was incubated with poly dIdC and 32P-labelled NFkB Oligonucleotide (Promega) for 20 min at 4 C. For supershift assays, reactions were incubated in the presence of 2 mg of anti-ag or anti-myc as a negative control (Sigma) for 16 h at 4 C. EMSA and supershift reaction mixtures were resolved by electrophoresis on an 8% non-denaturing polyacrylamide gel (37:5:1).

Enzyme-linked immunosorbent assay. WT and nfkb1 / hepatocytes were cultured overnight before protein extraction from the pelleted cells. Twenty micrograms of protein was used for CXCL2/MIP2 ELISA. The ELISA was performed according to the manufacturers instructions (R&D Quantikine ELISA kit).

Histological stains. Formalin xed, parafn-embedded liver sections were stained with haematoxylin and eosin and 0.1% Sirius Red Picric solution following standard procedures. Diastase PAS, Reticulin and Sirius red/fast green, keratin 19 and glutamine synthetase were immunostained at the Department of Cellular Pathology, Royal Victoria Hospital, Newcastle Upon Tyne.

Immunohistochemistry. Staining was performed on formalin-xed, parafn-embedded liver sections. Endogenous peroxidase activity was blocked with hydrogen peroxide and antigen retrieval was achieved using proteinase K(20 mg ml 1) for detection of F480 1:100 (ab6640 Abcam), 0.01% pronase for neutrophil elastase 1:200 (ab21595 Abcam) and NIMP-R14 1:200 (ab2557 Abcam), and antigen unmasking solution for active caspase-3 1:200 (9664 Cell Signaling), S100A9 1:100 (ab73987 Abcam), 4HNE 1:50 (MHN-020P JalCA), aSMA 1:1000 (F3777 Sigma) and Collagen IV 1:100 (Abcam). EDTA (1 mM) for gH2A.X 1:100 (9718 Cell Signaling) and trypsin for PCNA 1:250 (ab2426 Abcam). Tissue was blocked using an Avidin/Biotin Blocking Kit (Vector Laboratories) followed by 20% swine serum in PBS and then incubated with primary antibodies overnight at 4 C. The next day, slides were washed and incubated with biotinylated swine anti-rabbit 1:200 (eo353 Dako), biotinylated goat anti-uorescein 1:300 (BA-0601 Vector) or goat anti-rat 1:200 (STAR80B Serotec), followed by Vectastain Elite ABC Reagent. Antigens were visualized using DAB peroxidase substrate kit and counterstained with Mayers haematoxylin. For 4HNE, amplication was achieved using the mouse on mouse kit (Vector) Immuno-stained cells were manually counted and expressed as the mean number of positive cells in 15 high power( 20) elds. Image analysis of a minimum of ten elds ( 10) was performed

using Leica Qwin for F4/80. All stained histological sections were blinded before counting/analysis.

Histological assessment of liver disease and cancer. Histological examination of aged WT and nfkb1 / liver sections was performed by a liver pathologist (DGT) and immunostains were evaluated by two observers (DGT and FO).

Steatosis severity, the presence of steatohepatitis and extent of brosis were assessed according to Kleiner et al.54 Evaluation of hepatocellular dysplasia and neoplasms was based according to established histological criteria55.

Telomere immuno-FISH. Formalin-xed parafn-embedded liver sections were dewaxed and hydrated, washed with water and then citric saline antigen retrieval was performed. Slides were blocked with normal goat serum (1:60) in BSA/PBS and incubated with rabbit g-H2A.X (1:250) at 4 C overnight. The next day, slides were washed three times in PBS, incubated with secondary antibody for 30 min, washed three times in PBS and then incubated with Avidin DCS (1:500) for 20 min. After g-H2A.X immunouorescence, slides were washed three times in PBS, cross-linked with 4% paraformaldehyde for 20 min and dehydrated in graded ethanol. Sections were denatured for 10 min at 80 C in hybridization buffer (70% formamide (Sigma), 25 mM MgCl2, 0.1 M Tris pH 7.2, 5% blocking reagent (Roche) containing2.5 mg ml 1 Cy-3-labelled telomere-specic (50-CCCTAA-30) peptide nuclei acid probe (Panagene), followed by hybridization for 2 h at room temperature in the dark. Slides were washed twice with 70% formamide in 2 SSC for 15 min, fol

lowed by 2 SSC and PBS washes for 10 min. Next, sections were incubated with

DAPI, mounted and imaged. In-depth Z stacking was used (a minimum of 40 optical slices with 63 objective, Leica DM5500B) followed by Huygens (SVI)

deconvolution. Whole-image stacks were used to count TAFs.

Immunocytochemistry. Hepatocytes were cultured on sterile collagen-coated coverslips, then xed in 2% paraformaldehyde for 10 min and then blocked cells for 45 min with 0.2% Fish skin Gelatine, 0.5% BSA and 0.5% Triton X-100 in PBS. Slides were then incubated with the rabbit polyclonal anti-53BP1 4 mg ml 1 (NB100-904 Novus Biologicals) overnight at 4 C. The next day, the cells were washed and incubated with Alexa Fluor 594 secondary antibody (Invitrogen) for 60 min before mounting with DAPI mountant (Vector).

RNA isolation and real-time PCR. Total RNA was isolated from mouse liver or cultured cells using the Total RNA Purication Kit (QIAGEN) and then treated with DNAse and used as a template in rst-strand complementary DNA synthesis using random primers (Promega). SYBR Green quantitative reverse transcriptase PCR was performed using the primers listed in (Supplementary Table 5).

Generation of p50 constructs. Flag-p50 and HA-p50 was sub-cloned into pCDNA3 using HindIII and ApaI restriction enzyme digestion and used as a template to generate the following Flag- and haemagglutinin (HA)-tagged p50 mutant constructs: p50 T145A mutant, p50 S210A mutant, p50 T315A mutant, p50 S337A mutant and p50 S342A mutant by two-step site-directed mutagenesis. A common Flag tag or HA tag forward primer was used with the respective mutant reverse primer. A common reverse primer was used for the second PCR reaction. Final PCR products were restriction enzyme digested with HindIII and ApaI, separated on 1% agarose gel, excised, puried and sub-cloned them into pCDNA3. The primer sequences used are listed in (Supplementary Table 6).

Co-transfection and immunoprecipitation. To assess p50:RelA interactions, Cos-7 cells were co-transfected with 0.5 mg RelA-eGFP and 1 mg Flag-tagged WT-p50 or mutant-p50. For p50:p50 interactions, Cos-7 cells were co-transfected with 0.5 mg of Flag-tagged WT-p50 or mutant-p50, and 1 mg of HA-tagged WT-p50 construct. The Flag immunoprecipitation kit (Sigma) was used according to the manufacturers instructions. At 36 h post transfection, cells were PBS washed and then lysed into RIPA buffer (input control) or lysed with Flag lysis buffer (50 mM Tris HCl, pH 7.4, with 150 mM NaCl, 1 mM EDTA and 1% Triton X-100). Flag beads were washed in wash buffer (50 mM Tris HCl pH 7.4 with 150 mM NaCl) and then 1 mg of cell lysate was added to the beads. Cell lysates and beads were agitated overnight at 4 C and then centrifuged at 13,000 r.p.m. for 30 s. The supernatant was removed and the remaining beads were washed 3 in wash

buffer. Protein was then eluted by adding 100 ml of 3 Flag peptide and incubated

at 4 C for 30 min with gentle agitation. The beads were centrifuged and the supernatant removed. Thirty micrograms of input control and 50 ml of IP eluate was loaded onto a 9% SDSPAGE gel for western blotting.

Isolation of whole-tissue lysates. Tissue (B5 mg) was lysed and homogenized in 300 ml RIPA buffer supplemented with protease and phosphatase inhibitor cocktails. Lysates were passed through a QIAshredder and then spun at 16,000g for 2 min. Flow through was collected and sonicated for 10 s. Debris was pelleted by centrifugation at 16,000g at 4 C for 15 min and the supernatant was collected. Protein concentration was measured using the detergent-compatible protein assay kit purchased from Bio-Rad.

SDSpolyacrylamide gel electrophoresis. Total protein was fractionated by 9% SDSPAGE, transferred onto nitrocellulose and then blocked blots with Tris-buffered saline and Tween 20 (0.1%) containing 5% BSA before incubation overnight with primary antibodies (1:1,000) to S100A9 (ab73987 Abcam), PCNA (ab2426 Abcam), Cyclin D1 (ab16663 Abcam), p105/p50 (ab7971 Abcam), RelA (ab7970 Abcam), c-Rel (sc71 Santa Cruz), CyP2E1 (ab28146 Abcam) or GAPDH (Abcam), total p38a (9218 Cell Signaling), P-p38a (9216 Cell Signaling), HA or

Flag-HRP conjugate (Sigma). Next day, membranes were washed in T-TBS and then incubated with horseradish peroxidase-conjugated mouse anti-rabbit IgG.

NATURE COMMUNICATIONS | 6:6818 | DOI: 10.1038/ncomms7818 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 11

& 2015 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms7818

Blots were washed and antigen was detected using enhanced chemiluminescence (Amersham Biosciences). Images have been cropped for presentation. Full-size images are presented in Supplementary Fig. 10.

Model of p50 homodimer binding DNA. Molecular graphics images were generated using PyMOL (DeLano Scientic) and solvent-accessible surface areas calculated with Stride56.

Statistical analysis. Data are expressed as meanss.e.m. GraphPad Instat was used to perform unpaired Students t-test or analysis of variance with a Tukeys post-hoc test and *Po0.05, **Po0.01 or ***Po0.001 was considered signicant.

References

1. Schutte, K., Bornschein, J. & Malfertheiner, P. Hepatocellular carcinoma--epidemiological trends and risk factors. Digest. Dis. 27, 8092 (2009).

2. Coulouarn, C. & Clement, B. Stellate cells and the development of liver cancer: Therapeutic potential of targeting the stroma. J. Hepatol. 60, 13061309 (2014).

3. Taniguchi, K. & Karin, M. IL-6 and related cytokines as the critical lynchpins between inammation and cancer. Semin. Immunol. 26, 5474 (2014).

4. Lefkowitch, J. H. Morphology of alcoholic liver disease. Clin. Liver Dis. 9, 3753 (2005).

5. McDonald, B. & Kubes, P. Neutrophils and intravascular immunity in the liver during infection and sterile inammation. Toxicol. Pathol. 40, 157165 (2012).

6. Shi, J., Gilbert, G. E., Kokubo, Y. & Ohashi, T. Role of the liver in regulating numbers of circulating neutrophils. Blood 98, 12261230 (2001).

7. Moles, A. et al. A TLR2/S100A9/CXCL-2 signaling network is necessary for neutrophil recruitment in acute and chronic liver injury in the mouse. J. Hepatol. 60, 782791 (2014).

8. Saito, J. M., Bostick, M. K., Campe, C. B., Xu, J. & Maher, J. J. Inltrating neutrophils in bile duct-ligated livers do not promote hepatic brosis. Hepatol. Res. 25, 180191 (2003).

9. Vesselinovitch, S. D., Koka, M., Mihailovich, N. & Rao, K. V. Carcinogenicity of diethylnitrosamine in newborn, infant, and adult mice. J. Cancer Res. Clin. Oncol. 108, 6065 (1984).

10. Maeda, S., Kamata, H., Luo, J. L., Leffert, H. & Karin, M. IKKbeta couples hepatocyte death to cytokine-driven compensatory proliferation that promotes chemical hepatocarcinogenesis. Cell 121, 977990 (2005).

11. Naugler, W. E. et al. Gender disparity in liver cancer due to sex differences in MyD88-dependent IL-6 production. Science 317, 121124 (2007).

12. Daley, J. M., Thomay, A. A., Connolly, M. D., Reichner, J. S. & Albina, J. E. Use of Ly6G-specic monoclonal antibody to deplete neutrophils in mice. J. Leuk. Biol. 83, 6470 (2008).

13. Campbell, S. J. et al. Hepatic nuclear factor kappa B regulates neutrophil recruitment to the injured brain. J. Neuropathol. Exp. Neurol. 67, 223230 (2008).

14. Oakley, F. et al. Nuclear factor-kappaB1 (p50) limits the inammatory and brogenic responses to chronic injury. Am. J. Pathol. 166, 695708 (2005).

15. Rajendrasozhan, S., Chung, S., Sundar, I. K., Yao, H. & Rahman, I. Targeted disruption of NF-{kappa}B1 (p50) augments cigarette smoke-induced lung inammation and emphysema in mice: a critical role of p50 in chromatin remodeling. Am. J. Physiol. Lung Cell. Mol. Physiol. 298, L197L209 (2010).

16. Iyoda, K. et al. Involvement of the p38 mitogen-activated protein kinase cascade in hepatocellular carcinoma. Cancer 97, 30173026 (2003).

17. Fridlender, Z. G. et al. Polarization of tumor-associated neutrophil phenotype by TGF-beta: N1 versus N2 TAN. Cancer Cell 16, 183194 (2009).

18. Houghton, A. M. et al. Neutrophil elastase-mediated degradation of IRS-1 accelerates lung tumor growth. Nat. Med. 16, 219223 (2010).

19. Jamieson, T. et al. Inhibition of CXCR2 profoundly suppresses inammation-driven and spontaneous tumorigenesis. J. Clin. Invest. 122, 31273144 (2012).

20. Nozawa, H., Chiu, C. & Hanahan, D. Inltrating neutrophils mediate the initial angiogenic switch in a mouse model of multistage carcinogenesis. Proc. Natl Acad. Sci. USA 103, 1249312498 (2006).

21. Pekarek, L. A., Starr, B. A., Toledano, A. Y. & Schreiber, H. Inhibition of tumor growth by elimination of granulocytes. J. Exp.Med. 181, 435440 (1995).

22. Jurk, D. et al. Chronic inammation induces telomere dysfunction and accelerates ageing in mice. Nat. Commun. 2, 4172 (2014).

23. Nemeth, J. et al. S100A8 and S100A9 are novel nuclear factor kappa B target genes during malignant progression of murine and human liver carcinogenesis. Hepatology 50, 12511262 (2009).

24. Wiechert, L. et al. Hepatocyte-specic S100a8 and S100a9 transgene expression in mice causes Cxcl1 induction and systemic neutrophil enrichment. Cell Commun. Signal. 10, 40 (2012).

25. Acharyya, S. et al. A CXCL1 paracrine network links cancer chemoresistance and metastasis. Cell 150, 165178 (2012).

26. Ichikawa, M., Williams, R., Wang, L., Vogl, T. & Srikrishna, G. S100A8/A9 activate key genes and pathways in colon tumor progression. Mol. Cancer Res. 9, 133148 (2011).

27. Schnekenburger, J. et al. The calcium binding protein S100A9 is essential for pancreatic leukocyte inltration and induces disruption of cell-cell contacts.J. Cell. Physiol. 216, 558567 (2008).28. McNeill, E. & Hogg, N. S100A9 has a protective role in inammation-induced skin carcinogenesis. Int. J. Cancer 135, 798808 (2014).

29. Elsharkawy, A. M. et al. The NF-kappaB p50:p50:HDAC-1 repressor complex orchestrates transcriptional inhibition of multiple pro-inammatory genes.J. Hepatol. 53, 519527 (2010).30. Kuprash, D. V., Udalova, I. A., Turetskaya, R. L., Rice, N. R. & Nedospasov, S.A. Conserved kappa B element located downstream of the tumor necrosis factor alpha gene: distinct NF-kappa B binding pattern and enhancer activity in LPS activated murine macrophages. Oncogene 11, 97106 (1995).31. Udalova, I. A. et al. Functional consequences of a polymorphism affecting NF-kappaB p50-p50 binding to the TNF promoter region. Mol. Cell. Biol. 20, 91139119 (2000).

32. Zhong, H., May, M. J., Jimi, E. & Ghosh, S. The phosphorylation status of nuclear NF-kappa B determines its association with CBP/p300 or HDAC-1. Mol. Cell 9, 625636 (2002).

33. Ziegler-Heitbrock, H. W. et al. Tolerance to lipopolysaccharide involves mobilization of nuclear factor kappa B with predominance of p50 homodimers.J. Biol. Chem. 269, 1700117004 (1994).34. Collins, P. E., Kiely, P. A. & Carmody, R. J. Inhibition of transcription by B Cell Leukaemia 3 (Bcl-3) requires interaction with Nuclear Factor (NF)-kappaB p50.J. Biol. Chem. 289, 70597067 (2014).35. Palmer, S. & Chen, Y. H. Bcl-3, a multifaceted modulator of NF-kappaB-mediated gene transcription. Immunol. Res. 42, 210218 (2008).

36. Cheng, C. S. et al. The specicity of innate immune responses is enforced by repression of interferon response elements by NF-kappaB p50. Sci. Signal. 4, ra11 (2011).

37. Amulic, B., Cazalet, C., Hayes, G. L., Metzler, K. D. & Zychlinsky, A. Neutrophil function: from mechanisms to disease. Ann. Rev. Immunol. 30, 459489 (2012).

38. Fumagalli, M. et al. Telomeric DNA damage is irreparable and causes persistent DNA-damage-response activation. Nat. Cell Biol. 14, 355365 (2012).

39. Hewitt, G. et al. Telomeres are favoured targets of a persistent DNA damage response in ageing and stress-induced senescence. Nat. Commun. 3, 708 (2012).

40. Farazi, P. A. et al. Differential impact of telomere dysfunction on initiation and progression of hepatocellular carcinoma. Cancer Res. 63, 50215027 (2003).

41. Rudolph, K. L., Millard, M., Bosenberg, M. W. & DePinho, R. A. Telomere dysfunction and evolution of intestinal carcinoma in mice and humans. Nat. Genet. 28, 155159 (2001).

42. Begus-Nahrmann, Y. et al. Transient telomere dysfunction induces chromosomal instability and promotes carcinogenesis. J. Clin. Invest. 122, 22832288 (2012).

43. Jaeschke, H. Mechanisms of Liver Injury. II. Mechanisms of neutrophil-induced liver cell injury during hepatic ischemia-reperfusion and other acute inammatory conditions. Am. J. Physiol. Gastrointest. Liver Phyisiol. 290, G1083G1088 (2006).

44. el Ghissassi, F., Barbin, A., Nair, J. & Bartsch, H. Formation of 1,N6-ethenoadenine and 3,N4-ethenocytosine by lipid peroxidation products and nucleic acid bases. Chem. Res. Toxicol. 8, 278283 (1995).

45. Dumitru, C. A., Lang, S. & Brandau, S. Modulation of neutrophil granulocytes in the tumor microenvironment: mechanisms and consequences for tumor progression. Semin. Cancer Biol. 23, 141148 (2013).

46. Campbell, L. M., Maxwell, P. J. & Waugh, D. J. Rationale and means to target pro-inammatory interleukin-8 (CXCL8) signaling in cancer. Pharmaceuticals 6, 929959 (2013).

47. Fridlender, Z. G. & Albelda, S. M. Tumor-associated neutrophils: friend or foe? Carcinogenesis 33, 949955 (2012).

48. Pero, R. W., Sheng, Y., Olsson, A., Bryngelsson, C. & Lund-Pero, M. Hypochlorous acid/N-chloramines are naturally produced DNA repair inhibitors. Carcinogenesis 17, 1318 (1996).

49. Sheedfar, F., Di Biase, S., Koonen, D. & Vinciguerra, M. Liver diseases and aging: friends or foes? Aging Cell 12, 950954 (2013).

50. Cheng, C. W. et al. Effects of NFKB1 and NFKBIA gene polymorphisms on hepatocellular carcinoma susceptibility and clinicopathological features. PLoS One 8, e56130 (2013).

51. Duan, W. et al. Association between the NFKB1-94ins/del ATTG polymorphism and cancer risk: an updated meta-analysis. Cancer Invest. 32, 311320 (2014).

52. Lo, S. S., Chen, J. H., Wu, C. W. & Lui, W. Y. Functional polymorphism of NFKB1 promoter may correlate to the susceptibility of gastric cancer in aged patients. Surgery 145, 280285 (2009).

53. Mohd Suzairi, M. S. et al. The functional -94 insertion/deletion ATTG polymorphism in the promoter region of NFKB1 gene increases the risk of sporadic colorectal cancer. Cancer Epidemiol. 37, 634638 (2013).

54. Kleiner, D. E. et al. Design and validation of a histological scoring system for nonalcoholic fatty liver disease. Hepatology 41, 13131321 (2005).

12 NATURE COMMUNICATIONS | 6:6818 | DOI: 10.1038/ncomms7818 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2015 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms7818 ARTICLE

55. Goodman, Z. D., Terracciano, L. M. & Wee, A. 14Tumours and tumour-like lesions of the liver. in MacSweens Pathology of the Liver (Sixth Edition). (ed. Ferrell, A.D.B.C.P.D.) 761851 (Churchill Livingstone, 2012).

56. Frishman, D. & Argos, P. Knowledge-based protein secondary structure assignment. Proteins 23, 566579 (1995).

Acknowledgements

We thank Professor Nancy Hogg (CRUK London Research Institute) for kindly providing s100a9 / mice, Professor Alistair Burt (The University of Adelaide) for access to human patient tissue and the Newcastle University Bioinformatics Unit for assistance with in silico analysis of p50 protein structure. This work was funded by a European Commission FP7 grant INFLA-CARE (EC Contract No. 223151; http://inflacare.imbb.forth.gr/

Web End =http:// http://inflacare.imbb.forth.gr/

Web End =inacare.imbb.forth.gr/ ) and supported by grants from the UK Medical Research Council (Grant G0700890, MR/K0019494/1 and M501700 to D.A.M. and G0900535 to F.O.) and the Wellcome Trust (WT086755MA to D.A.M.). The IVIS system was purchased under a Wellcome Trust Equipment Grant (087961) awarded to D.A.M and others.

Authors contributions

C.L.W. carried out the majority of the laboratory-based work and analyses presented in the manuscript. D.J., N.F., P.B., S.L., A.M.E., R.G.G., J.B.C., C.F., A.L., J.F.P., G.S., K.C., A.P., C.R. and J.M. performed a portion of the laboratory experiments and their related analyses. A.J.M. was responsible for in silico analyses of the p50 protein. G.E.B. and

N.F. produced the p50-expressing adenoviruses. F.O., J.M. and C.L.W. carried out all of the in vivo experiments. D.T. performed the routine histological examinationand scoring of livers from the ageing mice and F.O. carried out histological and statistical analyses. J.M. produced all of the nal gures. F.O. and D.A.M. conceived the studies, designed the experiments, were chiey responsible for data interpretation and wrote the manuscript. All authors read and commented on the nal manuscript.

Additional information

Supplementary Information accompanies this paper at http://www.nature.com/naturecommunications

Web End =http://www.nature.com/ http://www.nature.com/naturecommunications

Web End =naturecommunications

Competing nancial interests: The authors declare no competing nancial interests.

Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions

Web End =http://npg.nature.com/ http://npg.nature.com/reprintsandpermissions

Web End =reprintsandpermissions/

How to cite this article: Wilson C. L. et al. NFkB1 is a suppressor of neutrophil-driven hepatocellular carcinoma. Nat. Commun. 6:6818 doi: 10.1038/ncomms7818 (2015).

This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the articles Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

Web End =http://creativecommons.org/licenses/by/4.0/

NATURE COMMUNICATIONS | 6:6818 | DOI: 10.1038/ncomms7818 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 13

& 2015 Macmillan Publishers Limited. All rights reserved.