Composition and functionality of the intrahepatic innate lymphoid cell-compartment in human nonfibrotic and fibrotic livers.

Clinical

1280

HIGHLIGHTS FRONTLINE ࢯ Research Article

Composition and functionality of the intrahepatic innate lymphoid cell-compartment in human

nonfibrotic and fibrotic livers

Marianne Forkel

, Lena Berglin

, Eliisa Kek¨ al¨ ainen

, Adrian Carlsson

, Emma Svedin

, Jakob Micha¨elsson

, Maho Nagasawa

, Jonas S Erjef¨ alt

, Michiko Mori

, Malin Flodstr¨ om-Tullberg

, Annika Bergquist

,

Hans-Gustaf Ljunggren

, Magnus Westgren

, Ulrik Lindforss

, Danielle Friberg

, Carl Jorns

, Ewa Ellis

, Niklas K Bj¨ orkstr¨ om

and Jenny Mj¨ osberg

1Center for Infectious Medicine, Department of Medicine, Karolinska Institute Stockholm, Sweden

2Department of Cell Biology and Histology, Academic Medical Center of the University of Amsterdam, Amsterdam, The Netherlands

3Unit of Airway Inflammation, Department of Experimental Medical Sciences, Lund University, Lund, Sweden

4Department of Molecular Medicine and Surgery, Karolinska Institute and Center for Digestive Diseases, Karolinska University Hospital, Stockholm, Sweden

5Center for Fetal Medicine, Department of Clinical Science, Intervention and Technology (CLINTEC), Karolinska Institute, Stockholm, Sweden

6Department of Otorhinolaryngology, CLINTEC, KI, Stockholm, Sweden

7Division of Transplantation Surgery, CLINTEC, Karolinska Institutet (KI), Stockholm, Sweden

8Department of Clinical and Experimental Medicine, Link¨oping University, Sweden

Human innate lymphoid cells have been described to exist in different organs, with func- tional deregulation of these cells contributing to several disease states. Here, we per- formed the first detailed characterization of the phenotype, tissue-residency properties, and functionality of ILC1s, ILC2s, and ILC3s in the human adult and fetal liver. In addition, we investigated changes in the ILC compartment in liver fibrosis. A unique composition of tissue-resident ILCs was observed in nonfibrotic livers as compared with that in mucosal tissues, with NKp44⁻ILC3s accounting for the majority of total intrahepatic ILCs. The frequency of ILC2s, representing a small fraction of ILCs in nonfibrotic livers, increased in liver fibrosis and correlated directly with the severity of the disease. Notably, intra- hepatic ILC2s secreted the profibrotic cytokine IL-13 when exposed to IL-33 and thymic stromal lymphopoetin (TSLP); these cytokines were produced by hepatocytes, hepatic stellate cells (HSCs), and Kupffer cells in response to TLR-3 stimulation. In summary, the present results provide the first detailed characterization of intrahepatic ILCs in human

Correspondence:Jenny Mj ¨osberg e-mail: jenny.mjosberg@ki.se

adult and fetal liver. The results indicate a role for ILC2s in human liver fibrosis, implying that targeting ILC2s might be a novel therapeutic strategy for its treatment.

Keywords: Innate lymphoid cells ^rLiver ^rFibrosis ^rHuman ^rTissue-residency



Introduction

Innate lymphoid cells (ILCs) are a heterogeneous group of innate immune cells with a lymphoid morphology, comprising NK cells and the more recently described subsets of non-NK ILCs. These noncytotoxic ILCs commonly express the IL-7 receptor α-chain (CD127) and depend on IL-7 signaling for their development [1].

CD127⁺ILCs are divided into three subpopulations based on their cytokine secretion-profile and the expression of transcription factors, mirroring the classification of T helper cells [1]. Group 3 ILCs (ILC3s) are dependent on RORγt for their development and produce the signature cytokines IL-22 and/or IL-17, where IL-22 is produced mainly by the ILC3 subset expressing the natural cyto- toxicity receptor NKp44 [2–4]. ILC3s also include lymphoid tissue inducer cells (LTi), which are important in formation of lymph nodes during fetal development [3]. Group 1 ILCs (ILC1s) are defined by the expression of Tbet and production of IFN-γ [5].

This group also includes cytotoxic ILC1s such as conventional nat- ural killer cells (cNK) and a subset of CD103⁺intraepithelial ILC1s in the gut [6]. Group 2 ILCs (ILC2s) depend on the transcription factors GATA3 [7, 8] and RORα [9] for their development and function. ILC2s secrete the type 2 cytokines IL-13 and IL-5 upon stimulation with IL-25, IL-33, or thymic stromal lymphopoetin (TSLP) [8, 10–13]. Recently, ILCs have been characterized in var- ious human organs and are increasingly being recognized to play important roles during homeostasis and disease [5, 6, 10–16].

Liver fibrosis is the consequence of sustained liver dam- age accompanied by a progressive accumulation of extracellular matrix (ECM) proteins. The major causes underlying liver fibrosis are chronic hepatitis C virus infection, alcohol abuse, and nonal- coholic steatohepatitis [17]. The end stage of hepatic fibrosis, liver cirrhosis, is accompanied by an increased risk for portal hyperten- sion, liver failure, and liver cancer [18]. Activated hepatic stellate cells (HSCs) have been suggested as the main producers of ECM proteins, thus playing a key role in liver fibrosis [19, 20]. To date, several profibrotic cytokines have been identified. One of the most potent is TGF-β1, which acts as an inducer of collagen produc- tion in HSCs [21]. Moreover, IL-13 is known to promote tissue fibrosis [22] and increased IL-13 levels were found in fibrotic liver tissue of patients infected withS. japonicum [23]. Interest- ingly, in mouse models ofS. mansoni infection ILC2s mediated lung fibrosis [24] and IL-13 was identified as a driver of hepatic fibrosis [25]. Moreover, a mechanism involving ILC2s in fibro- sis development in the mouse liver was recently suggested [26].

Although animal models of liver fibrosis have previously provided

important insights into disease mechanisms, the human and mouse liver differ substantially, particular with respect to immune cell composition [27]. Since the role for ILCs in the develop- ment of human liver fibrosis remains obscure, studies on patient material provide valuable insights into disease pathogenesis in humans.

Here, we investigated the distribution, phenotype, and func- tion of ILC subsets in the human adult and fetal liver. Addition- ally, we addressed the presence of ILC subsets in human fibrotic liver tissue. Our results demonstrate that the composition of ILCs in the liver is unique as compared with gut and tonsil and that intrahepatic ILCs are most likely tissue-resident cells. Further- more, the phenotype of adult and fetal liver differs markedly, as only fetal liver ILC3s express the LTi marker NRP1. In addition, we demonstrate that the frequency of intrahepatic ILC2s corre- lates with the degree of liver fibrosis, and IL-33 and TSLP, pro- duced by TLR3-stimulated hepatocytes, HSCs and Kupffer cells promote production of potentially profibrotic IL-13 from ILC2s.

These first data on human intrahepatic ILC characteristics in health and fibrosis might have clinical implications, as targeting ILCs may represent a therapeutic strategy for the treatment of liver fibrosis.

Results

All groups of innate lymphoid cells are present in the human liver

We set out to investigate the composition of ILCs with special emphasis on non-NK ILCs in fresh liver cell suspensions. Cells were isolated from tumor-free areas of nonfibrotic human livers.

Within the fraction of CD45⁺cells lacking lineage markers (Lin⁻), the majority of cells expressed CD56 (Fig. 1A). Most of these cells lacked expression of CD127. CD45⁺Lin⁻CD127⁻CD56⁺cells were readily identified as intrahepatic NK cells. The majority of these cells were CD56^bright(Fig. 1A), expressed NKG2A but lacked CD16 (data not shown), which is in sharp contrast to the periph- eral blood compartment where cytotoxic CD56^dimNKG2A⁻CD16⁺ cells are more prevalent [28, 29]. In addition, we also identi- fied CD45⁺Lin⁻cells expressing CD127 and lacking the NK cell markers NKG2A and CD16, indicating that these cells belong to the non-NK ILC lineages (Fig. 1A). Similar to ILCs that were pre- viously described in skin, gut, and upper airways [5, 13, 15], the CD45⁺Lin⁻CD127⁺CD16⁻NKG2A⁻ILCs could be subgrouped into four distinct populations on the basis of CRTH2, CD117, and

1282

Figure 1. Identification of ILC subsets in the human liver by flow cytometry. (A) Gating strategy for the flow cytometry analysis of ILCs and NK cells within mononuclear cells in the human liver. Lineage staining contains TCRα/β, TCR γ/δ, CD19, CD1a, CD123, BDCA2, CD14, FcεR1, CD34 as well as dead cell marker (DCM). Values in the gates indicate the percentage of cells in each gate or quadrant. Data are representative of six independent experiments performed with one donor each. (B) Expression of Eomes, Tbet, GATA3 and RORγ(t) within the ILC1, ILC2, and ILC3 subsets compared with that in NK cells from the liver, as measured by flow cytometry. Data are representative of 3–5 independent experiments performed with one donor each. (C) Expression of Helios and AHR in NKp44⁻and NKp44⁺ILC3s compared with that in NK cells. Data are representative of two independent experiments performed with one donor each.

NKp44 expression, identifying ILC2s [13], ILC1s [5], NKp44⁺, and NKp44⁻ ILC3s [2, 4, 5]. All ILCs expressed the pan-ILC marker CD161 (Fig. 1A).

By analyzing transcription factors specifically associated with the development and function of NK cells (Eomes and Tbet), ILC1s (Tbet), ILC2s (GATA3), and ILC3s (RORγ(t), Helios, and AHR), we could confirm the identity of the ILC subpopulations (Fig. 1B). NK cells expressed Tbet and Eomes. The majority of intrahepatic ILC1s expressed Tbet and intermediate levels of RORγ(t) (Fig. 1B). Within the Tbet⁺ ILC1s a small population of cells additionally expressed Eomes. ILC2s expressed GATA3 and intermediate levels of RORγ(t). ILC3s expressed RORγ(t) (Fig. 1B) and Helios (Fig. 1C), which was previously shown to be expressed by ILC3s in human tonsils [6, 13]. Interestingly, in contrast to ILC3s in tonsils (Supporting Information Fig. 1A), intrahepatic ILC3s lacked detectable expression of AHR (Fig. 1C).

Despite this, when analyzing intrahepatic ILC3s expanded in vitro, ILC3s produced IL-22 at frequencies comparable to tonsil ILC3s (Supporting Information Fig. 1B).

In summary, these data demonstrate that all currently described noncytotoxic ILC populations can be found in the human liver and that intrahepatic ILCs share a similar transcription factor profile with ILC subsets found in mucosal tissues such as gut and tonsil.

The intrahepatic ILC compartment is uniquely composed and differs between adult and fetal livers

We next set out to evaluate the overall composition of the ILC fam- ily in the human liver compared to gut and tonsil, mucosal sites where ILCs have previously been characterized [5, 30]. CD56⁺ NK cells dominated in the liver compared to the other organs (Fig. 2A). Lin⁻CD127⁺ILCs were present at comparable frequen- cies in the liver and tonsil, with the highest frequency found in gut (Fig. 2B). Overall, the liver had a distinct ILC subset composition compared to gut and tonsil. ILC1s and ILC2s were more frequent in the liver, especially compared to the gut where these cells were barely detectable (Fig. 2C). In contrast, NKp44⁺ILC3s were rel- atively rare in the liver as compared to gut and tonsil. Instead, NKp44⁻ILC3s was the most frequent ILC population in the liver (Fig. 2D).

To analyze intrahepatic ILCs in more detail we performed an extensive characterization of the phenotype of the ILC pop- ulations including markers of tissue residency. All ILC subsets exhibited hallmarks of tissue-resident cells with high expres- sion of CD69 and absence of CD62L (Fig. 2E). Interestingly, CD49a and CD103, previously shown to be expressed by dis- tinct subsets of intrahepatic NK cells [29] and intraepithelial ILC1s and ILC3s [6, 31] were only expressed by a minor pro- portion of each ILC subset (Fig. 2E). On average 5% of the intra- hepatic NK cells expressed CD49a. Furthermore, CD103, which is expressed by epithelium-associated ILC3s [31] and intraep- ithelial ILC1s [6] was absent on all intrahepatic ILC subsets.

These data indicate that intrahepatic ILCs, although likely tissue-

resident, do not utilize CD49a or CD103 for retention in the tissue.

Furthermore, the majority of intrahepatic non-NK ILCs lacked expression of the NK-cell receptors NKG2D expressed only low levels of NKp46 (Supporting Information Fig. 2). CD11a, which has been suggested to discriminate between NK cells and ILC3s [32] was expressed at high levels by NK cells, ILC1s, and ILC2s, whereas it was expressed at an intermediate level on intrahep- atic ILC3s (Supporting Information Fig. 2). The chemokine recep- tor CXCR3 was expressed by a fraction of intrahepatic ILC1s (Supporting Information Fig. 2), in accordance with a previous report of CXCR3-expression on ILC1s found in Crohn’s disease ileum [5]. A small fraction of ILC3s in the human tonsil and gut expresses the MHC class II molecule HLA-DR and mouse ILC3s expressing MHC-II regulate homeostais of gut T cells [30, 33].

In the liver, a substantial proportion of all intrahepatic ILC sub- sets expressed HLA-DR (Fig. 2F) suggesting that they could be involved in the regulation of intrahepatic T cells. Finally, we examined the expression of the cell surface receptor neuropilin- 1 (NRP1), a marker of lymphoid tissue inducer (LTi) cell activ- ity in mice and humans [34]. NRP1 was absent from all adult intrahepatic ILC subsets (Fig. 2F), indicating that not even NKp44⁻ ILC3s, which are phenotypically similar to LTi cells, possess LTi activity.

Next, we investigated whether the composition and phenotype of intrahepatic ILCs in adult livers was similar to that in fetal livers, or if substantial changes occur after birth. Fetal livers collected at gestational week (gw) 6–20 contained almost exclusively NKp44⁻ ILC3s (Fig. 3A and B). The other ILC subsets, ILC1s, ILC2s, and NKp44⁺ ILC3s, were absent through gw 6–10 and only started to be detectable from gw 15 onwards (data not shown). In con- trast to adult livers, a smaller proportion of fetal ILCs expressed CD69 (Fig. 3C). Moreover, HLA-DR was expressed at lower levels as compared to the adult intrahepatic ILCs. As expected, consid- ering the pronounced lymphoid organogenesis occurring during fetal development, the majority of fetal ILC3s expressed NRP1, likely representing LTi cells (Fig. 3D). Similar to the adult liver, CD49a expression could only be detected on a small fraction of ILCs.

Taken together, these data show that all ILC populations are present in the human liver with a composition significantly differ- ent from the one found in gut and tonsil. Furthermore, the differ- ences observed in the adult and fetal ILC compartment highlight potential discrepancies in ILC function dependent on age.

The frequency of intrahepatic ILC2s correlates with the severity of liver fibrosis

Several cytokines produced by subsets of ILCs, including IL-13 and IL-22 have been associated with the development of liver fibrosis [22, 35]. Therefore, we assessed the composition of ILC subsets in fibrotic livers. Although no difference in the total fre- quency of NK cells or Lin⁻CD127⁺ILCs could be seen between livers with no or mild fibrosis (grade 0–2) and severely fibrotic

1284

Figure 2. Distribution of ILC subsets in the adult liver compared with that in gut and tonsil. (A–C) The frequency of indicated cell subsets in liver, gut, and tonsil was measured by flow cytometry. (A, B) The percentages of (A) NK cells and (B) total ILCs among CD45⁺cells are shown. (C) The percentages of ILC1s, ILC2s, NKp44⁻ILC3s, and NKp44⁺ILC3s of total ILCs are also shown. (D) ILC subset distribution in the adult liver as measured by flow cytometry. (A–D) Results are displayed as median and interquartile range of the indicated numbers of donors. Independent experiments were performed for 4–10 donors (tonsil, n= 5; liver, n = 10; gut, n = 4). (E, F) The surface expression of (E) CD69, CD62L, CD103, and CD49a, and (F) HLA-DR and NRP1 on ILC1s, ILC2s, ILC3s, and NK cells was determined by flow cytometry. Unstained control lymphocytes or FMO control (CD49a) were used as controls. Plots are representative ofࣙ4 independent experiments each performed with one donor.

or cirrhotic livers (grade 3–4) (Fig. 4A), important differences were detected within the Lin⁻CD127⁺ILC compartment. The fre- quency of ILC2s was more than doubled, whereas the frequency of NKp44⁻ILC3s was reduced in severe fibrosis (Fig. 4B). Notewor- thy, the frequency of ILC2s directly correlated with the severity of liver fibrosis (Fig. 4C). The frequency of ILC2s was also increased in severe fibrosis when evaluating the percentage of ILC2s out of total CD45⁺ and total CD3⁻ cells (Fig. 4D and E). This was not seen for NKp44⁻ILC3s, where the frequency expressed as a percentage of CD45⁺ and CD3⁻ cells was similar between mild and severe fibrosis (Supporting Information Fig. 3). Interestingly, no change in the frequency of CD3⁺CRTH2⁺ cells, representing

mainly Th2 cells [36], was observed (Fig. 4F). To further validate our findings from fresh liver samples, we repeated the analysis on a cohort of frozen mononuclear cells from human liver tissue.

Also in this cohort, we detected a good correlation between ILC2 frequency and fibrosis score (Supporting Information Fig. 4).

The transcription factor expression of ILC2s from severely fibrotic livers, with expression of GATA3, intermediate levels of RORγ(t), but no Tbet (Fig. 4G), was consistent with the profile of ILC2s from nonfibrotic livers (Fig. 1B). Intrahepatic ILC2s from livers with fibrosis expressed high levels of CD69 and low lev- els of CD62L, as observed in nonfibrotic livers, but distinct from peripheral blood ILC2s (Fig. 4H). These data excluded that the

Figure 3. ILC distribution and phenotype of ILCs in fetal liver. (A) ILC subset distribution in the fetal liver was determined as for adult liver by flow cytometry, and displayed as median and interquartile range (n= 6 donors). (B) Transcription factor profile of NKp44⁻ILC3s and NK cells was measured by intracellular flow cytometry. Plots shown are representative of two independent experiments. (C, D) Surface expression of (C) CD69, CD62L, and HLA-DR, and (D) CD49a and NRP1 on intrahepatic ILC1s, ILC2s, ILC3s, and NK cells determined by flow cytometry. Unstained cells were used as controls. Representative plots from four individuals stained in two independent experiments.

1286

Figure 4. Phenotypic characterization of ILCs in fibrotic liver. (A) The frequency of NK cells and total ILCs among CD45⁺cells (left) and CD3⁻CD45⁺ cells (right) in patients with no or mild fibrosis, and patients with severe fibrosis, as determined by flow cytometry. (B) The frequency of ILC1s, ILC2s, NKp44⁻, and NKp44⁺ILCs among total ILCs is shown. (C) The frequency of ILC2s among total ILCs versus fibrosis score (Pearson correlation analysis). (D–E) The frequency of ILC2s among (D) CD45⁺cells and (E) CD3⁻cells in patients with no or mild fibrosis and patients with severe fibrosis.

(F) The frequency of CRTH2⁺T cells among CD45⁺cells in patients with no or mild fibrosis and patients with severe fibrosis. Mild fibrosis, score 0–2, n= 11; severe fibrosis, score 3–4, n = 9. Symbols represent individual donors and lines indicate median values. Comparisons between groups were made with two-tailed Mann–Whitney test. Normality in data distribution was evaluated by D’Agostino and Pearson omnibus normality testing.^*p

ࣘ 0.05,^**pࣘ 0.01. (G) Flow cytometry analysis of GATA3, RORγ(t), and Tbet expression in CRTH2⁺ILC2s from fibrotic livers (black). The expression in NK cells (gray) is shown for comparison. (H, I) CD69 and CD62L expression on ILC2s from (H) peripheral blood and (I) liver. Plots are representative of independent experiments performed using 3–5 livers and PBMC from two donors.

enrichment of ILC2s in severely fibrotic/cirrhotic livers was due to contamination by ILCs from peripheral blood.

IL-33 and TSLP are produced in the liver and triggered by TLR3 challenge

In other organs, including blood, tonsil, skin, and airways [8, 13, 15], ILC2s are activated by cytokines, including IL-25, IL-33, and TSLP. In order to identify the potential trigger of ILC2 activity in fibrotic livers, we assessed the expression ofIL25, IL33, andTSLP in whole liver lysates. Only IL33 expression could be detected (Fig. 5A) and was confirmed on the protein level by immunohistochemistry showing that IL-33 was produced in both nonfibrotic and fibrotic livers (Fig. 5B). A quantification of these stainings showed no significant difference in the expression levels of IL-33 in nonfibrotic compared to fibrotic liver tissue (data not shown). Additionally, no significant difference in IL-33 expression between fibrotic and nonfibrotic areas of fibrosis-affected livers was observed (data not shown). However, since in severe fibrosis much of the tissue is replaced by ECM proteins, we wanted to study the expression of IL33 in specific liver cell subsets under the influence of a fibrosis-relevant stimulus. IL-33 is produced and released upon cellular stress caused by, for example, activation by virally encoded TLR-ligands [8]. Hence, as a model stimuli we chose poly(I:C), a TLR3 agonist, that to some degree mimics the effects of a hepatitis C infection [37], a known cause of liver fibrosis.

As IL-33 production by hepatocytes and HSCs has been shown [37, 38], we assessedIL33 expression upon poly(I:C) stimulation in primary hepatocytes from nonfibrotic livers and the hepatic stellate cell line LX2 [39]. Indeed, poly(I:C)-stimulation led to an increase in IL33 expression in both cell types (Fig. 5C and D). Furthermore, IL-33 production has been reported in lung macrophages [40]. We therefore sorted CD14⁺CD163⁺ Kupf- fer cells from nonfibrotic livers and exposed them to poly(I:C).

WhereasIL33 was not detected in resting Kupffer cells, poly(I:C) stimulation lead to a highly variable (4–96-fold), yet consistent increase inIL33 expression (Fig. 5E). Poly(I:C) also boosted TSLP expression in all analyzed cell types (Fig. 5C, D, and E), but its expression was possibly too low to be detected in the whole liver lysates (Fig. 5A).

Intrahepatic ILC2s produce IL-13 in response to IL-33 and TSLP

We next investigated the effects of IL-33 and TSLP on intrahepatic ILC2s. To assess the function of this small cell population with lim- ited availability of primary human tissues, we used expanded cell lines of primary intrahepatic ILC2s [13]. After 10–14 days of cul- ture, intrahepatic ILC2s maintained their surface phenotype with expression of CRTH2, CD127 (albeit downregulated), CD161, and CD117. NK/ILC3-associated markers NKp44 and CD56 were absent or expressed by a minority of the expanded ILC2s, respec- tively (Fig. 6A). Upon PMA/ionomycin stimulation, ILC2s showed

the archetypical cytokine profile, producing IL-13 and TNFα (Fig. 6B). Interestingly, most ILC2s also produced IL-2 and GM- CSF, the latter being implicated in Kupffer cell maturation and function [41]. Surprisingly, most expanded ILC2s produced IFN-γ and a small fraction of cells produced some IL-22. As observed in other organs [8, 13], intrahepatic ILC2s responded to IL-33 or TSLP with substantial IL-13 production and the combination exerted a synergistic effect (Fig. 6C). Although PMA/ionomycin potently triggered IFN-γ production, only low levels of IFN-γ were secreted upon the more physiological IL-33 and TSLP stimula- tion. Of note, intrahepatic ILC2s did not produce TGF-β1, nei- ther at resting nor IL-33/TSLP activated conditions (data not shown). Additionally, we validated the cytokine profile of intra- hepatic ILC2s by assessing production of IL-13, IL-5, and IL-4 from freshly isolated, noncultured, intrahepatic ILC2s (on average 3000 ILC2s/analysis) in response to PMA/ionomycin stimulation (Fig.

6D). IL-13, IL-5, and IL-4 was detected from all analyses of ILC2s, whereas only one out of three ILC3 samples showed production of these cytokines at much lower levels.

Discussion

In this study, we investigated the distribution, phenotype, and function of the hitherto uncharacterized ILC compartment in the human adult and fetal liver. In the adult liver, we identified all described populations of human Lin⁻ CD127⁺ ILCs, including ILC1s, ILC2s, NKp44⁻ ILC3s as well as NKp44⁺ ILC3s, speak- ing for a possible role of ILCs in liver homeostasis and disease.

The phenotypes of these ILC populations largely corresponded to those of previously characterized ILC1-3s in human mucosal tis- sues such as the gut and tonsil [5, 13]. Within the Tbet⁺ ILC1s a small fraction of cell expressed the transcription factor Eomes.

This transcription factor profile would be in line with intraepithe- lial ILC1s or NK cells [6]. If these intrahepatic Eomes⁺Tbet⁺cells belong to the bona-fide ILC1s or if they represent a contamination by other cells, due to the still poor definition of ILC1s by surface markers, remains unclear.

Interestingly, there were large differences in the distribution of the ILC subsets in the various organs. The gut and tonsil are dominated by NKp44⁺ILC3s, shown to have a protective role in mucosal barriers via production of IL-22 [42]. The liver however, harboured few NKp44⁺ILC3s, and was instead home to a large population of NKp44⁻ILC3s. These cells have been shown to be precursors of, or show plasticity toward all other ILC populations, including NKp44⁺ILC3s, ILC1s, and ILC2s [5, 15, 43], therefore possibly representing a na¨ıve ILC3 population. This argument is supported by the observation that in contrast to the adult liver, the fetal liver harbored almost exclusively NKp44⁻ILC3s, with ILC1s, ILC2s, and NKp44⁺ILC3s being detectable only at later gestational age. This indicates that the signals required for development, dif- ferentiation, or recruitment of ILC1s, ILC2s, and NKp44⁺ILC3s in the liver appears later in life. Of note, the NKp44⁻ILC3s in the fetal liver were different from the corresponding population in the adult since fetal ILC3s expressed NRP1. Transcripts ofNrp1 have

1288

Figure 5. IL-33 and TSLP production in the human liver. (A) IL33, TSLP and IL25 expression in total liver lysates was measured by qPCR. Results are displayed as mean± SD; n = 23; ND: not detectable. (B) Immunohistochemistry staining for IL-33. Representative stainings of nonfibrotic (n

= 6) and cirrhotic liver tissue (n = 6) including isotype controls for the same samples. Magnification 25 ×, scale bar 50 μm. (C–E) IL33 and TSLP mRNA expression was induced by poly(I:C) stimulation of different intrahepatic cell types, and measured by qPCR at the indicated time-points (ns: nonstimulated). (C) Primary hepatocytes, (D) HSCs (LX2 cells), and (E) primary Kupffer cells. Results are displayed as fold change compared with that of nonstimulated cells at the corresponding time-points, normalized to b-actin expression. Data shown as mean± SD of at least three independent experiments, each performed with 1–2 liver donors (C and E). Comparisons were made with two-tailed Student’s paired t-test^*pࣘ 0.05,^**pࣘ 0.01,^***pࣘ 0.001.

previously been shown to be expressed by ILC3s in mouse and humans [30, 44] and NRP1 was demonstrated to be a marker of cells with LTi activity [34]. Our data suggest that fetal, but not adult liver NKp44⁻ILC3s are LTi cells contributing to lymphoid organogenesis.

Recently, it was shown that mouse ILCs are tissue-resident cells [45] and a population of tissue-resident human NK cells marked by CD49a expression was identified in human liver [29].

Intrahepatic ILCs, although showing only low expression of CD49a on a small fraction of cells, expressed the activation, and tissue- residency marker CD69 and did not express CD62L, indicating a tissue-resident phenotype. CD69 interferes with tissue egress of lymphocytes [46] and CD62L is a marker of lymphoid tissue hom- ing, which is generally absent on lymphocytes within tissues [47].

Interestingly, CD69 was only lowly expressed on a minority of fetal intrahepatic ILCs, indicating that fetal ILCs might not be tissue- resident. It is therefore possible that the human fetal liver is a source of LTi cells seeding other organs during development.

Our study of disease-affected fibrotic livers points toward a role for ILC2s in human liver fibrosis. We identified a direct correla- tion between the frequency of intrahepatic ILC2s and the degree of liver fibrosis. Importantly, the frequency of the adaptive coun- terpart of ILC2s, CRTH2⁺ T cells, was not altered in liver fibro- sis, suggesting that type 2 skewing of the lymphocyte compart- ment in liver fibrosis might be an innate, rather than an adaptive phenomenon.

Recently, it was shown that severe-combined immunodefi- ciency patients are lacking ILCs before and after nonmyeloablative hematopoietic stem cell transplantation [48]. This lack of ILCs had no obvious impact on the susceptibility to infections or inflam- mation. Even though these data are pointing toward a redundant function of ILCs under normal conditions it might be that ILCs play a crucial role in specific pathogenic settings, for instance infection with helminths or certain bacteria that are encountered rarely in countries with a high hygienic standard. Furthermore, studies that systematically investigate the very long-term effects of absence of ILCs or the kinetics of infections in these patients would be needed to clarify the role of ILCs for health and disease. Our observation that only the ILC2 but not the Th2 frequency is altered in liver fibrosis points toward a unique function of ILC2s in this disease setting.

In this study,IL33 mRNA and IL-33 protein, a known activator of ILC2s, were found to be expressed in both healthy and cirrhotic livers. In humans, IL-33 expression has previously been shown in hepatocytes [37] and HSCs [38]. Hence, we set out to deter- mine the potential trigger for IL-33 secretion in these particular hepatic cell types. We could identify primary hepatocytes, HSCs, and Kupffer cells as cellular sources ofIL-33 and TSLP following TLR3-activation, as a model for hepatitis C infection. Production of these cytokines could potentially cause the accumulation of ILC2s in fibrotic livers. Although we could not detect TSLP in liver tissue lysates, the induction of TSLP after poly(I:C) stimulation

Figure 6. Cytokine production by intrahepatic ILC2s. (A) The surface phenotype of in vitro-expanded intrahepatic ILC2s (dark gray) was measured by flow cytometry. NK cells (light gray) are shown for comparison. (B) Intracellular cytokine staining of expanded intrahepatic ILC2s, NK cells, and freshly isolated T cells from human tonsil after 6 h of PMA/ionomycin stimulation. Plots are representative of three independent experiments with one donor each. (C) Intrahepatic ILC2s were expanded for 3 days with IL-33 or TSLP, either alone or in combination. IL-13 and IFN-γ in supernatants was measured by ELISA. Data presented as mean± SEM, n = 6 cultures. Wilcoxon signed-rank test^*pࣘ 0.05. (D) Freshly isolated intrahepatic ILC2s and ILC3s were stimulated for 6 h with PMA/ionomycin. The levels of IL-13, IL-5, and IL-4 in supernatants were measured by ELISA (IL-13) or luminex (IL-5, IL-4). Data presented as mean± SEM, n = 3 cultures.

in the different liver cell types could indicate a role for TSLP in human liver fibrosis via ILC2 activation. Indeed, there are reports suggesting a profibrotic effect of TSLP [49–51]. In addition to cytokine-induced proliferation of ILC2s in liver fibrosis, we are also raising the possibility that the accumulation of ILC2s might be the result of plasticity among ILCs. The observed simultaneous reduction of NKp44-ILC3s in fibrotic livers could indicate a plastic change into ILC2s. This hypothesis is also supported by the overall frequency of ILCs that remains unchanged in healthy and fibrotic livers.

IL-33 and TSLP are known to induce IL-13 secretion from ILC2s in other tissues [8, 13, 15]. We confirmed this effect in intrahepatic ILC2s, where stimulation with both cytokines synergistically induced high levels of IL-13. This finding is rele- vant since IL-13 has been suggested to be a profibrotic cytokine in liver fibrosis [23, 25, 26]. In one of these previous studies [26], IL-33 overexpression in liver injury was accompanied by an accu- mulation of ILC2s and increased hepatic collagen, which was attributed to ILC2-derived IL-13 acting on HSCs. The increased frequency of ILC2s in fibrotic human liver supports the find- ings in mice, indicating the possibility of a similar mechanism in humans.

In conclusion, we provide a first detailed characterization of intrahepatic ILCs in nonaffected and fibrotic adult, as well as in fetal liver. Adult intrahepatic ILCs display markers of activation and tissue-residency whereas the exclusive expression of NRP1 and low expression of CD69 on fetal cells indicate LTi functionality and a nontissue-resident phenotype in fetal liver. Furthermore, we show that human intrahepatic ILC2s are increased in fibrotic livers and make up a previously unrecognized source for IL-13, which might act as an important factor in human liver fibrosis.

Hence, these findings could have important clinical implications, as targeting ILC2s may present a novel therapeutic strategy when considering treatments options for liver fibrosis.

Material and methods

Collection of patient material

Nonfibrotic liver material was freshly obtained from tumor resec- tion surgeries mostly for colorectal carcinoma metastasis, cholan- giocellular carcinoma, or hepatocellular carcinoma. Samples were taken from nontumor affected areas. Severely fibrotic or cirrhotic

1290

Table 1. Antibodies for flow cytometry

Fluorochrome Antigen Clone Company

AF647 CD49a TS2/7 Biolegend

AF700 CD16 3G8

CD45 HI30

CD3 UCHT1 BD

IFNγ B27

APC CD103 B-Ly7 eBioscience

IL-13 JES10-5A2 Biolegend

APC-Cy7 CD62L DREG-56

CXCR3 G025H7

Biotin NKp46 9E2

CD14 63D3

CD1a HI149

CD123 6H6

CD19 HIB19

CD34 581

TCRα/β IP26 TCRγ/δ B1

CD94 DX22

FcER1 AER-37 (CRA-1)eBioscience

BDCA2 AC144 Miltenyi

BV421 IL-2 5344-111 BD

BV510 streptavidin Biolegend

BV605 CD161 HP-3G10

CD163 GHI/61

BV650 streptavidin

CD45 HI30

BV711 CD56 HCD56

BV785 HLA-DR L243

CD3 OKT3

efluor 660 GATA3 TWAJ eBioscience

AHR FF3399

Eomes WD1928

FITC FcεR1 AER-37 (CRA-1)Biolegend

CD34 581

CD94 DX22

CD123 6H6

CD1a HI149

TCRα/β IP26 TCRγ/δ B1

CD19 4G7 BD

BDCA2 AC144 Miltenyi

NRP1 AD5-17F6

CD11c KB90 Dako

CD14 T ¨UK4

Pacific Blue Helios 22F6 Biolegend

PE NKp44 Z231 Beckman

Coulter

NKG2A Z199

IL-22 142928 R&D Systems RORγ(t) AFKJS-9 eBiosciences

CD14 61D3

HLA-DR L234

PE-CF594 NKG2D 1D11 BD

Tbet O4-46

GM-CSF BVD2-21C11

(Continued)

Table 1. Continued

Fluorochrome Antigen Clone Company

CD69 FN50

PE-Cy5 NKp44 Z231 Beckman

Coulter

CD3 UCHT1

PE-Cy5.5 CD117 104D2D1

PE-Cy7 CD127 R34.34

TNF MAb11 BD

V450 CRTH2 BM16

V500 CD45 HI30

CD16 3G8

livers were obtained from liver transplantations mainly due to primary sclerosing cholangitis, but also due to alcoholic cirrho- sis, nonalcoholic steatohepatitis, cryptogenic cirrhosis, angiomy- olipoma, leiomyosarcoma, Adams Olivier Syndrome, or alpha-1 antitrypsin deficiency. Fibrosis scores were judged according to the Batts/Ludwig classification. Additionally, completely healthy livers of deceased organ donors were obtained. Gut mucosa was obtained from intestinal resection surgeries due to colorectal car- cinoma and was taken distally from the tumor. Tonsils were received from adult tonsillectomies due to obstructive sleep disor- ders. All surgeries were performed at Karolinska University Hos- pital, Huddinge. Fetal livers from the first trimester (gestational weeks 6–10) were obtained from the Department of Neurobiol- ogy, Care Sciences and Society, Karolinska Institutet, Huddinge.

In addition, frozen fetal liver cells (gestational week 15–20) were obtained from the Academic Medical Center, Amsterdam, the Netherlands.

Blood from healthy donors was obtained from the Blood Bank at Karolinska University Hospital, Huddinge. For all tissue and blood collection, permission was obtained from the regional eth- ical board at Karolinska Institutet or from the Medical Ethical Committee of the Academic Medical Center of the University of Amsterdam, the Netherlands. All patients gave their informed consents.

Liver tissue from the deceased organ donors was included in the study according to the regulations for the organ transplanta- tion law of Sweden (1995:831), that is, the donor’s prior written declaration was followed as well as the written and informed con- sent from next of kin.

Immunohistochemistry (IHC) analysis of IL-33

Five micrometers tissue sections from frozen samples were placed on SuperFrost Ultra Plus slides (Histolab) and stored at –80˚C until staining. Next, sections were air-dried 10 min and fixed in 4% paraformaldehyde (Sigma Aldrich) for 20 min on ice. Sec- tions were incubated with Bloxall (Vector Laboratories) for 10 min and then in Innovex background Buster (Innovex Biosciences) for 20 min at room temperature. Samples were incubated with pri- mary antibody (mouse anti-human IL-33 from Enzo Life Sciences,

clone; Nessy-1), at a concentration of 2 μg/mL, overnight in a moisture chamber at 4˚C. Subsequently, sections were incubated with ImmPRESS mouse secondary antibody (Vector Laboratories) for 30 min at room temperature and specific staining was detected by incubation with ImmPACT DAB (Vector Laboratories). Finally, tissue sections were counterstained with Hematoxylin (Histolab) and mounted with Kaisers´s glycerol gelatine (Merck Millipore).

Positive staining was visualized by light microscopy (Leica DM 4000B).

IHC-stained specimens were analyzed in a blinded fashion by acquired computerized image analysis (ACIA). Twenty consecu- tive photomicrographs were taken from each tissue section, which corresponded to the majority of the sample area. The results were analyzed both as ACIA-values, calculated as the percentage of the area that stained positively multiplied by the mean intensity of positive staining, and as frequency of expression, calculated as percentage of positively stained area divided by the total sample area.

Cell isolation

Sections of liver tissue or gut mucosa were cut into small pieces and digested in enzyme buffer (1 mM HEPES in RPMI plus DNase and collagenase II) (both 0.25 mg/mL, Sigma) at 37°C for 30–45 min under magnetic stirring. Digestion was stopped by adding IMDM (Life Technologies) supplemented with 10% fetal calf serum (FCS; Sigma). The cell suspension was filtered through a 100μm cell strainer and mononuclear cells were enriched using lymphoprep (Axis Shield).

Alternatively, intrahepatic lymphocytes were isolated using a three-step collagenase perfusion protocol on livers obtained after partial or complete hepatectomy. Livers were perfused with warm (37°C) Hank’s Buffered Salt Solution (Cambrex invitro) supplemented with Ethylene Glycol Tetraacetic Acid (Sigma);

pure Hank’s Buffered Salt Solution or Eagle’s Minimum Essen- tial Medium with Earle’s salts (Cambrex invitro) supplemented with Collagenase XI (Sigma). After digestion, the tissue was trans- ferred into 4°C cold Eagle’s Minimum Essential Medium (EMEM), cut in small pieces and filtered through sterile gauze. Collagenases were removed by centrifugation and cell pellets were washed twice in cold EMEM. Hepatocytes were separated by three con- secutive centrifugations at 50 g for 5 min at 4°C. The lym- phocyte fraction was collected in the supernatant after the first centrifugation step and mononuclear cells were isolated using lymphoprep.

Fetal livers were placed immediately into RPMI medium con- taining 10% FCS, penicillin, streptomycin, and LZ glutamine (R10 medium) for transport. Subsequently, the tissue was cut into small pieces and incubated in 3–4 mL RPMI containing collagenase II (Sigma; 0.25 mg/mL) and DNAse (Roche; 0.2 mg/mL) under constant magnetic stirring at 37°C for 30–45 min. Cell suspen- sions were filtered through 5 mL filter top tubes (BD Falcon).

Cells were washed twice with 4 mL R10. Alternatively, single cell suspensions were prepared from whole fetal liver tissues as previ- ously described [52].

Isolated cells were either directly used or frozen in FCS con- taining 10% DMSO (Sigma), stored in liquid nitrogen for later flow cytometry analysis.

For isolation of lymphocytes from whole tonsils the tissue was minced through a 100μm cell strainer. The cell suspension was filtered through a 40μm cell strainer and mononuclear cells were separated using lymphoprep (Axis Shield).

PBMCs were isolated using lymphoprep (Axis Shield).

Flow cytometry analysis of surface and intracellular markers and cell sorting

Cells in single-cell suspensions were analyzed for surface pro- tein expression. In addition, for transcription factor expres- sion by intracellular staining the Foxp3 staining kit (eBio- science) was used. Additionally, the LIVE/DEAD^R Fixable Green Dead Cell Stain Kit (Life Technologies) was used. For pheno- typical analysis, cells were acquired on a LSR Fortessa (BD Bioscience) using FACS Diva software Version 8 (BD Bio- sciences). Flow cytometry data were analyzed with FlowJo Software Version 9.6 (TreeStar).

For cell sorting, lymphocytes were predepleted of CD3⁺ and CD19⁺ cells using FITC-labeled primary antibodies, anti-FITC microbeads, and MACS depletion columns (Miltenyi Biotech). Cell sorting was performed on an Aria III with the FACS Diva 7 soft- ware.

For Intracellular cytokine detection, ILC2s were stimulated with PMA (10 ng/mL) (Sigma) and ionomycin (500 nM) (Life Technologies) for 6 h. After 2 h, Golgiplug (BD Biosciences) was added. Intracellular staining of cytokines was performed using Cytofix/Cytoperm (BD Biosciences). All flow cytometry antibod- ies used are stated in Table 1.

ILC culture

Sort-purified primary ILC2s from human liver, defined as Lin⁻CD45⁺CD127⁺CRTH2⁺NKp44⁻CD161⁺, were cultured in Yssel’s medium (in-house) containing 1% normal human serum (NHS, Invitrogen) and 500 U/mL IL-2 (Novartis). Irradiated feeder cells consisting of pooled PBMC from three donors (30 Gy) and JY-cells (60 Gy) were added together with 1μg/mL phytohaemag- glutinin (Oxoid, Fisher Scientific) as previously described [13].

Medium was refreshed every three days by replacing half of the volume with 1000 U/mL IL-2 in Yssel’s media 1% NHS. After 14 d of culture, ILC2s were rested overnight in 1 U/mL IL-2 in Yssel’s media 1% NHS and cultured for 3 d in Yssel’s media 1% NHS containing combinations of IL-2 (1 U/mL) (Novartis), IL-33 (50 ng/mL), and TSLP (50 ng/mL) (both from Peprotech).

Culturing and stimulation of Kupffer cells, HSCs, and hepatocytes

Kupffer cells were sorted from liver single-cell suspensions as Lin⁻ (CD3⁻CD34⁻CD94⁻TCRα/β⁻TCRγ/δ⁻CD19⁻BDCA2⁻)

1292

Table 2. Primer sequences used in the study

mRNA Forward primer Reverse primer

ACTB CAC CAT TGG CAA TGA GCG GTT C AGG TCT TTG CGG ATG TCC ACG T

IL33 GCC TGT CAA CAG CAG TCT ACT G TGT GCT TAG AGA AGC AAG ATA CTC

CD45⁺CD14⁺CD163⁺ cells. Cells were seeded in RPMI supple- mented with 10% FCS in 96-well plates (BD Biosciences) and cultured for 12 h with or without poly(I:C) (50μg/mL; Sigma) before lysing cells in RLT buffer (Qiagen).

LX-2 hepatic stellate cells (a kind gift from Prof. Scott Fried- man, Icahn School of Medicine at Mount Sinai, NY, USA) were cultured in DMEM (Thermo Scientific) containing 5% FCS. For stimulation experiments, 100.000 cells per well were seeded 24- well plates (BD Biosciences) and rested for 36 h in serum-free DMEM before stimulation with poly(I:C). After 6, 12, and 24h, cells were lysed in RLT buffer.

For culturing primary hepatocytes, six-well plates were coated for 30 min with collagen isolated from rat tails (in house). Hepato- cytes were plated at 1.5× 10⁶cells per well in Williams E medium (Sigma), containing 1% Glutamine (Sigma), 1 M HEPES, 10⁻⁸M insulin, 50 mg/mL gentamycin (Lonza), 250μg/mL amphotericin B and 5% FCS. After 2 h, medium was changed to WE medium without FCS. Cells were rested overnight, washed two times with 5 mL PBS and stimulated with 50μg/mL poly(I:C) in 2 mL WE medium. After 3, 6, 12, and 24 h cells were lysed in RLT buffer.

ELISA and multi-plex Immunoassay

For determining cytokine concentrations in supernatants, the PeliKine compact human IL-13 or IFN-γ kit (both from Sanquin) and the human TGF-β1 DuoSet ELISA kit (R&D systems) were used according to the manufacturer’s instructions. Plates were read on an iMark Microplate Absorbance Reader (Biorad) and analyzed using Microplate Manager V6.3 software. For measurement of IL- 13, IL-5, and IL-4 in supernatants of freshly isolated intrahepatic ILCs a customized 6-plex Human Magnetic Luminex Assay (R&D Systems) was used according to manufacturer instructions and measurements were taken on a MagPix Instrument (BioRad). For display all samples containing nondetectable concentrations were set to half of the lowest detected concentration value.

Quantitative real-time PCR

RNA isolation was performed using the RNeasy Mini Kit (Qia- gen) according to the manufacturer’s instructions. RNA concen- tration was measured with a Nanodrop 1000 (Thermo Scientific).

Reverse transcription was performed with the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems) according to the manufacturer’s protocol. The qPCR reaction was performed in a 7500 Real time PCR system (Applied Biosystems) using the Taq- Man universal mastermix II with the TaqMan Gene Expression

Assays Hs00263639 m1 forTSLP total and Hs03044841 m1 for IL25 or SYBR^R Select Master Mix (all from Applied Biosystems) with primers forIL33, ACTB (Table 2). Relative gene expression was analyzed with the Ct method, providing a fold- change of expression normalized to the house-keeping geneACTB (Ct) as well as to the nonstimulated control (Ct). To be able to assess fold-changes also in cases where the gene expression was undetectable in the nonstimulated condition, the nonstimulated condition was set to Ct= 45 that was the maximum number of PCR cycles.

Statistical analysis

For statistical analysis Prism 6 from GraphPad Prism was used.

Differences between groups were calculated with Student’st-test or Wilcoxon signed-rank test. Correlation analysis was performed using Pearson’s correlation coefficient. D’Agostino and Pearson omnibus normality testing was used for evaluating normality in data distribution. All statistical analysis was confirmed by Statsoft Scandinavia AB (Uppsala, Sweden).

Acknowledgments: We thank Prof. Hergen Spits, Kees Weijer, and Esther Siteur van Rijnstra from the Academic Medical Center, Amsterdam, for their help in fetal liver cell acquisition and tissue processing. We also thank Dr. Iyadh Douagi, Karolinska Institutet, Stockholm, for support with cell sorting and Prof. Scott Friedman, Icahn School of Medicine at Mount Sinai, NY, USA for providing the LX2 cell line.

Grant support

J. Mj¨osberg and M. Forkel are supported by grants from the Swedish Research Council (524-2010-6770, 524-2012-2682, and 521-2013-2791), the Swedish Cancer Society (120108 and 130396), the Swedish Society for Medical Research and the Foun- dation for Strategic Research (ICA12-0023).

Authors’ contributions

M.F. designed the study, performed experiments, analyzed data, and wrote the manuscript; L.B., A.C. performed immunohisto- chemistry stainings and contributed to manuscript writing; E.K.,

E.S., J.S.E., and M.M. performed experiments; Ja.M. and M.F.T.

interpreted data and contributed to manuscript writing; H.G.L.

contributed to manuscript writing; N.B. and M.N. contributed to material collection and manuscript writing; A.B., M.W., U.L., D.F., C.J., and E.E. provided clinical samples and contributed to manuscript writing; Je.M. designed the study, performed experi- ments, analyzed data, and wrote the manuscript.

Conflict of interests: The authors declare no financial or com- mercial conflict of interest.

References

1 Spits, H., Artis, D., Colonna, M., Diefenbach, A., Di Santo, J. P., Eberl, G., Koyasu, S. et al., Innate lymphoid cells—a proposal for uniform nomen- clature. Nat. Rev. Immunol. 2013. 13: 145–149.

2 Crellin, N. K., Trifari, S., Kaplan, C. D., Cupedo, T. and Spits, H., Human NKp44+IL-22+ cells and LTi-like cells constitute a stable RORC+ lineage distinct from conventional natural killer cells. J. Exp. Med. 2010. 207: 281–

290.

3 Eberl, G., Marmon, S., Sunshine, M. J., Rennert, P. D., Choi, Y.

and Littman, D. R., An essential function for the nuclear receptor RORgamma(t) in the generation of fetal lymphoid tissue inducer cells.

Nat. Immunol. 2004. 5: 64–73.

4 Hoorweg, K., Peters, C. P., Cornelissen, F., Aparicio-Domingo, P., Papazian, N., Kazemier, G., Mjosberg, J. M. et al., Functional differences between human NKp44(-) and NKp44(+) RORC(+) innate lymphoid cells.

Front Immunol. 2012. 3: 72.

5 Bernink, J. H., Peters, C. P., Munneke, M., te Velde, A. A., Meijer, S. L., Weijer, K., Hreggvidsdottir, H. S. et al., Human type 1 innate lymphoid cells accumulate in inflamed mucosal tissues. Nat. Immunol. 2013. 14:

221–229.

6 Fuchs, A., Vermi, W., Lee, J. S., Lonardi, S., Gilfillan, S., Newberry, R. D., Cella, M. et al., Intraepithelial type 1 innate lymphoid cells are a unique subset of IL-12- and IL-15-responsive IFN-gamma-producing cells. Immu- nity 2013. 38: 769–781.

7 Hoyler, T., Klose, C. S., Souabni, A., Turqueti-Neves, A., Pfeifer, D., Rawl- ins, E. L., Voehringer, D. et al., The transcription factor GATA-3 controls cell fate and maintenance of type 2 innate lymphoid cells. Immunity 2012.

37: 634–648.

8 Mjosberg, J. and Spits, H., Type 2 innate lymphoid cells-new members of the "type 2 franchise" that mediate allergic airway inflammation. Eur. J.

Immunol. 2012. 42: 1093–1096.

9 Wong, S. H., Walker, J. A., Jolin, H. E., Drynan, L. F., Hams, E., Camelo, A., Barlow, J. L. et al., Transcription factor RORalpha is critical for nuocyte development. Nat. Immunol. 2012. 13: 229–236.

10 Moro, K., Yamada, T., Tanabe, M., Takeuchi, T., Ikawa, T., Kawamoto, H., Furusawa, J. et al., Innate production of T(H)2 cytokines by adipose tissue-associated c-Kit(+)Sca-1(+) lymphoid cells. Nature 2010. 463: 540–

544.

11 Neill, D. R., Wong, S. H., Bellosi, A., Flynn, R. J., Daly, M., Langford, T. K., Bucks, C. et al., Nuocytes represent a new innate effector leukocyte that mediates type-2 immunity. Nature 2010. 464: 1367–1370.

12 Price, A. E., Liang, H. E., Sullivan, B. M., Reinhardt, R. L., Eisley, C. J., Erle, D. J. and Locksley, R. M., Systemically dispersed innate IL-13-expressing cells in type 2 immunity. Proc. Natl. Acad. Sci. USA 2010. 107: 11489–11494.

13 Mjosberg, J. M., Trifari, S., Crellin, N. K., Peters, C. P., van Drunen, C. M., Piet, B., Fokkens, W. J. et al., Human IL-25- and IL-33-responsive type 2 innate lymphoid cells are defined by expression of CRTH2 and CD161.

Nat. Immunol. 2011. 12: 1055–1062.

14 Kim, B. S., Siracusa, M. C., Saenz, S. A., Noti, M., Monticelli, L. A., Sonnen- berg, G. F., Hepworth, M. R. et al., TSLP elicits IL-33-independent innate lymphoid cell responses to promote skin inflammation. Sci. Transl. Med.

2013. 5: 170ra116.

15 Teunissen, M. B., Munneke, J. M., Bernink, J. H., Spuls, P. I., Res, P. C., Te Velde, A., Cheuk, S. et al., Composition of innate lymphoid cell subsets in the human skin: enrichment of NCR ILC3 in lesional skin and blood of psoriasis patients. J. Invest. Dermatol. 2014. 134: 2351–2360.

16 Villanova, F., Flutter, B., Tosi, I., Grys, K., Sreeneebus, H., Perera, G. K., Chapman, A. et al., Characterization of innate lymphoid cells in human skin and blood demonstrates increase of NKp44+ ILC3 in psoriasis.

J. Invest. Dermatol. 2014. 134: 984–991.

17 Bataller, R. and Brenner, D. A., Liver fibrosis. J. Clin. Invest. 2005. 115:

209–218.

18 Xu, R., Zhang, Z. and Wang, F. S., Liver fibrosis: mechanisms of immune- mediated liver injury. Cell Mol. Immunol. 2012. 9: 296–301.

19 Senoo, H. H. R., Nagai, Y., and Wake, K., Stellate cells (vitamin A-storing cells) are the primary site of collagen synthesis in non-parenchymal cells in the liver. Biomed. Res. 1984. 5: 451–458.

20 Blomhoff, R. and Wake, K., Perisinusoidal stellate cells of the liver: impor- tant roles in retinol metabolism and fibrosis. FASEB J. 1991. 5: 271–277.

21 Lee, U. E. and Friedman, S. L., Mechanisms of hepatic fibrogenesis. Best Pract. Res. Clin. Gastroenterol. 2011. 25: 195–206.

22 Wynn, T. A., IL-13 effector functions. Annu. Rev. Immunol. 2003. 21: 425–

456.

23 Long, X., Chen, Q., Zhao, J., Rafaels, N., Mathias, P., Liang, H., Potee, J.

et al., An IL-13 promoter polymorphism associated with liver fibrosis in patients with Schistosoma japonicum. PLoS One 2015. 10: e0135360.

24 Hams, E., Armstrong, M. E., Barlow, J. L., Saunders, S. P., Schwartz, C., Cooke, G., Fahy, R. J. et al., IL-25 and type 2 innate lymphoid cells induce pulmonary fibrosis. Proc. Natl. Acad. Sci. USA 2014. 111: 367–372.

25 Fallon, P. G., Richardson, E. J., McKenzie, G. J. and McKenzie, A. N., Schistosome infection of transgenic mice defines distinct and contrast- ing pathogenic roles for IL-4 and IL-13: IL-13 is a profibrotic agent.

J. Immunol. 2000. 164: 2585–2591.

26 McHedlidze, T., Waldner, M., Zopf, S., Walker, J., Rankin, A. L., Schuch- mann, M., Voehringer, D. et al., Interleukin-33-dependent innate lym- phoid cells mediate hepatic fibrosis. Immunity 2013. 39: 357–371.

27 Liedtke, C., Luedde, T., Sauerbruch, T., Scholten, D., Streetz, K., Tacke, F., Tolba, R. et al., Experimental liver fibrosis research: update on animal models, legal issues and translational aspects. Fibrogenesis Tissue Repair 2013. 6: 19.

28 Moroso, V., Famili, F., Papazian, N., Cupedo, T., van der Laan, L. J., Kazemier, G., Metselaar, H. J. et al., NK cells can generate from pre- cursors in the adult human liver. Eur. J. Immunol. 2011. 41: 3340–3350.

29 Marquardt, N., Beziat, V., Nystrom, S., Hengst, J., Ivarsson, M. A., Kekalainen, E., Johansson, H. et al., Cutting edge: identification and char- acterization of human intrahepatic CD49a+ NK cells. J. Immunol. 2015.

194: 2467–2471.

30 Bjorklund, A. K., Forkel, M., Picelli, S., Konya, V., Theorell, J., Friberg, D., Sandberg, R. et al., The heterogeneity of human CD127(+) innate lymphoid cells revealed by single-cell RNA sequencing. Nat. Immunol.

2016. 17: 451–460.

1294

31 Marquardt, N., Ivarsson, M. A., Sundstrom, E., Akesson, E., Mar- tini, E., Eidsmo, L., Mjosberg, J. et al., Fetal CD103+ IL-17-producing group 3 innate lymphoid cells represent the dominant lympho- cyte subset in human amniotic fluid. J. Immunol. 2016. 197: 3069–

3075.

32 Ahn, Y. O., Blazar, B. R., Miller, J. S. and Verneris, M. R., Lineage relation- ships of human interleukin-22-producing CD56+ RORgammat+ innate lymphoid cells and conventional natural killer cells. Blood 2013. 121:

2234–2243.

33 Hepworth, M. R., Fung, T. C., Masur, S. H., Kelsen, J. R., McConnell, F. M., Dubrot, J., Withers, D. R. et al., Immune tolerance. Group 3 innate lym- phoid cells mediate intestinal selection of commensal bacteria-specific CD4(+) T cells. Science 2015. 348: 1031–1035.

34 Shikhagaie, M. M., Bj ¨orklund, ˚A. K., Mj ¨osberg, J., Erjef ¨alt, J. S., Cornelis- sen, A. S., Ros, X. R., Bal, S. M. et al., Neuropilin-1 is expressed on lym- phoid tissue residing LTi-like group 3 innate lymphoid cells and associ- ated with ectopic lymphoid aggregates. Cell Rep 2017. 18: 1761–1773.

35 Zhao, J., Zhang, Z., Luan, Y., Zou, Z., Sun, Y., Li, Y., Jin, L. et al., Patholog- ical functions of interleukin-22 in chronic liver inflammation and fibrosis with hepatitis B virus infection by promoting T helper 17 cell recruitment.

Hepatology 2014. 59: 1331–1342.

36 Nagata, K., Tanaka, K., Ogawa, K., Kemmotsu, K., Imai, T., Yoshie, O., Abe, H. et al., Selective expression of a novel surface molecule by human Th2 cells in vivo. J. Immunol. 1999. 162: 1278–1286.

37 Arshad, M. I., Patrat-Delon, S., Piquet-Pellorce, C., L’Helgoualc’h, A., Rauch, M., Genet, V., Lucas-Clerc, C. et al., Pathogenic mouse hepati- tis virus or poly(I:C) induce IL-33 in hepatocytes in murine models of hepatitis. PLoS One 2013. 8: e74278.

38 Marvie, P., Lisbonne, M., L’Helgoualc’h, A., Rauch, M., Turlin, B., Preisser, L., Bourd-Boittin, K. et al., Interleukin-33 overexpression is associated with liver fibrosis in mice and humans. J. Cell. Mol. Med. 2010. 14: 1726–

1739.

39 Xu, L., Hui, A. Y., Albanis, E., Arthur, M. J., O’Byrne, S. M., Blaner, W.

S., Mukherjee, P. et al., Human hepatic stellate cell lines, LX-1 and LX-2:

new tools for analysis of hepatic fibrosis. Gut 2005. 54: 142–151.

40 Chang, Y. J., Kim, H. Y., Albacker, L. A., Baumgarth, N., McKenzie, A.

N., Smith, D. E., Dekruyff, R. H. et al., Innate lymphoid cells medi- ate influenza-induced airway hyper-reactivity independently of adaptive immunity. Nat. Immunol. 2011. 12: 631–638.

41 Naito, M., Hasegawa, G. and Takahashi, K., Development, differentia- tion, and maturation of Kupffer cells. Microsc. Res. Tech. 1997. 39: 350–364.

42 Satoh-Takayama, N., Vosshenrich, C. A., Lesjean-Pottier, S., Sawa, S., Lochner, M., Rattis, F., Mention, J. J. et al., Microbial flora drives interleukin 22 production in intestinal NKp46+ cells that provide innate mucosal immune defense. Immunity 2008. 29: 958–970.

43 Lim, A. I., Li, Y., Lopez-Lastra, S., Stadhouders, R., Paul, F., Cas- rouge, A., Serafini, N. et al., Systemic human ILC precursors provide a substrate for tissue ILC differentiation. Cell 2017. 168: 1086–1100, e1010.

44 Robinette, M. L., Fuchs, A., Cortez, V. S., Lee, J. S., Wang, Y., Durum, S. K., Gilfillan, S. et al., Transcriptional programs define molecular

characteristics of innate lymphoid cell classes and subsets. Nat. Immunol.

2015. 16: 306–317.

45 Gasteiger, G., Fan, X., Dikiy, S., Lee, S. Y. and Rudensky, A. Y., Tissue residency of innate lymphoid cells in lymphoid and nonlymphoid organs.

Science 2015. 350: 981–985.

46 Mackay, L. K., Braun, A., Macleod, B. L., Collins, N., Tebartz, C., Bedoui, S., Carbone, F. R. et al., Cutting edge: CD69 interference with sphingosine- 1-phosphate receptor function regulates peripheral T cell retention.

J. Immunol. 2015. 194: 2059–2063.

47 Sallusto, F., Lenig, D., Forster, R., Lipp, M. and Lanzavecchia, A., Two subsets of memory T lymphocytes with distinct homing potentials and effector functions. Nature 1999. 401: 708–712.

48 Vely, F., Barlogis, V., Vallentin, B., Neven, B., Piperoglou, C., Ebbo, M., Perchet, T. et al., Evidence of innate lymphoid cell redundancy in humans. Nat. Immunol. 2016. 17: 1291–1299.

49 Oh, M. H., Oh, S. Y., Yu, J., Myers, A. C., Leonard, W. J., Liu, Y. J., Zhu, Z.

et al., IL-13 induces skin fibrosis in atopic dermatitis by thymic stromal lymphopoietin. J. Immunol. 2011. 186: 7232–7242.

50 Datta, A., Alexander, R., Sulikowski, M. G., Nicholson, A. G., Maher, T.

M., Scotton, C. J. and Chambers, R. C., Evidence for a functional thymic stromal lymphopoietin signaling axis in fibrotic lung disease. J. Immunol.

2013. 191: 4867–4879.

51 Christmann, R. B., Mathes, A., Affandi, A. J., Padilla, C., Nazari, B., Bujor, A. M., Stifano, G. et al., Thymic stromal lymphopoietin is up-regulated in the skin of patients with systemic sclerosis and induces profibrotic genes and intracellular signaling that overlap with those induced by interleukin-13 and transforming growth factor beta. Arthritis Rheum. 2013.

65: 1335–1346.

52 van Lent, A. U., Centlivre, M., Nagasawa, M., Karrich, J. J., Pouw, S. M., Weijer, K., Spits, H. et al., In vivo modulation of gene expression by lentiviral transduction in "human immune sys- tem" Rag2-/- gamma c -/- mice. Methods Mol. Biol. 2010. 595: 87–

115.

Abbreviations: ECM: extracellular matrix · gw: gestational week · HSC: hepatic stellate cell· IHC: immunohistochemistry · ILC: Innate lymphoid cell · LTi cells: lymphoid tissue inducer cells · poly(I:C):

polyinosine-polycytidylic acid· TSLP: thymic stromal lymphopoetin

Full correspondence:Dr. Jenny Mj ¨osberg, Karolinska Institutet, Center for Infectious Medicine, F59 Karolinska University Hospital Huddinge, Stockholm 14186, Sweden

e-mail: jenny.mjosberg@ki.se

Received: 17/12/2016 Revised: 8/5/2017 Accepted: 7/6/2017

Accepted article online: 14/6/2017