Respiratory Influenza Virus Infection Induces Memory-Like Liver NK Cells

Respiratory Influenza Virus Infection Induces Memory-like Liver NK Cells in Mice Tingting Li, Jian Wang, Yanshi Wang, Yongyan Chen, Haiming Wei, Rui Sun and Zhigang Tian This information is current as of September 25, 2021. J Immunol published online 28 December 2016 http://www.jimmunol.org/content/early/2016/12/21/jimmun ol.1502186 Downloaded from

Supplementary http://www.jimmunol.org/content/suppl/2016/12/21/jimmunol.150218 Material 6.DCSupplemental

Why The JI? Submit online. http://www.jimmunol.org/

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists

• Fast Publication! 4 weeks from acceptance to publication

*average by guest on September 25, 2021

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 2016 by The American Association of Immunologists, Inc. All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. Published December 28, 2016, doi:10.4049/jimmunol.1502186 The Journal of Immunology

Respiratory Inﬂuenza Virus Infection Induces Memory-like Liver NK Cells in Mice

Tingting Li,* Jian Wang,* Yanshi Wang,* Yongyan Chen,* Haiming Wei,* Rui Sun,*,† and Zhigang Tian*,†

Although NK cells are classified as innate immune cells, recent studies have demonstrated the transformation of NK cells into long-lived memory cells that contribute to secondary immune responses in certain mouse models. However, whether NK cells mount an Ag- specific memory response to acute influenza virus infection has not yet been examined. Here, we show that, consistent with previous studies, lung NK cells play an important role in controlling viral proliferation after primary influenza virus infection. However, although lung NK cells display a memory phenotype at the late stage of infection, these cells do not protect mice against secondary influenza virus infection. Interestingly, liver NK cells from influenza virus–infected mice possess a memory phenotype and protect 2 mice against secondary influenza virus infection. Memory-like liver NK cells display a CD49a+DX5 phenotype, and the adoptive Downloaded from transfer of purified liver CD49a+DX52 NK cells into naive mice followed by viral infection results in protective immunity and decreased viral titer. Moreover, we demonstrate that primary inactivated influenza virus induces memory NK cells residing in the liver of Rag12/2 mice. Collectively, these data suggest that liver CD49a+DX52 NK cells remember encountered Ag from influenza virus after primary infection and are more protective upon subsequent infection. The Journal of Immunology, 2017, 198: 000–000.

mmune memory is a feature of the adaptive immune system. memory NK cells provide effective and specific protection against http://www.jimmunol.org/ However, recent studies have demonstrated that NK cells, these viruses. However, the sensitization of NK cells to influenza A I important components of the innate immune system, also virus does not require NKp46-hemagglutinin (-HA) interactions. mediate Ag-specific recall responses, similar to Tand B cells (1–4). With respect to cytokine-induced memory-like NK cells, Cooper Three major viewpoints of NK cell memory initially arose from et al. have shown that the response to restimulation ex vivo either the studies of NK cell memory to recall to skin contact hyper- with cytokines or through their activating receptors Ly49H or sensitive chemical hapten and noninfectious virus-like particles (1, NK1.1, indicates IFN-g production is more robust than resting NK 2), cytokines (3), and mouse CMV (MCMV) (4). In chemical cells (2, 3). This enhanced response could be detected even 3 wk hapten-induced contact hypersensitivity (CHS) mouse models, after transfer into naive hosts. Using an MCMV infection mouse by guest on September 25, 2021 RAG-deficient mice lacking T and B cells exhibit CHS responses model, Sun et al. (4) demonstrated that NK cells bearing the Ly49H elicited through hapten exposure, and hapten-specific memory- receptor specifically recognize MCMV-infected cells expressing like NK cells typically reside in the liver and display a CD49a+ 2 the viral glycoprotein m157 and eventually form MCMV-specific DX5 phenotype (1, 2, 5). Using noninfectious virus-like parti- immune memory. Together, these studies demonstrate the existence cles, Silke Paust and colleagues showed that liver NK cells re- of NK cell memory. Moreover, memory NK cells mediate an ef- member influenza virus, vesicular stomatitis virus, and HIV, and fective protection against vaccinia virus (6) and HSV-2 (7) in the absence of T and B cells. More recently, robust, durable, Ag-specific NK cell memory can be induced in primates after both SIV (8) *The Chinese Academy of Sciences Key Laboratory of Innate Immunity and Chronic infection and Ad26 vector vaccination. Therefore, there is mounting Disease, School of Life Sciences and Medical Center, University of Science and Technology of China, Hefei, Anhui 230027, China; and †Collaborative Innovation evidence for memory NK cell memory response in different models Center for Diagnosis and Treatment of Infectious Diseases, State Key Laboratory for of viral disease in both mice and primates (9, 10). As more pathogen Diagnosis and Treatment of Infectious Diseases, First Affiliated Hospital, College of Medicine, Zhejiang University, Hangzhou, Zhejiang 310003, China components recognized by NK cell receptors, including activating ORCIDs: 0000-0003-2754-9784 (T.L.); 0000-0002-4127-322X (J.W.). and inhibitory receptors, are characterized (11, 12), additional infectious disease models, especially natural pathogen infection Received for publication October 8, 2016. Accepted for publication November 22, 2016. mouse models, will be required to further analyze the mechanisms This work was supported by Ministry of Science and Technology of China 973 Basic underlying NK cell memory formation during infection. Science Project Grants 2013CB944902 and 2013CB530506 and National Natural Influenza virus infection is a common respiratory disease. The Science Foundation of China Grants 81361120388, 31400783, 91542114, first line of defense against influenza virus infection is mediated 31300727, and 31300753. by the innate immune system, which controls the initial response to Address correspondence and reprint requests to Dr. Zhigang Tian, Dr. Rui Sun, and Dr. Jian Wang, School of Life Sciences and Medical Center, University of Science viral infection and regulation of the adaptive immune response (13, and Technology of China, 443 Huangshan Road, Hefei, Anhui 230027, China. E-mail 14). As an important part of the innate immune system, NK cells addresses: [email protected] (Z.T.), [email protected] (R.S.), and [email protected]. edu.cn (J.W.) have a higher frequency in the lungs than in any of the other The online version of this article contains supplemental material. tissues (15, 16). Most of the data from animal models suggest that NK cells may be directly antiviral, and able to recognize and kill Abbreviations used in this article: AsGM1, asialo-GM1; CHS, contact hypersensitivity; HA, hemagglutinin; Hopx, homeobox-only protein; i.n, intranasal(ly); MCMV, virus-infected cells through interactions with influenza hemag- mouse CMV; MDCK, Madin-Darby canine kidney; PR8, mouse-adapted influenza A/ glutinin and the receptor, NKp46 (17–19). However, NK cells can PR/8/34 strain; RSV, respiratory syncytial virus; WT, wild-type. also contribute to pathological damage of host tissues with high Copyright Ó 2016 by The American Association of Immunologists, Inc. 0022-1767/16/$30.00 dose influenza infection (20, 21). These studies raise the interesting

www.jimmunol.org/cgi/doi/10.4049/jimmunol.1502186 2 INFLUENZA INFECTION INDUCES INNATE IMMUNE MEMORY possibility that NK cells may play dual roles during influenza in- i.n. with influenza virus. The immunized mice were i.v. administered anti- fection, conferring protection or inducing pathology, depending NK1.1 (PK136) or PBS 1 d prior to infection. on the viral dose. Whether memory NK cells can provide protection Determination of viral loads against subsequent influenza infection and the related immune mechanisms is poorly understood. Thus, a live influenza virus The viral loads in the lungs were determined through quantitative real-time infection mouse model might provide an adequate platform to PCR analysis at the indicated times after PR8 infection. Briefly, total RNA was extracted from whole lung tissue, and cDNA was subsequently syn- study the immune memory of NK cells. thesized. The viral loads in each sample were calculated through relative In the current study, to determine whether NK cells could mount quantification, and the ratio of viral RNA to b-actin mRNA 1 d after PR8 an Ag-specific memory response to acute respiratory influenza infection in wild-type (WT) mice was used as a control. The following virus infection, we first explored the existence of memory-like NK primers were specific for the matrix protein of influenza virus (23): 59- GGACTGCAGCGTAGACGCTT-39 (forward) and 59-CATCCTGTTGTA- cells after viral infection and subsequently characterized the TATGAGGCCCAT-39 (reverse). Alternatively, the infectious viral titers phenotype, function, and location of memory-like NK cells. We (viral PFU) were determined using a Madin-Darby canine kidney (MDCK) observed that influenza virus infection–induced memory-like NK cell determination assay. This method takes advantage of the production of cells share similar phenotypes with MCMV-specific Ly49H+ viral nucleoprotein in infected cells for the quantification of viral titers. memory NK cells. Similar to hapten-specific memory NK cells, At various times after influenza infection, lungs were harvested in 1 ml of DMEM (Thermo Scientific, Beijing, China), homogenized, serially diluted, memory-like NK cells induced through influenza virus infection and 200 ml samples were added to 24-well plates containing confluent also reside in the liver, and the adoptive transfer of liver CD49a+ monolayers of MDCK cells (1.4 3 105 cells per well) for 5–6 h at 37˚C. A DX52 NK cells into naive mice followed by viral infection results 1:1 mixture of 2% methylcellulose/23 DMEM plus 10% FCS was pre- in protective immunity and decreased viral titer. Furthermore, pared at room temperature or 4˚C for at least 2 h, and 300 mlofthe Downloaded from methylcellulose/medium-mixture overlay was added to each well. The plate primary inactivated mouse-adapted influenza A/PR/8/34 strain (PR8) was gently shaken from side to side and incubated overnight at 37˚C. 2/2 treatment could protect Rag1 mice against secondary PR8 in- Subsequently, these cells were fixed and stained with an Ab specific for the fection–induced death and reduced the viral titer in the lung. Com- influenza virus nucleoprotein. The signal was detected using a secondary Ab pared with Rag12/2Tbx212/2 mice that lack CD49a+DX52 liver conjugated with HRP, which amplifies the signal of the antinucleoprotein Ab. NK cells, inactivated influenza-treated Rag12/2Nfil32/2 could resist Subsequent exposure to the Trueblue peroxidase substrate (KPL Company) turned only the infected cells blue. The titer in the lung homogenate was secondary live influenza virus infection. Thus, the results of the determined after counting the number of blue cells in a given area. Each http://www.jimmunol.org/ current study suggest that influenza virus infection induces the stained cell corresponded to a single infectious unit. The calculation for PFU/ml formation of memory-like NK cells, which reside in the liver and lung homogenate was PFU/ml = number of spots 3 dilution/0.2 ml. protect mice against secondary influenza virus infection, and the Depletion of NK cells live influenza virus infection mouse model might be a new natural pathogen infection mouse model after the MCMV infection model Mice were i.v. injected with 30 ml rabbit anti-mouse asialo-GM1 (AsGM1) to study memory-like NK cells. Ab for NK cells depletion 24 h before and 72 h after influenza virus infection. Mice were i.v. or i.p. injected with 200 mg of anti-mouse NK1.1 Ab (clone PK136) 24 h before influenza virus infection. Control mice Materials and Methods received 200 ml of PBS alone. Mice by guest on September 25, 2021 NK cell enrichment and adoptive transfer

C57BL/6 mice (6–10 wk of age) were obtained from the Shanghai Ex- 2/2 perimental Animal Center (Chinese Academy of Sciences, Shanghai, NK cells were enriched from C57BL/6 or Rag1 mice through negative China). Rag12/2 (B6) mice were obtained from the Model Animal Re- selection using a mixture of Abs, including anti-CD19, anti-CD4, anti- search Center (Nanjing, China), and these mice were originally obtained CD8a, anti-CD5, anti-Gr1, and anti–Ter-119 (Miltenyi Biotec). NK cells from The Jackson Laboratory. Nfil3+/2 (B6) mice were provided from Tak were 60–95% pure based on NK1.1 staining. Then, a FACSAria cell sorter W. Mak (University of Toronto, Toronto, ON, Canada). Nfil32/2 mice were (BD Biosciences) was used to purify NK cells or different NK cell subsets. The purity of the sorted cell populations was more than 95%, as deter- bred in house and maintained at an animal facility under specific pathogen- ∼ 3 5 3 6 free conditions. Tbx212/2 mice were obtained from The Jackson Labo- mined through postsort flow cytometry. After sorting, 2 10 or 1 10 2/2 2/2 2/2 2/2 NK cells were i.p. injected into neonatal mice or i.v. injected into adult ratory. Rag1 Nfil3 and Rag1 Tbx21 mice were crossbred in 2/2 ∼ house. Animal care and experimental procedures were performed in ac- Rag1 mice 1 or 30 d prior to influenza virus infection. What is more, 1 3 105 liver DX5+ NK and DX52 NK cells were sorted from PR8- cordance with the experimental animal guidelines of the University of + Science and Technology of China. infected CD45.2 C57BL/6 mice and adoptively transferred into sublethally irradiated CD45.1+CD45.2+ C57BL/6 or CD45.1+ C57BL/6 mice Viral strains and growth conditions i.v. The dynamic change of NK cells in the lung and liver from recipient mice were analyzed at the indicated time points (12 h, 4 d, 15 d, and 30 d) PR8 (H1N1) was a gift from Hong Meng (Institute of Basic Medicine, after transfer by flow cytometry. Shandong Academy of Medical Sciences, China), and this strain was amplified in the allantoic cavities of 10-d-old specific pathogen-free em- Cell preparation bryonated chicken eggs (Merial Vital Laboratory Animal Technology, Beijing, China). Allantoic fluids were harvested 72 h after inoculation and Liver mononuclear cells and splenocytes were isolated as previously de- stored at 280˚C until further analysis. The hemagglutination titer of each scribed (24). Lung mononuclear cells were isolated as previously de- sample was 10,240 HA units/ml. Virus of the respiratory syncytial virus scribed, with minor modifications (25). (RSV) long strain (subtype A) was grown in Hep-2 cells, frozen at 280˚C, and assayed for infectivity. In some experiments, the virus was inactivated Flow cytometry assay through incubation in a 56˚C water bath for 35 min prior to use for After blocking Fc receptors, single-cell suspensions were incubated with the stimulation or immunization (22). indicated fluorescently labeled mAbs at 4˚C for 30 min in PBS (containing Mouse infection and immunization 0.1% sodium azide and 1% BSA) and subsequently washed twice. To assess IFN-g production, single-cell suspensions were cultured with plate- C57BL/6 mice were i.p. anesthetized using sodium pentobarbital (50 mg/g bound anti-NK1.1 (clone PK136) for 5 h, plate-bound anti-NKp46 (clone body weight) prior to intranasal (i.n.) infection with influenza virus (H1N1, 29A1.4, BioLegend) for 5 h, inactivated virus (∼200 ml PR8 or RSV) for PR8; lethal dose, 0.5 HA; sublethal dose, 0.1 HA) in 50 ml of sterile saline. 24 h, or 50 ng/ml PMA and 1 mg/ml ionomycin for 4 h in the presence of Control mice received 50 ml of sterile saline alone. Rag12/2 mice were i.n. GolgiPlug (Brefeldin A, BD Biosciences). Meanwhile, the single-cell infected with influenza virus (H1N1, PR8; lethal dose, 0.2 HA; low dose, suspensions without in vitro stimulation were nonstimulation control. 0.005 HA) in 50 ml of sterile saline. Rag12/2, Rag12/2 Nfil32/2 and The cells were subsequently collected and stained for surface markers. Rag12/2 Tbx212/2 mice were s.c. injected with inactivated PR8 or PBS in After fixation and permeabilization, the cells were stained with PE- the abdomen on days 230 and 225. After 1 mo, the mice were infected conjugated anti–IFN-g. The samples were analyzed by using a BD LSR The Journal of Immunology 3

II and FlowJo 7.6 software. All Abs were obtained from BD Biosciences, also showed that lung NK cells from PR8-infected mice expressed eBioscience, or BioLegend. higher levels of IFN-g than lung NK cells from uninfected control Cytometric bead array assay mice (Supplemental Fig. 2C). Further studies showed that lung NK cells from PR8-infected and uninfected control mice expressed sim- 5 Freshly purified NK cells at 1 3 10 per well were treated with inactivated ilarlevelsofIFN-g after stimulation with inactivated RSV (Fig. 1D), PR8 or RSV overnight in a humidified atmosphere of 5% CO2 at 37˚C. anti-NK1.1 Ab (Fig. 1E), or anti-NKp46 Ab (Fig. 1F). Thus, these Cytokines were determined in cell-free supernatants using mouse Th1/ Th2/Th17 cytokine kit (IFN-g/IL-2/IL-4/IL-6/TNF/IL-17A/IL-10; BD results suggest that lung NK cells from influenza virus–infected mice Biosciences) according to the manufacturer’s protocol. Data were analyzed possess Ag-specific characteristics, consistent with reports in the by using BD LSR II and FlowJo 7.6 software. memory T cell field (29). To further investigate the memory functions of lung NK cells from influenza virus–infected mice, we sort-purified Statistical analysis lung NK cells (Supplemental Fig. 2B) and subsequently transferred Two-tailed Student t tests were used for statistical analyses. The data were 2 3 105 purified lung NK cells into neonatal C57BL/6 mice (7– 6 expressed as the means SEM, and the data are representative of two to 10 d old). After 30 d, the recipient mice were infected i.n. with a four independent experiments. The log-rank (Mantel–Cox) test was used to evaluate survival rate. All p values , 0.05 were considered statistically lethal dose of PR8 (Fig. 1G). Although the recipient mice showed a significant. slightly slow weight loss (Fig. 1H), the survival rate of the recipient mice was not significantly increased after PR8 infection when com- Results pared with control mice (Fig. 1I). In order to preclude the influence of the number of transferred lung NK cells, we increased the transferred Lung NK cells control viral proliferation after primary number of lung NK cells from 2 3 105 to 1 3 106, but this still Downloaded from influenza virus infection could not protect recipient mice against influenza-induced death As previously reported (16), i.n. influenza virus (PR8) infection (Supplemental Fig. 3A–B). Furthermore, transferring 2 3 105 or 1 3 promoted the accumulation of NK cells in the lungs during early 106 purified splenic NK cells from PR8-infected mice also could not stages of infection in C57BL/6 mice (Supplemental Fig. 1A). protect recipient mice against influenza-induced death (Supplemental Compared with control mice, lung NK cells upregulated expres- Fig. 3C–D), which means that there were also no memory NK cells

sion of the activating receptor NKG2D, activating marker CD69, existing in the spleen after influenza virus infection. Therefore, al- http://www.jimmunol.org/ and functional molecule IFN-g after PR8 infection (Supplemental though lung NK cells from influenza virus–infected mice possess a Fig. 1B, 1C). To examine the function of NK cells in controlling memory-like phenotype, they cannot protect mice from death medi- 2 2 influenza virus infection, we used Nfil3 / mice (26) that lack NK ated by secondary influenza virus infection. cells in the lung, and we observed that PR8-infected Nfil32/2 mice showed more weight loss postinfection when compared with PR8- Liver NK cells possess a memory phenotype and protect mice infected WT mice (Supplemental Fig. 1D). We also observed against influenza-induced death 2 2 higher viral loads in the lungs of Nfil3 / mice when compared Liver NK cells have been reported to develop immunological with PR8-infected WT mice (Supplemental Fig. 1E). Furthermore, memory against haptens or vaccines containing viral Ags, such as we treated mice i.p. with PK136 or i.v. with AsGM1 to deplete NK influenza virus, vesicular stomatitis virus, and HIV type 1 (HIV-1) by guest on September 25, 2021 cells before influenza virus infection and found higher viral loads (2). To determine whether respiratory influenza virus infection in the lungs from both PK136-treated and AsGM1-treated mice could induce memory-like NK cells in the liver, we analyzed the when compared with PR8-infected control mice (Supplemental phenotype and function of liver NK cells after primary influenza Fig. 1F, 1G). Taken together, these data suggest that NK cells in virus infection. Although the number of NK cells in the liver was the lung are activated and are involved in controlling viral pro- slightly decreased 3 d after PR8 infection (Supplemental Fig. 4A), liferation after primary influenza virus infection. the expression of CD69, NKG2D, and IFN-g was increased in liver NK cells (Supplemental Fig. 4B, 4C), suggesting that liver Lung NK cells possess a memory phenotype but do not protect NK cells were also activated after respiratory influenza virus in- mice against influenza-induced death fection. In order to preclude the possibility that influenza virus To explore whether memory-like NK cells generated in the lungs after entered the liver and activated NK cells directly, we detected the primary influenza virus infection and provided protection against viral titer in the lung and liver after PR8 infection by MDCK cell subsequent influenza virus infection, C57BL/6 mice were infected i.n. determination assay. The results showed that influenza virus could with a nonlethal dose of PR8. After 30 d, the number of NK cells in the be detected in lung but not in liver from day 1 to day 3 after PR8 lungsofinfectedmicereturnedtonormalwhencomparedwithun- infection (Supplemental Fig. 4D). Type I IFNs have been reported infected control mice (Fig. 1A). As previously reported, memory NK to promote NK cells to express IFN-g during influenza virus in- cells from MCMV-infected mice expressed higher levels of KLRG1 fection (30, 31). In our study, we found that the level of IFN-a in and Ly6C but lower levels of CD62L compared with naive NK cells both bronchoalveolar lavage fluid and serum was significantly (4, 27, 28). Consistently, in the current study, the expression of increased at the early days after PR8 infection (Supplemental Fig. KLRG1 and Ly6C on the surface of lung NK cells was significantly 4E), which might be responsible for the activation of NK cells in increased, whereas CD62L expression was decreased 30 d after in- liver during influenza virus infection. fluenza virus infection (Fig. 1B). Thus, lung NK cells from influenza Moreover, as expected, the number of NK cells in the liver did virus–infected mice exhibit a memory NK cell-like phenotype. not increase 30 d after PR8 infection (Fig. 2A), but liver NK cells To determine whether lung NK cells from influenza virus–infected showed upregulated expression of the memory-related markers mice possessed memory cell-like functions, we stimulated lung Ly6C and KLRG1 and downregulated expression of CD62L lymphocytes with inactivated PR8 andobservedthatlungNKcells (Fig. 2B). Functional analysis showed that liver NK cells from from PR8-infected mice expressed higher levels of IFN-g than lung PR8-infected mice expressed higher levels of IFN-g than liver NK NK cells from uninfected control mice (Fig. 1C, Supplemental Fig. cells from control mice stimulated with inactivated PR8 (Fig. 2C 2A). To rule out the influence of memory T and B cells, we sort- and Supplemental Fig. 2C). Similar to lung NK cells, liver NK purified lung NK cells from PR8-infected mice (Supplemental Fig. cells from PR8-infected and control mice also expressed the same 2B) and stimulated them with inactivated PR8 in vitro. The results level of IFN-g after stimulation with inactivated RSV (Fig. 2D), 4 INFLUENZA INFECTION INDUCES INNATE IMMUNE MEMORY Downloaded from http://www.jimmunol.org/ by guest on September 25, 2021

FIGURE 1. Lung NK cells possess a memory phenotype but do not protect mice against influenza-induced death. (A–F) C57BL/6 mice were i.n. infected with 0.05 HA of PR8. After 30 d, the number of NK cells in the lung was counted (A), the expression of memory markers (Ly6c, KLRG1 and CD62L) on NK cells was determined by flow cytometry (B), and the expression of IFN-g in NK cells from control and PR8-infected mice was determined by flow cytometry after stimulating with inactivated PR8 (C), inactivated RSV (D), plate-bound anti-NK1.1 Ab (E) and plate-bound anti-NKp46 Ab (F). (G–I) Schematic of experimental design. C57BL/6 mice were i.n. infected with PBS or PR8, 30 d later, 2 3 105 lung NK cells were isolated and adoptively transferred into neonatal C57BL/6 mice (7–10 d), and then the recipient mice were infected with a lethal dose of influenza virus 30 d later (G). Changes in body weight (H) and survival rate of recipient mice (I)were analyzed. Data were pooled from three experiments. Data are expressed as mean 6 SEM by a Student t test. Data represent three independent experiments with at least three mice per group(A–F). The log-rank (Mantel–Cox) test was used to evaluate the survival rate (I). *p , 0.05, **p , 0.01. ns, not significant. anti-NK1.1 Ab (Fig. 2E), or anti-NKp46 Ab (Fig. 2F), suggesting memory function after influenza virus infection, we compared the that liver NK cells from influenza virus–infected mice also possess phenotype and function of CD49a+DX52 and CD49a2DX5+ liver Ag-specific characteristics. However, unlike the lung NK cells, the NK cells. The results showed that the expression of KLRG1 and 2 adoptive transfer of liver NK cells from PR8-infected mice pro- Ly6C was significantly increased on CD49a+DX5 liver NK cells 2 tected recipient mice against PR8 infection–induced weight loss but not on CD49a DX5+ liver NK cells 30 d after PR8 infection 2 and death (Fig. 2G–I). Altogether, these data suggest that respi- (Fig. 3A, 3B), suggesting that the CD49a+DX5 NK cell subset ratory influenza virus infection induces memory-like NK cell might be the main source of memory-like NK cells in the liver. To 2 production in the liver, which protects mice against secondary address this hypothesis, the expression of IFN-g in CD49a+DX5 2 + influenza virus infection–induced death. liver NK cells and CD49a DX5 liver NK cells was analyzed after stimulating with inactivated PR8 or inactivated RSV. The results Liver influenza-specific memory-like NK cells display a showed that CD49a+DX52 liver NK cells but not CD49a2DX5+ + 2 CD49a DX5 phenotype liver NK cells from PR8-infected mice expressed higher levels of Liver NK cells can be divided into two distinct subsets based on the IFN-g than liver NK cells from control mice stimulated with cell surface expression of CD49a and DX5: CD49a+DX52 and inactivated PR8 (Fig. 3C), and there was no difference between CD49a2DX5+ NK cells and liver resident CD49a+DX52 NK cells them after stimulating with inactivated RSV (Fig. 3D). Furthermore, 2 2 have been reported to possess memory potential and to confer we sort-purified CD49a+DX5 liver NK cells and CD49a DX5+ hapten-specific CHS responses upon hapten rechallenge (5, 32). liver NK cells (Supplemental Fig. 2B) and then measured genes To determine which subset of liver NK cells possessed immune involved in immune memory and observed that the expression of The Journal of Immunology 5 Downloaded from http://www.jimmunol.org/ by guest on September 25, 2021

FIGURE 2. Liver NK cells possess a memory phenotype and protect mice against influenza-induced death. (A–F) C57BL/6 mice were i.n. infected with 0.05 HA of PR8. After 30 d, the number of NK cells in the liver was counted (A), the expression of memory markers (Ly6c, KLRG1 and CD62L) on NK cells was determined by flow cytometry (B), and the expression of IFN-g in NK cells from control and PR8-infected mice was determined by flow cytometry after stimulating with inactivated PR8 (C), inactivated RSV (D), plate-bound anti-NK1.1 Ab (E) and plate-bound anti-NKp46 Ab (F). (G–I) Schematic of experimental design. C57BL/6 mice were i.n. infected with PBS or PR8, 30 d later, 2 3 105 liver NK cells were isolated and adoptively transferred into neonatal C57BL/6 mice (7–10 d), and then the recipient mice were infected with a lethal dose of influenza virus 30 d later (G). Changes in body weight (H) and survival rate of recipient mice (I)were analyzed. Data were pooled from three experiments. Data are expressed as mean 6 SEM by a Student t test. Data represent three independent experiments with at least three mice per group (A–F). The log-rank (Mantel–Cox) test was used to evaluate the survival rate (I). *p , 0.05, **p , 0.01, ***p , 0.001, ****p , 0.0001. ns, not significant. homeobox-only protein (Hopx), which upregulated in memory cells over, CD49a+DX52 liver NK cells but not CD49a2DX5+ liver NK and might promote the maintenance of memory NK cells after cells from PR8-infected mice could protect recipient mice infected MCMV infection (28, 33), in CD49a+DX52 liver NK cells was with PR8 at day 30 after adoptive transfer (Fig. 3I). Altogether, much higher than in CD49a2DX5+ liver NK cells (Fig. 3E). Thus, these results demonstrate that liver CD49a+DX52 NK cell is the these results suggest that CD49a+DX52 liver NK cells have more primary source of memory NK cells against influenza virus. potential as memory-like NK cell candidates than CD49a2DX5+ Next, to explore the mechanism of how CD49a+DX52 liver NK liver NK cells. To further measure the immune memory function of cell transfer protects recipient mice against influenza-induced both subsets, we adoptively transferred CD49a+DX52 liver NK cells death, we i.v. transferred CD49a+DX52 liver NK cells or and CD49a2DX5+ liver NK cells from PR8-infected mice into naive CD49a2DX5+ liver NK cells from PR8-infected CD45.2+ C57BL/ adult Rag12/2 mice or neonatal C57BL/6 mice followed by infec- 6 mice into sublethally irradiated CD45.1+ or CD45.1+CD45.2+ tion with a lethal dose of PR8 1 or 30 d later (Fig. 3F). We detected C57BL/6 mice and then detected the dynamic change of donor viral titers in the lungs 4 d after PR8 infection and observed that NK cells in the lung and liver of recipient mice (Fig. 4A). The only the CD49a+DX52 liver NK cell transfer significantly decreased results showed that both donor-derived DX52 and DX5+ NK cells the viral load in the lungs of recipient Rag12/2 mice infected with mainly existed in the lung a short time after adoptive transfer PR8 both 1 and 30 d after adoptive transfer (Fig. 3G, 3H). More- and then decreased gradually. Interestingly, as time goes on, the 6 INFLUENZA INFECTION INDUCES INNATE IMMUNE MEMORY Downloaded from http://www.jimmunol.org/

FIGURE 3. CD49a+DX52 liver NK cells express high levels of memory markers and inhibit influenza virus proliferation. (A–D) C57BL/6 mice were i.n. by guest on September 25, 2021 infected with 0.05 HA of PR8. After 30 d, the expression (A) and the percentage (B) of memory markers (Ly6c and KLRG1) on liver DX5+ NK cells and liver DX52 NK cells were determined by flow cytometry. The expression of IFN-g in NK cells was determined by flow cytometry after stimulating with inactivated PR8 (C) and inactivated RSV (D). (E) The level of the memory-specific gene (Hopx) in liver DX5+ NK and DX52 NK cells from control mice or PR8-infected C57BL/6 mice was detected by real-time PCR. (F–H) Schematic of the experimental design. Liver DX5+ NK cells and DX52 NK cells from control mice or PR8-infected C57BL/6 mice were adoptively transferred into Rag12/2 mice, and recipient Rag12/2 mice were subsequently infected with a lethal dose of influenza virus 1 d later (G) or 30 d later(H). The viral titer in the lung of recipient Rag12/2 mice was detected 4 d after PR8 infection by MDCK determination assay. (I) Liver DX5+ NK cells and DX52 NK cells from control mice or PR8-infected C57BL/6 mice were adoptively transferred into neonatal C57BL/6 mice (7–10 d), and then the recipient mice were infected with a lethal dose of influenza virus 30 d later as shown in the schematic of the experimental design in (F). Survival rate of recipient mice was analyzed. Data are expressed as mean 6 SEM by a Student t test. The data represent three independent experiments with at least three mice per group (A–H). The log-rank (Mantel–Cox) test was used to evaluate survival rate. Data were pooled from two experiments. *p , 0.05, **p , 0.01. ns, not significant. donor-derived DX52 NK cells gradually increased in the liver whereas Ly6C was also markedly increased on CD49a+DX52 liver NK cells the donor-derived DX5+ NK almost had no increase (Fig. 4B, 4C). but not on CD49a2DX5+ liver NK cells in Rag12/2 mice (Fig. 5C). Therefore, these results would explain why CD49a+DX52 liver Next, to determine whether primary influenza virus infection– 2 2 NK cells, but not other NK cells including lung NK cells and induced memory-like NK cells in Rag1 / mice could protect mice 2 2 CD49a2DX5+ liver NK cells, provided protection against influ- against secondary influenza virus infection, we i.n. infected Rag1 / enza virus infection in recipient mice. mice with a lethal dose of PR8 30 d after primary PR8 infection. However, we observed that primary PR8 infection–induced Development of influenza-specific memory NK cells is T cell memory-like NK cells could not protect Rag12/2 mice against and B cell independent death and weight loss caused by a secondary lethal dose of PR8 Previous studies have shown that hapten and HSV-2 induce the infection, and the depletion of NK cells using an anti-NK1.1 Ab generation of memory-like NK cells in a T cell– and B cell– (PK136) prior to secondary PR8 infection did not influence sur- independent manner (1, 7). To determine whether memory-like vival when compared with the untreated group (Fig. 5D–F). Fur- 2 NK cell generation after influenza virus infection depended on ther studies showed that both CD49a+DX5 liver NK cells and 2 2 2 T and B cells, we i.n. infected Rag12/2 mice that lack T and CD49a DX5+ liver NK cells from PR8-infected Rag1 / mice not B cells with a nonlethal dose of PR8 and subsequently assessed only failed to protect recipient mice against influenza-caused death the phenotype of NK cells in the lungs and livers 30 d postinfec- but also promoted the death of the recipient mice after influenza virus tion. Consistent with findings in C57BL/6 mice, NK cells in Rag12/2 infection (Fig. 5G, 5H). Thus, these data suggest that influenza virus mice also showed increased KLRG1 expression 30 d after PR8 in- infection could not induce the production of the memory NK fection (Fig. 5A–C). Importantly, the expression of both KLRG1 and cells in Rag12/2 mice. We thought the reason was that, unlike in The Journal of Immunology 7 Downloaded from http://www.jimmunol.org/ by guest on September 25, 2021

FIGURE 4. CD49a+DX52 liver NK cells accumulate and amplify in liver of recipient mice after adoptive transfer. (A–C) Schematic of experimental design. A total of 1 3 105 liver DX5+ NK and DX52 NK cells were sorted from PR8-infected CD45.2+ C57BL/6 mice and adoptively transferred into sublethally irradiated CD45.1+CD45.2+ C57BL/6 or CD45.1+ C57BL/6 mice i.v. (A). The dynamic change of NK cells in the lung and liver from recipient mice were analyzed at the indicated time points (12 h, 4 d, 15 d, 30 d) after transfer by ﬂow cytometry (B and C). Data are expressed as mean 6 SEM by a Student t test. The data represent two independent experiments with at least three mice per group.

WTmice,PR8primaryinfectionresultedinpersistentweight showed that inactivated PR8 priming could significantly reduce loss and worsened conditions in Rag12/2 mice (Fig. 5I), likely viral titer after secondary influenza virus infection in Rag12/2 resulting from the lack of T and B cells to completely eliminate the Nfil32/2 mice rather than in Rag12/2Tbx212/2 mice (Fig. 6F, virus or the loss of tolerance to infection-induced tissue damage (34). 6G). Taken together, these results suggest that primary influenza To control the influence of primary influenza virus infection, we virus treatment protects Rag12/2 mice against secondary influ- treated Rag12/2 mice s.c. with inactivated PR8 as a primary enza virus infection–induced death through inducing memory-like immunization according to a previous report (2) and then infected NK cell generation in a T cell– and B cell–independent manner. mice i.n. with a lethal dose of PR8 30 d later (Fig. 6A). The results showed that primary inactivated PR8 treatment protected Rag12/2 Discussion mice against secondary PR8 infection–induced death and de- Immune memory is a classical feature that distinguishes innate and creased the viral titer in the lungs when compared with the control adaptive immunity. However, the finding of memory-like NK cells group; in addition, the depletion of NK cells using an anti-NK1.1 might obscure the concept of innate and adaptive immune systems Ab prior to secondary influenza virus infection significantly de- (35, 36). In the past several years, memory-like NK cells have creased the survival rate of Rag12/2 mice and increased the viral been studied using various mouse models (37, 38). However, titer in the lungs after secondary PR8 infection (Fig. 6B, 6C). whether NK cells mount an Ag-specific memory response after Furthermore, adoptive transfer of CD49a+DX52 liver NK cells but influenza virus infection has not yet been examined. In the current not CD49a2DX5+ liver NK cells from inactivated PR8-treated study, we used a well-established mouse model of respiratory Rag12/2 mice could reduce the viral titer in recipient mice influenza virus infection to characterize Ag-specific memory-like (Fig. 6D, 6E). Moreover, to further determine whether CD49a+ NK cells, and we observed that lung NK cells possess a memory- DX52 liver NK cells in Rag12/2 mice perform a protective effect, like phenotype but do not protect mice against secondary influenza we used Rag12/2Nfil32/2 mice in which CD49a+DX52 NK cells virus infection. Further studies showed that liver-resident CD49a+ are dominant in liver and Rag12/2Tbx212/2 mice that lack DX52 NK cells also display a memory-like phenotype and protect CD49a+DX52 liver NK cells as the study objects. The results mice against influenza virus reinfection. Thus, consistent with the 8 INFLUENZA INFECTION INDUCES INNATE IMMUNE MEMORY Downloaded from http://www.jimmunol.org/

FIGURE 5. NK cells from Rag12/2 mice cells do not possess memory functions after primary influenza virus infection. (A) Rag12/2 mice were i.n. infected with 0.005 HA of PR8. After 30 d, the expression of memory markers (Ly6c and KLRG1) on NK cells from the lung and liver was determined by flow cytometry. (B) The frequencies of Ly6c+ and KLRG1+ NK cells in the lung and liver NK cells (CD32NK1.1+) were analyzed. (C) The frequencies of Ly6c+ and KLRG1+ NK cells in liver DX5+ and DX52 NK cells were analyzed. (D–F) Schematic of experimental design. Rag12/2 mice were i.n. infected with 0.005 HA of PR8. After 30 d, mice were i.n. infected with 0.1 HA of PR8 after anti-NK1.1 Ab (PK136) treatment or not (D). The bodyweight change by guest on September 25, 2021 (E) and the survival rates (F)ofRag12/2 mice were monitored after subsequent influenza virus infection. (G and H) Schematic of experimental design. Liver DX5+ and DX52 NK cells isolated from control or PR8-infected Rag12/2 mice were adoptively transferred into neonatal C57BL/6 mice (7–10 d), and the recipient mice were subsequently infected with a lethal dose of influenza virus 30 d later (G). Survival rate of recipient mice was monitored (H). (I) Rag12/2 mice were i.n. infected with 0.005 HA of PR8. Changes in body weight was analyzed. Data are expressed as mean 6 SEM by a Student t test. Data represent three independent experiments with at least three mice per group in (A–C). The log-rank (Mantel–Cox) test was used to evaluate survival rate (D– H). Data were pooled from two experiments. *p , 0.05, **p , 0.01. ns, not significant. results from studies using noninfectious influenza virus particles, although the activating receptor NKp46 binds to the influenza memory-like NK cells are observed in vivo after active influenza protein HA (39), Silke Paust and colleagues showed that HA was virus infection, thereby providing an adequate platform to investi- not required for NK cell–mediated recall responses to influenza gate the mechanisms underlying the generation, migration, and virus-like particles (2). Moreover, in the current study, we ob- function of memory-like NK cells. served that influenza virus–specific memory-like NK cells reside Influenza virus–specific memory-like NK cells share several in the liver but not in the lung. So, we speculated that the gen- characteristics with MCMV-specific memory NK cells (4) and eration of influenza virus–specific memory-like NK cells in the hapten-specific memory NK cells (1, 2). Similar to MCMV- current study might depend on the interaction between other li- specific memory NK cells, influenza virus–specific memory-like gands and receptors from influenza virus and NK cells, respec- NK cells also upregulate expression of the receptors KLRG1 and tively, and Ag ligands from influenza virus might enter the liver Ly6C and the transcription factor Hopx. Consistent with findings through the bloodstream, although influenza virus could not be in previous studies concerning hapten-specific memory NK cells, detected in the liver during infection (40). influenza virus–specific memory-like NK cells reside in the liver The generation of MCMV-specific memory NK cells requires and display a CD49a+DX52 phenotype. Although influenza virus– simultaneous stimulation from both specific Ags and proin- specific memory-like NK cells have been detected after active flammatory cytokine signaling, and the clonal expansion of Ly49H+ influenza virus infection, several questions remain. For example, NK cells following MCMV infection is dependent on IL-12 and what kinds of ligand and receptor interactions induce the gener- STAT4 signaling (41). The key NK cell survival factor IL-15 is ation of memory-like NK cells? Why do influenza virus–specific critical to the maintenance of the memory NK cell population (42). memory-like NK cells reside in the liver but not in the lung? Increased cytokine expression is induced during influenza virus In studies of MCMV-specific memory NK cells (4), Sun et al. infection (31, 43, 44), and these proteins can enter the liver through demonstrated that the interaction between the Ly49H receptor on the bloodstream. According to the findings in other memory-like NK cells and the glycoprotein m157 from MCMV mediates NK cell studies, the generation of influenza virus–specific memory- the generation of MCMV-specific memory NK cells. However, like NK cells might also need the cytokine stimulation. The Journal of Immunology 9 Downloaded from http://www.jimmunol.org/

FIGURE 6. Inactivated PR8 treatment inhibits viral proliferation and protects mice against secondary influenza-induced death. (A–C) Schematic of experimental design. The inactivated PR8 treated Rag12/2 mice by s.c. immunization at day 230 and 225. At day 0, Rag12/2 mice were i.n. infected with activated PR8 after anti-NK1.1 Ab (PK136) treatment or not (A). Survival rates of Rag12/2 mice after activated PR8 infection (B). Viral titer in the lung was by guest on September 25, 2021 detected at day 4 and 7 after activated PR8 infection (C). (D and E) Schematic of experimental design. Liver DX5+ and DX52 NK cells isolated from control or inactivated PR8 immunized Rag12/2 mice were adoptively transferred into naive Rag12/2 mice, and then the recipient Rag12/2 mice were infected with a lethal dose of influenza virus 1 d later. Viral titers in the lung of recipient mice were detected at day 4 after PR8 infection (E). (F) Flow cytometry assay determined the CD49a+DX52 NK cells and CD49a2DX5+ NK cells in the liver of Rag12/2 Nfil32/2 mice and Rag12/2 tbx212/2 mice. (G) Rag12/2 Nfil32/2 mice and Rag12/2 tbx212/2 mice immunized with inactivated PR8 and then challenged with activated PR8 30 d later. Viral titers in the lung were detected at day 4 after activated PR8 infection. Data are expressed as mean 6 SEM by a Student t test. The log-rank (Mantel–Cox) test was used to evaluate survival rate. Data were pooled from three experiments (B). Data represent three independent experiments with at least three mice per group(C–G). *p , 0.05, **p , 0.01. ns, not significant.

The results of a previous study showed that chemical hapten- and the downregulation of CD62L expression could indicate induced memory NK cells reside in the liver, and chemokine memory-like NK cells, according to previous studies of MCMV- receptor CXCR6 on NK cells mediates the migration of NK cells specific memory NK cells (28). In the current study, we ob- into the liver (2). However, we previously showed that liver- served that, at the late stages of influenza virus infection, both lung resident CD49a+DX52 NK cells mediate hapten-specific recall and liver NK cells upregulate Ly6C and KLRG1 expression and responses, and CD49a+DX52 NK cells rarely circulate or emi- downregulate CD62L expression, and both lung and liver NK cells grate from the liver (5, 45). In addition, CD49a+DX52 liver NK from influenza virus–infected mice express higher levels of IFN-g cells differentiate from liver-resident hepatic progenitor cells and after stimulation through inactivated influenza virus in vitro sense Ags in the liver and acquired memory function locally. compared with NK cells from noninfected mice. However, the Thus, the liver may have unique precursors for memory NK cells, adoptive transfer of liver CD49a+DX52 NK cells, but not lung NK and they are developmentally distinct from NK cells derived from cells or liver CD49a2DX5+ NK cells, protected recipient mice bone marrow (45). Thus, how haptens stimulate CD49a+DX52 against influenza virus infection. However, why lung NK cells NK cells and how CD49a+DX52 NK cells exert recall responses display a memory-like phenotype but fail to protect mice against outside the liver remain unknown. In the current study, we also influenza virus infection remains unanswered. We speculate that observed that liver-resident CD49a+DX52 NK cells mediate the concentration of inflammatory cytokines determines the ef- memory effector functions, although lung NK cells also display a fector function of memory-like NK cells. The concentration of memory-like phenotype. However, whether CXCR6 plays a role inflammatory cytokines in the liver but not in the lung is optimal in the generation of influenza-specific memory-like NK cells for memory-like NK cell generation. As reported in previous needs further study. memory T cell studies (46–49), strong stimulation and excessive Although the specific phenotype of memory-like NK cells re- activation are not suitable for the generation and function of mains unknown, the upregulation of Ly6C and KLRG1 expression memory T cells. Thus, the cytokine concentration in the lung 10 INFLUENZA INFECTION INDUCES INNATE IMMUNE MEMORY during influenza virus infection might be too high for memory-like 8. Reeves, R. K., H. Li, S. Jost, E. Blass, H. Li, J. L. Schafer, V. Varner, NK cell generation. C. Manickam, L. Eslamizar, M. Altfeld, et al. 2015. Antigen-specific NK cell memory in rhesus macaques. Nat. Immunol. 16: 927–932. Consistent with previous studies (1, 6, 7), the generation of 9. Hendricks, D. W., G. Min-Oo, and L. L. Lanier. 2016. Sweet is the memory of influenza-specific memory-like NK cells in the current study was past troubles: NK cells remember. Curr. Top. Microbiol. Immunol. 395: 147–171. 2/2 10. Cerwenka, A., and L. L. Lanier. 2016. Natural killer cell memory in infection, also independent of T and B cells. We used Rag1 mice that inflammation and cancer. Nat. Rev. Immunol. 16: 112–123. naturally lack T and B cells to study the function of memory-like 11. Raulet, D. H. 2004. Interplay of natural killer cells and their receptors with the NK cells after active influenza virus infection. However, although adaptive immune response. Nat. Immunol. 5: 996–1002. 2/2 12. Sun, C., H. Sun, C. Zhang, and Z. Tian. 2015. NK cell receptor imbalance and we observed memory-like NK cells in Rag1 mice at the late NK cell dysfunction in HBV infection and hepatocellular carcinoma. Cell. Mol. stage of influenza virus infection, these cells did not protect mice Immunol. 12: 292–302. against secondary influenza virus infection. In contrast with WT 13. Valkenburg, S. A., J. A. Rutigliano, A. H. Ellebedy, P. C. Doherty, P. G. Thomas, and K. Kedzierska. 2011. Immunity to seasonal and pandemic influenza A mice, primary influenza virus infection resulted in persistent viruses. Microbes Infect. 13: 489–501. 2/2 weight loss and worsening conditions in Rag1 mice, as these 14. La Gruta, N. L., K. Kedzierska, J. Stambas, and P. C. Doherty. 2007. A question mice lack the adaptive immunity of T and B cells to completely of self-preservation: immunopathology in influenza virus infection. Immunol. Cell Biol. 85: 85–92. eliminate viruses. Moreover, as reported in previous memory 15. Gre´goire, C., L. Chasson, C. Luci, E. Tomasello, F. Geissmann, E. Vivier, and T cell studies (46–49), strong and persistent stimulation are not T. Walzer. 2007. The trafficking of natural killer cells. Immunol. Rev. 220: 169–182. suitable for the generation and function of memory T cells, so the 16. Wang, J., F. Li, M. Zheng, R. Sun, H. Wei, and Z. Tian. 2012. Lung natural killer 2/2 cells in mice: phenotype and response to respiratory infection. Immunology 137: condition in Rag1 mice after influenza virus infection might be 37–47. not suitable for the production of memory NK cells. Otherwise, 17. Stein-Streilein, J., and J. Guffee. 1986. In vivo treatment of mice and hamsters one study showed that RAG-deficient NK cells exhibit a cell- with antibodies to asialo GM1 increases morbidity and mortality to pulmonary Downloaded from influenza infection. J. Immunol. 136: 1435–1441. intrinsic hyperresponsiveness and a diminished survival capacity 18. Liu, B., I. Mori, M. J. Hossain, L. Dong, K. Takeda, and Y. Kimura. 2004. after virus stimulation as a result of reducing expression of DNA Interleukin-18 improves the early defence system against influenza virus in- damage response mediators and defects in the repair of DNA fection by augmenting natural killer cell-mediated cytotoxicity. J. Gen. Virol. 85: 423–428. breaks (50), which might be another reason why memory NK cells 19. Gazit, R., R. Gruda, M. Elboim, T. I. Arnon, G. Katz, H. Achdout, J. Hanna, 2 2 could not be detected in Rag1 / mice after influenza virus in- U. Qimron, G. Landau, E. Greenbaum, et al. 2006. Lethal influenza infection in the absence of the natural killer cell receptor gene Ncr1. Nat. Immunol. 7: 517– fection. To further explore this idea, we used inactivated influenza http://www.jimmunol.org/ 2/2 523. virus to immunize Rag1 mice and observed that memory-like 20. Abdul-Careem, M. F., M. F. Mian, G. Yue, A. Gillgrass, M. J. Chenoweth, NK cells decreased the viral titer during secondary active influ- N. G. Barra, M. V. Chew, T. Chan, A. A. Al-Garawi, M. Jordana, and enza virus infection. A. A. Ashkar. 2012. Critical role of natural killer cells in lung immunopathology during influenza infection in mice. J. Infect. Dis. 206: 167–177. Our study provides another adequate platform to investigate the 21. Zhou, G., S. W. Juang, and K. P. Kane. 2013. NK cells exacerbate the pathology mechanisms underlying the generation, migration, and function of of influenza virus infection in mice. Eur. J. Immunol. 43: 929–938. 22. Dou, Y., B. Fu, R. Sun, W. Li, W. Hu, Z. Tian, and H. Wei. 2015. Influenza memory-like NK cells. Studies to reveal the mechanisms for NK vaccine induces intracellular immune memory of human NK cells. PLoS One 10: cell memory may lead to the potential use of a vaccine that spe- e0121258. cifically induces protective memory NK cells, which would provide 23. van der Sluijs, K. F., L. J. van Elden, M. Nijhuis, R. Schuurman, J. M. Pater,

S. Florquin, M. Goldman, H. M. Jansen, R. Lutter, and T. van der Poll. 2004. by guest on September 25, 2021 a new strategy to protect humans against influenza virus infection. IL-10 is an important mediator of the enhanced susceptibility to pneumococcal pneumonia after influenza infection. J. Immunol. 172: 7603–7609. 24. Wang, J., R. Sun, H. Wei, Z. Dong, B. Gao, and Z. Tian. 2006. Poly I:C prevents Acknowledgments T cell-mediated hepatitis via an NK-dependent mechanism. J. Hepatol. 44: 446–454. +/2 We thank Dr. Tak W. Mak for providing the Nfil3 mice. We thank H. 25. Ishikawa, E., M. Nakazawa, M. Yoshinari, and M. Minami. 2005. Role of tumor Meng (Shandong Academy of Medical Sciences) for the influenza A/PR/8/ necrosis factor-related apoptosis-inducing ligand in immune response to influ- 34 strain. We thank Jinsheng He (Department of Immunology, Anhui enza virus infection in mice. J. Virol. 79: 7658–7663. 26. Kamizono, S., G. S. Duncan, M. G. Seidel, A. Morimoto, K. Hamada, Medical University) for providing the long strain of RSV. G. Grosveld, K. Akashi, E. F. Lind, J. P. Haight, P. S. Ohashi, et al. 2009. Nfil3/ E4bp4 is required for the development and maturation of NK cells in vivo. J. Exp. Med. 206: 2977–2986. Disclosures 27. Sun, J. C., S. Lopez-Verges, C. C. Kim, J. L. DeRisi, and L. L. Lanier. 2011. NK The authors have no financial conflicts of interest. cells and immune “memory”. J. Immunol. 186: 1891–1897. 28. Immunological Genome Project Consortium. 2012. Molecular definition of the identity and activation of natural killer cells. Nat. Immunol. 13: 1000–1009. References 29. McKinstry, K. K., T. M. Strutt, Y. Kuang, D. M. Brown, S. Sell, R. W. Dutton, and S. L. Swain. 2012. Memory CD4+ T cells protect against influenza through 1. O’Leary, J. G., M. Goodarzi, D. L. Drayton, and U. H. von Andrian. 2006. multiple synergizing mechanisms. J. Clin. Invest. 122: 2847–2856. T cell- and B cell-independent adaptive immunity mediated by natural killer 30. Price, G. E., A. Gaszewska-Mastarlarz, and D. Moskophidis. 2000. The role of Nat. Immunol. cells. 7: 507–516. alpha/beta and gamma interferons in development of immunity to influenza A 2. Paust, S., H. S. Gill, B. Z. Wang, M. P. Flynn, E. A. Moseman, B. Senman, virus in mice. J. Virol. 74: 3996–4003. M. Szczepanik, A. Telenti, P. W. Askenase, R. W. Compans, and U. H. von 31. Hwang, I., J. M. Scott, T. Kakarla, D. M. Duriancik, S. Choi, C. Cho, T. Lee, Andrian. 2010. Critical role for the chemokine receptor CXCR6 in NK cell- H. Park, A. R. French, E. Beli, et al. 2012. Activation mechanisms of natural mediated antigen-specific memory of haptens and viruses. Nat. Immunol. 11: killer cells during influenza virus infection. PLoS One 7: e51858. 1127–1135. 32. Peng, H., E. Wisse, and Z. Tian. 2016. Liver natural killer cells: subsets and roles 3. Cooper, M. A., J. M. Elliott, P. A. Keyel, L. Yang, J. A. Carrero, and in liver immunity. Cell. Mol. Immunol. 13: 328–336. W. M. Yokoyama. 2009. Cytokine-induced memory-like natural killer cells. 33. Albrecht, I., U. Niesner, M. Janke, A. Menning, C. Loddenkemper, A. A. Kuhl,€ Proc. Natl. Acad. Sci. USA 106: 1915–1919. I. Lepenies, M. H. Lexberg, K. Westendorf, K. Hradilkova, et al. 2010. Persis- 4. Sun, J. C., J. N. Beilke, and L. L. Lanier. 2009. Adaptive immune features of tence of effector memory Th1 cells is regulated by Hopx. Eur. J. Immunol. 40: natural killer cells. Nature 457: 557–561. 2993–3006. 5. Peng, H., X. Jiang, Y. Chen, D. K. Sojka, H. Wei, X. Gao, R. Sun, 34. Jamieson, A. M., L. Pasman, S. Yu, P. Gamradt, R. J. Homer, T. Decker, and W. M. Yokoyama, and Z. Tian. 2013. Liver-resident NK cells confer adaptive R. Medzhitov. 2013. Role of tissue protection in lethal respiratory viral-bacterial immunity in skin-contact inflammation. J. Clin. Invest. 123: 1444–1456. coinfection. Science 340: 1230–1234. 6. Gillard, G. O., M. Bivas-Benita, A. H. Hovav, L. E. Grandpre, M. W. Panas, 35. Vivier, E., D. H. Raulet, A. Moretta, M. A. Caligiuri, L. Zitvogel, L. L. Lanier, M. S. Seaman, B. F. Haynes, and N. L. Letvin. 2011. Thy1+ NK [corrected] cells W. M. Yokoyama, and S. Ugolini. 2011. Innate or adaptive immunity? the ex- from vaccinia virus-primed mice confer protection against vaccinia virus chal- ample of natural killer cells. Science 331: 44–49. lenge in the absence of adaptive lymphocytes. [Published erratum appears in 36. Sun, J. C., and L. L. Lanier. 2009. Natural killer cells remember: an evolutionary 2011 PLoS Pathog. 7.] PLoS Pathog. 7: e1002141. bridge between innate and adaptive immunity? Eur. J. Immunol. 39: 2059–2064. 7. Abdul-Careem, M. F., A. J. Lee, E. A. Pek, N. Gill, A. E. Gillgrass, M. V. Chew, 37. Min-Oo, G., Y. Kamimura, D. W. Hendricks, T. Nabekura, and L. L. Lanier. S. Reid, and A. A. Ashkar. 2012. Genital HSV-2 infection induces short-term NK 2013. Natural killer cells: walking three paths down memory lane. Trends cell memory. PLoS One 7: e32821. Immunol. 34: 251–258. The Journal of Immunology 11

38. Cooper, M. A., and W. M. Yokoyama. 2010. Memory-like responses of natural 44. Liu, Q., Y. H. Zhou, and Z. Q. Yang. 2016. The cytokine storm of severe in- killer cells. Immunol. Rev. 235: 297–305. fluenza and development of immunomodulatory therapy. Cell. Mol. Immunol. 39. Mandelboim, O., N. Lieberman, M. Lev, L. Paul, T. I. Arnon, Y. Bushkin, 13: 3–10. D. M. Davis, J. L. Strominger, J. W. Yewdell, and A. Porgador. 2001. Recog- 45. Jiang, X., Y. Chen, H. Peng, and Z. Tian. 2013. Memory NK cells: why do they nition of haemagglutinins on virus-infected cells by NKp46 activates lysis by reside in the liver? Cell. Mol. Immunol. 10: 196–201. human NK cells. Nature 409: 1055–1060. 46. Mescher, M. F., J. M. Curtsinger, P. Agarwal, K. A. Casey, M. Gerner, 40. Wang, J., F. Li, H. Wei, Z. X. Lian, R. Sun, and Z. Tian. 2014. Respiratory C. D. Hammerbeck, F. Popescu, and Z. Xiao. 2006. Signals required for pro- influenza virus infection induces intestinal immune injury via microbiota- gramming effector and memory development by CD8+ T cells. Immunol. Rev. 211: 81–92. mediated Th17 cell-dependent inflammation. [Published erratum appears in 47. Welsh, R. M. 2001. Assessing CD8 T cell number and dysfunction in the 2014 J. Exp. Med. 211: 2683.] J. Exp. Med. 211: 2397–2410. presence of antigen. J. Exp. Med. 193: F19–F22. 41. Sun, J. C., S. Madera, N. A. Bezman, J. N. Beilke, M. H. Kaplan, and 48. Fuller, M. J., and A. J. Zajac. 2003. Ablation of CD8 and CD4 T cell responses L. L. Lanier. 2012. Proinflammatory cytokine signaling required for the gener- by high viral loads. J. Immunol. 170: 477–486. ation of natural killer cell memory. J. Exp. Med. 209: 947–954. 49. Ochel, A., M. Cebula, M. Riehn, U. Hillebrand, C. Lipps, R. Schirmbeck, 42. Firth, M. A., S. Madera, A. M. Beaulieu, G. Gasteiger, E. F. Castillo, H. Hauser, and D. Wirth. 2016. Effective intrahepatic CD8+ T-cell immune K. S. Schluns, M. Kubo, P. B. Rothman, E. Vivier, and J. C. Sun. 2013. Nfil3- responses are induced by low but not high numbers of antigen-expressing he- independent lineage maintenance and antiviral response of natural killer cells. J. patocytes. Cell. Mol. Immunol. 13: 805–815. Exp. Med. 210: 2981–2990. 50. Karo, J. M., D. G. Schatz, and J. C. Sun. 2014. The RAG recombinase dictates 43. Lee, S. H., T. Miyagi, and C. A. Biron. 2007. Keeping NK cells in highly reg- functional heterogeneity and cellular fitness in natural killer cells. Cell 159: 94– ulated antiviral warfare. Trends Immunol. 28: 252–259. 107. Downloaded from http://www.jimmunol.org/ by guest on September 25, 2021 D ersn he needn xeiet iha es he ieprgroup. per mice three least at with experiments independent three represent A Kclsi h ugwscutdA,teepesoso D9adNGDo Kclswr eemndb Flow by determined were cells NK on NKG2D and of CD69 number the of later, expressions days Three the IFN- (PR8). counted(A), of virus A was expression the influenza lung and of the HA Cytometry(B), 0.1 in with cells intranasally infection. infected NK virus were influenza mice primary C57BL/6 after (A-C) proliferation viral inhibit cells NK lung Activated 1. Figure Supplemental fP8ifcin h ia ie nteln a eemndb DKclsdtriainasya as3ad7after 7 and 3 days at assay determination cells MDCK 0.5HA by after injected determined were hours were mice 72 was mice mean C57BL/6 and lung C57BL/6 as (G) before the (F) infection. expressed hours the in after 24 and (E). are titer 7 depletion infection, Data infection and viral cells PR8 infection. 3 after The NK of days 7 for infection. 0.5HA at antibody and before PR8 PCR AsGM1 hours 3 of Real-time anti-mouse 24 by days rabbit depletion determined at with cells was PCR intravenously lung NK the for Real-time in (PK136) by titer antibody viral determined anti-NK1.1 was with lung i.p. the injected in titer viral Nfil3 20 40 60 80 100 110 -/- Control 0 60 70 80 90 0 iewr netditaaal ih05H fP8 oyegtwsmntrddiyatrifcin() n the and (D), infection after daily monitored was Bodyweight PR8. of HA 0.5 with intranasally infected were mice **

1 PR8 Days after PR8 infection PR8 after Days 2 Nfil3 mice C57BL/6 B 3 -/- mice Isotype *** CD69 4 ** 5 Control PR8 6 * igigLi Tingting 7 * NKG2D Relative concentration of E influenza matrix protein ± Ȗ E yaStudent’s a by SEM nN elwsdtrie yFo yoer() DE 5B/ ieand mice C57BL/6 (D-E) Cytometry(C). Flow by determined was cell NK in et al. et Control 10 20 30 40 0 *** PR8 Supplemental Figure 1

Control F t Relative concentration of PR8 et ,P<00;*,P<00;** .0.Data 0.001. < P ***, 0.01; < P **, 0.05; < P *, test. influenza matrix protein C IFN-Ȗ Control PR8 NK1.1 я G

PFU/ml in lung(*106) я 10 15 20

Control 0 5 * PR8 Tingting Li et al. Supplemental Figure 2

A B Pre-sort Post-sort non-stimulation 0.4 ns Control PR8 0.3 0.2 Lung NK 0.1 0.0 Lung

PR8

Ȗ Control Liver

IFN- 0.8 ns DX5+ NK 0.6 Liver 0.4 0.2 0.0 Liver DX5- NK NK1.1 PR8 CD3 DX5 Control NK1.1 NK1.1

C Lung Liver inactivated PR8 inactivated RSV inactivated PR8 inactivated RSV

Control Control

PR8 PR8

IFN-Ȗ IFN-Ȗ

Supplemental Figure 2. Purified NK cells from influenza virus-infected mice expressed higher level of IFN-Ȗ after inactivated influenza virus stimulation in vitro. (A) C57BL/6 mice were intranasally infected with 0.05 HA of PR8. 30 days later, the expression of IFN-Ȗ in NK cells from control and PR8-infected mice was determined by Flow Cytometry without any stimulation. (B) Purity determination of lung and liver NK cells after sorting. DX5 expression of donor NK cells before transfer is shown pre- and post-sort. (C) Lung and liver NK cells from PR8-infected mice and control mice were purified and then stimulated with inactivated PR8 and RSV. IFN-Ȗ in cell- free supernatants was determined by using cytometric bead array (CBA, BD Biosciences) according to the manufacturer’s protocol. Data are expressed as mean ± SEM by a Student’s t test. ns: not significant. Data represent three independent experiments with at least three mice per group. Tingting Li et al. Supplemental Figure 3

A C E 2 105 Control mice-derived spleen NK cells (n=4) PR8-infected mice-derived spleen NK cells (n=6) 110 100 90 80 70 60 % of initial bodyweight % of initial body weight 0 2 4 6 8 10 Days after PR8 infection B D F Control mice-derived spleen NK cells (n=7) ns PR8-infected mice-derived spleen NK cells(n=8) 100 80 60 40 20 Percent survival Percent Percent survival Percent 0 survival Percent 0 3 6 9 12 Days after PR8 infection

Supplemental Figure 3. Adoptive transfer of lung and splenic NK cells from PR8-infected mice fails to protect recipient mice against influenza-caused death.

(A-F) 1×106 Lung and 1×106 or2×105 splenic NK cells isolated from control or PR8-infected C57BL/6 mice were adoptively transferred into neonatal C57BL/6 mice (7-10 d), and subsequently the recipient mice were infected with a lethal dose of influenza virus 30 days later. Changes in the body weight (A, C, E) and survival rate of recipient mice (B, D, F) were analyzed. Data are expressed as mean ± SEM by a Student’s t test. The Log-rank (Mantel-Cox) test was used to evaluate survival rate. ns: not significant. Data were pooled from 3 experiments. Tingting Li et al. Supplemental Figure 4

A ) B C 4 15 * 25 ** Control PR8 20 *** 20 10 15

15 Ȗ NK cells NK

+ 10

10 Ȗ 5 IFN- 5 5

0 0 %IFN- 0

Number of NK cells(x10 ol PR8 PR8 PR8 NK1.1 PR8 Control Contr Control Control D Blank 1dpi 2dpi 3dpi E

Lung homogenate (ng/ml) (ng/ml) IFN- Liver IFN- homogenate

Supplemental Figure 4. Liver NK cells were activated by influenza virus infection in an indirect way.

(A-C) C57BL/6 mice were intranasally infected with 0.1 HA of PR8. Three days later, the number of NK cells in the liver was counted (A), the expressions of CD69 and NKG2D on NK cells were determined by Flow Cytometry (B), and the expression of IFN-Ȗ in NK cells was determined by Flow Cytometry(C). (D-E) C57BL/6 mice were intranasally infected with 0.5 HA of PR8. PR8 titers in the lung and liver were determined by MDCK cells determination assay at the indicated times after infection (D), and IFN-Į levels in BALF and serum were detected by ELISA (E) . Data are expressed as mean ± SEM by a Student’s t test. *, P < 0.05; **, P < 0.01; ***, P < 0.005; ****, P < 0.001; ns: not significant. Data represent three independent experiments with at least three mice per group.