Spontaneous focal activation of invariant natural killer T (iNKT) cells in mouse liver and kidney
© Zeng and Howard. 2010
Received: 17 June 2010
Accepted: 30 November 2010
Published: 30 November 2010
Skip to main content
© Zeng and Howard. 2010
Received: 17 June 2010
Accepted: 30 November 2010
Published: 30 November 2010
Invariant natural killer T (iNKT) cells differ from other T cells by their hyperactive effector T-cell status, in addition to the expression of NK lineage receptors and semi-invariant T-cell receptors. It is generally agreed that the immune phenotype of iNKT cells is maintained by repeated activation in peripheral tissues although no explicit evidence for such iNKT cell activity in vivo has so far been reported.
We used an interferon (IFN)-γ-inducible cytoplasmic protein, Irga6, as a histological marker for local IFN-γ production. Irga6 was intensely expressed in small foci of liver parenchymal cells and kidney tubular epithelium. Focal Irga6 expression was unaffected by germ-free status or loss of TLR signalling and was totally dependent on IFN-γ secreted by T cells in the centres of expression foci. These were shown to be iNKT cells by diagnostic T cell receptor usage and their activity was lost in both CD1 d and Jα-deficient mice.
This is the first report that supplies direct evidence for explicit activation events of NKT cells in vivo and raises issues about the triggering mechanism and consequences for immune functions in liver and kidney.
Invariant natural killer-like T (iNKT) cells are placed ambiguously between adaptive and innate immune systems (reviewed in [1, 2]). Derived from the thymus, expressing rearranged T-cell receptor (TCR) alpha and beta chains, they seem to belong to the adaptive immune system, while their receptor homogeneity, their continuous state of activation, their rapid secretion of large amounts of interferon (IFN)-γ and interleukin (IL)-4, their presumed recognition of invariant glycolipid self-ligands associated with the non-classical major histocompatibility (MHC) class I molecule, CD1 d, recall various aspects of innate immune recognition. Many features of iNKT cell behaviour are puzzling: their thymic ontogeny and relation to the classical pathways of T-cell mediated differentiation; the relative importance of endogenous and exogenous ligands in activation; and the polarity of their cytokine profile towards IFN-γ or IL-4 in relation to the activating ligand. However, in this report we address the basis for another characteristic of these enigmatic cells, namely their constitutive state of readiness to respond with massive cytokine production (reviewed in ). Using a sensitive endogenous reporter for IFN-γ production, we show that iNKT cells are constitutively and endogenously activated to IFN-γ production in the liver and kidney of normal mice. The activation is apparently restricted to these tissues, focal, spontaneous and independent of signals derived from bacteria or viruses. It is, however, dependent on the expression of CD1 d and on the presence of the classical iNKT cell receptor. The results suggest that the constitutive state of iNKT readiness may be maintained by intermittent local stimulation with endogenous ligands.
Secondly, if IFN-γ is responsible for the focal induction of Irga6, there should be activation in the high expression patches of the IFN-γ-dependent promoter of the Irga6 gene, with the use of exon 1A, in contrast to the general hepatic parenchyma in which the liver-specific promoter is constitutively expressed with usage of exon 1B . We therefore isolated high expression patches and areas of general expression level by laser microdissection and used quantitative real-time polymerase chain reaction (qRT-PCR) in order to identify the use of the IFN-γ-dependent 5'-untranslated 1A exon and the constitutively expressed 5'-untranslated 1B exon in the two tissue sites (Figure 2C). Expression of the IFN-γ-dependent exon 1A was 27.5× higher by qRT-PCR in the high expression patches than in the general liver, while 1B exon expression was not altered (1.5×). However, when the Irga6-1A PCR products from the general liver were cloned and sequenced, only 5.9% of were specific (a consequence of the very low absolute expression of the IFN-γ-inducible transcript in non-focal hepatic parenchyma), while 100% of those cloned from the high expression patches were specific. Thus, the true excess of Irga6-1A in the high expression patches relative to the general parenchyma was 100 × 27.9/5.9 or about 470×, confirming the intense local induction by IFN-γ.
Thirdly, if IFN-γ is being secreted locally, affected cells should express not only Irga6 but also other members of the IFN-γ-inducible IRG protein family that are not constitutively expressed in hepatic parenchymal cells . We therefore stained adjacent histological sections of liver and kidney for Irga6 and for Irgm3. Indeed, focal expression of Irgm3 coincided accurately with the high Irga6 expression patches (Additional file 3 : Figure S2).
Direct evidence for the expression of IFN-γ in the patches with mononuclear cell cores, but not in the coreless patches, was obtained by RT-PCR for IFN-γ transcripts in microdissected patch material. A strong signal for IFN-γ was recovered from only the patches with mononuclear cell cores (Figure 2D). These experiments established beyond doubt that the liver and kidney contain reactive foci of local, T-cell-dependent IFN-γ production identified by the local high expression of Irga6.
The character of the small IFN-γ-secreting reactive foci in the liver and kidney suggested that they could be due to local immune activity stimulated by microbial material. We therefore analysed the liver and kidney for reactive foci from mice deficient in components of the Toll-le receptor (TLR) system (Figure 2B). No effects were seen in mice deficient in TLR2, 4 or 9 or MyD88, arguing against bacterial components as the initiators of the foci. In order to test this conclusion further, we examined the liver and kidney from germ-free mice from three independent sources. No significant reduction in numbers of reactive foci in liver or spleen was detected (Figure 2B). Bacterial infection is, thus, unlikely to be a stimulus for the reactive foci
If the Irga6 high expression foci with mononuclear cores associated with high IFN-γ production were, indeed, due to iNKT cells, they should also be absent from mice deficient in the essential ligand-presenting molecule CD1 d [20, 21] and in mice deficient in iNKT cells as a result of deletion of the diagnostic Jα18 T cell receptor segment . Both these predictions were fulfilled, with a substantial loss of high Irga6 expression patches from the kidney and cored patches from the liver in both mutants (Figure 4B). The patches of high Irga6 expression in the liver without mononuclear cores were not affected in either case. In order to find out whether CD4+ or CD8+ T cells play any additional role in the induction of high Irga6 expression patches, organs from mice lacking MHC Aβb (CD4+ T cell deficient ) and TAP1 (CD8+ T cell deficient ) were also examined. Patches of high Irga6 expression were normal in liver and kidney from both strains. Thus, iNKT cells alone appear to be responsible for the high Irga6 expression foci.
We present direct evidence for the constitutive activation of mouse iNKT cells in two organs, the liver and the kidneys, generating small foci of high IFN-γ expression detected by the local high expression of an IFN-γ-inducible cytoplasmic protein, Irga6. In a previous study, we surveyed the pattern of constitutive Irga6 expression in a number of other organs and high expression foci were not seen . The liver and kidneys thus appear to be special in this respect. Our results suggest a local activation event that triggers production of type II interferon by iNKT cells. The mouse liver sinusoids are known to contain a major population of NKT cells and NKT cells have been reported in the kidney, probably corresponding to the T cells identified in the tubulo-interstitial spaces . Dendritic cells (DC) and hepatic stellate cells (Ito cells)  can both present exogenous α-GalCer via CD1 d to iNKT cells. Stellate cells of the liver sinusoids have also been shown to be IL-15-dependent activators of iNKT cells to proliferation in the absence of exogenous antigen . An apparently similar cell type is also found in the tubular cortex of the kidney essentially co-localizing with dendritic cells [5, 27]. Both DC and stellate cells are strong candidates for the activating cell implied by our experiments, but this does not, by itself, account for the focal triggering event. Our experiments with germ-free and TLR-deficient mice make bacterial products acting on DC TLRs [28, 29] an unlikely cause. More plausible would, perhaps, be a locally reactivating endogenous retrovirus triggering DC or stellate cells and, thereby, iNKT cells locally. Arguing against a primary viral stimulus, we could show no deficit in liver cored high expression patches or kidney patches in organs from mice deficient in MAVS/CARDIF , the common activation pathway of the cytosolic RIG helicases that act as viral sensors (Figure 4C).
It will be interesting to know how the spontaneous activation of iNKT cells and consequent constitutive IFN-γ production in liver and kidney of major inbred strains of mouse affects immunological function in these organs. NKT cells have been identified very recently as a minority population among normal resident lymphocytes in mouse kidney and both an excitatory role in ischaemia-reperfusion injury  and an inhibitory role in experimental glomerular nephritis induction by anti glomerular basement membrane antibody [32, 33] have been described. Likewise, the large population of NKT cells that patrol the liver sinusoids  have been implicated in various liver disorders , but their spontaneous, focal activation has not been reported. Spontaneous focal activation of NKT cells in the liver and kidney with high local levels of IFN-γ will have to be considered in future studies of immune function in these organs.
Invariant natural killer T (iNKT) cells are activated spontaneously in mouse liver and kidney. These activation events are local, probably extravascular, and result in the local secretion of IFN-γ. This in turn results in the local induction of IFN-γ-responsive genes in non-immune tissue cells, namely hepatic parenchymal cells and renal tubular epithelial cells. These activation events are spontaneous in the sense that they appear to be independent of exogenous pathogenic material of bacterial or viral origin. The cause of individual activation events is still unclear. In general, however, the results are consistent with a widely held view that iNKT cells are rendered spontaneously active by interaction with an endogenous ligand. In this study we demonstrate these activation events histologically. Further investigations may reveal the nature of the stimulatory event that lies at the heart of each activating focus. It will be important to generalise these findings to the human case, and to assess the implications of these findings for liver and kidney immunopathology.
C57BL/6J mice were obtained from the specific pathogen-free mouse facility in the Institute for Genetics Uni. Köln. We are enormously indebted to the many colleagues and institutions that provided mice or organs for this study. TLR2-/- , TLR4-/- , TLR9-/- , TLR2/4-/- and MyD88-/-  mice were provided by Professor Dr Marina Freudenberg (Max Planck Institute for Immunobiology, Freiburg). IFN-γ-/- mice , IFN-receptor deficient mice (IFNAR, ), (IFNGR, ) and STAT-1-/- mice  were provided by Dr Thomas Kolbe (University of Veterinary Medicine, Vienna). RAG-1-/- mice  were provided by Dr Heike Weighardt (Technische Universitaet Muenchen, Munich). JHT mice  were provided by Dr Ari Waisman (Johannes Gutenberg-Universität Mainz, Mainz, Germany). Germ-free C57BL/6 mice were provided from three independent facilities, namely, University of Zürich, Zürich, Switzerland (Mr Rudolf Jörg), the Gulbenkian Institute of Science, Oeiras, Portugal (Dr Jocelyne Demengeot) and the Karolinska Institute, Stockholm. CD1d-/- mice  were provided by Professor Luc van Kaer (Vanderbilt University Medical Center, TN, USA). Jα18-/- mice  were provided by Dr Maria C. Leite-de-Moraes, Hôpital Necker, Paris, France. MAVS/CARDIF-deficient mice  were provided by Dr Jürg Tschopp (Lausanne, Switzerland). MHC Aβ-/- mice  were provided by Dr Jocelyne Demongeot (Gulbenkian Institute of Sciences, Oeiras, Portugal). TAP1-/- mice  were purchased from the Jackson Laboratory (Maine, US). For all assays, wild-type C57BL/6 mice from the same donor facility were used as controls. To control germ-free C57BL/6 mice, C57BL/6 mice from colonies maintained at SPF level from the same suppliers were used. All the above mice were on the C57BL/6J background. SWR/J, SJL, C57BR/J, C57L/J, RIIIS/J, A/J and C57BR/cdJ strains were purchased from The Jackson Laboratory (Maine, USA). In most cases, liver and kidneys were removed at the donor animal facility, immediately fixed in cold TBS/4% paraformaldehyde and shipped at 4°C by the quickest route to the Institute for Genetics in Cologne, Germany.
For paraffin sections, mouse tissues were fixed in TBS/4% paraformaldehyde at 4°C and dehydrated through an ethanol series at 4°C (50%, 70%, 90% and 96%). Tissues were then transferred into isopropanol and finally into a paraffin:isopropanol (1:1) solution at 60°C. The isopropanol was evaporated and fresh paraffin was then replenished several times at 60°C before the tissues (in paraffin) were moved to room temperature. The embedded tissues were cut with a microtome RM 2065 (Leica Microsystems, Wetzlar, Germany) into 6 μm thick serial sections. For cryosections, mouse liver was snap-frozen in liquid nitrogen and cut into consecutive serial sections (6 μm) using a cryotome CM 3050 S (Leica Microsystems).
Paraffin sections were de-waxed in xylene and post-fixed in 4% paraformaldehyde (1 h, RT). For Irga6 or Irgm3 staining, protein epitopes were demasked (10 min, boiling) in 10 mM citrate buffer (pH 6.0). For F4/80 or CD3 staining, protein epitopes were demasked by 0.1% Trypsin (SIGMA Type II) solution (0.1% CaCl2, pH7.8) for 30mins at 37°C or in 1 mM EDTA (pH8.0) for 10 min at 100°C, respectively. Sections were then saturated with quenching buffer (0.3% H2O2, 20 min). After phosphate buffered saline (PBS) washing, the sections were probed either with rabbit anti-Irga6 antiserum 165/3  or mouse anti-Irgm3 mAb (BD Transduction Laboratory) in DAKO diluent. HRP-staining was carried out with the horseradish peroxidase (HRP)-substrate kit HistoGreen (Linaris, Wetzlar, Germany). Nuclei were counterstained with Nuclear Fast Red. For Irga6 and F4/80 or CD3 double staining, Irga6 was first probed with 165/3 and stained with HRP-substrate AEC (Sigma, MO, USA). The sections were then washed with PBS (1 h) and subsequently probed with anti-F4/80 mAb (Serotec, NC, USA) or anti-CD3 mAb (Vector Laboratories, Linaris). The AP-staining was then carried out with AP-substrate kit III (Vector laboratory, Linaris). For staining of cryo-sections, sections were fixed first with cold acetone (10 min), before probing with anti-Irga6 antiserum 165/3 or anti-TCR Vβ8.1/8.2 mAb (BD Pharmingen, CA, USA). The sections were then stained with the HRP-substrate kit HistoGreen for Irga6 or with AP-substrate kit III for TCR Vβ 8. Samples were analysed using Zeiss Axioplan II microscope (Zeiss, Jena, Germany) equipped with SPOT RT slider digital camera (Diagnostic Instruments, MI, USA).
Irga6 high expression foci were counted on stained 6 μm histological sections using 100× total magnification. Sections 300 μm apart were examined in order to avoid repeated counts of the same high expression patch. A total of 30 fields were counted from each organ, from which the mean number of high expression patches per field was estimated. The raw data from all such determinations are presented in Additional file 1 : Table S1.
Consecutive serial cryosections (6 μm) for liver were prepared using a cryotome CM 3050 S (Leica Microsystems) and sections were carried on PALM MembraneSlides (P.A.L.M. Microlaser Technologies, Bernried, Germany), which was covered with a polyethylene naphthalate membrane. In order to avoid the RNA degradation caused by the staining of sections with antibody, the following procedure was followed. Consecutive sections were numbered (1, 2, 3, 4...). All sections with odd-numbers (1, 3...) were on one slide (named 'A') and the sections with even numbers (2, 4...) on another slide (named 'B'). Both slides were then fixed in 70% ethanol (-20°C). The B slide was stored at -80°C and the A slide was stained for Irga6 with 165/3 using histogreen substrate. Geographical locations of Irga6 focal expression on the A slide were recorded photographically. The B slide was then stained only with Nuclear Fast Red, dehydrated in an ethanol series (70%, 96%, 100% 2 min each) and dried at 50°C. Irga6 focal expression was located on the B slide by referring to the photos of adjacent sections on the A slide. The Irga6 expression foci and non-focus region on the B slides were then collected by laser microdissection (LMD) equipment generously made available by Dr Margarete Odenthal (Institute for Pathology, University of Cologne, Cologne, Germany) consisting of an Interface Microbeam Mini laser (P.A.L.M Microlaser Technologies) and an Axiovert 135 microscope (Zeiss). Total RNA from the dissected samples (50-100 foci) was extracted with the RNeasy Micro kit (QIAGEN, Hilden, Germany) and cDNA was synthesized with half of the total RNA with the Superscript First-Strand Synthesis System for RT-PCR kit (Invitrogen, CA, USA).
PCR was carried out in a Light Cycler I System (Roche, Berlin, Germany) using a LightCycler SYBR Green I PCR kit (Roche). The 5' primers specific for Irga6 1A and 1B together with the common 3' primer on the coding exon were used to detect expression level of Irga6-1A and Irga6-1B transcript forms as target genes . Mouse GAPDH gene was used as a reference gene and input control (for primers see Additional file 4 : Table S2). 2-4 μL of cDNA from materials collected by LMD was used as templates. Primer efficiency for Irga6-1A, -1B or mGAPDH was determined using liver cDNA dilution series (1, 1/10, 1/100...) as templates as described previously . The proportional increase in Irga6-1A or -1B expression for 'foci' to 'non-foci' was then determined according to . Melting curve analyses were performed in order to verify the amplification specificity. Each sample was tested in duplicate or triplicate. In order to determine whether the final PCR products amplified for Irga6-1A were -1A specific, 1A PCR products were cloned in to pGEM-T-easy (Promega, CA, USA) vector and sequenced. The percentage of -1A specific clones was then calculated. For Irga6 high expression foci, all clones amplified for Irga6-1A were -1A, while for the non-focal material, only 5.9% of clones amplified for Irga6-1A were -1A. The directly determined enrichment factor of 27.9 for Irga6-1A in Irga6 high expression foci (Figure 2C) can, therefore, be multiplied a factor of 16.9 (100/5.9) in order to give a relative enrichment of over 470 in Irga6-1A compared with non-focal tissue.
Total RNA from mouse liver or lymph nodes was extracted using the RNeasy Mini kit (QIAGEN) and cDNA was synthesized using the Superscript First-Strand Synthesis System for RT-PCR kit (Invitrogen). One mcrolitre cDNA from the liver and T cells or 2-4 μL of cDNA from laser dissected materials were used in each of the following RT-PCR reactions as templates. Primers located in different exons were used to generate specific products only from cDNA but not from genomic DNA (Additional file 4 : Table S2). For nested-PCR, 1 μL PCR product of the first round of PCR was used as template for the second PCR, making use of a nested primer and the common 3' or 5' primer. The cycle number was always 45 for each round of PCR. All final PCR-products were verified by sequencing.
In order to determine the TCR Vβ usage and Vα14 junctional diversity, nested-RT-PCR was performed. cDNA from more than 50 Irga6 cored patches collected by LMD or cDNA of 1000 lymphocytes from mesenteric and cervical lymph nodes were synthesized and nested-RT-PCR was performed as described in  The Vβ8 and Cβ primers  Vα14 and Cα primers , Vα2, Vα8, Vα17 primers  were described before (Additional file 4 : Table S2). The final products of nested-PCR were cloned into pGEM-T-easy (Promega) and sequenced. The sequences were compared to classical TCR V sequences in the database online using IMGT/V-QUEST http://imgt.cines.fr/. Clones with a sequence identity of more than 95% to known TCR V sequences were identified as correct TCR V clones. Clones were classified into groups with same junctions. Representatives from each group were selected aligned and analysed with free software GeneDoc (Version 2.6.002; http://www.nrbsc.org) and Vector NTI (Version 9, Invitrogen). See Additional file 5 : Table S3 for complete junction sequence data.
invariant natural killer-like T cell
IFN type I receptor
IFN type II receptor
mouse mammary tumour virus
phosphate buffered saline
polymerase chain reaction
quantitative reverse transcription PCR
We are very much indebted to Dr Margarete Odenthal, Institute of Pathology, University of Cologne, Cologne, Germany, for allowing us to make extensive use of her LMD facility. This work would not have been possible without the generosity of many scientists who provided knock-out and wild-type mice (itemized in Materials and Methods) and in some cases also enabled us to harvest tissues in their facilities. For this our thanks are due to Marina Freudenberg (Freiburg, Germany), Maria Leite de Moraes (Paris, France), Luc Van Kaer (Vanderbilt, TN, USA), Jocelyne Demengeot (Oeiras, Portugal), Jurg Tschopp (Lausanne, Switzerland), Rudolf Jörg (Zürich, Switzerland), Thomas Kolbe (Vienna, Austria), Heike Weighardt (Munich, Germany) and Ari Waisman (Mainz, Germany). We are grateful to two anonymous referees for suggesting that we identify iNKT cells more explicitly by examining tissues from mice deficient in Jα18, TAP and class II MHC. We owe special thanks to Pia Scholl, Institute for Genetics, for her devoted attention recently to the immunohistology of these as well as of the MAVS/CARDIF deficie.
This work was supported in part by the University of Cologne, Cologne, Germany, and in part by grants in the Schwerpunktprogramm 1110 and the collaborative research programme, SFB670 'Cell-autonomous Immunity', both from the Deutsche Forschungsgemeinschaft. Jia Zeng was supported in part from the International Graduate School in Genetics and Functional Genomics.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.