Developmental control of the Nlrp6 inflammasome and a substrate, IL-18, in mammalian intestine

Animal methods.

All experimental procedures were carried out in accordance with the UK Animals (Scientific Procedures) Act 1986 under an appropriate Home Office License and with approval from the local ethical review process.

Rats.

Twenty time-mated pregnant Wistar rats were euthanized by CO₂ overdose at embryonic days 16 (E16; n = 10) and 20 (E20; n = 10), where day 0 was defined as the day of plug and term is 21 days. Pups were delivered by caesarean section or born naturally for postnatal experiments and euthanized by cervical dislocation; tissue including lung and whole intestine was removed under a dissecting microscope. Tissue from multiple siblings was snap frozen in liquid nitrogen prior to storage at −80°C for subsequent RNA or protein analysis. For immunohistochemistry some siblings were fixed by immersion in ice cold neutral buffered formalin for 6 h, then washed twice with PBS, transferred to 70% ethanol and embedded in paraffin wax.

Sheep.

Twenty Welsh Mountain pregnant ewes of known gestational age were kept in individual pens and maintained on 200 g/day concentrates with free access to hay, water, and a saltlick block. Fetuses were delivered for tissue collection by caesarean section under general anesthesia (20 mg/kg sodium pentobarbitone iv) at 100 (n = 5), 114–116 (n = 5), 129–131 (n = 5), and 144–145 (n = 5) days gestation where term is 145 ± 2 days. Immediately following delivery, lambs were euthanized with sodium pentobarbitone (200 mg/kg). During necropsy, samples of fetal jejunum were collected midway between the pyloric sphincter and ileocecal junction. Tissues were frozen in liquid nitrogen and stored at −80°C until analysis.

Microarray analysis.

Transcript profiling using microarray analysis was performed by using lung and intestine RNA from a single animal from each of 10 litters at a given gestational age. Array experiments were performed by the Genomics CoreLab, Cambridge Biomedical Research Centre. Briefly, the total RNA was processed by using Affymetrix one-cycle target labeling protocol (Santa Clara, CA) and hybridized to Affymetrix Rat Genome 230 2.0 GeneChips. Raw data from transcript profiling experiments are available on the Gene Expression Omnibus Database (http://www.ncbi.nlm.nih.gov/geo/), accession ID: GSE16849.

Data were normalized by robust multiarray averaging (RMA) and quantile normalization using LIMMA (27) (http://bioinf.wehi.edu.au/limma/). Normalized transcript abundance data were compared between E16 and E20 by two independent methods: the Cyber-T algorithm (15) and Rank Product Analysis (3). The Cyber-T algorithm is an unpaired t-test, modified by the inclusion of a Bayesian prior based on the variance of other transcripts in the data set. Transcripts that were significantly regulated in both Cyber-T (Bayes P value <0.001, posterior probability of differential expression >0.99) and Rank Product Analysis (P < 0.0001) and showed an absolute fold change of more than five were defined as differentially expressed. Microarray data were annotated by using the NetAffx Analysis Center (Affymetrix) files. To generate the list of up-or downregulated genes only Entrez Genes or UniGene clusters were considered if at least one probe set gave an unambiguous match.

Innate immunity gene lists.

Gene ontology (GO) analysis was used to determine whether genes in a given functionally related group were up- or downregulated. Enrichment of GO terms among the significant genes was studied by using FatiGo, part of Babelomics 4.0 suite (18) (http://www.babelomics.org), a two-tailed Fisher exact test was used with statistical significance set at P < 0.05. However, the GO categories “innate immune response” (GO:0045087) or “inflammatory response” (GO:0006954) are not comprehensive and do not include genes with a clear role in defense responses (e.g., several interleukins and pattern recognition receptors). Therefore, we obtained a curated nonredundant list of 5,070 human and mouse innate immune system genes from the Innate Database [InnateDB, URL: http://www.innatedb.ca (16), accessed on 12/06/09]. The rat orthologs of the innate immune system genes were identified by using Biomart (www.ensembl.org), resulting in a list of 4,185 rat Ensembl genes. The Affymetrix array employed in this study included a total of 11,392 Ensembl genes. Of the 4,185 rat Ensembl genes, 2,709 were represented on the array. The significant reduction in numbers is because many rat Entrez Genes, and all UniGene clusters, have no rat Ensembl genes linked to them.

Mapping of transcription factor binding sites.

The Affymetrix probe sets upregulated in the rat intestine were first converted to Ensembl gene IDs by using Biomart. The Ensembl IDs of the mouse and human homologs were also obtained from Biomart by inputting the rat Ensembl gene IDs. To identify overrepresented transcription factor binding sites (TFBSs) the Ensembl gene IDs for rat, mouse or human were subjected to TFBS enrichment analysis using FactorY (4, 10). A 1,000 bp region upstream of the transcription start site (TSS) was analyzed. Both Jaspar (http://www.genereg.net) and Transfac (http://www.gene-regulation.com/) databases were used for searches. FactorY uses hypergeometric distribution to calculate the P value and a false discovery rate (FDR) of 0.05 was used for multitest-correction. Hierarchal clustering analysis of TFBSs was performed using Partek Genomic Suite 6.5 (Partek). TFBSs that occurred in all three species with p < 0.05 were considered significant.

To locate peroxisome proliferator-activated receptor γ-retinoid X receptor-α (PPARG-RXRA) and chicken ovalbumin upstream promoter transcription factor 1 (coup-TF/NR2F1) binding sites 1,000 bp upstream of the TSS, rat, mouse, and human genes were extracted by use of Biomart (Ensembl 58; rat:RGSC3.4, mouse:NCBIM37, human:GRCh37) and aligned by use of ClustalW2 (http://www.ebi.ac.uk/clustalW). Sequences were scanned for TFBSs using matrices from the Jaspar database for PPARG-RXRA (MA0065) and coup-TF (MA0017) by using a threshold score of 80%.

RNA extraction.

Total RNA was purified from rat and sheep frozen tissue by using TRIzol (Invitrogen, Paisley, UK) and treatment with DNaseI (Promega, Southampton, UK) for 30 min at 37°C. The reaction was stopped by addition of stop solution and heating at 65°C for 10 min. RNA was extracted from Caco2 cells using spin columns with on-column DNaseI digestion according to the manufacturer's instructions (Macherey Nagel, Düren, Germany). RNA integrity was confirmed by using RNA 6000 Nano chips on an Agilent 2100 Bioanalyzer (Agilent, Stockport, UK). Only RNA with an optical density 260/280 >1.8 was used for quantitative RT-PCR (qRT-PCR) and RNA with a RNA integrity number greater than 7 was used for array hybridization.

Sheep cDNA cloning.

The sheep genome has not yet fully been sequenced but some sequence information is available in the NCBI expressed sequence tag (EST) database (13). A BLASTN search of the sheep EST database revealed that, for some genes (caspase-1, IL-18, and sucrase-isomaltase), sequences were already available and these sequences were used to design primers for PCR. In the case of pycard, where no sheep EST was available, the human, mouse, rat, and bovine sequences were aligned and degenerate primers were designed to amplify a partial ovine coding sequence. All primers are listed in Table 1. PCR (caspase-1, IL-18, sucrase-isomaltase, and pycard) was performed in 25 μl containing a final concentration of 1.25 mM MgCl₂, 200 μm of each dNTPs, 800 nM of each primer pair, and 0.1 units/μl BioTaq DNA polymerase (Bioline, London, UK) and sheep intestinal cDNA as template. The PCR conditions were initial denaturation at 95°C for 5 min followed by 45 cycles of 95°C for 45 s, 55°C for 45 s, 72°C for 90 s, and a final extension step of 72°C for 10 min. In some cases (caspase-1, IL-18, and sucrase-isomaltase) 5% DMSO was added to reduce secondary structures. PCR products were separated on 1.5% agarose gels and bands of the expected size were excised and purified by using the Qiaquick Gel Extraction Kit (Qiagen, Crawley, UK) then cloned into the pGEM-T-Easy plasmid (Promega). For each gene, at least two clones were sequence verified.

qRT-PCR.

Total rat, sheep, or human RNA was reverse transcribed to cDNA by use of Superscript II (Invitrogen) and used as template for qRT-PCR with an ABI Prism 7900HT system (Applied Biosystems, Warrington, UK). For rat transcripts, qRT-PCR was performed by using cDNA derived from 100 ng of RNA as template with Absolute Fast qPCR mix (Abgene, Epsom, UK) and primers and probes for Nlrp6, Pycard, Caspase-1, IL-18, sucrase-isomaltase, and 18S. These primers and probes were predesigned and preoptimized (Applied Biosystems: Rn00592690_m1, Rn00597229_g1, Rn00562724_m1, Rn00564957_m1 and Rn00824548_m1, respectively). All samples were assayed in triplicate. Expression levels were quantified by the standard curve method with normalization to 18S (32a). Sheep sequences were used to design TaqMan probes and primers by use of Applied Biosystems file builder 3.1 (http://marketing.appliedbiosystems.com/mk/get/FB3_login?isource=fr_WWW_Filebuilder_SfweDnload_Filebuilder). Sheep probe and primer sets are listed in Table 2. For sheep qRT-PCR relative quantification was performed against the geometric mean of four housekeeping genes (18S, GAPDH, cyclophilin A, and β-actin) since there was more interanimal variability in the sheep and this method has been reported to be more accurate than any single reference gene or 18S RNA (33).

qRT-PCR was performed on cDNA from Caco2 cells with primers and probe sets to detect NLRP6 and retinol binding protein 2 (RBP2) mRNAs (Applied Biosystems: Hs00373246_m1 and Hs00188160_m1, respectively) and normalized to 18S (Hs99999901_s1) by the delta delta cycle threshold method.

Immunohistochemistry.

Paraffin wax-embedded sections were cut and mounted onto Superfrost Plus slides. Sections were dewaxed and hydrated in graded ethanol prior to antigen retrieval by microwaving in 10 mM citrate buffer (pH 6.0) for 10 min. Endogenous peroxidase activity was inactivated with 3% H₂O₂ in methanol for 20 min. Sections were blocked with 5% goat serum in Tris-buffered saline. Anti-human PYCARD antibody was added at 1:200 (Alexis, Exeter, UK) or anti-mouse IL-18 receptor-α antibody at 1:200 (R&D Systems, Abingdon, UK), and sections were incubated overnight at 4°C. Immunoreactivity was detected by biotinylated secondary antibodies at 1:500 (DAKO, Ely, UK) and avidin biotin complex (Vector Laboratories, Peterborough, UK), and visualized with diaminobenzidine (Sigma, Bookham, UK). Control sections were incubated with rabbit or goat immunoglobulin fraction (DAKO) in place of the primary antibody. Slides were counterstained with hematoxylin, dehydrated, and mounted with DPX mountant.

Western blotting.

Fetal or neonatal rat frozen tissue was homogenized in PTN50 buffer [50 mM Na₃PO₄, pH 7.4, 50 mM NaCl, 1% Triton X-100] with protease inhibitors (P8340, Sigma). Protein concentration was determined by use of the BCA protein assay reagent kit (Pierce, Cramlington, UK), and 20 μg of protein (per well) was separated by electrophoresis on 18% SDS polyacrylamide gels to enable separation of the 24-kDa inactive and 18-kDa active forms of IL-18. Proteins were transferred to a polyvinylidene difluoride membrane (Invitrogen), blocked with 5% skimmed milk in PBS-Tween-20 (0.05%) for 1 h, and subsequently probed with an antibody to IL-18 (catalog no. sc-6179, Santa Cruz Biotechnology) at 1:1,000 on a shaker at 4°C overnight. Binding was detected by horseradish peroxidase-conjugated rabbit anti-goat secondary antibody at 1:5,000 (DAKO). ECL substrate (Amersham Biosciences, Little Chalfont, UK) was then used to detect binding. Membranes were stripped with Restore Western blot stripping buffer (Pierce) for 15 min at room temperature. To correct for loading variations the membrane was reprobed with a mouse monoclonal antibody to β-actin (AC-15, Ambion, Warrington, UK) at 1:5,000; the secondary antibody was rabbit anti-mouse at 20,000 (DAKO).

Cell culture.

The human epithelial colorectal adenocarcinoma cell line, Caco2, was obtained from the European Collection of Animal Cell Culture and maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, penicillin-streptomycin, glutamate, and nonessential amino acids.

Rosiglitazone (Cayman Chemicals, Tallinn, Estonia) was dissolved in ethanol and diluted before use in media; control was vehicle only. Cells were seeded at 1.2 × 10⁵/cm² and left overnight to settle before addition of rosiglitazone at 1, 10, or 20 μm for 6 h. Cells were then washed with PBS, trypsinized from the well, and snap frozen before RNA extraction and qRT-PCR. Experiments were performed in triplicate on three occasions (n = 9).

Statistics.

Comparison of observed and expected proportions was performed by χ² test with Yates correction. The association between gestational age and gene expression level was analyzed by the Pearson correlation coefficient. Means were compared by two-tailed Student's t-test. Statistical significance was assumed at P < 0.05.

Transcript profile.

We performed expression array analysis of intestinal samples and lung samples obtained from 10 E16 and 10 E20 rats. A single animal from each litter was used. Rat Affymetrix 230 2.0 arrays have a total of 31,099 probe sets, which map to 28,532 rat genes as defined by Entrez Gene and/or UniGene. In the intestine 769 probe sets (666 genes) were upregulated and 178 probe sets (159 genes) were downregulated greater than fivefold between E16 and E20 embryos (see Supplementary Table S1); 23 genes showed over 100-fold upregulation, including small molecule transporters (Slc2a2, Slc7a9, Slc26a3, Clca3) and lipid binding proteins (Apoa1, Apoa4, Apoc2, Mttp, Fabp2, Rbp2) that play a role in intestinal absorption and transfer. Mucins 3 and 13, as well as a mucin-like protein (Mucdhl), were also upregulated more than 100-fold. In the lung 382 probe sets (306 genes) were upregulated more than fivefold between E16 and E20 and 73 probe sets (61 genes) were downregulated more than fivefold (Supplementary Data S1; the online version of this article contains supplemental data). The largest increase was seen for transcripts encoding surfactant proteins A, B, and D (318-, 132-, and 89-fold, respectively). We then compared gene expression in the lung and the intestine. A Venn diagram representing the intersections between the genes and changes in their expression levels in lung and intestine is shown in Fig. 1.

We assessed the enrichment of GO biological process terms (level 3) of the significantly increased probe sets using FatiGO, Babelomics 4.0. Of the 31,099 probe sets on the array 3,134 had associated GO biological processes. Of the 769 probe sets with significantly increased signal in the rat intestine between E16 and E20, 136 had associated GO biological processes, and of the 382 similarly performing probe sets in the lung, 73 had associated GO biological processes. Overrepresented GO terms in rat intestine and lung are in Supplementary Table S2. In both the intestine and lung two GO terms were significantly overrepresented; “immune response” (P = 0.001 and P = 0.005, respectively) and “antigen processing and presentation” (P < 0.001 and P = 0.041, respectively) and within these categories there were few overlapping genes. In the intestine “cytokine production,” which may also relate to the immune system, was overrepresented (P = 0.019) whereas in the lung “respiratory gaseous exchange” was overrepresented (P = 0.031). Probe sets downregulated in the rat intestine (21 of 178 had associated GO biological processes) showed an overrepresentation of GO terms “developmental maturation” (P = 0.013) and “tissue remodeling” (P = 0.024). The lung had no significantly overrepresented GO terms associated with downregulated probe sets.

GO analysis clearly showed that genes related to the immune system were upregulated in the developing lung and intestine. The immune response, innate immune response, and inflammatory response categories were not well annotated in GO; therefore we assessed whether the lists of up- or downregulated genes that map to Ensembl Gene IDs were enriched for elements of the innate immune system using the InnateDB as our reference dataset (16). We found that of 382 rat Ensembl genes differentially expressed in the intestine, 128 (33.5%) were listed in InnateDB compared with 23.8% (2,709 of 11,392) of all Ensembl genes represented on the array (P < 0.0001). In the lung, 35.1% (66 of 188) of the upregulated rat Ensembl genes were present in InnateDB, which was significantly greater than the proportion in the whole array (P = 0.0003). Table 3 lists genes in the InnateDB list which were upregulated more than fivefold in both lung and intestine. Tables 4 and 5 list genes in the InnateDB list that were specifically upregulated more than 20-fold in the intestines or the lungs, respectively; genes involved in inflammasome formation are in bold.

Validation of changes in inflammasome-related transcripts.

Analysis of the data in Table 4 for functionally related transcripts demonstrated greater than 20-fold upregulation of three transcripts that form the Nlrp6 inflammasome complex (caspase-1, pycard, nlrp6). Moreover, there was a greater than fivefold upregulation of a gene which encodes one of the substrates of the complex IL-18 (Supplementary Table S1). These changes were not observed in the lung. Further analysis therefore focused on this system. Out-of-sample validation was performed by qRT-PCR with intestinal samples obtained from littermates of the animals used for the array analysis. These analyses confirmed the significant upregulation of nlrp6, pycard, caspase-1, and IL-18 mRNAs in the intestinal samples (88-, 216-, 45-, and 22-fold, respectively) (Fig. 2). The magnitude and intestinal specificity of the increase in expression of this system was comparable to that observed for the digestive enzyme sucrase-isomaltase. Nlrp6 and caspase-1 demonstrated small increases in the lung, but the absolute changes were much lower than observed in the intestine (Fig. 2).

Analysis of transcripts in fetal sheep tissues.

For validation in a second species qRT-PCR was performed by using cDNA from sheep jejunum at various gestational ages (100, 115, 130, and 145 days). There was considerable interanimal variability in the sheep specimens, and no clear pattern between levels of expression of pycard or caspase-1 and gestational age was found (Fig. 3). However, there was a striking increase in the mRNA encoding IL-18 in late gestation (4.6-fold, P = 0.006), the magnitude of which was comparable to that observed for the intestinal digestive enzyme, sucrase-isomaltase (11.6-fold, P = 0.048).

Analysis at the protein level.

Pycard, the central adaptor protein of the inflammasome, was at the lower level of limit of detection by immunohistochemistry in the intestine of E16 embryos but was clearly detectable in the intestinal epithelium in E20 whole mount embryos. Pycard was also detectable in the liver and skin at E20 (Fig. 4) and remained clearly detectable in the intestinal epithelium from animals at postnatal days 1, 2, and 5 (data not shown). The IL-18 receptor subunit α was not detectable in the intestine of E16 embryos but was expressed on the apical surface of intestinal epithelial cells at E20 and also on cells within the intestinal villi (Fig. 5).

Western blotting using an antibody specific for IL-18 demonstrated no apparent signal at E16 (Fig. 6). A single band with an estimated molecular weight of 24-kDa was observed at E20. Samples from three animals obtained on the first day of life demonstrated bands at both 24-kDa (inactive precursor of IL-18) and 18-kDa (active form).

Mapping of transcription factor binding sites.

Six Jaspar and seven Transfac motifs were significantly overrepresented in the 1,000 bp upstream of the transcription start sites of genes upregulated in rat fetal intestine between E16 and E20 in all three species (rat, mouse, and human) (Table 6). Hierarchical clustering revealed that the TFBSs PPARG-RXRA and coup-TF commonly occurred together. Nineteen genes upregulated at E20 contained PPARG-RXRA and coup-TF binding sites (Table 7). These genes had various functions including protection against oxidative stress (PON3, GDA), roles in digestion or metabolism (CYP2D6, MEP1A, NAALADL1, RBP2, SLC5A11), and also immunity (CD74, NLRP6). NLRP6 contained both these motifs, is involved in the innate immune response, and was therefore selected for further investigation. Alignment of the promoter regions of human, rat, and mouse Nlrp6 revealed two potential PPARG-RXRA binding sites that were conserved across all three species. Both PPARG-RXRA sites also contained a coup-TF binding site (human nt −221 to −207 and −805 to −791 from TSS) (Fig. 7B). The promoter regions for the positive control gene, retinol binding protein (RBP2), were also aligned and predicted to contain two PPARG-RXRA binding sites (−72 to −86 and −258 to −244) and a further coup-TF binding site (−124 to −111) (Fig. 7D).

Regulation of NLRP6 in vitro.

Caco2 cells showed a significant increase in NLRP6 mRNA that was dose dependent in response to the PPAR-γ agonist rosiglitazone (Fig. 7A). RBP2, which is known to be regulated by PPARs in Caco2 cells (29), was also increased significantly (Fig. 7C). Treatment of Caco2 cells with rosiglitazone (20 μm) for 6 h significantly increased NLRP6 and RBP2 mRNA above that of vehicle control (P = 0.023 and 0.025, respectively).