Report

HIN-200 Proteins Regulate Caspase Activation in Response to Foreign Cytoplasmic DNA

See allHide authors and affiliations

Science  20 Feb 2009:
Vol. 323, Issue 5917, pp. 1057-1060
DOI: 10.1126/science.1169841

Abstract

The mammalian innate immune system is activated by foreign nucleic acids. Detection of double-stranded DNA (dsDNA) in the cytoplasm triggers characteristic antiviral responses and macrophage cell death. Cytoplasmic dsDNA rapidly activated caspase 3 and caspase 1 in bone marrow–derived macrophages. We identified the HIN-200 family member and candidate lupus susceptibility factor, p202, as a dsDNA binding protein that bound stably and rapidly to transfected DNA. Knockdown studies showed p202 to be an inhibitor of DNA-induced caspase activation. Conversely, the related pyrin domain–containing HIN-200 factor, AIM2 (p210), was required for caspase activation by cytoplasmic dsDNA. This work indicates that HIN-200 proteins can act as pattern recognition receptors mediating responses to cytoplasmic dsDNA.

Recognition of viral dsDNA is important in the initiation of antiviral responses, and is thought to occur in the cytoplasm (13). Transfection of DNA into the cytoplasm induces interferon-β (IFN-β) production, inflammasome activation, and cell death (37), all of which are specific responses to double-stranded DNA (dsDNA) and not single-stranded DNA (ssDNA). Several pathogen products and endogenous danger signals activate the inflammasome, a complex causing the clustering and activation of caspase 1, which subsequently cleaves pro–interleukin-1β (proIL-1β) and proIL-18 to their mature forms (8). Caspase 1 activation can also contribute to cell death in response to bacterial infection, in a process termed pyroptosis (9). Inflammasome responses to dsDNA require the adapter protein apoptosis-associated speck-like protein containing a caspase recruitment domain (ASC), which has a pyrin domain as well as a caspase recruitment domain (CARD) through which it binds caspase 1 (5). ASC itself is normally recruited via interactions between its pyrin domain and other pyrin domains on pathogen-sensing proteins such as Nod-like receptors. Thus, a pyrin domain–containing protein is likely to be involved in recognition of cytosolic dsDNA leading to inflammasome activation.

Primary macrophages and some macrophage cell lines die in response to transfection of dsDNA, and we proposed this as an antiviral defense (fig. S1) (6, 7). In investigating the mechanism of DNA-induced cell death, we found that transfected DNA rapidly activated both caspase 3 and caspase 1, and members of the HIN-200 family of DNA binding proteins initiated and controlled this response.

The only well-characterized receptor for foreign DNA is Toll-like receptor 9 (TLR9) (10), which recognizes unmethylated CpG-containing DNA within the endosome (11). DNA-dependent IFN-β induction and inflammasome activation are independent of TLR9 (35), as is the DNA-dependent death of bone marrow–derived macrophages (BMMs) (Fig. 1A). The observed toxicity was a specific response to dsDNA and not ssDNA (Fig. 1B) (6), was abolished by deoxyribonuclease I (DNase I) treatment (Fig. 1C), and depended on the length of transfected DNA (Fig. 1D). Short ds oligonucleotides did not efficiently induce cell death (6), but 44–base pair (bp) DNA could kill cells when transfected at high concentration (fig. S2). In investigating the mechanism of cytoplasmic DNA-induced cell death, we found that an executioner caspase, caspase 3, was activated within 5 min of introduction of DNA (Fig. 1E) in cell types sensitive to DNA-induced cell death (fig. S3). DNA-dependent caspase 1 activation was also observed (Fig. 4A), but was not accompanied by IL-1β secretion, because there is negligible IL-1β mRNA in unprimed BMMs (fig. S4).

Fig. 1.

Transfected DNA–induced BMM death is dependent on DNA length and strandedness and is independent of TLR9. Mitochondrial activity was measured by cleavage of the MTT reagent as an index of cell viability. Except as noted, the assay was performed at 24 hours after transfection, and bars show the mean and error bars the range of duplicate electroporations. Results are representative of two to five experiments. (A) Wild-type and TLR9–/– BMMs were electroporated either without DNA or with 10 μg of calf thymus genomic DNA (CT DNA). Results were normalized to the “no DNA” samples. (B) Cell death was caused by transfected dsDNA but not ssDNA. C57BL/6 BMMs were electroporated without DNA or with 10 μg of poly(dA) or poly(dA):(dT). (C) Loss of mitochondrial function occurred as early as 3 hours after transfection of poly(dA):(dT). TLR9–/– BMMs were electroporated either without DNA, with the indicated amounts of poly(dA):(dT), with 3 μg of DNase I–digested poly(dA):(dT) (“DNase 3μg”), or with DNase I in digestion buffer as a control (“DNase alone”). The assay was perfomed at 3 hours after electroporation. Bars show the mean and error bars the range of values from two independent experiments. (D) The length of electroporated DNA fragments determines the degree of cell death. BALB/c BMMs were electroporated without DNA or with 10 μg of polymerase chain reaction (PCR) products of various length, as indicated, or with CT DNA. (E) Caspase 3 is rapidly activated after electroporation with dsDNA. BALB/c BMMs were either untreated (“no zap”) or electroporated alone (“–DNA”) or with 20 μg of CT DNA (“+DNA”). Cells were harvested after 5, 15, or 20 min. Total caspase 3 was detected by Western blotting. Pro–caspase 3 is 35 kD in size and activated cleaved forms are 17 and 19 kD in size.

To identify candidate DNA receptors, we analyzed cytoplasmic extracts from BMMs for binding to a ds 44-bp oligonucleotide by electromobility shift assay (EMSA). A single strong DNA binding complex was found in the ultracentrifuged pellet of cytoplasmic extract (Fig. 2A). The fact that the DNA binding protein sedimented with ultracentrifugation (fig. S5A) suggests that it is attached to a membrane structure or cytoskeleton, but is released upon DNA binding; hence, the discrete band on EMSA. The candidate receptor bound specifically to dsDNA, with no competition by either ssDNA or dsRNA (Fig. 2A and fig. S5B). dsDNA binding showed no requirement for a sequence motif (Fig. 2B), but like the biological response to DNA, was length dependent; when tested in equal nanogram amounts, a 100-bp fragment of DNA was a more effective competitor than 44-bp DNA, and 22-bp DNA was a poor ligand (Fig. 2C and fig. S5C).

Fig. 2.

p202 is a dsDNA-specific binding protein found in the ultracentrifuge pellet fraction of cytoplasmic extract. (A) EMSA shows that cytoplasmic extract contains a dsDNA-specific binding protein (arrow). Protein extract was bound to 44-bp dsDNA or 44-base ssDNA probe alone (“no comp.”) or with a 70-fold molar excess of unlabeled ds or ss probe, showing that ssDNA did not compete for binding. “Probe” indicates probe without extract. (B) The dsDNA binding protein requires no specific sequence motif. Cytoplasmic protein was incubated with probe and 7- or 13-fold molar excess of competitors, either unlabeled probe (ds44bp) or a sequence-scrambled version (dsSCR). (C) Binding depends on DNA length. Cytoplasmic protein was incubated with 0.77 ng of 44-bp probe and 1.2 ng of unlabeled 22-, 44-, or 100-bp DNA (1.6-fold ng excess). (D) Purification of dsDNA-binding protein. Cytoplasmic extract was incubated with beads bound to either dsDNA, ssDNA, or no DNA (“beads”). Analysis of bound proteins revealed a 52-kD dsDNA binding protein (arrow). (E) EMSA supershift showing that the dsDNA-specific binding protein is p202. Cytoplasmic protein and probe were incubated with or without antisera against p202, p204, or vimentin. (F) p202 stably interacts with cytosolic DNA. Biotinylated plasmid DNA (“biotin-dsDNA”), unlabeled plasmid (“dsDNA”), or biotinylated 44-base oligonucleotide (“biotin-ssDNA”) was electroporated into RAW264 cells stably expressing p202-V5. After 1 hour, cells were lysed and proteins associated with the biotinylated DNA were isolated by binding to streptavidin-Sepharose (“pellet”). The location of p202 in the biotinylated DNA-bound pellet fraction or unbound supernatant (S/N) was assessed by Western blotting for V5. (G) Purified recombinant SUMO-p202 binds directly to plasmid DNA. N-terminally SUMO/His-tagged p202 was incubated with streptavidin beads with or without bound biotinylated dsDNA. Bound protein was analyzed by Western blotting with anti-His. Analysis of input protein (right) shows full-length SUMO-p202 (arrow) and two C-terminal truncations (asterisks). A contaminating cross-reactive protein of 15kD bound nonspecifically to beads. (H) Purified SUMO tag does not bind to DNA. The experiment was performed as in (G).

We purified the DNA binding protein from the ultracentrifuged pellet fraction by using dsDNA linked to Sepharose beads. Protein eluted from the beads showed one major 52-kD dsDNA-specific band, when binding to dsDNA-Sepharose and ssDNA-Sepharose was compared (Fig. 2D). This band was analyzed by mass spectrometry and identified from three tryptic fragments (VFNMDLK, LFTYDSIK, and VMVFEENLEK) (12) as the interferon-inducible protein p202. An antibody to p202 supershifted the protein-DNA complex on EMSA, whereas two control antibodies had little effect (Fig. 2E). When biotinylated DNA was electroporated into cells, p202 within the cell lysate was stably bound to plasmid, but not ss oligonucleotide (Fig. 2F).

Although p202 was previously shown to be a DNA binding protein, no function has been ascribed to this activity (13). To determine whether p202 can bind DNA alone, we expressed p202 with an N-terminal small ubiquitin-like modifier (SUMO)–His tag in Escherichia coli and purified the SUMO-p202 fusion protein. Full-length SUMO-p202 bound to dsDNA-beads, but truncated p202 products lacking the C-terminal end of the protein and a purified SUMO control did not (Fig. 2, G and H), showing the C-terminal to be essential for high-affinity binding to dsDNA. When analyzed by EMSA, recombinant p202 showed a dsDNA-specific binding activity (fig. S5D), consistent with the result for endogenous p202 (Fig. 2A).

p202 colocalization with microinjected Cy3–CT DNA (calf thymus DNA) occurred within 5 min (fig. S6 and Fig. 3). A 44-bp ds oligonucleotide also colocalized with p202, whereas there was little discernable colocalization of p202 with a 22-bp ds oligonucleotide (Fig. 3). This result corresponds with the higher affinity of p202 for long DNA (Fig. 2C). p202 also colocalized with electroporated Cy3-labeled plasmid (fig. S7).

Fig. 3.

Colocalization of microinjected Cy3-labeled CT DNA with p202. One hour after microinjection with labeled DNA, NZB BMMs were methanol fixed and antibody stained to show p202 localization. p202 (green, left panels) completely colocalized with CT DNA (red, center top panel) and slightly with ds44bp (red, center panel) and not at all with ds22bp (red, center bottom panel). Colocalization is indicated in yellow (right panels).

In the absence of introduced DNA, p202 was diffuse in the cytoplasm (fig. S8). p202 was enriched in regions of the cell that stained strongly for the late endosomal/lysosomal marker LAMP-1 (lysosomal-associated membrane protein 1), but did not exclusively reside in this location (fig. S9A). After introduction of DNA, colocalization with LAMP-1 was lost (fig. S9B). Thus, a proportion of p202 may be bound to the cytoplasmic face of late endosomes/lysosomes and dissociate from this site upon DNA binding. This would explain the presence of p202 in the cytosolic ultracentrifuge pellet fraction and subsequent release to form a discrete band on EMSA.

To examine the function of p202, we knocked down (reduced) its expression using three different small interfering RNAs (siRNAs) (p202#1 to #3) and examined the response to electroporated DNA 24 hours later (Fig. 4, A and B, and fig. S10). Activation of both caspase 3 and caspase 1 was more pronounced with reduced p202, whereas use of siRNAs against the related protein p204, as well as TLR9, had no effect. Thus, p202 is not a DNA receptor mediating caspase activation, but instead antagonizes this pathway.

Fig. 4.

HIN-200 factors regulate DNA-dependent caspase activation. (A) Knockdown of p202 enhances DNA-induced caspase 3 and caspase 1 activation. BALB/c BMMs were electroporated with the indicated siRNAs and left for 24 hours before electroporation with either no DNA (“cont.”) or 1 μg of poly(dA):(dT) (“AT”). After 20 min, cells were lysed and protein extracts analyzed by Western blotting for cleaved caspase 3, S6 ribosomal protein as loading control, and caspase 1. A separate gel with the same samples is shown for caspase 1. The large arrow indicates full-length pro-caspase 1 (45kD), a small arrow the active cleaved product (10 kD), and the asterisk a nonspecific band. Results are representative of five experiments with three different siRNAs against p202 (fig. S10). (B) p202 and p204 mRNA expression relative to hypoxanthine-guanine phosphoribosyltransferase (HPRT) measured by real time PCR, 24 hours after electroporation with siRNAs used in (A). Bars show the mean of duplicate assays and error bars the SD as defined in the online methods. (C) Knockdown of AIM2 prevents DNA-induced caspase 3 and caspase 1 activation. BALB/c BMMs were electroporated with the indicated siRNAs and left for 24 hours before electroporation with either no DNA (“cont.”), 10 μg of CT DNA (“CT”), or 1 μg of poly(dA):(dT) (“AT”). Samples were analyzed as in (A). Results are representative of four experiments. (D) AIM2 mRNA expression relative to HPRT measured by real time PCR as in (B), 24 hours after electroporation of siRNAs used in (C). (E) Expression of p202 and AIM2 mRNAs in BMMs from C57BL/6, BALB/c, and NZB mouse strains. Shown are the mean and range of results for two independent RNA preparations. (F) Activation of caspases in BMMs from three mouse strains electroporated with DNAs as in (C). Arrows and asterisk are as in (A).

p202 is a member of the hematopoietic interferon-inducible nuclear protein HIN-200 family, a cluster of 13 or more interferon-inducible genes on mouse chromosome 1 (1416). This family of proteins are characterized by the presence of one or two 200–amino acid HIN domains of poorly understood function, although the C-terminal HIN domain of p202 is required for dsDNA binding (Fig. 2G). All HIN-200 factors, apart from p202, contain an N-terminal pyrin domain, making them candidates for a DNA-dependent activator of caspase 1. Among the HIN-200 family, the pyrin domain of absent in melanoma 2 (AIM2 or p210) is the most similar to inflammasome-related pyrin domains (17). Further, AIM2 is the only family member with a clear human ortholog and is known to heterodimerize with p202 (15, 18). Knockdown of AIM2 completely prevented activation of caspase 3 and caspase 1, even though the mRNA knockdown was only partial (Fig. 4, C and D). Sequestration of DNA by p202, or heterodimerization between p202 and AIM2, would inhibit AIM2-mediated responses, because the lack of a pyrin domain on p202 would reduce clustering of ASC and subsequent caspase 1 activation (fig. S11). Consequently, in the presence of p202, when the amount of AIM2 protein drops below a threshold, it would no longer signal.

Expression of p202 varies greatly between mouse strains (19). p202 mRNA was highly expressed in BMMs from NZB mice but barely detectable in those of C57BL/6 mice, whereas AIM2 expression was similar in all three strains tested (Fig. 4E). Consistent with the hypothesis that p202 regulates DNA-dependent caspase activation initiated by AIM2, caspase activation correlated inversely with the abundance of p202 in the three strains (Fig. 4F). p202 is suggested as a susceptibility factor for systemic lupus erythematosus (SLE), because it falls within the major susceptibility locus in NZB and BSXB mice and is overexpressed in these strains (19, 20). SLE is an autoimmune disease in which DNA may act both as antigen and adjuvant (21), and results here suggest that p202 contributes to the SLE phenotype by modifying responses to cytoplasmic DNA. As well as DNA deriving from viruses or retrotransposons, cytoplasmic DNAs could be derived from phagocytosed self-DNA that has escaped into the cytoplasm.

In summary, the HIN-200 family proteins are a class of pattern recognition receptors, in which AIM2 promotes and p202 represses the activation of caspases in response to cytoplasmic dsDNA. The function of other members of the family awaits definition. The conservation seen among the HIN domains suggests they all have the potential to mediate responses to nucleic acids, perhaps differing in sequence specificity, localization, or effector function.

Supporting Online Material

www.sciencemag.org/cgi/content/full/1169841/DC1

Methods

Figs. S1 to S11

References

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article