CARD8 and NLRP1 undergo autoproteolytic processing through a ZU5-like domain

Abstract

The "Function to Find Domain" (FIIND)-containing proteins CARD8 (Cardinal; Tucan) and NLRP1 (NALP1; NAC) are well known components of inflammasomes, multiprotein complexes responsible for activation of caspase-1, a regulator of inflammation and innate immunity. Although identified many years ago, the role of the FIIND is unknown. Here, we report that CARD8 and NLRP1 undergo autoproteolytic cleavage at a conserved SF/S motif within the FIIND. Using bioinformatics and computational modeling approaches, we detected striking structural similarity between the FIIND and the ZU5-UPA domain present in the autoproteolytic protein PIDD. This allowed us to generate a three-dimensional model and to gain insights in the molecular mechanism of the cleavage. Site-directed mutagenesis experiments revealed that the second serine of the SF/S motif is required for CARD8 and NLRP1 autoproteolysis. Furthermore, we discovered an important function for conserved glutamic acid and histidine residues, located in proximity of the cleavage site in regulating the autoprocessing efficiency. Altogether, these results identify a function for the FIIND and show that CARD8 and NLRP1 are ZU5-UPA domain-containing autoproteolytic proteins, thus suggesting a novel mechanism for regulating innate immune responses.

Figure 1. CARD8 and NLRP1 undergo cleavage within the FIIND.

(A) Schematic representation of NLRP1 and CARD8 domains structure. The CARD8 region for which numerous isoforms arise through alternative mRNA splicing is indicated in red. The regions of the CARD8 and NLRP1 proteins that contain the FIIND and CARD and that were expressed in cells are shown. (B) HEK293T cells were transiently transfected with plasmids encoding the FIIND-CARD region of CARD8 (left) or NLRP1 (right) with N-terminal flag tags. After 24 hours, cell lysates were prepared and analyzed by SDS-PAGE/immunoblotting using anti-flag antibody. Unexpected smaller forms of the flag-tagged proteins are indicated by arrow heads. Molecular weight markers are indicated in kilo-Daltons (kDa). (C) Analysis of endogenous CARD8. Cell lysates from the indicated cancer and leukemia cell lines were normalized for total protein content and subjected to SDS-PAGE/immunoblot analysis using rabbit antiserum directed against the C-terminal portion of CARD8. The smaller form of CARD8 is indicated by arrowhead. (D) HCT116 cells were stably transduced with lentiviruses encoding either two different shRNAs or scrambled control. Cell lysates were analyzed by immunoblotting using a rabbit anti-CARD8 antibody. Full-length CARD8 migrates at molecular mass of ∼60 kD, whereas cleaved CARD8 appears at ∼28 kD (arrowhead). As a loading control, membranes were re-probed with anti-ß-actin antibody. Asterisks indicate non-specific bands. (E) HEK293T cells stably expressing SBP-tagged proteins were lysed and incubated with magnetic streptavidin beads, followed by elution with biotin. Eluted proteins were subjected to SDS-PAGE and stained with Comassie Blue. The bands indicated by an arrow (band 1; band 2) were subjected to tryptic digestion and analyzed by LC-MS to reveal peptides corresponding to a novel cleavage site between Phe-296 and Ser-297.

Figure 2. CARD8 three-dimensional model.

(A) Charge distribution and predicted salt bridges in the interface between the ZU5-like and the CARD domain. Only interface forming segments of the protein chain are shown as a C-alpha atom connecting trace. Atoms of the ZU5-like domain are colored light blue, those belonging to the CARD domain are colored green. Putative salt bridges are indicated by a dashed line connecting the C-beta atoms of the involved residues, labeled with their distance in units of Ångströms. The left side of the figure presents the electrostatic surface potential of the CARD domain facing the ZU5-like domain, whereas the right side is obtained by a 180-degree rotation and displays the surface of the ZU5-like domain. Surfaces are rendered semitransparent such that the backbone of the opposite domain is visible. The charge distribution ranges from -5 (red) to +5 (blue) in units of K_bT/e_c, where K_b is Boltzmann's constant; T = 310, temperature; and e_c, charge of an electron. Compared to the ZU5/death domain interface of the template structure, this model lacks a clear hydrophobic patch and exhibits a higher degree of electrostatic interactions. (B) Hydrophobic pocket for the phenyl ring of Phe-296. Burial of the phenyl ring is an important structural feature of the SFS motif as it positions its neighboring serine side-chains for activation. Seven of the eight hydrophobic residues shown in this figure are thoroughly conserved in all FIIND domains. The exception is Ala-246, which, according to the multiple sequence alignment (see below), can be substituted by a variety of amino acids. However, each of them is principally able to contribute hydrophobic interactions with their carbon side-chain atoms. The C-alpha trace of the protein chain is rendered in dark grey cartoon representation. Atoms are colored by their element, carbon, white; nitrogen, blue; and oxygen, red. The buried phenylalanine residue is highlighted in light blue and important side-chains are rendered as a ball-and-stick model. (C) Multiple sequence alignment (MSA) for selected protein sequences containing the FIIND domain. Proteins with sequence identities higher than 90% are not shown. As a query human CARD8 FIIND domain (Human_card8) was taken to detect homologs in the RefSeq database using NCBI Blast. The resulting sequences were trimmed to highlight important residues from the FIIND domain. (The complete MSA is provided in Methods S1.) Protein sequences are sorted by sequence identity to Human_Card8, giving two distinct clusters of FIIND domains, one for CARD8 proteins and the other for NLRP1 proteins, which is also reflected when phylogenetic trees are constructed from these sequences (data not shown). A grey box highlights the SFS motif. See Methods S1 for additional details. Human_card8 residues selected for mutagenesis experiments are colored orange. Amino acids printed on blue background signify columns with residue identities greater than or equal to 90%. Secondary structure prediction for Human_card8 was performed with Psipred shown in the row labeled “SSpred” containing the following symbols: dash (−), predicted coil; E, predicated strand; and H, predicted helix. In a second row termed “Phe-296 pocket” residues are marked with an asterisk (*), which form the hydrophobic pocket for the phenyl ring of Phe-296. Looking at these columns, this pocket is highly conserved across all FIIND domains, except for Ala-246, which is replaced by various non-hydrophobic amino acids. Protein sequence names provide common names for the organism and the type of protein. Organisms are abbreviated as follows: Human, Homo sapiens; Monkey, Macaca mulatta; Marmoset, Callithrix jacchus; Panda, Ailuropoda melanoleuca; Mouse, Mus musculus; Opossum, Monodelphis domestica; Rat, Rattus norvegicus; Horse, Equus caballus; Dog, Canis familiaris; Suffices can be card8, CARD8-like proteins; nlrp1, NLRP1-like proteins; and ded, death effector domain containing proteins.

Figure 3. Mutagenesis analysis of CARD8 and NLRP1 autoproteolytic mechanism.

In (A), (B), and (D), HEK293T cells were seeded in 6-well plates at a density of 1×10⁶ cells per well and transiently transfected with 2 µg of indicated expression plasmids. After 24 hours, cells were collected and lysed. Cleared lysates were normalized for total protein content and subjected to SDS-PAGE/immunoblot anlaysis using anti-flag antibody. Plasmids transfected were (A) flag-NLRP1 and flag-NLRP1(S1213A), (B) flag-CARD8 and flag-CARD8(S297A), and (D) various flag-tagged CARD8 mutants. WT  =  wild-type CARD8; NC  =  non-transfected control. (C) In vitro translation of myc-tagged CARD8 and CARD8(S297A) was performed and equal-volume aliquots of the reaction were subjected to SDS-PAGE/immunoblotting using anti-myc antibody. Arrowheads indicate cleaved protein fragments.