Three-dimensional Structure of the NLRP7 Pyrin Domain

The innate immune system provides an initial line of defense against infection. Nucleotide-binding domain- and leucine-rich repeat-containing protein (NLR or (NOD-like)) receptors play a critical role in the innate immune response by surveying the cytoplasm for traces of intracellular invaders and endogenous stress signals. NLRs themselves are multi-domain proteins. Their N-terminal effector domains (typically a pyrin or caspase activation and recruitment domain) are responsible for driving downstream signaling and initiating the formation of inflammasomes, multi-component complexes necessary for cytokine activation. However, the currently available structures of NLR effector domains have not yet revealed the mechanism of their differential modes of interaction. Here, we report the structure and dynamics of the N-terminal pyrin domain of NLRP7 (NLRP7 PYD) obtained by NMR spectroscopy. The NLRP7 PYD adopts a six-α-helix bundle death domain fold. A comparison of conformational and dynamics features of the NLRP7 PYD with other PYDs showed distinct differences for helix α3 and loop α2-α3, which, in NLRP7, is stabilized by a strong hydrophobic cluster. Moreover, the NLRP7 and NLRP1 PYDs have different electrostatic surfaces. This is significant, because death domain signaling is driven by electrostatic contacts and stabilized by hydrophobic interactions. Thus, these results provide new insights into NLRP signaling and provide a first molecular understanding of inflammasome formation.

caspase-1. Recent reports have also shown that this inhibition is mediated by a direct interaction with both pro-IL-1␤ and procaspase-1 (17). Furthermore, the NLRP7 PYD has also been reported to bind Fas-associated factor 1 (FAF-1), a protein previously recognized to activate apoptosis and inhibit NF-B activation (19,20).
NLRP7 regulates multiple important biological processes, especially during embryonic development (21). Various NLRP7 mutations were linked to familial recurrent hydatidiform moles, which is a rare embryonic misdevelopment resulting in early pregnancy loss (22)(23)(24)(25)(26)(27)(28). Abnormal regulation of NLRP7 was also detected in endometrial cancer tissues (29) and testicular seminomas (30). Most recently, mutations in NLRP7 were identified in patients with autoimmune conditions such as Hashimoto disease, lupus, vitiligo, and Crohn disease (28). The cause(s) of these disorders remain elusive. Understanding how NLRP7 mediates signal transduction at a molecular level may provide new therapeutic approaches to these diseases. Despite the abundance of PYDs in several protein families, structural information about PYDs is very limited (31)(32)(33)(34). Currently, only a few structures of effector domains involved in NLR signaling are known. Furthermore, no structural data are available for complexes formed between NLRs and their effector proteins (i.e. ASC or caspases). Both are essential for understanding the molecular mechanisms of NLR signaling.
Here we report the three-dimensional structure of the NLRP7 PYD, as well as an analysis of its dynamics using NMR spectroscopy. Importantly, the NLRP7 PYD shows structural and dynamic differences when compared with the PYDs of NLRP1, as well as ASC and ASC2. Recently, a completely disordered helix (helix ␣3) was reported in the NLRP1 PYD structure (34). This has subsequently inspired speculation about the possibility that large conformational changes might occur upon PYD-PYD binding. Canonical PYD-PYD and CARD-CARD (35)(36)(37) interactions are driven primarily by strong electrostatic interactions and likely stabilized by the subsequent formation of hydrophobic interactions. For example, based on the total buried surface area in the APAF-1 CARD-caspase-9 CARD complex (35), it was estimated that 70% of the interaction was due to electrostatic interactions, and 30% was due to hydrophobic interactions. Here we identify structural, hydrophobic, and electrostatic surface charge differences between the different NLRP PYDs. These observed differences now begin to explain the differential downstream signaling of these NLRPs, critical for understanding the formation of inflammasomes. We also probe these and other protein-protein interactions of the NLRP7 PYD by NMR titration interaction studies in vitro. Taken together, we provide the first critical insights into PYD signaling and fundamentally enhance our molecular understanding of the NLR-based innate immune system.

Protein Expression and Purification
Protein expression and purification of the human NLRP7 PYD (residues 1-96) was recently described (38).
ASC PYD-ASC PYD was cloned into pET28 (Novagen). The plasmids were transformed into Escherichia coli BL21 (DE3) cells. The cultures were grown to an A 600 of ϳ0.8 at 37°C, and expression was induced by the addition of 0.4 mM isopropyl ␤-D-thiogalactopyranoside. Proteins were expressed for 5 h at 25°C. Cells were harvested by centrifugation, resuspended in 30 ml of buffer (50 mM Tris, pH 8.0, 100 mM NaCl), transferred to a 50-ml Falcon tube, and stored at Ϫ20°C overnight. For purification, the cells were lysed by sonication (5 pulses of 70 W for 35 s) on ice. Immediately following sonication, 5 l of DNase (1 unit/l) was added to the lysed sample, and it was incubated for 20 min on ice. The cell debris was harvested by centrifugation (50,000 ϫ g, 20 min, 4°C). The ASC-containing pellet was resuspended in 25 ml of urea buffer (50 mM Tris, pH 8.0, 8 M urea) and incubated overnight at room temperature. Following centrifugation (50,000 ϫ g, 20 min, 4°C), the supernatant containing denatured ASC was loaded onto a nickel-nitrilotriacetic acid column. The column was then washed with 20 ml of wash buffer 1 (20 mM Tris, pH 8.0, 6 M urea, 5 mM imidazole) followed by 20 ml of wash buffer 2 (20 mM Tris, pH 8.0, 1 M NaCl, 20 mM imidazole). The protein was eluted three times with 1.2 ml of elution buffer (50 mM citric acid, pH 4.0, 1 M NaCl, 5 mM dithiothreitol). The purity of ASC was verified by SDS-PAGE, and its folded state was verified by NMR spectroscopy.
FAF-1 FID-The FAF-1 FID domain (residues 1-180) was cloned into RP1B (39). Freshly transformed E. coli BL21-Codon-Plus (DE3)-RIL (Stratagene) cells were grown at 37°C to an A 600 of ϳ0.6, and expression was induced with 1 mM isopropyl ␤-D-thiogalactopyranoside. The proteins were expressed for 18 h at 18°C. The cultures were harvested by centrifugation, and the cell pellets were stored at Ϫ80°C. The cell pellets were resuspended in lysis buffer (50 mM Tris-HCl, pH 8.0, 5 mM imidazole, 500 mM NaCl, 0.1% Triton X-100) supplemented with a Complete EDTA-free protease inhibitor tablet (Roche Applied Science). The cells were lysed by high pressure homogenization (Avestin C-3 Emulsiflex), and the cell debris was removed by centrifugation (50,000 ϫ g, 40 min, 4°C). The clarified lysate was loaded onto a His-Trap HP Column (GE Healthcare), and the protein was eluted with a 5-500 mM imidazole gradient. Fractions containing the FAF-1 FID domain were pooled, incubated with TEV protease to remove the N-terminal Thio 6 His 6 tag, and dialyzed against fresh buffer (20 mM Tris-HCl, pH 7.5, 200 mM NaCl). The cleaved sample, containing the now untagged FAF-1 FID domain, was loaded onto a nickelnitrilotriacetic acid column (Invitrogen), from which it eluted in the flow through. For the final purification step, the protein was loaded onto a Superdex 75 26/60 size exclusion column (GE Healthcare) equilibrated with 20 mM sodium phosphate, pH 6.0, 200 mM NaCl, 0.5 mM tris(2-carboxyethyl)phosphine. Fractions containing the pure FAF-1 FID domain, as confirmed by SDS-PAGE, were pooled and concentrated.

NMR Spectroscopy
NMR measurements were performed at 298 K on a Bruker AvanceII 500 MHz spectrometer. A 13 C-resolved 1 H, 1 H NOESY spectrum was recorded on a Bruker AvanceII 800 MHz spectrometer. Both spectrometers are equipped with TCI HCN z-gradient cryoprobes. All of the NMR experiments were performed with either a 15 N-or a 15 N/ 13 C-labeled NLRP7 PYD sample at a final concentration of 1 mM dissolved in 20 mM sodium phosphate, pH 6.0, 50 mM NaCl, 0.25 mM tris(2-carboxyethyl)phosphine.
Established NMR methods were employed for the sequencespecific backbone and side chain assignments of NLRP7 PYD (40). Semi-automated programs were used to evaluate the data (CARA). We used the following spectra for the structure calculation: three-dimensional 15 N-resolved 1 H, 1 H NOESY, threedimensional 13 C-resolved 1 H, 1 H NOESY (mixing time, 80 ms), and two-dimensional 1 H, 1 H NOESY (mixing time, 80 ms; D 2 O solution). NOESY peak picking, NOESY peak assignment, and three-dimensional structure calculation were performed automatically using the ATNOS/CYANA software package (41)(42)(43). The input for the structure calculations of NLRP7 PYD was the amino acid sequence, the complete chemical shift lists, and the three-and two-dimensional NOESY spectra. Constraints for backbone dihedral angles derived from 13 C chemical shifts were used only in the initial structure calculation.
A total of 1670 NOESY-derived distance constraints (ϳ17.5 NOE constraints/residue) were used for the structure calculation of NLRP7 PYD. The 20 conformers from the final CYANA cycle with the lowest residual CYANA target function values were energy-minimized in a water shell with CNS (44) using the RECOORD (45) script package ( Table 1). The NLRP7 PYD model has excellent stereochemistry, with 98.1% of the residues in the most favored and additionally allowed regions of the Ramachandran diagram, 1.3% in the generously allowed region, and 0.6% in the disallowed region. The quality of the structures was assessed by the programs WHATCHECK (46), AQUA (47), NMR-PROCHECK (47), and MOLMOL (48). All of the structure comparisons throughout the manuscript were performed using the lowest energy conformer of the bundle of structures. 15 N longitudinal (R 1 ) and transverse (R 2 ) relaxation rates and the 15 N{ 1 H}-NOE (hetNOE) were measured as described previously (49 -51) at 500 MHz. All of the spectra were recorded with 2048 points in the direct dimension and 256 points in the indirect dimension. The sweep widths of the direct and indirect dimensions were 6010 and 1242 Hz. T 2 was sampled at 20, 80, 110, 140, 180, 220, 250, and 350 ms; T 1 was measured at 20, 100, 180, 250, 320, 380, 450, and 650 ms. For the hetNOE measurement, an experiment with proton presaturation was interleaved with an experiment without proton presaturation. The relaxation delay for all T 1 and T 2 experiments was 3 and 5 s for all of the hetNOE experiments. For accurate error estimations, two complete sets for T 1 and hetNOE and three complete sets of T 2 spectra were recorded.
All of the relaxation spectra were processed using NMRPipe version 97.027.12.56 (53), and peak intensities were analyzed with NMRView version 5.2.2.01 (54). The exponential decay function I ϭ Ae (Ϫr/t) (where I is the peak intensity, A is the amplitude, t is the relaxation delay, and r ϭ R 1 or R 2 ) was fit to the peak intensities using the jitter function in NMRView to determine the R 1 , R 1 , and R 2 relaxation rates. hetNOEs were calculated by dividing the intensity of the peaks in the spectra without presaturation by the intensity of the peaks in the presaturated spectra. The data from overlapped peaks were excluded from the analysis. A Modelfree relaxation analysis (isotropic diffusion) was performed using the program TEN-SOR (55). 15 N chemical shift anisotropy and 15 N-1 H dipole- The loops are in gray. B, lowest energy conformer of NLRP7 PYD in an orientation identical to that shown in A. The helices are light blue. The N and C termini, as well as the helices, are labeled. C, top view of NLRP7 PYD. Hydrophobic residues essential for the formation of the hydrophobic core that stabilize helices ␣1, ␣2, and ␣4 -␣6 are highlighted in cyan stick representation. D, view of NLRP7 PYD focusing on helices ␣2 and ␣3. Hydrophobic residues essential for the formation of the second hydrophobic cluster, which stabilizes helix ␣3 to the core of the NLRP7 PYD, are highlighted in pink stick representations.

Chemical Shift Assignments and Coordinates
Chemical shift assignments of NLRP7 PYD were deposited in the Biological Magnetic Resonance Data Bank under accession number 16263 (38), and the coordinates were submitted to the Protein Data Bank under accession code 2KM6.

RESULTS
Protein Expression and Purification-Numerous NLRP7 PYD constructs with different C-terminal lengths were tested for overexpression. NLRP7 1-96 (NLRP7 PYD) showed the high-est overexpression of soluble and folded protein and was subsequently used for all structural studies. NLRP7 PYD is a monomer in solution as verified by size exclusion chromatography, where a sharp peak containing NLRP7 PYD corresponding to a molecular mass of a ϳ12 kDa was detected, as identified by SDS-PAGE analysis. The monomeric state of NLRP7 PYD was further confirmed by dynamic light scattering, which showed an R h of 1.77 nm, an expected radius for a 12.7-kDa globular protein. The ASC PYD domain was expressed and refolded as described under "Experimental Procedures." The FAF-1 FID domain (residues 1-180) was produced in a similar manner to the NLRP7 PYD. The FAF-1 FID domain is a monomer in solution as verified by size exclusion chromatography.
NLRP7 PYD and the PYDs of NLRP1, ASC, and ASC2 Show Distinct Structural Features-The PYD domains of NLRP7, NLRP1 (24% sequence identity; Protein Data Bank accession code 1PN5) (34), ASC (27% sequence identity; Protein Data Bank accession code 1UCP) (31,57), and ASC2 (25% sequence identity; Protein Data Bank accession code 2HM2) (32) only share modest primary sequence similarity (Fig. 2). ASC2 (also referred to as POP1 and ASCI) functions as a dominant negative inhibitor of ASC and other PYD-containing proteins FIGURE 2. A, primary sequence alignment of the NLRP7, ASC, ASC2, and NLRP1 PYDs. The experimental secondary structure of NLRP7 PYD is depicted. The gray boxes highlight conserved hydrophobic residues important for the formation of the hydrophobic core that stabilize helices ␣1, ␣2, and ␣4 -␣6. The differences in the number of hydrophobic residues in helices ␣2 and ␣3 as well as the ␣2-␣3 loop can be readily observed. B, primary sequence alignment of human NLRP pyrin domains (NLRP8 and NLRP13 PYDs have been excluded, because the primary sequence shows large amino acid insertions, making direct comparison with other NLRP domains difficult). Residues involved in the formation of the ␣2-␣3 loop hydrophobic cluster in NLRP7 are highlighted by gray boxes and underlined. The gray boxes highlight hydrophobic residues, the blue boxes highlight positively charged residues, and the red boxes highlight negatively charged residues. Most PYDs of human NLRPs have hydrophobic residues in identical positions as the ones identified as important for the stabilization of the ␣2-␣3 loop and helix ␣3 in the NLRP7 PYD. AUGUST 27, 2010 • VOLUME 285 • NUMBER 35

JOURNAL OF BIOLOGICAL CHEMISTRY 27405
involved in NF-B and caspase-1 activation. The structure of mouse NLRP10 has also been determined (Protein Data Bank accession code 2DO9), but because the structure determination of the mNLRP10 PYD has not been published and no dynamics data have been reported for this domain, we refer to our comparison with the mNLRP10 PYD in supplemental Figs. S1-S3.
Despite the low overall sequence similarity, multiple hydrophobic residues that are essential for the formation of the PYD hydrophobic core are highly conserved, not only among these PYD domains but also across the entire NLRP1-14 family. These residues include Leu 14 , Leu 17 , Phe 25 , and Leu 58 . Although the hydrophobic core of PYDs is conserved, the architecture of the helices, especially helix ␣3 and the loop connecting helices ␣2 and ␣3, differs significantly (Fig. 3A). The ␣2-␣3 loop and helix ␣3 are well ordered in NLRP7 PYD. The well ordered ␣2-␣3 loop in NLRP7 is strongly anchored by six hydrophobic residues (Trp 30 , Phe 32 , Leu 34 , Leu 38 , Thr 41 , and Trp 43 ; Fig. 3B, left panel). An ␣-helical turn can be identified for residues Val 37 -Lys 40 in NLRP7 PYD. Similarly, the ␣2-␣3 loops of ASC and ASC2 are also well ordered and adopt a similar structure, although helix ␣3 in ASC2 is shorter by one turn. Four or three hydrophobic residues anchor the ␣2-␣3 loop in ASC and ASC2, respectively (Leu 38 , Leu 32 , Tyr 36 , and Ile 39 in ASC; Leu 32 , Phe 36 , and Ile 39 in ASC2; Fig.  3B, middle panels). In addition, hydrophobic and polar interactions between helix ␣3 and ␣4 (Fig. 3, C and D) further stabilize helix ␣3 in the NLRP7, ASC, and ASC2 PYDs.
In contrast, no ␣2-␣3 loop hydrophobic residues or hydrophobic contacts between helices ␣2, ␣3, and ␣4 can be identified in the NLRP1 PYD that could contribute toward any conformational stability of this loop (Fig. 3, B and D, right  panels). As a result, helix ␣3 is missing in the NLRP1 PYD structure, and the ␣2-␣3 loop is highly dynamic, as shown by dynamics NMR measurements (34), and extended toward helix ␣4.
The significant increase of the number of hydrophobic residues in the ␣2-␣3 loop and helix ␣3 from NLRP1 to NLRP7 allows for the formation of an exceedingly stable second hydrophobic cluster in the NLRP7 PYD (Fig. 1D). These hydrophobic residues are also responsible for the differential flexibility of the ␣2-␣3 loop (described in the next section), as well as the formation of helix ␣3. We used the program AGADIR (59) to predict the helical behavior of the residues between the C termini of helices ␣2 and ␣3. The calculations show very low propensities for helix formation (Ͻ1%), confirming that the presence of helix ␣3 in the structures of NLRP7, ASC, and ASC2 PYDs is due primarily to tertiary contacts within the proteins.
Thus, although the conserved central hydrophobic cores of PYDs are expected to lead to a similar fold for helices ␣1, ␣2, and ␣4 -␣6 (Figs. 1C and 2) for all PYDs, further differences in the conformations of the ␣2-␣3 loop and helix ␣3 are expected The pairwise RMSD between NLRP7 PYD and all other PYDs was calculated by superposition of helices ␣1-␣6. The largest structural difference is localized to the ␣2-␣3 loop as well as helix ␣3, which is illustrated in the current orientation. B, the PYDs of NLRP7, ASC, ASC2, and NLRP1. Hydrophobic residues in the ␣2-␣3 loop as well as helix ␣3 are highlighted in pink (stick model) and labeled. The black arrow indicates increased hydrophobicity of helices ␣2 and ␣3 between the different PYDs. C and D, ϳ90°left rotation (C) relative to A, allowing for the identification of hydrophobic residues (highlighted in pink (stick model) and labeled) in helices ␣3 and ␣4 (D).
in other PYDs whose structures have not yet been determined. Therefore, the amino acid composition and, as a consequence, the structural and dynamics behavior of the ␣2-␣3 loop and helix ␣3 likely function as key regulators of the differential signaling by the NLRP receptor family.
NLRP7 PYD Auto-and Cross-correlated Motion Analysis-To further understand the differences between the PYDs, we measured auto-correlated 15 N-relaxation data: 15 N{ 1 H}-NOE, 15 N longitudinal (R 1 ), and transverse (R 2 ) relaxation rates. Similar measurements were reported for the ASC2 and NLRP1 PYD domains, as well as for full-length ASC (which includes both the N-terminal PYD and the C-terminal CARD domain). Comparison of the 15 N{ 1 H}-NOE data, which reports on fast time scale (ps to ns) backbone motions, is shown in Fig. 4A. Although the ␣2-␣3 loop in the PYDs of NLRP7, ASC, and ASC2 exhibit rigidities nearly identical to their neighboring ␣-helices, the ␣2-␣3 loop in the NLRP1 PYD shows substantially increased dynamics. The observed increase in flexibility is most likely a result of the absent second hydrophobic cluster in the NLRP1 PYD.
A R 2 /R 1 ratio plot identified that NLRP7 PYD residues Trp 30 , Leu 34 , Val 37 , Glu 39 , Thr 41 , Trp 43 , and Ser 44 , which are located in the ␣2-␣3 loop, have increased R 2 /R 1 ratios (supplemental Fig. S4). This shows that these residues undergo motions on a slower time scale than those of the rest of the protein. Indeed, in a model-free relaxation analysis ( c ϭ 6.4 ns), these residues required the inclusion of a R ex term to accurately model the data (supplemental Fig. S4). To further confirm the dynamics of this loop, we recorded R 2 rates at two static fields (11.7 and 18.8 T), as well as cross-correlation rates between the 15 N chemical shift ani-sotropy and 15 N-1 H dipole-dipole interactions ( xy ) (supplemental Fig. S4). The cross-correlation rates were used to calculate the exchange contribution using R ex ϭ R 2 Ϫ xy (Fig. 4B), which confirmed the R ex contribution for the aforementioned residues. Using on-resonance R 1 measurements (radio-frequency spin lock ϭ 2 kHz), we determined that the chemical exchange is in the microsecond time regime (supplemental Fig. S4).
Interestingly, two proline residues, Pro 33 and Pro 42 , flank the ␣2-␣3 loop of the NLRP7 PYD. Pro 42 is conserved in the PYDs of NLRP7, ASC, ASC2, and NLRP1. However, Pro 33 is only conserved in NLRP7, ASC, and ASC2 PYDs; it is not present in NLRP1 PYD. A number of cross-peaks from amino acids Nand C-terminal to these prolines show small satellite peaks in the two-dimensional 1 H, 15 N HSQC, indicating a minor conformation, most likely caused by a cis/trans-isomerization of Pro 33 and Pro 42 (supplemental Fig. S5). By comparison of the peak intensities, the minor conformation is estimated to be present at a level of ϳ10%. A similar 10% cis/trans-proline isomerization was identified for Pro 42 in NLRP1 PYD.
The NLRP7, NLRP1, ASC, and ASC2 PYDs Have Distinct Electrostatic Surfaces-Electrostatic surface charge plays a major role in the formation and regulation of death domain protein-protein interactions (60). Initial work on the RAIDD CARD, as well as caspase-9 and Apaf-1 CARD domains, showed the importance of highly charged surface patches for the specific interaction of CARD domains (35,61). A similar mode of electrostatic interaction was also identified for the PYD-PYD interaction of ASC PYD and ASC2 PYD (37). Critically, the PYD-PYD interaction between NLRP1 PYD and ASC PYD and, in turn, the CARD-CARD domain interaction between the ASC CARD and caspase-1 CARD domains are essential for the formation of the inflammasome. Only NLRP1, NLRP3, and NLRC4 have been identified so far to form inflammasomes with clear physiological roles (14).
The electrostatic surface of ASC PYD is distinctively twofaced. A highly positively charged surface is formed near helices ␣2 and ␣3, and a negatively charged surface is observed near helices ␣1 and ␣4 (Fig. 5, A and D, respectively). ASC2 PYD binds via its positively charged helices ␣2 and ␣3 surface to the negative charged surface formed by residues from helices ␣1 and ␣4 of ASC PYD (comprised of residues Lys 21 , Lys 22 , Arg 38 , and Arg 41 and residues Asp 6 , Asp 10 , Glu 13 , Asp 48 , Asp 51 , and Asp 54 from ASC2 and ASC, respectively). Thus oppositely charged electrostatic surfaces are important for mediating these PYD-PYD interactions (37). Fig. 5 compares the electrostatic surface of NLRP7 PYD with those of ASC, ASC2, and NLRP1. We used the Protein Interaction Property Similarity Analysis server (62) to quantify the similarities between the electrostatic surfaces of the PYDs from NLRP1, NLRP7, ASC, and ASC2. The Protein Interaction Property Similarity Analysis server uses the University of Houston Brownian Dynamics program to calculate the electrostatic potential of a protein. Pairwise calculations comparing the NLRP7 PYD to the PYDs of ASC, ASC2, and NLRP1 were used to assess overall electrostatic similarity, where 0 indicates identical electrostatic surfaces, and 2 indicates completely different electrostatic surfaces. Based on an overall electrostatic distance, NLRP7 PYD is most similar to ASC PYD (electrostatic distance, 1.115) followed by NLRP1 PYD (electrostatic distance, 1.145) and ASC2 PYD (electrostatic distance, 1.293), although none of the other PYDs are identical/highly similar to the NLRP7 PYD (i.e. an electrostatic distance between 0 and 0.75).
In fact, a closer look at specific surfaces, such as the ␣2 and ␣3 and ␣1 and ␣4 interaction surfaces (hereafter referred to as interaction patches) illustrates distinct differences in the electrostatic surfaces of these PYDs. ASC2 PYD has the most similar interaction patch electrostatic surface to ASC PYD. Conversely, the interaction patch electrostatic surfaces of NLRP1 PYD and NLRP7 PYD are highly dissimilar. NLRP1 PYD has only a small positively charged interaction patch, similar to ASC PYD and ASC2 PYD, formed mainly by Lys 22 on helix ␣2. This positively charged patch is completely absent on NLRP7 PYD. Indeed, this surface of the NLRP7 PYD is nearly completely hydrophobic with small negatively charged patches. Conversely, the negative charges on opposing surfaces, espe-cially Glu 15 and Glu 56 from helices ␣1 and ␣4, respectively, are much more similar between the PYDs of NLRP7, ASC, and ASC2. Indeed, the NLRP1 PYD has the least negatively charged electrostatic surface at this interaction patch.
The PYDs of ASC and ASC2 have structurally opposed positively and negatively charged electrostatic interaction patches. Critically, these oppositely charged patches have also been shown to be essential for mediating homotypic protein-protein interactions between these PYDs. These electrostatic surfaces of these interaction patches are significantly different for the PYDs of NLRP1 and NLRP7, with the positively charged surface patch necessary for a potential ASC PYD interaction completely missing in NLRP7 PYD. This observation predicts that a protein-protein interaction between the PYDs of ASC and NLRP7 will not occur. To confirm this hypothesis, we experimentally tested the interaction of the NLRP7 PYD with ASC PYD using NMR spectroscopy. Unlabeled ASC PYD was titrated into 15 N-labeled NLRP7 PYD. No chemical shift changes in the NLRP7 PYD two-dimensional 1 H, 15 N HSQC spectrum were identified upon titration with ASC PYD (1:1 molar ratio), confirming our hypothesis that the NLRP7 PYD does not bind directly to the ASC PYD (supplemental Fig. S6). NLRP7 PYD Does Not Interact Directly with FAF-1-FAF-1 is a multidomain protein and a conserved pro-apoptotic factor that also has a central role in the regulation of NF-B. Recently, it was reported that NLRP3, NLRP7, and NLRP12 interact with FAF-1 via its Fas-interacting domain (FID) (19). The reported interaction studies showed that the interaction must be mediated by residues 1-180 of the FAF-1 FID domain. The FAF-1 FID domain consists of an N-terminal UBA domain, known to interact with polyubiquitinated proteins, as well as a C-terminal adjacent UB1, which has a typical ubiquitin fold and is implicated in the inhibition of NF-B signaling.
We used NMR spectroscopy to experimentally test the potential direct interaction of NLRP7 PYD with the FAF-1 FID domain. This is of interest because it would be the first nonhomotypic PYD interaction ever confirmed in vitro. The unlabeled FAF-1 FID domain was titrated into 15 N-labeled NLRP7 PYD. No chemical shift changes in the NLRP7 PYD two-dimensional 1 H, 15 N HSQC spectrum were identified upon titration FIGURE 5. Helices ␣2 and ␣3 form a predominantly hydrophobic surface on NLRP7 PYD. A, electrostatic surface representation of PYD of NLRP7, ASC, ASC2, and NLRP1. Blue, positive charge; red, negative charge. B, residues responsible for these charges are highlighted on the lowest energy conformer of each structure in pink (NLRP7 (light blue), ASC (salmon), ASC2 (gold), and NLRP1 (green)). C, identification of the second PYD interaction site (rotation Y ϭ Ϫ130°; X ϭ 30°) formed by residues on helices ␣1 and ␣4. Residues responsible for these charges are highlighted on the lowest energy conformer of each structure in pink. D, corresponding electrostatic surface representation of the PYDs of NLRP7, ASC, ASC2, and NLRP1. The arrows indicate experimentally confirmed direct homotypic PYD-PYD interactions.

DISCUSSION
PYDs are the most recently identified members of the DD superfamily (60). Indeed, PYDs are now considered the most common DD. However, the currently available structures have not yet revealed their differential modes of interaction. Rather they have raised additional questions, especially because the region that corresponds to helix ␣3 in the DD fold was completely disordered in the structure of NLRP1 PYD (34), which in turn has led to the proposal that conformational change may play a role in PYD-PYD binding. The interest in PYDs stems from their central role in the control of signaling pathways in the immune system, as well as in many apoptotic pathways. Thus it has been suggested that cross-talk among effectors is modulating signaling between the immune system and apoptosis. How this is mediated at a molecular level is still poorly understood.
Here we report the three-dimensional structure of the human NLRP7 PYD. Conserved hydrophobic residues are essential for the formation of the DD fold of helices ␣1, ␣2, and ␣4 -␣6. The NLRP7 PYD has a significant second hydrophobic cluster that allows for the formation of helix ␣3 and anchors it to helix ␣2. Furthermore, this hydrophobic cluster also stabilizes the only extended loop between secondary structural elements, the ␣2-␣3 loop. This is important, because the threedimensional structure of the NLRP7 PYD is significantly different from the NLRP1 PYD and much more similar to the PYDs of ASC and ASC2, two essential proteins in innate immune system signaling. This second hydrophobic cluster is entirely absent in the NLRP1 PYD, resulting in a highly disordered ␣2-␣3 loop that prohibits the formation of helix ␣3. Notably, primary sequence comparison of the NLRP7 PYD with other PYDs (Fig. 2B) demonstrates that this second hydrophobic cluster is generally conserved, and we therefore predict that the NLRP7 PYD is the prevalent fold for most of the 14 identified NLRP PYDs.
Both electrostatic and, to a lesser extent, hydrophobic interactions are important for DD protein-protein interactions. For example, oppositely charged surfaces have been identified as a critical selector of homotypic CARD-CARD and PYD-PYD interactions (35,37), which likely become subsequently stabilized by additional hydrophobic interactions. In ASC PYD, lysine and arginine residues on helices ␣2 and ␣3 form a highly positively charged surface. Negatively charged residues (glutamic and aspartic acids) on helices ␣1 and ␣4 of ASC2 are complementary and allow for a strong and specific interaction between the ASC and ASC2 PYDs.
Interestingly, the NLRP7 PYD surface is unique compared with that of ASC, ASC2, and NLRP1 PYD. In the NLRP7 PYD, the positively charged residues on helices ␣2 and ␣3 are mostly missing. The ␣2 and ␣3 surface is predominantly hydrophobic, and consequently the NLRP7 PYD does not have the typical PYD positively/negatively charged surfaces on opposing sides of the protein. Indeed, for NLRP1 and ASC PYDs, either a highly charged fusion protein (GB1) or a low pH of 3.7, respec-tively, was required to produce a 1 mM sample for the subsequent successful structure determination. In contrast, for NLRP7 PYD, it was possible to use low salt conditions, a much more commonly used pH of 6.0, and no highly charged GB1 fusion protein to concentrate the sample to 1 mM, without the occurrence of aggregation or multimer formation. This behavior is similar to that seen for ASC2 PYD. Clearly, additional PYD structures must be determined to understand this behavior.
These significantly different electrostatic surfaces are likely one of the reasons for differential downstream signaling by the NLRP family. The NLRP1 PYD interacts with the ASC PYD to form the NLRP1 inflammasome (63). Conversely, no inflammasome formation has been reported so far for NLRP7. Based on the differential electrostatic surfaces, it seems unlikely that a charge-driven interaction such as NLRP1 PYD-ASC PYD can also be formed between NLRP7 PYD and ASC PYD. We confirmed the absence of such a direct interaction via an NMR titration experiment. Thus the inability of NLRP7 PYD to bind ASC PYD most likely prohibits the formation of the NLRP7 inflammasome. However, the differential electrostatic surface of NLRP7 PYD likely allows for differential downstream signaling, leading to a negative regulation of pro-IL-1␤ and procaspase-1 processing.