NLRP6 self-assembles into a linear molecular platform following LPS binding and ATP stimulation

NOD-like receptors (NLRs) localize in the cytosol to recognize intracellular pathogen products and initialize the innate immune response. However, the ligands and ligand specificity of many NLRs remain unclear. One such NLR, NLRP6, plays an important role in maintaining intestinal homeostasis and protecting against various intestinal diseases such as colitis and intestinal tumorigenesis. Here, we show that the major component of the outer membrane of gram-negative bacteria, lipopolysaccharide (LPS), binds NLRP6 directly and induces global conformational change and dimerization. Following stimulation by ATP, the NLRP6 homodimer can further assemble into a linear molecular platform, and ASC is recruited to form higher molecular structures, indicative of a step-by-step activation mechanism. Our study sheds light on the mystery of LPS-induced inflammasome initiation, reveals the architecture and structural basis of potential pre-inflammasome, and suggests a novel molecular assembly pattern for immune receptors.


Screening of NLRP6 potential ligands and LPS binding experiments
To identify the potential ligand of NLRP6, a panel of microbial components were selected to be incubated with NLRP6 monomer, including NOD1/NOD2 agonists 2 , Ra/Rc forms of LPS 3 , RNA 4 , and both natural and synthetic DNA sequences 5 (Table 1). Prior to incubation, the ligand candidates were pre-treated as follows: (1) Fig. 2A-C). However, the EM image and measurement of all the LPS chemotypes treated NLRP6 "aggregates" showed homogeneous dimer formation, confirmed oligomerization of NLRP6 did not occur (Supplementary Fig. 2A), which is in contrast with the ATP and LPS induced peak shift.
Referring to previous work, Ra LPS forms small and homogeneous micelles and was successfully used in crystallization of TLR4 3 . Thus we chose the Ra form of for the electron microscopic experiments and SPR experiments.

Electron microscopy and image processing
NLRP6 monomer, dimer (incubated with LPS) and oligomer (incubated with LPS plus ATP) was subjected to Superdex 200 gel filtration chromatography (GE Healthcare). The NLRP6 protein fractions (Fig. 3a) were immediately diluted to 5-10 μg/ml and applied to grids after elution. Preparation of negatively stained samples and image acquisition were as described elsewhere 6 . About 100 particles of both monomers and tetramers were manually picked first.
These particles were then classified and averaged to generate five templates for automatic particle picking, using the GPU-accelerated program GautoMatch (http://feilab.ibp.ac.cn/LBEMSB/AutoMatch.html). Raw particles of dimers, tetramers and hexamers were picked individually (Table S1). The defocus value of each micrograph was determined by CTFFIND3 7 and the CTF was corrected by applying phase flipping to each micrograph using applyctf in EMAN 8 . The reference-free two-dimensional classification was performed using e2refine2d.py in EMAN2 9 . Initial models for monomers, dimers, tetramers and hexamers were built using random conical tilt method in EMAN2 9 . Before model refinement, multi-reference refinements (e2refinemulti.py in EMAN2) were taken for the data sets for monomers and tetramers to coarsely classify them into sub-classes. Only the best classes of monomers and tetramers were kept in order to reduce heterogeneity during further refinement. Since there were fewer particles for dimers and hexamers this procedure was not carried out on these data sets. Model refinements were then performed for all the data sets separately using e2refine.py. All models were refined without any symmetric restraint at first.
Analysis of the symmetry-free models by self-rotation suggested that dimer and tetramers adopted good C2 symmetry. Therefore, both the dimer and tetramer were refined under restraint of C2 symmetry. All models were validated by projection matches and tilt-pair parameter plots. All parameters for image processing, including imaging magnification, box sizes, binning sizes, pixel or voxel sizes, particle numbers and final resolutions are listed in Table S1.

SPR experiment
In order to investigate the interaction between LPS and NLRP6, a kinetics assay was performed at 298 K using a Surface Plasmon Resonance (SPR) Biacore3000 machine (GE Healthcare, Uppsala, Sweden). A running buffer containing 50 mM MES pH 6.5, 500 mM NaCl, 0.005% tween-20 was prepared, vacuum filtered and degassed immediately prior to the experiment. NLRP6 monomer from gel filtration chromatography was dissolved in 10 mM sodium acetate pH 4.5 with a concentration of 5 μg/ml and immobilized on a CM5 sensor chip with 1300 RU. The Ra chemotype of E. coli LPS (Sigma, L9641) was dissolved in running buffer at a concentration of 1 mg/ml, after sonicated on ice so as to separate LPS into small and homogeneous micelles.
Kinetic profiling was performed using the single cycle kinetics method 10 . In each analysis cycle, increased LPS concentrations were injected consecutively over the NLRP6 surfaces and a reference blank flow cell at a flow rate of 10 μl/min. LPS was diluted in the running buffer to four different concentrations in the range 0.9 μM to 25 μM and injected for 60 s at each concentration. After the last injection, the running buffer was run alone for 10 min. Data were analyzed using Biacore 3000 and fit to a 1:1 Langmuirbinding model.