The PRY/SPRY Domain of Pyrin/TRIM20, The Familial Mediterranean Fever Protein, Interacts with β2-Microglobulin to Promote Inammasome Formation

Background. Pyrin/TRIM20 is expressed in the neutrophils and monocytes/macrophages and regulates caspase-1 activation and interleukin-1β maturation. Although the mutations in the PRY/SPRY domain of pyrin cause familial Mediterranean fever (FMF), the mechanism of how mutated pyrin provokes excessive inammation in FMF patients is not well understood. This study was undertaken to explore the new pyrin PRY/SPRY domain-binding protein and to investigate how the interaction affected the pyrin function. Methods. We carried out the yeast two-hybrid screening for pyrin PRY/SPRY domain-binding proteins. Then, protein interactions were investigated using Lenti-X 293T cells expressing pyrin-associated proteins by co-immunoprecipitation analysis. We also used human embryonic kidney (HEK) 293 cells expressing pyrin-associated proteins and human neutrophils stimulated with monosodium urate crystal for immunouorescence staining analysis.

β2MG-pyrin association is suppressed by the p20 subunit of caspase-1. Pyrin M694V mutation promotes the β2MG-pyrin association in the presence of p20. The mutation itself does not affect pyrin-PSTPIP1 interaction in the presence of β2MG. Taken together, our results point to the importance of β2MG-pyrin interaction in the molecular mechanism of pyrin in ammasome activation in FMF.

Methods
Yeast two-hybrid library screening Yeast two-hybrid screening was performed using the Matchmaker Gold Two-hybrid System (Clontech Laboratories, Palo Alto, CA). To construct bait plasmids, full-length human pyrin (FL-pyrin) and the Cterminal region of human pyrin (C-pyrin; aa 598-781) was cloned in frame with the GAL4 DNA-binding domain of pGBKT7 vector. The yeast strain AH109 was transformed with pGBKT7-FL-pyrin and pGBKT7-C-pyrin, respectively. As prey plasmids, human leukocyte cDNA library was inserted downstream of the GAL4 activation domain of pACT2 vector (Clontech Laboratories), and then the yeast strain Y187 was transformed with the plasmids. After mating the AH109 and Y187 transformants, they were spread on plates with high selectivity medium (synthetic drop-out medium without adenine, histidine, tryptophan, and leucine). After identifying the positive colonies by β-galactosidase activity, cDNAs inserted in the prey vector were ampli ed by colony-direct PCR. The positive inserts were sequenced and analyzed using the BLAST program (National Center for Biotechnology Information, Bethesda, MD).

Antibodies (Abs)
Anti-FLAG monoclonal and polyclonal Abs were obtained from Sigma-Aldrich. Anti-V5 monoclonal Ab was obtained from Invitrogen. Anti-pyrin rabbit polyclonal Ab and anti-β2MG mouse monoclonal Ab were from Sigma-Aldrich.
Monosodium urate (MSU) crystal preparation MSU crystals were prepared by the method described by Seegmiller et al. (21). Brie y, uric acid (Sigma-Aldrich) was dissolved in boiling water, and the pH of the solution was adjusted to 7.2 by adding NaOH.
Then the solution was cooled slowly while stirring at room temperature until MSU crystals precipitated. The crystals were washed with ethanol, autoclaved, suspended in PBS at a concentration of 10 mg/ml, and then sonicated to obtain rod-shaped crystals.
Human neutrophils were isolated from the peripheral blood of healthy donors by centrifugation using Polymorphprep (Abbot diagnostic technologies AS) as a manufacturer's instruction. Puri ed neutrophils were resuspended in RPMI medium with 10% FBS and cultured in 35-mm dishes at the concentration of 2 × 10 5 cells per dish. Neutrophils were primed with 500 ng/ml Salmonella enteritica lipopolysaccharide (LPS; Sigma-Aldrich) for 3 h and then stimulated with 200 µg/ml MSU crystals for 2 h, which were subjected to immuno uorescence analyses.

Transfection
For HEK293 cells, plasmid transfection was performed by lipofection using lipofectamine LTX (Invitrogen) for immuno uorescence analyses. For Lenti-X 293T cells, plasmid transfection was performed using calcium phosphate method and for immunoprecipitation assays and Western blotting.

Immunoprecipitation assay
Lenti-X 293T cells (2.0-4.0 × 10 6 cells) transfected with the expression plasmids (see Supplemental   Table 1 for usage) were lysed with the 500 µl of lysis buffer (20 mM Tris-HCl (pH 6.0), 120 mM NaCl, 5% glycerol, and 0.5% Triton X-100). After clari cation by centrifugation at 20,000×g for 30 min, each 20 µl of these lysates was used as supernatant, and the remaining were incubated with 5 µl of agarose beads conjugated with anti-FLAG Ab (Sigma-Aldrich) or 10 µl of magnet beads conjugated with anti-V5 Ab (Medical & Biological Laboratories) for 1h at 4℃. After extensive washing with the lysis buffer, the immunocomplexes were solubilized by adding SDS sample buffer to the beads and subjected to Western blot analysis using a chemiluminescence ECL system (Amersham Biosciences). Anti-FLAG and anti-V5 Abs were diluted to 1:5,000 for Western blot analyses. β-Actin was detected as a loading control using 1:10,000 dilution of Direct-Blot HRP anti-β-actin Antibody (BioLegend). We performed each experiment at least three times to con rm reproducibility. Each data in the gures show a representative of them.
Immuno uorescence staining HEK293 cells transfected with expression plasmids were cultured on human type collagen-coated glass chamber slides (BD biosciences). After washed with PBS, they were xed with 2% paraformaldehyde in PBS for 15 min at room temperature and then permeabilized with PBS containing 0.5% Triton X-100 for 10 min. After blocking with PBS containing 10% BSA, the cells were treated with appropriate primary Abs (rabbit anti-FLAG polyclonal Ab and mouse anti-V5 monoclonal Ab) for 1 h at 37℃ in a moist chamber, washed with PBS containing 0.05% Tween 20, and incubated with secondary Abs (Alexa Flour 488 donkey anti-mouse IgG and Alexa Fluor 555 goat anti-rabbit IgG Abs (Amersham Biosciences)). After washing, the samples were observed under a uorescence microscope, BZ9000 (Keyence).
Neutrophils with or without stimulation were collected by centrifugation at 500 × g for 10 min. After washed with PBS, xed, permeabilized, and blocked as described above, the cells were treated with primary Abs (rabbit anti-pyrin polyclonal Ab and mouse anti-β2MG Ab) diluted to 1:1,000. After washing, the cells were incubated with the secondary Abs. We performed each experiment at least three times to con rm reproducibility. Each data in the gures show a representative of them.

Results
Yeast two-hybrid screening for pyrin-binding protein.
Although certain bacterial toxins induce pyrin in ammasome formation (22), FMF attacks are not always triggered by infection but some other factor including stress and menstrual cycle. Therefore, we hypothesized that some endogenous proteins expressed in neutrophil may interact with pyrin and play an important role as a damage-associated molecular pattern. To identify a molecule that interacts with pyrin, we performed a yeast two-hybrid screening assay using human leukocyte cDNA library as prey. Our use of full-length pyrin (FL-pyrin; aa 1-781) as bait yielded 60 positive colonies (Fig. 1). To exclude false positives as much as possible before subsequent con rmatory binding assays, and to identify pyrinbinding proteins which interact with pyrin via its PRY/SPRY (B30.2) domain where mutations are often observed in patients with FMF, we also performed the screening using the C-terminal region of pyrin molecule (C-pyrin; aa 598-781) as bait. We found 124 positive colonies among which only 9 were isolated from both screenings using FL-pyrin bait and that using C-pyrin bait in common (Fig. 1). The colony-direct PCR and sequence revealed the 9 candidates for genes encoding pyrin-binding molecules, including FTH1, EEF1A1, TPT1, COX5A, CYBA, B2M, ACTB, FLOT1, and LTB (Table 1). Table 1 The genes obtained commonly from both yeast two-hybrid library screenings using FL-and C-pyrin baits.  (23). Two-thirds of β2MG expression amount in neutrophils is localized in gelatinase granules or speci c granules although the physiological role of this intragranular β2MG is still unknown (23). As we identi ed the B2M gene, which encodes β2MG protein, as the candidate for genes encoding pyrin-binding molecules by the yeast two-hybrid screening, we focused on β2MG and tried to con rm the interaction with pyrin by immunoprecipitation assay. We subcloned the full-length B2M cDNA into pcDNA3.1(-) vectors with V5 tag in the N-terminus and co-transfected them with FLAG-tagged pyrin expression vector into Lenti-X 293T cells. The co-immunoprecipitation analysis using anti-FLAG Ab revealed that V5-tagged β2MG was coprecipitated with FLAG-tagged pyrin ( Fig. 2A). Additionally, the reciprocal immunoprecipitation assay using anti-V5 Ab showed that FLAG-pyrin was co-precipitated with V5-β2MG (Fig. 2B). To further con rm the pyrin-β2MG interaction, the expression vectors for FLAG-pyrin and V5-β2MG were co-transfected into HEK293 cells and the localization of these fusion proteins was examined by immuno uorescence staining. As shown in Fig. 2C, the overexpressed V5-β2MG distributed homogeneously in the cytoplasm and did not show any speci c structure when FLAG-pyrin was not overexpressed. On the other hand, V5-β2MG co-localized with pyrin in the speck-like structures in the perinuclear position in more than 90% of the cells in which both FLAG-pyrin and V5-β2MG were overexpressed.
To con rm the localization of endogenous pyrin and β2MG, we examined human neutrophils isolated from the peripheral blood of healthy donors and stained with anti-pyrin and anti-β2MG Abs. β2MG staining was not co-localized with pyrin staining in neutrophils without any stimulation (Fig. 3, upper panels). Because interaction between neutrophils and MSU crystals has been shown to elicit various responses including IL-1β secretion (24), we next investigated their localization in the stimulated neutrophils. Interestingly, β2MG was co-localized with pyrin in neutrophils stimulated with LPS and MSU in the speck-like structures (Fig. 3, lower panels), suggesting that the pyrin-β2MG interaction is induced by neutrophil activation.
β2MG binds to the hotspot of FMF-associated mutations of pyrin.
Because the B2M gene was identi ed not only by yeast two-hybrid screening using FL-pyrin bait but also using C-pyrin bait, we expected that β2MG can bind to the PRY/SPRY domain of pyrin. To further investigate the binding site for β2MG in pyrin molecule, we constructed the vectors which express FLAGtagged pyrin lacking PRY/SPRY domain corresponding to exon 10 (ΔE10 pyrin) and pyrin PRY/SPRY domain (E10 pyrin) and transfected them with V5-β2MG expression vector into Lenti-X 293T cells.
These results indicate that pyrin interacts with β2MG via its PRY/SPRY domain.
β2MG-pyrin interaction recruits PSTPIP1 leading to the assembly of pyrin in ammasome.
PSTPIP1, which was identi ed as a genetic cause of another autoin ammatory disorder, PAPA syndrome, has been shown to interact with the B-box/coiled-coil domain of pyrin (26). Furthermore, PSTPIP binding has been shown to activate pyrin by unmasking the pyrin domain, promoting ASC-mediated oligomerization and in ammasome formation (14). Thus, we next investigated whether pyrin-β2MG interaction affects pyrin-ASC and pyrin-PSTPIP1 interactions. As shown in Fig. 6A, when HEK293 cells overexpressing FLAG-tagged pyrin and V5-tagged ASC were subjected to immuno uorescence staining, FLAG-pyrin and V5-ASC were co-localized in the cytoplasm with the speck formation (arrows in the light eld images), indicating pyrin-ASC interaction. To clarify the effect of pyrin-β2MG interaction on pyrin-ASC interaction, FLAG-tagged pyrin, and V5-tagged ASC expression vectors were transfected into Lenti-X 293T cells with or without V5-tagged β2MG expression vector. The co-immunoprecipitation analysis showed that pyrin did not interact with ASC (Fig. 6B) and β2MG overexpression did not induce pyrin-ASC interaction (Fig. 6C).
We next examined the effect of pyrin-β2MG interaction on pyrin-PSTPIP1 interaction. First, we investigated the interaction between pyrin, PSTPIP1, and ASC in the absence of β2MG overexpression by the co-immunoprecipitation analysis using FLAG-pyrin, V5-ASC expression vectors together with V5tagged wild-type PSTPIP1 or one of the most common PAPA-associated PSTPIP1 mutation, A230T. FLAG-pyrin co-precipitated A230T PSTPIP1 drastically as compared to the wild-type PSTPIP1 (Fig. 6D), consistent with the previous study showing that A230T mutation induces its hyperphosphorylation and increases the a nity to pyrin (26). Additionally, the pyrin-ASC interaction was also promoted by the mutation, indicating that the PSTPIP1 mutation increases the pyrin in ammasome formation (Fig. 6D).
To assess whether β2MG promotes ASC recruitment, we have overexpressed V5-β2MG together with pyrin, PSTPIP1, and ASC in the cells, and investigated the change of ASC recruitment by the β2MGmediated enhancement of pyrin-PSTPIP1 interaction. As a result, β2MG overexpression promoted ASC recruitment (Fig. 6F). The ability is PSTPIP1-dependent because the induction of ASC recruitment was not observed without PSTPIP1 overexpression (Fig. 6C). These results suggest that β2MG functions as a pyrin-binding protein inducing PSTPIP1-mediated pyrin in ammasome formation.
The activation of caspase-1 is caused by the autocleavage of pro-caspase-1, which leads to the formation of active caspase-1 p10/p20 tetramer (27). It is a pivotal process for IL-1β and IL-18 production after the assembly of the pyrin in ammasome. On the other hand, p20 and p10 have been shown to interact with the PRY/SPRY domain of pyrin directly, resulting in the inhibitory effect on IL-1β production (18). Thus, we hypothesized that the binding of β2MG and caspase-1 to pyrin may compete on the pyrin PRY/SPRY domain. We constructed the expression vectors for V5-tagged caspase-1 p20 and transfected them into the Lenti-X 293T cells with the expression vectors for FLAG-tagged ΔE10 or E10 pyrin. Coimmunoprecipitation analysis revealed that caspase-1 p20 interacts with the PRY/SPRY domain of pyrin as shown in the previous report (Fig. 7A) (18). To investigate the relationship between the FMFassociated pyrin mutations and caspase-1 binding, we co-transfected the vectors for FLAG-tagged pyrin deletion mutants, pyrin E10-1, E10-2 or E10-3 with V5-tagged p20 into Lenti-X 293T cells, and subjected them to co-immunoprecipitation analysis. As is the case with β2MG, FLAG-tagged E10-2, which includes the hotspot of FMF-associated mutations, co-precipitated V5-p20 best among the pyrin deletion mutants (Fig. 7B). To elucidate whether caspase-1 has an inhibitory effect on pyrin-β2MG interaction, FLAGtagged pyrin and V5-tagged β2MG expression vectors were co-transfected into Lenti-X 293T cells with or without V5-tagged caspase-1 p20 expression vector. Co-immunoprecipitation analysis revealed that overexpression of caspase-1 p20 inhibits the interaction between pyrin and β2MG (Fig. 7C). These results suggest that p20 acts as an inhibitor for β2MG-mediated promotion of pyrin in ammasome formation and suppresses the excess in ammation via a negative feedback mechanism.
Pyrin M694V mutation changes the balance of a nities toβ2MG and caspase-1 p20.
Homozygous M694V mutation can cause severer forms of FMF and AA amyloidosis in higher prevalence (28). To elucidate whether M694V mutation affects β2MG a nity for pyrin, we constructed expression vectors for FLAG-tagged full-length pyrin and E10-2 fragment with M694V mutation and compared the β2MG a nity between for M694V-mutated pyrin or E10-2 fragment and for the wild type by coimmunoprecipitation assay. As shown in Fig. 8A and 8B, the M694V mutation did not make any impact on the β2MG a nity in the absence of caspase-1 p20 overexpression. Because our data showed that caspase-1 p20 inhibits the association of β2MG with pyrin, we next investigated the effect of M694V mutation on inhibition of pyrin-β2MG interaction by p20. Unlike wild-type pyrin, M694V-mutated pyrin coprecipitated β2MG even in the presence of p20, suggesting that pyrin M694V mutation weakened negative feedback on PSTPIP1-mediated pyrin in ammasome formation by caspase-1 p20 (Fig. 8C).
Finally, we examined whether pyrin M694V mutation affects PSTPIP1 recruitment by coimmunoprecipitation analysis. As a result, M694V-mutation itself did not affect the pyrin-PSTPIP1 interaction in the presence of β2MG (Fig. 8D). These ndings suggest that pyrin M694V mutation leads to the overproduction of pyrin in ammasome not by promoting β2MG-induced PSTPIP1 recruitment but by reduction of the inhibitory effect of caspase-1 p20 on pyrin-β2MG interaction.

Discussion
Although the function of pyrin protein, encoded by the MEFV gene, has been investigated by many researchers since MEFV was identi ed as the responsible gene for FMF, it has not been completely understood how MEFV mutations affect the function of pyrin. In this study, we identi ed β2MG as a protein that binds to the pyrin PRY/SPRY domain, which was competitively blocked by caspase-1 p20 subunit. β2MG-pyrin interaction leads to the recruitment of PSTPIP1, a crucial component of pyrin in ammasome. Pyrin M694V mutation strengthened the β2MG-pyrin interaction.
Recently, several papers have shown that RhoA GTPase indirectly regulates pyrin in ammasome activation. RhoA-dependent serine/threonine protein kinases PKN1 and PKN2 phosphorylate pyrin at S208 and S242 residues, which are located between pyrin and B-box domains, and 14-3-3 protein traps the phosphorylated pyrin, preventing pyrin from in ammasome formation (29). They also elucidated that some bacterial toxins, such as TcdB from Clostridium di cile, inhibit the activity of RhoA, leading to a decrease of PKN1/2 activities and pyrin phosphorylation, which in turn make the pyrin dissociate from 14-3-3 and facilitate active pyrin in ammasome formation. These ndings are supported by the reports on the dominantly inherited disorder called pyrin-associated autoin ammation with neutrophilic dermatosis (PAAND) (30,31). In patients with PAAND, pyrin S242A or G244L mutation cause a decrease of PKN1/2-dependent pyrin phosphorylation and binding to 14-3-3, triggering the spontaneous pyrin activation. Although this RhoA-dependent pyrin activation model well describes the physiological function of pyrin as a surveillance protein against bacterial toxins without directly recognizing the pathogen-associated molecular patterns, it is di cult to explain the pathological signi cance of this model since FMF-associated mutations are clustered in PRY/SPRY domain of pyrin. Considering that RhoA-dependent pyrin activation is observed in murine macrophage whose pyrin lacks PRY/SPRY domain (32) and that there are differences in clinical symptoms between FMF and PAAND, it is plausible that PRY/SPRY domain of human pyrin has a function for in ammasome formation in a certain manner other than the RhoA-dependent pathway. As expected, we have discovered the new PRY/SPRY domainbinding regulator, β2MG, for pyrin in ammasome formation in this study.
β2MG is ubiquitously expressed mainly as a component of MHC class I or CD1 molecules, contributing to antigen presentation. On the other hand, it has been reported that in neutrophil two-thirds of β2MG molecules are localized in granules, including gelatinase granules and speci c granules (33). The function and binding proteins of β2MG contained in granules of neutrophil had not been known before we identi ed β2MG as a pyrin-binding protein in this study. Under normal circumstances, intragranular β2MG and cytoplasmic pyrin never interact because of their different localizations. This raises the question of where and how pyrin interacts with β2MG in the cell. When neutrophils phagocytose foreign bodies resistant to phagosomal digestion such as MSU crystals, phagosomes containing them fuse with granules, eventually rupture, and are unable to digest its contents (34,35). This event is called "phagosomal destabilization" and results in the cytosolic distribution of intragranular proteins. Although the phagosome destabilization has been argued mainly for the activation of NLRP3 in ammasome so far, it can also cause the pyrin in ammasome activation as these two in ammasomes can be activated by the same stimulation (22,36,37). Thus, we suspect that the pyrin-β2MG interaction can be occurred by this phagosomal destabilization before pyrin in ammasome formation. Consistent with this idea, β2MG was not co-localized with pyrin in human neutrophils without stimulation while their co-localization was observed in neutrophils stimulated with LPS and MSU in the speck-like structures. Pyrin can be considered to function as a cytosolic sensor for phagosomal destabilization by detecting β2MG.
FMF is, as the name suggests, prevalent in Mediterranean descendants and the selective advantage for the MEFV mutation carrier is speculated (38). A recent study showed that FMF mutant pyrin interacts less rmly with a virulence factor YopM of Yersinia pestis, which is an intracellular parasite, than the wild-type pyrin, thereby attenuating YopM-induced interleukin-1β suppression (39). Thus, FMF mutations positively selected in Mediterranean populations increase resistance to Y. pestis. Considering other intracellular parasites such as Mycobacterium tuberculosis and Listeria monocytogenes are known to tolerate or escape from the phagolysosomal digestion of macrophages (40), not only crystal structures like MSU but also such pathogenic microorganisms can cause the phagosomal destabilization in neutrophils. Based on our hypothesis, the selective advantage for the MEFV mutation carrier can partly rely on the expected mechanism that β2MG can be released into the cytoplasm by the trigger due to infection and that following pyrin in ammasome activation may have a pivotal role in the control of infection. PSTPIP1 harboring PAPA syndrome-associated mutations (A230T and E250Q) presents a reduced a nity for PEST (rich in proline [P], glutamic acid [E], serine [S], and threonine [T])-type protein tyrosine phosphatase (PTP-PEST), resulting in increased phosphorylation of PSTPIP1. Subsequently, the hyperphosphorylated PSTPIP1 strongly binds to pyrin, leading to excessive formation of pyrin in ammasome (26). Although the importance of PSTPIP1 in pyrin in ammasome formation had been elucidated, it was unknown what causes pyrin-PSTPIP1 interaction other than the PAPA syndromeassociated PSTPIP1 mutations. Our ndings suggest that the β2MG binding to pyrin promotes pyrin-PSTPIP1 interaction, which leads to pyrin in ammasome formation. It indicates the possibility that there exists a RhoA-independent pyrin activation pathway.
The present study also suggests the important role of interaction between pyrin and caspase-1 p20 subunit for the control of pyrin in ammasome activation. There are con icting reports on pyrin-caspase-1 interaction. One study showed all full-length and p20, p10 subunits of caspase-1 interacted with the pyrin PRY/SPRY domain and that the FMF-associated mutation of pyrin decreased its a nity for caspase-1 (18). On the contrary, another paper reported that the interaction of caspase-1 with the pyrin PRY/SPRY domain with the M694V mutation is equally strong as the one observed with the wild-type, suggesting that the mutation does not affect the binding of pyrin to caspase-1 (20). In our study, the caspase-1 p20 subunit strongly bound to the pyrin PRY/SPRY domain, inhibiting pyrin-β2MG interaction, and M694V mutation modi ed this antagonism toward the interaction with β2MG. Thus, we consider that the FMF-associated pyrin mutation can cause stronger in ammasome activation via its decreased a nity for caspase-1 and increased interaction with β2MG.
Based on our ndings, here we propose a novel hypothesis for FMF pathogenesis (Fig. 9). β2MG functions as a pyrin ligand inducing pyrin in ammasome formation and subsequent production of active caspase-1 p20 and p10 subunits, followed by blocking of the β2MG-pyrin interaction by p20 in a negative feedback manner. Pyrin M694V mutation disrupts the negative feedback by caspase-1 p20 subunit, resulting in the promotion of pyrin in ammasome formation and the vicious cycle of in ammation.

Conclusion
We identi ed the novel pyrin-binding protein, β2MG and found that pyrin-β2MG interaction recruits the pyrin in ammasome components, PSTPIP1 and ASC. Although further studies are needed to con rm how β2MG meets pyrin in the cytoplasm of neutrophils, inhibition of the pyrin-β2MG interaction can be a promising candidate as a target for FMF treatment.

Declarations Ethics approval and consent to participate
The study protocol is approved by the ethics committee of Yokohama City University Hospital (B100701027, B191000004), and informed consent was obtained from the healthy controls. The study was conducted based on the Declaration of Helsinki.

Consent for publication
Not applicable.

Availability of data and materials
The datasets analyzed during the current study are available from the corresponding author on reasonable request.