|
ULK1/2 Restricts the Formation of Inducible SINT- Speckles, Membraneless Organelles Controlling the Threshold of TBK1 Activation
Abstract
Spatial and temporal organization of cellular signaling pathways heavily relies on compartmentalization. Structural surfaces for signal transduction can be formed by membranes, components of the cytoskeleton, membrane-surrounded organelles, or membrane-less organelles (MLOs). These MLOs occur in the cytoplasm and nucleoplasm of eukaryotic cells and have recently also been discovered in bacteria (Uversky, 2017; Brangwynne, 2013; Al-Husini et al., 2018) . These liquid-like subcellular compartments are believed to be formed by phase separation (Boeynaems et al., 2018) , a process that involves the spontaneous separation of a supersaturated solution into a dense and a dilute phase. The liquid-like nature of MLOs allows fusion and fission events and a dynamic exchange of components. MLOs comprise a variety of subcompartments including nucleoli, promyelocytic leukemia (PML) nuclear bodies, P bodies, and stress granules (SGs) (Boeynaems et al., 2018; Darling et al., 2018; Wheeler and Hyman, 2018) . The dense phase inside MLOs contains a high concentration of proteins, which may facilitate biochemical reactions and control signaling thresholds from these very crowded environments (Boeynaems et al., 2018; Gomes and Shorter, 2018; Woodruff et al., 2017) . In addition, MLOs can act as dynamic buffers for RNAs and proteins. This buffering function also serves as a passive noise filter and thus reduces the inherent randomness of chemical reactions (Saunders et al., 2012) . Different MLOs have a variety of functions ranging from the expression of rRNAs and pre-assembly of ribosomes (nucleoli) to the organization of the spindle apparatus (centrosomes) and mRNA splicing (splicing speckles) (Bernardi and Pandolfi, 2007; Boulon et al., 2010; Brangwynne, 2013; Hyman et al., 2014) . Aberrant forms of MLOs occur upon failure of the protein quality control system or by mutation of MLO-resident proteins, often causing age-related diseases such as amyotrophic lateral sclerosis, inclusion body myopathy, and frontotemporal dementia (Hock and Polymenidou, 2016; Malinovska et al., 2013; Ramaswami et al., 2013; Taylor et al., 2016) . Furthermore, various viruses hijack MLO proteins to aid in their replication (Dhillon and Rao, 2018; Mö ller and Schmitz, 2003; Nakagawa et al., 2018; Reineke and Lloyd, 2013) . MLOs are dynamically formed by an interplay between RNA and intrinsically disordered proteins (IDPs) that typically harbor low-sequence-complexity domains enriched in polar side chains (Arg, Gln, Glu, Ser, Lys) or structure-breaking amino acids (Gly, Pro) (Uversky and Dunker, 2010) . The dynamic formation of MLOs can be regulated by post-translational modifications such as acetylation, SUMOylation, or phosphorylation (Bernardi and Pandolfi, 2007; de la Vega et al., 2011; Saito et al., 2019; Wippich et al., 2013) or by environmental cues such as changes in temperature, pH, or osmolarity (Uversky, 2017) . Interestingly, a small interfering RNA (siRNA) screen identified several kinases as regulators of MLO formation, including the Ser/Thr kinase TBK1 as a regulator of splicing speckles (Berchtold et al., 2018) .
TBK1 has been initially identified as a component of the antiviral response based on its ability to phosphorylate and thus activate IRF3 or IRF7 transcription factors, which in turn leads to inducible expression of type I interferons (IFN) (Fitzgerald et al., 2003; Sharma et al., 2003) . TBK1, IRF3, and further components for induced type I IFN signaling are known to translocate to perinuclear punctate structures (Saitoh et al., 2009; Seo et al., 2016) . Recent years have witnessed the identification of many additional TBK1 functions, which range from regulation of mitotic microtubule dynamics to the regulation of tumor necrosis factorinduced cell death (Lafont et al., 2018; Pillai et al., 2015; Xu et al., 2018) . Furthermore, stress-regulated TBK1 functions comprise its role in autophagy, mitophagy, and xenophagy (Heo et al., 2015; Pilli et al., 2012; Wild et al., 2011) . More recent evidence shows that TBK1 directly inhibits the AMP-activated protein kinase to increase energy storage and to repress respiration in adipose tissue, thus mediating a cross talk between immune signaling and metabolism . Basal TBK1 signaling is also required for the development of KRAS-driven cancers (Barbie et al., 2009) . Given the involvement of TBK1 in so many stress signaling pathways, TBK1 is a frequent target of viral proteins affecting its localization or protein-protein interactions (Liu et al., 2018; Onorati et al., 2016) .
How can one single kinase such as TBK1 contribute to so many different signaling pathways? One possible answer to this conundrum might rely on the differential association of TBK1 to various adapter proteins, which co-determine its function (Goncalves et al., 2011) . These adapter proteins include SINTBAD (TBKBP1), TANK (I-TRAF), and AZI2 (NAP1) and serve to assist in substrate binding and also affect the subcellular localization of TBK1 (Helgason et al., 2013) . Although the adapter proteins lack any intrinsic enzymatic activity, they can affect biological functions, as exemplified by AZI2-and SINTBAD-regulated intracellular xenophagy of Salmonella (Thurston et al., 2009) . SINTBAD also facilitates activation of the autophagy-initiating kinase ULK1 to control interleukin (IL)-15-induced autophagy in natural killer T cells .
Here we have studied the intracellular distribution of SINTBAD and found its stress-regulated incorporation into cytosolic speckles not corresponding to the known MLO members that were tested. Determination of the SINTBAD interactome allowed the identification of proteins contained in constitutive and inducible SINT-speckles or as regulators of this dynamic process. The formation of inducible SINT-speckles is inhibited by the constitutive chaperone activity of heat shock proteins (HSPs) and ULK1 signaling, whereas KAT2A (also referred to as GCN5) promotes SINT-speckle formation. The SINT-speckle components AZI2 and SINTBAD determine the threshold of TBK1 activation, which partially occurs within SINT-speckles. Knockdown of ULK1/2 was sufficient to trigger SINT-speckle formation, but interfered with focal TBK1 activation in a pathway-specific fashion.
Many proteins participating in the induction of innate immune response signaling have a propensity to form inducible polymers or filaments, which show a reduced solubility in standard lysis buffers (David et al., 2018; Pellegrini et al., 2018; Vajjhala et al., 2017) . To test the possible impact of cellular stress on the solubility of SINTBAD, HeLa cells were exposed to heat shock for different periods, followed by fractionation into a cytosolic and a nuclear/insoluble fraction representing not only chromatin proteins but also poorly soluble or aggregated proteins. A western blot analysis revealed that treatment with heat shock triggered the time-dependent transition of SINTBAD from the soluble to the nuclear/insoluble fraction, whereas this dynamic redistribution between the fractions did not occur for the SINTBAD-related proteins AZI2 or TANK ( Figure 1A) . A similar inducible redistribution of SINTBAD also occurred in arsenite-treated 293T cells ( Figure 1B) , showing that various stressors can trigger the relocation of SINTBAD in different cell types. It was also interesting to test whether SINTBAD translocation to the insoluble fraction is also seen in response to further different stimuli representing inflammatory conditions or osmotic or proteotoxic stress. These results show that most adverse agents, with the exception of inflammatory stimuli, triggered translocation of SINTBAD to the nuclear/insoluble fraction ( Figure 1C ). This behavior was seen for the endogenous proteins and also for the adapter proteins when expressed at moderate levels (Figures S1A and S1B).
To test whether the regulated solubility change of SINTBAD is associated with alterations of its intracellular distribution, immunofluorescence studies were performed in U2OS cells that are ideally suited and widely used for the characterization of subcellular compartmentation. In the absence of a commercial antibody faithfully detecting the endogenous SINTBAD protein in immunofluorescence studies, we generated SINT-BAD-deficient U2OS cells by CRISPR/Cas9-mediated genome editing ( Figures S2A and S2B ) that were A B C D Figure 1 . Stress-Inducible Formation of SINTBAD-Containing Speckles (A) HeLa cells were exposed for the indicated periods to heat shock, and cells were harvested and fractionated into cytosolic (cyto.) and nuclear/insoluble (nucl./insol.) extracts. Western blot was performed to detect the dynamic relocalization of the endogenous adapter proteins SINTBAD, AZI2, and TANK. The detection of phosphorylated ERK1/2 (T202/Y204), tubulin, and histone H3 serves as a control for the treatment, cytosolic, and nuclear/insoluble fraction, respectively. The positions of molecular weight markers are indicated. (B) 293T cells were treated with 0.5 mM arsenite for the indicated times and further analyzed as in (A).
(C) Different cell lines were treated with various stressors; the exact conditions are indicated in the Transparent Methods section. Following these stimulations, cells were fractionated into cytosolic and nuclear/insoluble extracts, followed by the analysis of SINTBAD translocation as described in (A and B). The occurrence of SINTBAD translocation to the nuclear/ insoluble fraction is indicated by a +.
(D) Double-deficient (sgAZI2/sgSINTBAD) U2OS cells stably expressing FLAG-SINTBAD or transfected to express moderate amounts of FLAG-AZI2 or FLAG-TANK were treated with arsenite (0.5 mM, 1 h) or heat shock (43 C, 1 h) and analyzed by indirect immunofluorescence using anti-FLAG antibodies. Nuclear DNA was stained with Hoechst; scale bars, 10 mm. The percentage of cells showing a localization as displayed in the figure is indicated. Figures S3A and S3B ). These cells were used to investigate its distribution in cells exposed to arsenite, heat shock, sorbitol, or Earle's balanced salt solution (EBSS) starvation medium. Untreated control cells showed SINTBAD as a largely cytosolic protein, whereas the various cell stressors triggered the formation of speckles occurring mainly in the cytosol and to a minor extent in the nucleus ( Figures 1D and S4 ), reflecting the transition of SINTBAD from the cytosolic to the nuclear/insoluble fraction. Staining of FLAG-tagged TANK did not reveal any arsenite or heat shockinduced changes of its cytosolic localization, whereas AZI2 already showed a speckled distribution in unstressed cells that was not further influenced by the indicated stressors ( Figure 1D ).
It was then interesting to directly compare the intracellular colocalization of the three TBK1-binding adapter proteins in unstressed and stressed cells. Cell treatment with arsenite or thermal stress triggered the formation of speckles, whereas TANK remained largely cytosolic and showed only minor colocalization with focal SINTBAD (Figure 2A ). The speckled localization of AZI2 in control and stress-exposed cells showed no significant overlap with TANK ( Figure 2B ). Interestingly, the coexpression of AZI2 and SINTBAD triggered the formation of speckles even in unstressed cells and displayed a very high degree of colocalization within these MLOs ( Figure 2C ).
To test the reversibility of inducible speckle formation, cells were exposed to heat shock and then further grown at 37 C to follow the fate of speckles over time. The size and number of speckles decreased in a timedependent manner, and speckles were not detectable after 3 h of cell recovery, as revealed by fluorescence microscopy ( Figure 3A ) and its quantitative analysis ( Figure 3B ). The resolution of speckles occurred also in the absence of de novo protein synthesis ( Figure S5A ) and in the presence of different lysosome inhibitors ( Figure S5B ), suggesting that lysosomal degradation and autophagic processes do not significantly contribute to this process. As chaperone function prevents aberrant phase transition of SGs (Mateju et al., 2017) it was interesting to test whether well-characterized inhibitors of HSP70 (VER155008, Pifithrin-m) or HSP90 (geldanamycin, radicicol) affect SINTBAD localization (Massey et al., 2010; Roe et al., 1999) . Inhibition of HSP70 or HSP90 function resulted in the formation of speckles even in the absence of stress signals ( Figure 3C ), revealing the importance of continuous chaperone function for the cytosolic localization of SINTBAD. Coimmunoprecipitation experiments showed the interaction between SINTBAD and HSP70, which was impaired under thermal stress conditions ( Figure 3D ). The interaction between SINT-BAD and HSP90 was only detectable after coexpression of TBK1 ( Figure 3E ), suggesting a rather indirect interaction that can be controlled by the relative abundance of the known HSP90 interactor TBK1 (Yang et al., 2006) or that the interaction is phosphorylation dependent.
TBK1 and several of its interactors including optineurin and mitochondrial antiviral-signaling protein are known to be recruited to membrane-surrounded organelles such as mitochondria (Fang et al., 2017; Moore and Holzbaur, 2016) or upon RNA virus infection to the Golgi apparatus (Pourcelot et al., 2016) . Thus, we asked whether SINTBAD would be found in association with these organelles. Costaining experiments with untreated and arsenite-or heat shock-treated cells showed significant colocalization neither with mitochondria ( Figure 4A ) nor with Golgi, lysosomes, peroxisomes, endosomes, or endoplasmic reticulum (Figure S6) . In addition, we costained speckles with marker proteins of several known MLOs. These experiments showed colocalization of speckles neither with the SG marker proteins eIF4G ( Figure 4B ) or Ras GTPaseactivating protein-binding protein 1 (G3BP1) nor with the P body marker DCP1a ( Figures S7A and S7B) .
As none of these experiments related the SINTBAD-containing speckles to known subcellular structures, we aimed to identify the SINTBAD interactome by mass spectrometry. Cytosolic and nuclear/insoluble fractions were prepared from 293T cells, transfected with moderate levels of FLAG-tagged SINTBAD, followed by immunoprecipitation and identification of coprecipitating proteins by mass spectrometry as schematically shown in Figure 5A . For further bioinformatic analysis we considered proteins identified in two independent biological and technical replicates and defined high-confidence interactors by three different criteria as specified in detail in Figure S8A . This analysis revealed 150 high-confidence SINT-BAD interactors of which 27% were found in the cytosol, 58% in the nuclear/insoluble fraction, and 15% in both fractions. A list of the high-confidence interactors and the complete mass spectrometry dataset is given in Table S1 . To validate the interaction between SINTBAD and some of the interactors by (D) 293T cells were transfected to express FLAG-SINTBAD together with HA-HSP70. One day later, one cell dish was treated with heat shock as shown and cells were lysed. Lysates were split into two aliquots, one being used for immunoprecipitation (IP) using an anti-HA antibody. The other aliquot was used as an input control sample to ensure correct and comparable protein expression. The samples were further analyzed by western blot using appropriate antibodies as indicated.
(E) 293T cells were transfected to express FLAG-HSP90 together with HA-SINTBAD or Myc-TBK1 as shown. IP was performed using anti-FLAG antibodies, followed by western blot analysis of the IP and input samples.
an independent experimental approach we performed coimmunoprecipitation experiments. These experiments confirmed the results of the mass spectrometry analysis and are exemplified for the interaction of SINTBAD and the tyrosine phosphatase PTPN23 or the autophagy regulator AMBRA1 ( Figure 5B ). Also, binding of ABIN2 (A20-binding inhibitor of nuclear factor-kB activation 2, also referred to as TNIP2) to SINTBAD was confirmed, and mapping experiments revealed the importance of the N-terminal region of SINTBAD for this interaction ( Figure 5C ). Interestingly, expression of SINTBAD and all of its mutants led to an increase in ABIN2 protein levels by an unknown mechanism. The other tested interactions are listed in Table S1 . The identified interacting proteins are known to serve various biological functions, as displayed in Figure 5D . While the biggest group is represented by enzymes and regulators of metabolic processes, two very prominent groups represent proteins with relevance in membrane trafficking or autophagy and also mitosis and cytoskeleton dynamics. The two smallest groups are represented by regulators of innate immunity and proteins mediating transcription, probably assisting in expression of inflammatory gene expression. To test whether mutual interactions have already been described for some of the SINTBAD-interacting proteins, we analyzed the interactors in the STRING database. This analysis revealed the existence of known protein interaction networks between the 150 proteins, as visualized in Figure 5E . We then compared the intracellular localization of SINTBAD with that of some of its interacting proteins. These experiments revealed that some of the proteins such as PTPN23 exactly mirrored the behavior of SINTBAD, as shown in Figure 5F . PTPN23 localizes in the cytosol of control cells and forms speckles after induction of cell stress. Costaining with SINTBAD revealed a complete colocalization in speckles after treatment with heat shock ( Figure 5F ) or arsenite ( Figure S8B ), confirming PTPN23 as a bona fide constituent of speckles. According to the ability of SINTBAD to bind and colocalize with further proteins we term these speckles inducible SINT-speckles.
Further costaining experiments revealed that some interactors can trigger the formation of SINT-speckles even in unstressed cells. An example of such a positive regulator is the lysine acetyltransferase KAT2A. Expression of this predominantly nuclear protein triggered the formation of SINT-speckles even in unstressed cells and also increased the fraction of nuclear speckles ( Figure 6A ). Another interactor triggering the formation of SINT-speckles was ABIN2, as its expression resulted in SINT-speckle formation even in unstressed cells ( Figure 6B ). The dominant effect of ABIN2 also occurred for PTPN23 ( Figure 6B , lower) and resembles that of AZI2, which was sufficient to trigger incorporation of SINTBAD into speckles (see Fig Figure S10 ). A kinase-inactive ULK1 K46I point mutant (Chan et al., 2009) , in contrast to the ULK1 wild-type, formed speckles upon cellular stress when expressed alone ( Figure 6D ). Coexpression of the kinase-inactive mutant with SINTBAD resulted in the formation of SINTspeckles even in unstressed cells, emphasizing the importance of ULK1 kinase activity for controlling SINTspeckle formation. To determine the consequences of ULK knockdown on SINT-speckle formation we interfered with the expression of ULK1 and also ULK2, as both kinases can have redundant functions (Lee and Tournier, 2011; Li et al., 2016) . Loss of both kinases already triggered the formation of SINTspeckles even in unstressed cells ( Figure 6E ), supporting the finding that constitutive ULK signaling is important for restriction of SINT-speckle formation. These kinase-dependent regulatory processes could enable regulation of SINT-speckle formation, as MLOs are highly dynamic and their formation is typically also regulated during the cell cycle (Rai et al., 2018) . This also applies to SINT-speckles, which do not occur in unstressed or stressed mitotic cells ( Figure 6F ).
As expression of ULK1 can trigger TBK1 phosphorylation , we tested the effect of ULK1 expression on the localization of phosphorylated TBK1. ULK1-expressing cells lacked any areas with focal TBK1 phosphorylation, as the activated kinase was found in the cytosol ( Figure 7A ), revealing that ULK1 can control the distribution of phosphorylated TBK1. To test whether loss of ULK1 and ULK2 affects TBK1 phosphorylation in response to thermal stress, the expression of both kinases was downregulated with specific siRNAs ( Figure S11 ). Immunofluorescence analysis of control cells revealed the heat shock-activated TBK1 in focal structures mainly in the nucleus with a partial overlap with cytosolic SINT-speckles. Downregulation of ULK1/2 largely inhibited the heat shock-induced phosphorylation of TBK1 ( Figure 7B ), suggesting an important contribution of these kinases for this activation pathway. In contrast, arsenite-induced TBK1 phosphorylation was not changed by ULK1/2 knockdown and occurred to a significant part in SINTspeckles ( Figure 7B) , showing that the ULK kinases control phosphorylation of TBK1 in a pathway-specific manner.
It was then interesting to test whether cell stress also leads to changes in the intracellular distribution of TBK1. Treatment with arsenite or heat shock resulted in a partial recruitment of TBK1 to SINT-speckles (Figure 7C) , suggesting that this previously identified interaction (Ryzhakov and Randow, 2007) also occurs in MLOs. To investigate a possible contribution of SINTBAD for the activation of TBK1, SINTBAD and AZI2 double-deficient U2OS cells were treated for various periods with arsenite and TBK1 activation was assessed with a phospho-specific antibody by immunoblotting. Knockout of SINTBAD alone had no effect (data not shown), whereas cells lacking SINTBAD and AZI2 ( Figure S2 ) showed reduced TBK1 phosphorylation ( Figure 7D ). To investigate the contribution of SINTBAD and AZI2 for ULK1-induced TBK1 phosphorylation by an independent experimental approach, cells were transfected to express ULK1 together with SINTBAD and/or AZI2. Immunoblotting revealed that ULK1-triggered TBK1 phosphorylation was further enhanced by SINTBAD and AZI2 ( Figure 7E ), corroborating the finding that both adaptor proteins contribute to control of the TBK1 activation threshold. (B) 293T cells were transfected to express hemagglutinin (HA)-SINTBAD along with GFP-AMBRA1 or GFP-PTPN23. One day later, immunoprecipitation (IP) was performed using GFP-Trap beads. IP and input samples were further analyzed by western blot using appropriate antibodies as indicated.
(C) An expression plasmid encoding HA-tagged ABIN2 was transfected into 293T cells together with various FLAG-SINTBAD truncation mutants (DN1: 106-615 amino acid [aa], DN2: 165-615 aa, DC: 1-520 aa). After IP, samples were analyzed by western blot as shown.
(D) Known functions of the respective SINTBAD interactors were retrieved from databases and Pubmed searches. Proteins were assigned to the functional groups as displayed; proteins with several functions are found in more than one group. Only the five largest functional groups (>10 proteins per group) are shown.
(E) Visualization of interacting protein networks in the SINTBAD interactome using the STRING database (Version 11.0). Line thickness indicates the strength of data support. SINTBAD (TBKBP1) is highlighted in dark green and written in bold. Proteins that were tested for their interaction with SINTBAD by co-IP (data not shown) are marked in green or red, indicating a confirmed or unconfirmed interaction, respectively. (F) U2OS cells were transfected to express the SINTBAD interactor PTPN23 alone and were left untreated or exposed to heat shock, followed by the analysis of GFP-PTPN23 localization by fluorescence microscopy. In addition, cells cotransfected to express GFP-PTPN23 and FLAG-SINTBAD were treated the same way. A representative experiment is shown. Scale bar, 10 mm; the percentage of cells showing the displayed phenotype is indicated. The right part schematically summarizes the intracellular localization of the proteins. . Characterization and Regulation of SINT-Speckles (A) U2OS cells were transfected to express the SINTBAD interactor KAT2A alone and were left untreated or exposed to heat shock, followed by the analysis of FLAG-KAT2A localization by fluorescence microscopy. In addition, cells
The severe fever thrombocytopenia syndrome phlebovirus non-structural protein S (NSs) targets the ABIN2/p105 complex to activate proviral signaling cascades (Choi et al., 2019) . As ABIN2 is a core component of SINT-speckles it was interesting to investigate whether the NSs protein also can be recruited to these MLOs. Immunofluorescence analysis not only confirmed the described cytosolic localization of NSs (Choi et al., 2019) but also revealed a fraction of NSs in colocalization with ABIN2 ( Figure 8A ). Induction of cell stress by arsenite or heat shock resulted in an increased recruitment of NSs to SINT-speckles (Figure 8A ). The expression of NSs also increased UKL1-induced TBK1 phosphorylation ( Figure 8B ). In summary, these data show that a virus-encoded protein affecting cellular signaling pathways such as NSs can inducibly associate with SINT-speckles.
Here we identified the stress-induced translocation of SINTBAD to SINT-speckles, as revealed by immunofluorescence and cell fractionation experiments. These SINT-speckles form a subcellular compartment with no significant overlap to other characterized MLOs (Darling et al., 2018) . SINTBAD contains two lowcomplexity domains containing 12.9% Gln (residues 106-346) and 27.7% Pro (residues 340-535). Core components of constitutive SINT-speckles such as ABIN2 and AZI2, and also the inducible interactor SINTBAD, are predicted to harbor long intrinsically disordered regions ( Figure 9A , Table S1 ), which are typical for MLO-resident proteins (Elbaum-Garfinkle et al., Nott et al., 2015; Uversky and Dunker, 2010) . Accordingly, about one-third of the SINT-speckle proteins (47 from 150) are predicted to have >40% disordered regions (Table S1 ). The mechanisms controlling the formation of constitutive SINT-speckles are not known and might involve post-translational modifications such as ULK1/2-mediated phosphorylation. Another possible mechanism could involve the bridging of two ABIN2 dimers by binding to M1-linked tri-ubiquitin chains, which might facilitate ABIN2 assembly to higher-order signaling complexes (Lin et al., 2017) . Also, changes in the relative expression levels of ABIN2 speckle components affect the formation of these MLOs, explaining the abundant finding that expression of a given protein such as ABIN2 or AZI2 can affect the intracellular localization of its interactors. This behavior is characteristic for MLO formation, and accordingly, also overexpression of SG components such as TIA1 or G3BP1 is sufficient to trigger formation of SGs (Kedersha and Anderson, 2007) . This implies that physiological variations in the amounts of SINTspeckle proteins can already affect speckle formation. Regulation of ABIN2 protein levels occurs in the presence of increased glucose levels or by the kinases TPL2 or IKKa/b (Chen et al., 2013; Leotoing et al., 2011; Nanda et al., 2018) , and it will thus be interesting to investigate whether these situations will affect the formation of SINT-speckles. The formation of inducible SINT-speckles is regulated by several mechanisms, as schematically shown in Figure 9B . SINT-speckle formation is triggered by the acetyltransferase KAT2A. A recent study showed the relevance of acetylation of low-complexity domains for the formation of SGs (Saito et al., 2019) , and it will be interesting to investigate whether the enzymatic activity of KAT2A contributes to its ability to promote inducible SINT-speckle formation. Formation of inducible SINT-speckles is antagonized by the kinase activity of its component ULK1, raising the possibility of an autoregulatory control of speckle homeostasis. Interestingly, a recent study showed that, vice versa, SINTBAD Figure 6 . Continued cotransfected to express FLAG-KAT2A and hemagglutinin (HA)-SINTBAD were treated the same way. Scale bar, 10 mm. The right part schematically summarizes the intracellular localization of the proteins.
(B) Upper: The experiment was done as in (A) with the difference that HA-ABIN2 was expressed either alone or together with FLAG-SINTBAD as shown. Lower: Cells were cotransfected to express GFP-PTPN23 and HA-ABIN2, followed by exposure to heat shock and the analysis of colocalization by immunofluorescence as shown. Scale bars, 10 mm.
(C) U2OS cells were transfected to express HA-ULK1 alone or together with FLAG-SINTBAD. Cells were left untreated or exposed to heat shock, stained, and analyzed by fluorescence microscopy. Scale bars, 10 mm. The right part schematically summarizes the results. (D) The experiment was performed as in (A), except that a kinase-inactive HA-ULK1-K46I mutant was transfected into U2OS cells.
(E) Reconstituted SINTBAD-deficient U2OS cells were treated for three days with siRNAs specifically targeting ULK1 and ULK2 or alternatively with a control siRNA. Cells were then analyzed by fluorescence microscopy for the intracellular distribution of SINTBAD as shown. Scale bars, 10 mm.
(F) Reconstituted SINTBAD-deficient U2OS cells were exposed to arsenite or heat shock, and FLAG-SINTBAD was costained with antibodies against phosphorylated TBK1 (S172) as a centrosome marker (Pillai et al., 2015) . The DNA was stained with Hoechst, and only mitotic cells were analyzed. Scale bars, 5 mm.
iScience 19, 527-544, September 27, 2019 537
A B E C D is required for ULK1 phosphorylation occurring during the cell type-specific induction of autophagy , suggesting a mutual cross talk between these SINT-speckle components. Depletion of ULK1/2 leads to the formation of SINT-speckles in unstimulated cells. As expression of a kinase-inactive ULK1 mutant results in the spontaneous formation of SINT-speckles it is reasonable to assume an important contribution of the kinase function for this process. Also, other protein kinases function to restrict the formation of MLOs. The related kinases HIPK1 and HIPK2 decrease the size and number of PML-NBs during the cell cycle (Berchtold et al., 2018) , and active DYRK3 facilitates the dissolution of several types of MLOs during mitosis (Rai et al., 2018) . Kinase-dependent mechanisms might also contribute to the elimination of SINT-speckles during mitosis, a process that is accompanied by massive SINTBAD phosphorylation (Figures S12A and S12B). MLO composition can be modulated in response to stress (Boulon et al., 2010; Dellaire and Bazett-Jones, 2004) , and accordingly this study reveals the recruitment of SINTBAD to SINT-speckles in response to a variety of cell stresses or after inhibition of the chaperone function of HSPs. Heat shock-induced speckle incorporation of SINTBAD can probably also be explained by the temperature-regulated SINTBAD-HSP70 interaction. The relevance of HSP70 was also shown for the prevention of aberrant SG formation (Mateju et al., 2017) , and it will be interesting to reveal whether HSPs play a general role in the control of MLO integrity, as recently implicated by the analysis of the cellular chaperone network (Rizzolo et al., 2017) . This study revealed the existence of a large functional chaperone supercomplex and its preferential interaction with proteins forming foci or condensates under stress conditions.
SINTBAD together with AZI2 controls the threshold of TBK1 phosphorylation, as revealed by loss-of-function and gain-of-function experiments. Active and phosphorylated TBK1 in arsenite-treated cells was largely occurring in the cytosol and showed considerable colocalization with SINT-speckles. In contrast, heat shock-induced TBK1 phosphorylation was mainly nuclear and showed only a partial overlap with SINT-speckles. These differential intracellular localizations together with their distinct dependency on upstream UKL1/2 signals suggest that several pathways lead to TBK1 phosphorylation. Thus it is conceivable that SINT-speckles serve as sites of TBK1 phosphorylation in a stimulus-and context-specific manner. The occurrence of phosphorylated TBK1 in SINT-speckles and also outside from these MLOs can be explained by the fact that only a fraction of TBK1 is found in SINT-speckles at a given time point. A further possible explanation is derived from the mechanism of TBK1 activation, where the initial activation of the kinase leads to rapid interdimer trans-autophosphorylation of its activation loop (Ma et al., 2012) . This implies that after primary activation of the kinase (probably facilitated by the high local protein density in SINTspeckles) active TBK1 can rapidly spread to create high local concentrations at substrate sides. This local enrichment of phosphorylated TBK1 is frequently seen by immunofluorescence and can occur in diverse subcellular localizations (Moharir et al., 2018; Pourcelot et al., 2016) . The localization of TBK1 is also controlled by differential interaction with adaptor proteins including SINTBAD, TANK, and AZI2, which compete for binding to a C-terminal interaction domain in TBK1 (Goncalves et al., 2011) . Contrary to the initial assumption that the TBK1 adaptor proteins control the antiviral function of the kinase, recent publications have shown their dispensability for IRF3 activation (Fang et al., 2017) .
Further possible functions of inducible SINT-speckles might be derived from a set of SINTBAD interactors, which fall into several categories (see Figure 5D ). The smallest group of SINTBAD interactors comprises the expected group of innate immune regulators, but interestingly the largest group is formed by enzymes and Figure 7 . Regulation of TBK1 Phosphorylation by Inducible SINT-Speckles (A) U2OS cells were transfected to express hemagglutinin (HA)-ULK1 and stained for the localization of HA-ULK1 and phosphorylated TBK1 (S172) as shown.
(B) U2OS cells stably expressing FLAG-SINTBAD were treated for 3 days with siRNAs specifically targeting ULK1 and ULK2 or alternatively with a scrambled siRNA (siSCR). Cells were left untreated or exposed to arsenite or heat shock and analyzed by immunofluorescence microscopy for the intracellular distribution of SINTBAD and phosphorylated TBK1 with specific antibodies. Areas of colocalization are shown by arrows. Scale bar, 10 mm; the percentage of cells showing the displayed phenotype is given.
(C) U2OS cells stably expressing FLAG-SINTBAD were treated with arsenite or heat shock and stained for the intracellular localization of SINTBAD and TBK1 with specific antibodies. Scale bar, 10 mm; the percentage of cells showing the displayed phenotype is given.
(D) U2OS wild-type (WT) cells and two U2OS cell clones (DKO #1 and #2) lacking SINTBAD and AZI2 expression due to CRISPR/Cas9-mediated gene editing were treated for the indicated periods with 0.5 mM arsenite. Phosphorylation of TBK1 and MAP kinases (p38, ERK1/2) was determined by immunoblotting with phospho-specific antibodies as shown, to ensure successful cell stimulation. The position of a non-specific band is indicated by an asterisk.
(E) 293T cells were transfected to express HA-tagged ULK1 WT along with FLAG-tagged SINTBAD or AZI2 as shown. After 1 day cell lysates were prepared and analyzed by immunoblotting for the phosphorylation of TBK1.
regulators of cell metabolic pathways controlling glucose phosphorylation, pyruvate decarboxylation, fatty acid synthesis, and amino acid metabolism. This might indicate a role of SINT-speckles in metabolic regulation as a mediator of the cross talk between innate immunity and metabolism (Hotamisligil, 2017; Joseph et al., 2018; Jung et al., 2019) . Interestingly, the ULK kinases have also been implicated in the regulation of glucose metabolic fluxes and lipid metabolism (Ro et al., 2013) . In addition, metabolic processes co-determine effector functions and cell fate decisions of cells from the innate and adaptive immune systems (Ganeshan and Chawla, 2014; Odegaard and Chawla, 2013) . Vice versa, immunomodulatory signals such as cytokines directly regulate metabolic hormones or pathways (Kö nner and Brü ning, 2011; Matsuki et al., 2003) . The second largest group is formed by proteins involved in vesicle trafficking and autophagy, which is consistent with a previous study identifying ABIN2 as a hub protein binding to components of the endosomal sorting complex (Banks et al., 2016) . This set of interactors might be also relevant for the recently uncovered role of SINTBAD as a regulator of IL-15-induced autophagy , adding to the emerging role of innate immune signaling proteins such as TBK1 and TRAF6 for the formation of autophagosomes Pilli et al., 2012; Shi and Kehrl, 2010; Thurston et al., 2009) . The third largest group of the SINTBAD interactome comprises regulators of mitosis and components of the cytoskeleton, in line with previous reports, documenting a function of TBK1 as a centrosome-associated regulator of mitotic microtubule dynamics (Pillai et al., 2015; Kim et al., 2013) . Although this study focuses on the identification and cell biological characterization of SINT-speckles, future studies must comprehensively characterize the function(s) of constitutive and inducible SINT-speckles.
SINT-speckles might also be of pathophysiological relevance in virus infections or protein aggregation diseases. This study shows that a significant fraction of the phlebovirus NSs protein colocalize with ABIN2. It will be interesting to investigate the functional consequences of this association as well as the molecular mechanisms leading to this interaction, as the NSs protein is a structured protein (Barski et al., 2017) . ABIN2 is also bound by the rabies virus-encoded M protein, which in turn controls the expression of inflammatory target genes (Besson et al., 2017) , and also other components of SINT-speckles assigned to the functional group ''inflammation and infection'' (see Figure 5D ) is targeted by viruses or involved in the antiviral response (Schmitz et al., 2014) . Another possible pathophysiological scenario is due to the function of MLOs as signaling hubs in a crowded microenvironment, which can come at the cost of unwanted protein aggregation. Interestingly, partial loss of TBK1 causes protein misfolding diseases, namely, familial amyotrophic lateral sclerosis and frontotemporal dementia (Freischmidt et al., 2015; Gijselinck et al., 2015) . Also, mutations in TBK1 and its substrate protein optineurin contribute to frontotemporal lobar degeneration (Le Ber et al., 2015; Pottier et al., 2015) . This disease is also associated with an elevated SINTBAD expression (Broce et al., 2018) , and it will be very interesting to reveal in future studies whether the formation and function of SINT-speckles is affected in neurodegenerative diseases.
In the present study we define components of constitutive and inducible SINT-speckles and reveal their regulation by ULK1/2, HSPs, and KAT2A. However, we have not identified all biological functions exerted by this dynamically regulated protein assembly.
Supplemental Information can be found online at https://doi.org/10.1016/j.isci.2019.08.001.
We thank Yvonne Horn, Markus Schwinn, and Georgette Stovall for excellent technical assistance. We thank the following colleagues for providing plasmids:
The authors declare no competing interests. Elbaum-Garfinkle, S., Kim, Y., Szczepaniak, K., Chen, C.C., Eckmann, C.R., Myong, S., and Brangwynne, C.P. (2015) . The disordered P granule protein LAF-1 drives phase separation into droplets with tunable viscosity and dynamics. Proc. Natl. Acad. Sci. U S A 112, 7189-7194. Fig. S5 . Disaggregation of heat shock-induced SINT-speckles is independent of de novo protein expression and autophagy, Related to Figure 3 . (A) SINTBAD-deficient U2OS cells reconstituted with Flag-SINTBAD were left untreated or exposed to heat shock. Indicated cells were allowed to recover from heat shock at 37 °C for 3 h in the absence or presence of the translation inhibitor anisomycin (5 µg/ml). (B) SINTBAD-deficient U2OS cells stably expressing Flag-SINTBAD were left untreated or exposed to heat shock, followed by a 3 h long recovery period at 37 °C. Treatments and recovery were performed in the presence of the indicated autophagy inhibitors (0.5 µM Bafilomycin A, 20 mM NH 4 Cl, 20 µM Chloroquine). The percentage of cells showing the displayed phenotype is indicated. et al., 2013)) into the pEGFP-C2 vector (Clontech); PCR-amplified ABIN2 from pCAGGS-E-hABIN2, that was kindly provided by Dr. R. Beyaert (VIB-Ghent University, Belgium, (Van et al., 2001) ) was subcloned with a N terminal HA-tag into pcDNA6 (Invitrogen); pcDNA3.1-HA-ULK1 was obtained from Dr. S. H. Tooze (The Francis Crick Institute, London, UK, (Joachim et al., 2015) ) and its kinase-inactive K46I mutant was generated by site-directed mutagenesis (QuikChange II XL, Agilent, (Chan et al., 2009) ). Flag-KAT2A was from Dr.
Ezra Burstein (UT Southwestern, Dallas, USA, (Mao et al., 2009) ) and pIRES-3x-Flag-NSs (SFTSV) from Dr. J. U. Jung (University of Southern California, Los Angeles, USA, (Choi et al., 2019) ).
293T, HEK-TLR4, HeLa and U2OS cells were cultured in DMEM (Gibco) containing high glucose (4.5 g/l), L-alanyl-glutamine (4 mM) and sodium pyruvate (110 mg/l) supplemented with 10 % fetal bovine serum (FBS), 100 U/ml penicillin and 100 µg/ml streptomycin. Cells were grown in a humidified incubator at 37 o C under 5 % CO 2 . Transient transfections of plasmids were performed using linear polyethylenimine (PEI, Polyscience Inc.) as described previously (Saul et al., 2015) . Unless indicated otherwise, cells were treated with 0.5 mM arsenite for 1 h or were exposed to heat shock at 43 °C in a humidified cell culture incubator with a 5 % CO 2 atmosphere for 1 h. Inhibitors were incubated 1 h prior cell stimulation. Cell stimulations for the experiments displayed in Fig. 1C were done as follows: HeLa cells were treated with 0.5 M sorbitol (30, 60, 90 min), 0.5 mM arsenite (20, 40, 60 min) or 4 % (v/v) ethanol (15, 30 and 60 min). In addition, HEK-TLR4 cells were treated with 1 µg/ml LPS (1, 2, 4 h), U2OS cells with 10 ng/ml IL-1β (0.5, 1.5, 5 h) and HEK293 cells with 20 ng/ml TNFα (30, 60, 90 min).
In order to generate SINTBAD-and/or AZI2-deficient U2OS cells, CRISPR/Cas9-mediated genome editing technology was performed as described (Ran et al., 2013) . The target site for human SINTBAD was designed as an anti-sense sgRNA (5´-CGTAGACTTTGAGGCGGCGT-3´) within the first exon of the SINTBAD gene. The target site for sgAZI2 was designed within the second exon as 5´-GGCCTATCATGCATATCGAG-3´. Oligos were ligated into px459 V2.0 vector (Addgene plasmid #62988) using standard protocols and verified by sequencing. U2OS cells, seeded in a 6 cm dish, were transfected with 1 µg of empty px459 or px459-sgRNA vector. One day after transfection, cells were selected for 30 h using 1 µg/ml Puromycin (Invivogen), diluted and further grown to allow the formation of single-cell clones. These clones were picked and analyzed for SINTBAD, AZI2
and Cas9 expression and verified by sequencing of the genomic DNA at the appropriate locus.
To obtain SINTBAD/AZI2 double-deficient cells, a SINTBAD-deficient cell clone was transfected with px459-sgAZI2 and selected as described above. To reconstitute SINTBADdeficient U2OS cells, cells were transfected with an pcDNA3.1/zeo-Flag-SINTBAD expression construct and continuously selected using 400 µg/ml Zeocin (Invivogen). Singlecell clones were picked an analyzed by Western blotting.
Reconstituted U2OS cells stably expressing Flag-SINTBAD were treated with 1 µM ULK1 and ULK2-specific Accell siRNAs (Dharmacon #A-005049-13-0005 and #A-005396-14-0005) or with a scrambled control RNA (Seibert et al., 2019) according to the protocol using Accell siRNA Delivery Media (Dharmacon #B-005000-500). Three days after siRNA transfection, half of the cells were seeded on coverslips and used for immunofluorescence staining and the remaining cells were analyzed for efficient knockdown by real time qPCR, using the following primers specific for ACTB (5´-CATGTACGTTGCTATCCAGGC-3´, 5´-CTCCTTAATGTCACGCACGAT-3´), ULK1 (5´-ACCCCATTACTGCGAACCTGGA-3´, 5´-GCACGAACAGCAGCGTGAAGC-3´) and ULK2 (5´-TCTGCATCACCATGTGCAAGAA-3´, 5´-AACATCTCATCCAGGGCT-3´).
To prepare cell lysates under native conditions, cells were lysed on ice for 20 min in IGEPAL buffer (20 M Tris/HCl pH 7.5, 150 mM NaCl, 1 % IGEPAL CA-630 (Sigma-Aldrich), 5 % glycerol and freshly added 10 mM NaF, 0.5 mM Na 3 VO 4 , 1 mM PMSF, 5 µg/ml leupeptin and 5 µg/ml aprotinin). The lysates were cleared by centrifugation and the supernatants were transferred into a fresh tube and either used for coimmunoprecipitation studies or mixed with sample buffer for Western blot analysis. To lyse cells under denaturing conditions, the washed cell pellets were resuspended in 1 SDS sample buffer and sonicated two times for 20 sec with a Branson sonifier to shear the genomic DNA. After boiling the samples for 5 min, the lysates were analyzed by Western blotting. For subcellular fractionation experiments, cells, grown and treated in a 6 cm dish, were lysed in 160 µl low-salt buffer (10 mM Hepes, pH 7.9, added PMSF) on ice for 10 min. NP-40 (Roche) was added to a final concentration of 0.25 %, samples were briefly vortexed and centrifuged for 10 sec at 16 000 × g. The supernatants
representing the cytoplasmic fractions were collected in fresh tubes and mixed with 5 × SDS sample buffer. The remaining pellets representing the nuclear/insoluble fractions were washed twice in low-salt buffer and then resuspended in 180 µl 1 × SDS sample buffer, boiled and sheared two times for 20 sec with a sonifier. The purity of the cellular fractions was confirmed by Western blotting, detecting Tubulin and histone H3 as markers for the cytosolic or nuclear/insoluble fraction, respectively.
For coimmunoprecipitation, cleared cell extracts lysed in IGEPAL lysis buffer were filled up to a volume of 600 μl with lysis buffer and supplemented with 1 μg precipitating antibody or control IgG. After adding 20 μl protein A/G agarose (Millipore), the samples were incubated for 4 h at 4°C on a rotating wheel. Alternatively, covalently antibody-coupled affinity gels were used for coimmunoprecipitation (anti-Flag M2 affinity gel, A2220, Sigma-Aldrich;
GFP-Trap_A, Chromotek). Agarose beads were then washed four times with 1 ml cold IGEPAL buffer. Precipitated proteins were eluted by adding 1.5 × SDS sample buffer. Equal amounts of protein were separated by SDS-PAGE, followed by semidry blotting to a PVDF membrane (IPVH00010, Millipore). Further analysis was performed using standard methods.
After blocking the membrane, primary antibodies, diluted in 2 % nonfat dry milk or 5 % BSA (Sigma-Aldrich) in TBST, were incubated overnight at 4 °C. Peroxidase-coupled antibodies (Jackson ImmuoResearch) were diluted 1:5000 in 2 % dry milk in TBST and incubated 2 h at room temperature. Immunoreactive bands were detected using the Western Lightning Plus-ECL reagent (Perkin Elmer) and visualized on a ChemiDoc Imaging System (Bio-Rad).
U2OS cells or their derivatives were grown on coverslips in 12-or 24-well plates. Cells were transfected and/or treated as indicated and described in the figure legend. After washing the cells with PBS, cells were fixed for 1 min with ice-cold methanol/acetone (1:1). After rehydration, the cells were blocked with 5 % BSA in PBS for 1 h at room temperature.
Coverslips were subsequently incubated with the indicated primary antibodies, diluted in PBS containing 1 % BSA and 0.1 % Triton X-100, for 90 min at room temperature or at 4 °C overnight. After washing three times with PBS cells were incubated with the appropriate secondary Alexa488-or Cy3-conjugated antibodies (Jackson ImmunoResearch) diluted 1:3000 in 1 % BSA in PBS for 90 min in the dark. After incubation, cells were washed three times in PBS and the nuclear DNA was stained with Hoechst 33324 (Invitrogen). The samples were mounted with Mowiol mounting medium and stored at 4 °C. Analysis of the stained cells was done using an Eclipse TE2000-E microscope (Nikon) and a 63 × oil-immersion lens. For each condition >30 healthy individual cells were analyzed and pictures of one representative cell were taken with an OCRA-spark digital CMOS camera (C11440-36U, Hamamatsu). For the quantification of protein localizations and cellular phenotypes, at least 100 cells for each condition were analyzed.
To identify SINTBAD interactors by mass spectrometry, 293T cells were either transfected with the pcDNA3.1-Flag-SINTBAD expression construct or the empty vector. Four 10 cm dishes for each condition were taken for large-scale immunoprecipitation. Washed cells were harvested and lysed in 1 ml low-salt buffer as described above. The received cytosolic fraction was mixed with 1 ml IGEPAL lysis buffer, containing Complete Protease Inhibitor
Cocktail and PhosSTOP (Roche). The remaining cell pellet was washed two times in low-salt buffer and then resuspended in 2 ml IGEPAL lysis buffer and sonified two times for 20 sec.
The two fractions were precleared by incubating them 1 h with 3 µg control mouse IgG (Santa Cruz Biotechnology) and 40 µl Protein A/G Agarose. Afterwards, immunoprecipitation was performed by incubating the lysates with 6 µg anti-Flag M2 antibodies together with 80 µl Agarose beads for 4 h at 4 °C on a rotating wheel. Beads were washed five times with 2 ml IGEPAL lysis buffer, transferred to a fresh tube and proteins were eluted at 70 °C for 10 min in LDS sample buffer (Invitrogen). Samples were separated on a 4-12 % Bis-Tris gradient-gel (NuPAGE, Invitrogen), stained with colloidal Coomassie (Invitrogen) and cut into small pieces (7 slides each lane). In-gel digestion of the proteins with trypsin and purification of the peptides was performed as described (Seibert et al., 2019) . Peptide solutions were desalted by stop and go extraction (STAGE) tips (Rappsilber et al., 2003) . The purification and mass spectrometry of two individual experiments was performed with a time lag and therefore analyzed with different instrumental settings. Dissenting setting parameters of analysis 1 and 2 are indicated by a slash. Samples were eluted from STAGE tips with acetonitrile and applied to the UHPLC system (EASY-nLC 1000, Thermo Fisher Scientific) in 0.1 % formic acid. Separation of peptides by hydrophobicity was performed with 50/18 cm in-house packed C18 columns (1.9 µm C18 beads, Dr. Maisch GmbH). Peptide elution was achieved with a binary solvent system (solvent A: 0.1% formic acid; solvent B: 80% acetonitrile, 0.1% formic acid) by increasing the relative amount of B from 10 % to 38 % in a linear gradient within 35/20 min, followed by 5/3 min up to 60 % and another 5/2 min to 95%. Re-equilibration was done within 5 min at 5 %. The samples were transferred to an in line coupled QExactive orbitrap/QExactive HF mass spectrometer (Thermo Fisher Scientific) using a nano electrospray ionization source. Full MS spectra were acquired with a data-dependent Top10/15 method that comprised a resolution of 70,000/60,000 at 200 m/z and an automatic gain control (AGC) target of 3e6 at a maximum injection time of 20 ms. The 10/15 most intense ions were further fragmented with higher-energy collisional dissociation (HCD) at a normalized collision energy of 25/27 and MS² spectra were generated at 35,000/30,000 resolution, AGC target of 5e5/1e5 and maximum injection time 120/64 ms.
Data from two biological replicates (2 28 raw files of IgG control and anti-Flag-SINITBAD) were analyzed using MaxQuant (v1.5.5.18) (Cox and Mann, 2008) and the implemented Andromeda search engine (Cox et al., 2011) . Protein assignment was accomplished with correlation of fragment spectra with the UniProt human database (July 2016). Common contaminants were excluded from the analysis. Data processing was performed with tryptic specifications and default settings for mass tolerance in MS and MS/MS spectra. The minimal peptide length was set to 7 amino acids by default and the false discovery rate on protein and peptide level was set to 1 %. Prior to further processing of the data, contaminants, reverse entries and proteins that were only identified by a modification site were filtered out. In order to define high confidence interactors, identified proteins were classified according to criteria given in suppl. Fig. 8A . Proteins from all three groups were combined and constitute the SINTBAD interactome.
The volcano plots for proteins of group 2a are displayed in suppl. Fig. S13 and were done using the Instant Clue program (Version 0.5.2) (Nolte et al., 2018) . The SINTBAD interactome was analysed by the STRING database version 11.0. Only interactions with a medium confidence score of 0.4 were shown. Line thickness indicates the strength of data support. The assignment of SINTBAD interactors to biological functions and processes was done by combined Genecards, Uniprot and Pubmed searches (www.genecards.org;
www.uniprot.org; www.ncbi.nlm.nih.gov/pubmed). The intrinsic disorder was analyzed using the PONDR prediction tool (http://www.pondr.com) (Peng et al., 2005) and Espritz (Walsh et al., 2012) .
|
