A Comparative Analysis of Paxillin and Hic-5 Proximity Interactomes

Abstract

Focal adhesions serve as structural and signaling hubs, facilitating bidirectional communication at the cell-extracellular matrix interface. Paxillin and the related Hic-5 (TGFβ1i1) are adaptor/scaffold proteins that recruit numerous structural and regulatory proteins to focal adhesions, where they perform both overlapping and discrete functions. In this study, paxillin and Hic-5 were expressed in U2OS osteosarcoma cells as biotin ligase (BioID2) fusion proteins and used as bait proteins for proximity dependent biotinylation in order to directly compare their respective interactomes. The fusion proteins localized to both focal adhesions and the centrosome, resulting in biotinylation of components of each of these structures. Biotinylated proteins were purified and analyzed by mass spectrometry. The list of proximity interactors for paxillin and Hic-5 comprised numerous shared core focal adhesion proteins that likely contribute to their similar functions in cell adhesion and migration, as well as proteins unique to paxillin and Hic-5 that have been previously localized to focal adhesions, the centrosome, or the nucleus. Western blotting confirmed biotinylation and enrichment of known interactors of Hic-5 and paxillin, FAK and vinculin, as well as several potentially unique proximity interactors of Hic-5 and paxillin, including septin 7 and ponsin, respectively. Further investigation into the functional relationship between the unique interactors and Hic-5 or paxillin may yield novel insights into their distinct roles in cell migration.

Introduction

Cell adhesion to the surrounding extracellular matrix (ECM) plays a fundamental role in numerous physiologic and pathologic processes including morphogenesis during development, tissue fibrosis and cancer cell invasion and metastasis (Alpha et al., 2020; Walma & Yamada, 2020; Wells, 2013). Focal adhesions are multi-protein complexes that physically connect the intracellular actin cytoskeleton to the ECM via transmembrane integrin receptors (Burridge et al., 1988). Focal adhesions can transduce mechanical forces from the extracellular environment to the cytoskeleton, and conversely, transmit tension generated by the cell to remodel the ECM (Burridge & Guilluy, 2016). Additionally, focal adhesions serve as sites of integration for signaling pathways involved in diverse cellular processes including proliferation, survival, differentiation, and migration (Humphries et al., 2019).

Paxillin (Turner et al., 1990; Turner & Miller, 1994) and the closely related family member, Hic-5 (TGFβ1i1) (Shibanuma et al., 1994; Thomas et al., 1999) function as molecular adaptor/scaffold proteins at focal adhesions, where they mediate mechanical and biochemical signaling through their interactions with an array of structural and regulatory proteins including vinculin, α-parvin (actopaxin), Focal Adhesion Kinase (FAK), and PTP-PEST, as well as Rho family GTPase regulatory and effector proteins (Deakin & Turner, 2008; Fujita et al., 1998; Nikolopoulos & Turner, 2000; Nishiya et al., 1999; Thomas et al., 1999; Turner, 2000a; Turner et al., 1990).

Paxillin and Hic-5 have similar molecular architectures, including several highly conserved N-terminal LD motifs, interspersed with inherently disordered regions and four contiguous C-terminal LIM domains (Brown et al., 1996; Fujimoto et al., 1999; Tumbarello et al., 2002; Turner, 1994). Accordingly, the two proteins share many common binding partners, including FAK, vinculin and the GIT2(PKL)/ARHGEF7(β-PIX)/PAK1 cassette, but each also have potentially unique interactors (Alpha et al., 2020; Deakin et al., 2012b). For example, paxillin has been reported to interact with the Crk-p130CAS-DOCK180 complex (Nishiya et al., 2001) to promote cell migration (Valles et al., 2004). Conversely, Hic-5 binds Smad3 to inhibit Smad3-induced changes in gene transcription (Wang et al., 2005).

In accordance with their interaction profiles, it is notable that paxillin and Hic-5 have both discrete and overlapping functions. For example, both paxillin and Hic-5 contribute to the regulation of planar cell migration, but play antagonistic roles in regulating cancer cell morphology and migratory characteristics in a three-dimensional (3D) ECM environment (Alpha et al., 2020; Deakin & Turner, 2011; Gulvady et al., 2018). Cancer cells with a high ratio of Hic-5 to paxillin protein expression exhibit phenotypic plasticity, switching between amoeboid and mesenchymal motility, while favoring more mesenchymal migration characteristics that are Rac1-mediated and require stable cell adhesions (Deakin & Turner, 2011; Gulvady et al., 2018). In contrast, cells with low Hic-5 to paxillin ratios tend to only exhibit an amoeboid migratory phenotype, which requires RhoA-mediated contractility, involving few or no adhesions (Deakin & Turner, 2011; Gulvady et al., 2018). Differential roles for the two proteins have also been observed during embryonic development (Crawford et al., 2003; Gao et al., 2007; Hagel et al., 2002; Rashid et al., 2017; Xu et al., 2019), epithelial-mesenchymal transition (Nakamura et al., 2000; Pignatelli et al., 2012; Tumbarello & Turner, 2007), wound repair (Dabiri et al., 2008), and in tumor fibroblast remodeling of the stromal matrix (Goreczny et al., 2017).

In addition to their roles at focal adhesions, both paxillin and Hic-5 can shuttle into and out of the nucleus (Ma & Hammes, 2018; Shibanuma et al., 2003; Wang & Gilmore, 2003), where they function as transcriptional coregulators, particularly as trans-activators for androgen and glucocorticoid receptors (Chodankar et al., 2014; Heitzer & DeFranco, 2006b; Kasai et al., 2003; Yang et al., 2000). Additionally, paxillin has been shown to localize to the centrosome and to function in the regulation of microtubule acetylation, Golgi cohesion and anterograde vesicle trafficking (Deakin & Turner, 2014; Dubois et al., 2017; Herreros et al., 2000; Rezey et al., 2019).

Proximity-dependent biotinylation utilizing an engineered biotin ligase fusion protein (Bio-ID), has been used in cells to identify both stable and transient, or context-dependent protein-protein interactions, as well as detect proteins that are in close proximity to each other (Kim et al., 2016; May & Roux, 2019; May et al., 2020; Roux et al., 2018; Roux et al., 2012; Sears et al., 2019). While several studies have used proximity-dependent biotinylation to document the proximity interaction profiles of several key focal adhesion proteins, including paxillin (Chastney et al., 2020a; Dong et al., 2016; He et al., 2023) and Hic-5 (Byron et al., 2022), a direct comparison of the paxillin and Hic-5 interactomes in the same cellular context has not been performed. Accordingly, in the current study we used paxillin and Hic-5 as baits in a BioID2 analysis to further interrogate the unique and overlapping roles of these closely related proteins both within and outside of focal adhesions.

Results and Discussion

Generation of BioID2-tagged Hic-5 and paxillin

Hic-5 and paxillin cDNA were cloned into the myc-BioID2 vector (Kim et al., 2016). These constructs, as well as the control BioID2 vector, were each stably expressed in U2OS cells. Myc staining demonstrated localization of the paxillin and Hic-5 fusion proteins to focal adhesions (Figure 1A). Treatment of the cells with biotin and subsequent staining with fluorescently-conjugated streptavidin (SA) demonstrated increased biotin labeling of focal adhesions, indicating that the fusion proteins were indeed biotinylating proximal proteins (Figure 1B). In contrast, no enrichment at focal adhesions was observed for the BioID2 vector control cells following either myc or streptavidin staining (Figure 1A–B).

Figure 1: Validation of the Hic-5- and paxillin-BioID2 constructs.

Representative maximum intensity projections of biotin-treated U2OS cells stably expressing the Hic-5, paxillin, or control-BioID2 constructs demonstrate (A) localization of the myc tagged constructs to focal adhesions, co-stained with zyxin and (B) streptavidin (SA) staining of biotinylated proteins at focal adhesions. (C) Western blots of biotin-treated U2OS lysates stably expressing the constructs probed for myc demonstrate bands at the appropriate molecular weight for Hic-5, paxillin, and control-BioID2 constructs (D) Probing with Neutravidin-HRP reveals biotinylation of proteins over a broad range of molecular weights (n = 3).

Western blotting of the respective cell lysates showed myc-tagged bands at the appropriate size for the BioID2-tagged paxillin (95 kDa), Hic-5 (77 kDa), and the control biotin ligase (27 kDa) (Figure 1C). Additionally, probing the membranes with neutravidin-HRP revealed biotinylation of proteins over a broad range of molecular weights in each lysate, with enhanced biotinylation of the respective bait protein (Figure 1D).

Mass spectrometry identified numerous Hic-5 and paxillin proximity interactors

In order to identify the proximity interactors for Hic-5 and paxillin, biotinylated proteins from each of the cell lysates were harvested by precipitation using neutravidin-conjugated agarose beads and subsequently analyzed by label-free quantitative mass spectrometry. A description of the cutoffs and filtering methods used to define the proximity interactors can be found in the Materials & Methods section. The main proximity interactor lists, depicted in the Venn diagram (Figure 2A) and Supplemental Table 1, include 68 identified proximity interactors: 13 specific to paxillin, and 23 unique to Hic-5.

Figure 2: Comparison of Hic-5- and paxillin-proximal proteins and their gene ontologies.

(A) Venn diagram showing unique and shared proximity interactors of Hic-5 and paxillin identified. Gene ontology enrichment analysis of Hic-5 and paxillin proximity interactors by (B) molecular function, (C) biological process, or (D) cellular compartment. The top 10 enriched terms per category are shown. (E) Network of all proximity interactors identified grouped by their function within the cell. Solid lines indicate proximity relationships determined in this study. Dashed lines indicate known interactions from the BioGRID and STRING interaction databases. Circled proteins are grouped by functional similarity and do not necessarily represent physical complexes.

The majority of the shared interactors are classical focal adhesion proteins with previously confirmed physical interactions with either, or both paxillin and Hic-5, including GIT1, GIT2 (PKL), ARHGEF7 (β-PIX), vinculin (VCL), FAK, ILK, α-parvin (actopaxin), β-parvin, talin (TLN 1/2), and tensin (TNS 1/2/3) (Brown et al., 2002; Goreczny et al., 2018; LaLonde et al., 2006; Nikolopoulos & Turner, 2000; Nikolopoulos & Turner, 2001; Nishiya et al., 2002; Nishiya et al., 2001; Petit & Thiery, 2000; Salgia et al., 1995; Turner et al., 1999; Turner & Miller, 1994; Zacharchenko et al., 2016). Zyxin (ZYX), the zyxin family member thyroid hormone receptor interactor 6 (TRIP6), and LIM domain containing 1 (LIMD1) do not bind directly to Hic-5 or paxillin but also belong to the LIM domain protein family of focal adhesion adaptor proteins that contain both LIM domains and inherently disordered regions (IDRs) (Sadler et al., 1992; Sharp et al., 2004; Thomas et al., 1999; Wang et al., 1999). The pseudokinases, PEAK1 and PEAK2 (Pragmin/PRAG1), are core adhesome proteins that were also identified in the previous proximity-dependent biotinylation screens with paxillin and Hic-5 (Byron et al., 2022; Chastney et al., 2020a; Dong et al., 2016; He et al., 2023; Tanaka et al., 2006; Wang et al., 2010).

PEAK1 and PEAK2 have a well-established role in regulating Src-dependent cell migration, partly through their interaction with tensin 3 (TNS3), which modulates β1-integrin activation (Bristow et al., 2013; Senda et al., 2016; Tactacan et al., 2015; Zuidema et al., 2022). PEAK1 and PEAK2 are recruited to the p130CAS/Crk signaling complex, known binding partners of paxillin and identified as proximity interactors in this screen, which leads to downstream Src-dependent phosphorylation of paxillin and focal adhesion turnover (Abu-Thuraia et al., 2020; Bristow et al., 2013; Patel et al., 2020; Petit et al., 2000b; Senda et al., 2016; Thomas et al., 1999; Wang et al., 2010; Webb et al., 2004). In addition to regulating Src activity, PEAK2 can also promote local RhoA activation through its interaction with Rnd2 (Rho7) (Tanaka et al., 2006). As noted earlier, differences in the relative expression of Hic-5 verses paxillin have been shown to induce distinct modes of migration through 3D ECMs, with Hic-5 promoting focal adhesion-dependent mesenchymal migration (Deakin & Turner, 2011; Gulvady et al., 2018), whereas, enrichment of paxillin promotes RhoA-dependent amoeboid migration which requires few to no adhesions (Deakin & Turner, 2011; Gulvady et al., 2018). PEAK1 and PEAK2 may differentially regulate Hic-5 and paxillin’s downstream signaling and focal adhesion dynamics to contribute to the distinct roles of Hic-5 and paxillin in 3D cell motility.

In the case of the paxillin-specific proximity interactors, Crk and KIND2 (kindlin) are well-established paxillin binding proteins that are part of the core adhesome (Rogalski et al., 2000; Schaller & Parsons, 1995; Theodosiou et al., 2016; Williams & Waterston, 1994). One previously undocumented proximity target for paxillin is RapGEF1 (C3G) (Gotoh et al., 1995; Sasi Kumar et al., 2015; Tanaka et al., 1994). Importantly, although not recognized as a core adhesome protein, this activator of the Rap1 GTPase performs a well-established role in regulating cell adhesion signaling (Rothenberg et al., 2023; Turner, 2000b; Voss et al., 2003), promoting the formation of stable paxillin and β1-integrin positive adhesions in part through an interaction with Crk/CrkL and p130CAS (Voss et al., 2003).

Pleckstrin homology domain-containing family H member 1 with MyTH4 domain (PLEKHH1) was also enriched in the paxillin-specific list (Huang et al., 2002). This protein plays a key role in cell repulsion through interactions at the plasma membrane with the ephrin type-B receptor 2 (EphB2) and the actin-based motor myosin IB (Prosperi et al., 2021; Zhong et al., 2006). Paxillin has previously been reported to interact with the closely related EphB1, via Nck, to regulate ephrin-mediated cell migration through activation of FAK (Vindis et al., 2004). Interestingly, EphB2 activation was shown to increase focal adhesion size in a PLEKHH1 dependent manner (Prosperi et al., 2021). Further investigation into the relationship between PLEKHH1 and paxillin in ephrin-mediated cell migration may yield a greater understanding into the mechanism behind how ephrin signaling affects focal adhesions.

The majority of Hic-5-specific proximity interactors, with the exception of LIM and SH3 protein 1 (LASP1) (Grunewald & Butt, 2008), α-actinin 1 (ACTN1) (Laukaitis et al., 2001), and β6-integrin (ITGB6) (Meecham & Marshall, 2020; Sheppard et al., 1990), are not canonical focal adhesion proteins, but are either perinuclear or related to endocytosis. They include nuclear pore complex 93 (NUP93) (Grandi et al., 1997), aldolase A (ALDOA) (Mukai et al., 1986; Saito et al., 2020), vacuolar protein sorting-associated protein 11 homolog (VPS11) (Dulic & Riezman, 1989; Kim et al., 2001), the Rab GDP-dissociation inhibitor 2 (GDI2) (Shisheva et al., 1994b), and striatin-interacting protein 2 (STRIP2), a member of the STRIPAK complex (Goudreault et al., 2009; Madsen et al., 2015). STRIP2 is a perinuclear protein that is involved in lung cancer growth and migration (Qiu et al., 2020). ALDOA, or fructose-bisphosphate aldolase, is a key glycolytic enzyme commonly overexpressed in cancer cells (Saito et al., 2020). However, while ALDOA is not enriched at focal adhesions, a recent study suggests that like Hic-5, it may promote EMT through a hypoxia-inducible factor 1α- (HIF-1α) dependent mechanism, and ALDOA overexpression inhibited expression of common cell-cell adhesion proteins, including β-catenin and E-cadherin (Saito et al., 2020; Zhu & Kyprianou, 2008).

Gene ontology enrichment analysis reveals enrichment of proteins related to cell adhesion and small GTPases

Gene ontology enrichment analysis was performed on the proximity interactor lists for paxillin and Hic-5 from Supplementary Table 1 and the top ten enriched gene ontologies were graphed (Figure 2B–D). Overall, paxillin and Hic-5 have similar enrichment profiles for molecular functions (Figure 2B), biological processes (Figure 2C), and cellular compartments (Figure 2D. For example, both the paxillin and Hic-5 proximity interactors had significant molecular function enrichments for cell adhesion molecule binding, cytoskeletal protein binding, and small GTPase binding. Both groups also had significant biological process enrichments for multiple adhesion processes, as well as cellular compartment enrichments for focal adhesions and actin cytoskeletal compartments. These similarities are not surprising considering their similar localizations and many overlapping functions (Alpha et al., 2020). One notable difference is the lack of enrichment for SH3 domain binding proteins in the Hic-5 list as compared to paxillin.

Enrichment of proximity interactors involved in endocytosis

Further investigation into the various functions and processes of the identified proximity interactors revealed a large subset of proteins involved in endocytosis (Figure 2E). It has been well described that endocytosis is required for integrin recycling and focal adhesion turnover which is critical for proper cell migration (Chao & Kunz, 2009; De Franceschi et al., 2015; Fourriere et al., 2019). Several studies have linked paxillin to the regulation of Rab5-mediated endocytosis and vesicle trafficking at focal adhesions (Chang et al., 2017; Mendoza et al., 2013; Zehrbach et al., 2023). Numerous Rab effectors were identified in our screen, including the paxillin-specific interactor, Erbin (Borg et al., 2000; Liu et al., 2018), and the shared interactors, RN-tre (Lanzetti et al., 2000; Matòsková et al., 1996) and rabenoysin-5 (RBSN) (Nielsen et al., 2000). RN-tre, a Rab GTPase activating protein (GAP), has been reported to localize weakly to focal adhesions, where it regulates integrin endocytosis, focal adhesion turnover, and directed cell migration through inhibition of Rab5 GTPase activity (Palamidessi et al., 2013). Interestingly, loss of paxillin leads to increased Rab5 activity, decreased speed of Rab5-positive vesicles transport, and decreased active β1-integrin at focal adhesions, possibly due to increased Rab-5 mediated endocytosis of β1-integrin (Zehrbach et al., 2023). This suggests that paxillin may regulate Rab5 activity and β1-integrin endocytosis through an interaction with RN-tre.

Although studies have yet to investigate the effects of Hic-5 on Rab5-mediated endocytosis, loss of Hic-5 in cancer associated fibroblasts has been shown to promote trafficking of β1-integrin to the lysosome (Goreczny et al., 2018). Many of the identified Hic-5-specific proximity interactors are involved in endocytosis and integrin recycling, including VPS11 (Dulic & Riezman, 1989; Kim et al., 2001), GDI2 (Shisheva et al., 1994a), coiled-coil domain containing protein 93 (CCDC93) (Phillips-Krawczak et al., 2015), CCZ1 homolog B (CCZ1B) (Kucharczyk et al., 2000), annexin A1 (ANXA1) (Seemann et al., 1996), and IST1 factor associated with ESCRT-III (IST1) (Dimaano et al., 2008). For example, VPS11, CCZ1B, and CCDC93 are known to play a key role in lysosome fusion with endocytic vesicles (Dulic & Riezman, 1989; Kucharczyk et al., 2000; Nordmann et al., 2010; Phillips-Krawczak et al., 2015; Wang et al., 2003). The E3 ubiquitin ligase, VPS11, is a key component of both the class C core vacuole and endosome transport (CORVET) complex on early endosomes and the homotypic fusion and protein sorting (HOPS) complex on late endosomes, which are involved in endosomal conversion and lysosomal fusion, respectively (Balderhaar & Ungermann, 2013; Dulic & Riezman, 1989; Kim et al., 2001; Peplowska et al., 2007; Peterson & Emr, 2001; Plemel et al., 2011; Segala et al., 2019). Similarly, CCZ1B, part of the Mon1-Ccz1 complex, facilitates lysosome fusion with late endosomes through its ability to activate Rab7 (Gao et al., 2018; Kucharczyk et al., 2000; Nordmann et al., 2010; Wang et al., 2003). Interestingly, the loss of some CORVET subunits impairs Mon1-Ccz1 localization to the early endosomes (Nordmann et al., 2010). CCZ1B localization to endosomes also depends on the percentage of phosphatidylinositol-3-phosphate (PI(3)P) in the membrane (Herrmann et al., 2023). Intriguingly, another identified Hic-5 proximity interactor, CCDC93 is known to regulate the concentration of PI(3)P in the endosomal membranes (Singla et al., 2019). CCDC93 is a part of the COMMD/CCDC22/CCDC93 (CCC) complex which regulates endosomal sorting and promotes recycling rather than lysosomal degradation of endosomal contents (Phillips-Krawczak et al., 2015). Previous studies have shown that knockdown of CCDC93 lead to decreased recycling of α5β1-integrins and increased trafficking of integrins to the lysosome, similar to what has been observed upon loss of Hic-5 (Goreczny et al., 2018; McNally et al., 2017; Singla et al., 2019). Given that many focal adhesion proteins are known to be recycled through endosomal compartments and Hic-5 has been shown to impact the trafficking of β1-integrin to the lysosome, Hic-5 may regulate lysosome fusion to endocytic vesicles through positive regulation of CCDC93 and negative regulation of VPS11 and CCZ1B (Caswell et al., 2008; Goreczny et al., 2018).

Hic-5 and paxillin localize to the centrosome

It is well established that many focal adhesion proteins have additional functions at the centrosome (Fielding et al., 2008a). Previous studies have demonstrated that paxillin is required for proper centriole cohesion, through the negative regulation of the histone deacetylase, HDAC6, leading to increased acetylated microtubules (Dubois et al., 2017). Multiple Hic-5 and paxillin binding partners have additional functions at the centrosome, including ILK and FAK, which regulate mitotic spindle organization (Fielding et al., 2008b; Park et al., 2009), and the GIT1/ARHGEF7/PAK1 complex, which promotes centrosome maturation (Zhao et al., 2005).

While centrosomal localization has been previously documented for paxillin (Dubois et al., 2017; Herreros et al., 2000; Rezey et al., 2019; Robertson & Ostergaard, 2011), it has not been reported for Hic-5. Co-staining of pericentrin and either Hic-5 or paxillin in parental U2OS cells and U2OS cells stably expressing either myc-BioID2-Hic-5 or Myc-BioID2-paxillin showed localization of both Hic-5 and paxillin to centrosomes (Figure 3A–B). A line profile drawn through the centrosome (Figure 3A–B) demonstrates co-occurrence of both Hic-5 and paxillin with pericentrin. Staining with myc and streptavidin further confirmed that the myc-BioID2-Hic-5 and myc-BioID2-paxillin constructs were able to localize and biotinylate proteins at the centrosome (Supplemental Figure 1A–B).

Figure 3: Hic-5 and paxillin localize to the centrosome and biotinylate centrosome-localized proximity interactors.

Representative images of parental U2OS cells (A) or U2OS cells stably expressing myc-BioID2-paxillin (B) stained with the centrosomal marker, pericentrin (red), and either Hic-5 or paxillin (green). The representative images are maximum intensity projections of sequential z-planes encompassing the pericentrin staining, in order to reduce background cytoplasmic signal. Intensity profiles of lines drawn through the centrosome show co-occurrence of pericentrin with both Hic-5 and paxillin. Arrowheads indicate the ends of the line used for the line profiles. (C) Centrosome-localized proximity interactors. Solid lines indicate proximity relationships determined in this study. Dashed lines indicate known interactions from the BioGRID and STRING interaction databases.

A closer examination of the Hic-5 and paxillin proximity interactors revealed a number of proteins that also localize to the centrosome (Figure 3C). The paxillin-specific interactor, NEDD1, is critical for γ-tubulin localization to centrosomes and microtubule nucleation (Manning & Kumar, 2007; Manning et al., 2010), while RAPGEF1 regulates centrosome duplication (Nayak & Radha, 2020). Both paxillin and Hic-5 also biotinylated the Dishevelled family members, DVL2 and DVL3, which localize to the centrosomal linker between duplicated centrioles and regulate centrosome cohesion during cell cycle-progression (Cervenka et al., 2016; Kikuchi et al., 2010). In fact, phosphorylation of DVL3 by the kinase NEK2 induces decoupling of DVL3 from centrosomal linker proteins, which is necessary for centrosomal separation and formation of the mitotic spindle (Cervenka et al., 2016). Interestingly, DVL has been shown to form complexes with adenomatous polyposis coli protein (APC) at both the centrosome and focal adhesions (Dikovskaya et al., 2004; Kikuchi et al., 2010; Lui et al., 2016; Olmeda et al., 2003; Peifer, 2022). Furthermore, APC binding to paxillin at focal adhesions, in a complex with DVL, has been reported to regulate focal adhesion dynamics (Cervenka et al., 2016). Paxillin may also promote centrosome cohesion through interaction with APC to regulate maintenance of DVL2/3 localization at the centrosome.

Although Hic-5 localizes to the centrosome (Figure 3A), its function at the centrosome remains unknown. Several centrosome-localized proteins were identified as Hic-5-specific proximity interactors, including nucleophosmin (NPM1) (Feuerstein & Mond, 1987; Okuda, 2002), IST1 (Renvoise et al., 2010), and Rab GDP-Dissociation Inhibitor 2 (GDI2) (Shisheva et al., 1995). Nucleophosmin localizes to the centrosome during G₀ and mitosis (Okuda, 2002). During S-phase, the dissociation of nucleophosmin from the centriole is required for centrosome duplication (Okuda, 2002). GDI2 has also been shown to localize to the centrosome, where it mediates the localization and recycling of GDP-bound Rab proteins (Pfeffer et al., 1995; Shisheva et al., 1995). Rabs play an important role in regulating the pericentriolar material and are required for proper centriole movement during cytokinesis (Hehnly et al., 2012; Krishnan et al., 2022; Shisheva et al., 1995). Further investigation into the functional relationship between Hic-5 and nucleophosmin or GDI2 may identify novel roles for Hic-5 at the centrosome.

Enrichment of nuclear-localized proximity interactors

Many focal adhesion components, particularly LIM domain-containing adhesome proteins, such as paxillin and Hic-5, are known to shuttle to the nucleus and act as transcriptional coregulators (Byron et al., 2022; Heitzer & DeFranco, 2006a; Ma & Hammes, 2018; Sathe et al., 2016). Previous studies have demonstrated that nuclear localization of Hic-5 and paxillin is required for androgen receptor, glucocorticoid receptor, and epidermal growth factor receptor (EGFR)-mediated changes in gene expression (Heitzer & DeFranco, 2007; Kasai et al., 2003; Ma & Hammes, 2018; Sen et al., 2012; Singhai et al., 2014; Yang et al., 2000). Among the list of Hic-5 and paxillin proximity interactors, there was a large subset of nuclear associated proteins (Figure 4) which consisted of numerous nuclear-specific proteins, such as lamin A/C (LMNA), NUP93, RAN, NPM1, NF-KB Inhibitor Epsilon (NFKBIE), KIAA0930, trinucleotide repeat-containing adaptor 6B (TNRC6B), and zinc finger protein 259 (ZPR1) (Dechat et al., 2008; Feuerstein & Mond, 1987; Galcheva-Gargova et al., 1998; Galcheva-Gargova et al., 1996; Grandi et al., 1997; Hicks et al., 2017; Ji et al., 2020; Ribbeck et al., 1998; Suzawa et al., 2017; Whiteside et al., 1997; Zhang et al., 2022), and core adhesome proteins that shuttle to the nucleus, including zyxin, TRIP6, LIM-containing preferred translocation partner in lipoma (LPP), LIMD1, ILK, FAK, CrkL, PEAK2, and LASP1 (Chun et al., 2005; Grunewald et al., 2007; Kadare et al., 2003; Nix & Beckerle, 1997; Petit et al., 2000a; Rhodes et al., 2000; Sharp et al., 2004; Wang & Gilmore, 2001; Weaver et al., 2014). The Hic-5-specific interactor, lamin A/C, is a component of the nuclear lamina, the matrix of proteins located adjacent to the inner nuclear membrane (Dechat et al., 2008). Hic-5 has been detected in nuclear matrix subcellular fractions, thereby placing it near lamin A/C (Yang et al., 2000). ZPR1, shuttles between the nucleus and the cytosol and is known to interact with elongation factor-1alpha (eEF-1a) which is required for cell proliferation and cell viability (Bohnsack et al., 2002; Khacho et al., 2008; Mingot et al., 2013; Vera et al., 2014). DDB1 and CUL4 associated factor 7 (DCAF7), also known as WDR68, has been shown to play an important role in transcription of multiple downstream genes including genes important for left-right patterning during development (Alvarado et al., 2016). Interestingly, one study found an increase in TGF-β signaling upon loss of DCAF7 (Alvarado et al., 2016). Perhaps DCAF7 may regulate TGF-β signaling via an interaction with Hic-5. The numerous transcription- and mRNA processing-related proteins identified as proximity interactors for Hic-5, further support the abundant evidence for Hic-5’s role as a transcriptional co-regulator (Heitzer & DeFranco, 2006b; Yang et al., 2000).

Figure 4: Hic-5 and paxillin biotinylated nuclear-localized proximity interactors.

(A) A network depicting nuclear-localized proximity interactors idenitified in this screen (solid lines) and their reported interactions from the BioGRID and STRING interaction databases (dashed lines).

Previous studies have established numerous roles for paxillin in the nucleus, ranging from regulating gene transcription to facilitating the nuclear export of mRNAs through its association with polyadenylation binding protein 1 (PABP1) (Dong et al., 2009; Ma & Hammes, 2018; Woods et al., 2005; Woods et al., 2002). Interestingly, several nuclear-localized proximity interactors of paxillin were identified, including RAPGEF1 (Shakyawar et al., 2017), Trinucleotide Repeat Containing Adaptor 6B (TNRC6B) (Hicks et al., 2017; Suzawa et al., 2017), Crk (Goh et al., 2000), KIND2 (Yu et al., 2012), and ribosomal protein SA (RPSA) (Jiang et al., 2023). TRNC6B is involved in RNA splicing and RNA silencing in the nucleus (Hicks et al., 2017; Suzawa et al., 2017). RAPGEF1 (C3G) also localizes to the nucleus upon Wnt signaling where it modifies chromatin structure to regulate gene expression (Shakyawar et al., 2017). Further investigation into the relationship between paxillin and TNRC6B or RAPGEF1 may yield a better understanding of the mechanisms for how paxillin regulates transcription and RNA processing in the nucleus.

Although both Hic-5 and paxillin lack a canonical nuclear import/localization signal, they do have nuclear export signals (Shibanuma et al., 2003; Woods et al., 2002). It is therefore assumed that they associate with a carrier complex for transport from the cytoplasm to the nucleus (Brown & Turner, 2004). However, a known carrier complex has yet to be identified. Hic-5 preferentially biotinylated the nuclear pore protein (NUP93) (Grandi et al., 1997) and RAN (Ribbeck et al., 1998). Interestingly, NUP93 has also been implicated in regulating both actin remodeling and extracellular matrix deposition in triple negative breast cancer; two key roles associated with Hic-5 function (Bersini et al., 2020; Goreczny et al., 2018; Goreczny et al., 2017; Gulvady et al., 2019; Nataraj et al., 2022; Pignatelli et al., 2012). RAN is known to translocate RNA and proteins through the nuclear pore complex (Ribbeck et al., 1998). Therefore, RAN and NUP93 could potentially be involved in shuttling Hic-5 between the cytoplasm and nucleus.

Validation of selected proximity interactors

In order to validate selected proximity partners for paxillin or Hic-5, biotinylated proteins were harvested from the respective cell lysates using neutravidin agarose beads and subjected to western blotting (Figure 5A). Probing first for paxillin and Hic-5 demonstrated robust enrichment of these proteins in the cells expressing their respective constructs, as well as their endogenous paxillin and Hic-5 counterparts. This was expected since these fusion proteins would likely cross-biotinylate other fusion protein molecules. Importantly, there was robust biotinylation and enrichment of known direct interactors of Hic-5 and paxillin, such as the focal adhesion proteins FAK and vinculin, but not of the abundant cytoskeletal protein α-tubulin (Fujita et al., 1998; Thomas et al., 1999; Turner et al., 1990; Turner & Miller, 1994).

Figure 5: Validation of selected proximity interactors of Hic-5 and paxillin.

(A) Western blots of the biotinylated proteins from U2OS cells stably expressing the Bio-ID2 constructs after pulldown using neutravidin beads. (B) The volume of sample was normalized based on the mount of biotinylated proteins per sample and the adjusted volume was run on a western blot and probed for septin 7 and α-tubulin. (C) Maximum intensity projections of U2OS cells stably expressing the BioID2 constructs stained for Septin 7, myc, and actin. Insets demonstrate septin fibers along actin stress fibers. Line profiles drawn along actin stress fibers were plotted. Arrowheads in the insets denote the ends of the line used for the line profile.

The focal adhesion protein ponsin (Mandai et al., 1999) and the cytoskeletal protein septin 7 (Sept7) (Mostowy & Cossart, 2012) were selected for further analysis. Ponsin, also known as Sorbin and SH3 domain-containing protein 1 (SORBS1) and C-Cbl-associated protein (CAP), is a member of the vinexin family of adaptor proteins known to regulate cell adhesion and cytoskeletal dynamics (Kioka et al., 2002; Mandai et al., 1999). In addition to its localization to focal adhesions which is mediated by vinculin (Mandai et al., 1999), ponsin has previously been shown to interact with paxillin at nascent costameres during muscle cell differentiation (Gehmlich et al., 2007). This interaction occurs between the second src homology domain 3 (SH3) domain of ponsin and paxillin’s proline-rich region (Gehmlich et al., 2007). Accordingly, ponsin was clearly enriched in the paxillin-BioID2 sample following pulldown with neutravidin beads and western blotting (Figure 5A). Hic-5 does not contain a similar proline-rich domain and ponsin was not detected in the western blot of the Hic-5 pulldown of biotinylated proteins, consistent with a lack of direct binding between ponsin and Hic-5 (Figure 5A). However, the mass spectrometry analysis detected ponsin as a shared proximity interactor of both Hic-5 and paxillin, with fold changes of 8 and 49, respectively (Figure 2A, Supplemental Table 1) (Thomas et al., 1999). Perhaps Hic-5 and ponsin’s mutual binding of vinculin brought them into close proximity, leading to the biotinylation of a smaller amount ponsin in the Hic-5-BioID2 samples than was biotinylated by the BioID2-tagged paxillin, which binds ponsin directly (Supplemental Table 1) (Deakin et al., 2012a; Mandai et al., 1999).

Septin 7 (Sept7), a member of the septin family of GTP-binding proteins, was detected in the mass spectrometry analysis as a novel proximity interactor of Hic-5. Western blotting of the neutravidin pulldown samples initially detected more biotinylated septin 7 in the control-BioID2 samples (Figure 5A). However, more biotinylated proteins are pulled down from the control-BioID2 samples than the Hic-5- or paxillin-BioID2. After performing densitometry on the neutravidin-HRP blot to determine the relative amounts of biotinylated proteins in each pulldown sample, we normalized the volume of sample loaded to the relative amount of biotinylated proteins in each sample and re-probed for septin 7 (Figure 5B). After normalizing for the amount of biotinylated proteins, there was clear enrichment of septin 7 in the Hic-5-BioID2 samples (Figure 5B). While it is well established that septin 7 copolymerizes with other septin isoforms to form septin filaments, we did not detect any enrichment of the other septin isoforms in the mass spectrometry analysis (Mostowy & Cossart, 2012).

The septin filament system, often called the fourth cytoskeleton, is emerging as an important mediator of both physical and biochemical cross-talk between the actin and microtubule networks (Mostowy & Cossart, 2012; Spiliotis, 2018; Spiliotis & Nakos, 2021). For example, septin 2 and septin 9 containing filaments have been shown to localize near focal adhesions and to promote focal adhesion maturation by crosslinking actin at transverse arcs and radial stress fibers, thereby increasing mechanical tension on focal adhesions (Dolat et al., 2014; Mostowy & Cossart, 2012). Furthermore, overexpression of septin 2 and septin 9 led to a significant increase in paxillin and FAK expression, as well as the number of paxillin-labeled focal adhesions, through their interaction with the Rho GAP ARHGAP4, leading to enhanced cell migration and invasion (Kang et al., 2021). Sept7 staining of non-transfected U2OS cells revealed localization along actin stress fibers as well as in dense, fibrillar networks where no actin stress fibers are present (Figure 5C). When analyzing fluorescence intensity along stress fibers using line profiles, paxillin and Hic-5 were both shown to localize just distal to the ends of the septin fibers with a small area of co-occurrence (Figure 5C). Given the role of septins in regulating focal adhesion maturation through their association with actin stress fibers, it is possible that septin is regulating Hic-5 turnover to promote focal adhesion maturation or conversely, Hic-5 may function to stabilize septin fibers.

In conclusion, a comparison of the proximity interactors for the closely related focal adhesion adaptor proteins paxillin and Hic-5 highlighted the significant overlap of their respective interactomes that likely contribute to their shared functions in cell adhesion and migration. Several potentially unique proximity interactors for the respective proteins were identified, including RapGEF1 and ponsin for paxillin, and Sept7 and Nup93 for Hic-5. In addition to describing an enrichment of endocytosis-, centrosomal-, and nuclear-related proximity interactors, we demonstrated a previously unreported localization of Hic-5 to the centrosome. Additional investigation into a potentially novel proximity interactor of Hic-5, septin 7, revealed co-occurrence between Hic-5 and septin 7 at the distal ends of actin stress fibers. Further characterization of these interactions may lead to novel insight into the distinct biologic functions of paxillin and Hic-5.

Materials & Methods

Cell lines & reagents

U2OS cells from American Type Culture Collection (RRID:CVCL_0042) were cultured in DMEM with 4.5 g/L glucose (Corning, cat# 15017CM) supplemented with 100 IU/mL penicillin, 100 μg/mL streptomycin (Corning, cat# 30002CI), 2 mM L-glutamine (Corning, cat# 23- 25-005-CI), 1 mM sodium pyruvate (Millipore Sigma, cat#107360), and 10% fetal bovine serum (Atlanta Biologicals, cat# S11150). Antibodies used in this study include antibodies against paxillin clone 165 (BD Biosciences, cat# 610620, RRID:AB_397952) and clone 349 (BD Biosciences, cat# 610052, RRID:AB_397464), FAK (clone 77/FAK, BD Biosciences, cat# 610088, RRID:AB_397495), Hic-5 (clone 34/Hic-5, BD Biosciences, cat# 611165, RRID:AB_398703), myc (clone 9B11, Cell Signaling Technology, cat# 2276S, RRID:AB_331783), vinculin (clone VIN-11-5, Sigma Aldrich, cat# V4505, RRID:AB_477617), zyxin (clone EPR4302, abcam, cat# ab109316, RRID:AB_3086768), SEPT7 (Immuno-Biological Laboratories, cat# 18991, RRID:AB_10705434), ponsin (SORBS1) (Novus Biologicals, cat# NBP1-86641, RRID:AB_11032588), pericentrin (Abcam, cat# ab4448, RRID:AB_304461), and α-tubulin (clone DM1A, Abcam, cat# ab7291, RRID:AB_2241126). Other reagents used for staining or blotting include neutravidin- horse radish peroxidase (HRP) (Thermo Scientific, cat# PI31001), streptavidin-AlexaFluor 488 (Molecular Probes, cat# S32354), Acti-stain 670 (Cytoskeleton, Inc., cat# PHDN1), anti-rabbit IgG DyLight 488 (Invitrogen, cat# 35552, RRID:AB_844398), anti-rabbit IgG DyLight 550 (Invitrogen, cat# 84541, RRID:AB_10942173), anti-mouse IgG AlexaFluor 488 (Jackson ImmunoResearch, cat# 715-545-150, RRID:AB_2340846), anti-mouse IgG DyLight 550 (Thermo Scientific, cat# 84540, RRID:AB_10942171), anti-mouse IgG HRP (Jackson ImmunoResearch, cat# 115-035-003, RRID:AB_10015289), and anti-rabbit IgG HRP (BioRad Laboratories, cat# 170-6515, RRID:AB_11125142). Additional reagents include Roche cOmplete Mini EDTA-free Protease Inhibitor Cocktail tablets (Krackeler Scientific, cat# 45-4693159001 and High Capacity NeutrAvidin Agarose (Fisher Scientific, cat# PI29202).

Construct generation, cell culture, and transfection

Paxillin (chicken) or Hic-5 (mouse) cDNA were subcloned into the multiple cloning site (MCS) of the myc-BioID2-MCS construct (addgene #74223, RRID:Addgene_74223) using Gibson Assembly (New England BioLabs, Inc., cat# E5510S) (Kim et al., 2016). Both paxillin and Hic-5 were C-terminally tagged. U2OS cells were transfected with these constructs using Lipofectamine LTX with Plus Reagent (Invitrogen, cat# 15-338-100). Mixed populations, stably expressing the constructs, were generated by selecting for 10 days in media without penicillin and streptomycin containing 0.2 mg/mL G418 (Fisher Bioreagents, cat# BP6735).

Western blotting

For analysis of the BioID2 construct products by western blot, U2OS cells expressing the constructs were incubated for 18 hours with 50 μM biotin and lysed with boiling 2X Laemmli sample buffer. Lysates were run on SDS-PAGE gels, transferred to a nitrocellulose membrane, and blotted with primary antibodies and the appropriate HRP-conjugated secondary antibodies or with neutravidin-HRP. Membranes were then detected by enhanced chemiluminescence using SuperSignal West Pico PLUS Chemiluminescent Substrate (Thermo Scientific, cat# PI34580) using a ChemiDoc MP Imaging System (Bio-Rad Laboratories, RRID:SCR_019037) and its associated Image Lab densitometry analysis software (Bio-Rad Laboratories, RRID:SCR_014210). To normalize samples by biotinylated protein amount, as in Figure 5B, the total intensity of the neutravidin signal per sample (Figure 1D) was first calculated using densitometry in FIJI (version 1.53t) (Image J, National Institutes of Health, RRID:SCR_002285) (Schneider et al., 2012) and the volume of each sample loaded was adjusted accordingly.

Immunofluorescence

U2OS cells were seeded on fibronectin-coated coverslips for 24 hours, fixed with 4% paraformaldehyde for 10 min, permeabilized with 0.1% TritonX-100, quenched with 0.1 M glycine, and blocked overnight at 4°C with 3% BSA in 1x PBS. For Figure 3A, U2OS cells were simultaneously fixed and permeabilized with 4% paraformaldehyde, 0.1% TritonX-100 for 15 min prior to quenching and blocking as above. Cells were then incubated for 2 hours at 37°C with primary antibodies diluted in PBS with 0.3% BSA, washed 3 times with 0.05% Tween-20 in PBS (PBST), incubated for 45 minutes at room temperature with secondary antibodies and phalloidin, washed for 3 times with PBST, when necessary counterstained with DAPI, washed twice with distilled H2O, and mounted using Prolong Glass (Life Technologies, cat# P36980) per manufacturer’s recommendations. Confocal Z-stacks were taken using a Leica SP8 laser scanning confocal microscope (RRID:SCR_024563) with an HPX Plan Apochromat 63x/1.4 NA oil λ BL objective. Detection was performed using three HyD (Hybrid Detectors) detectors and one PMT (photomultiplier tube). Images were acquired using the Leica LAS X software package (RRID:SCR_013673).

For analysis of immunofluorescence images, FIJI (version 1.53t) (Image J, National Institutes of Health, RRID:SCR_002285) (Schneider et al., 2012) was used to generate line profiles and assemble representative images. The line profile graphs were plotted using GraphPad Prism 10.1.0 (GraphPad Software, Boston, Massachusetts USA, RRID:SCR_002798).

Proximity Ligation analysis using BioID2

Biotin labeling and isolation of proximity partners for BioID2-tagged paxillin and Hic-5 was performed as previously described with minor modifications (n = 2) (Chastney et al., 2020b; Roux et al., 2018). Briefly, four 10-cm diameter cell culture dishes containing approximately 85% confluent U2OS cells for each BioID2 construct were biotinylated by adding 50 μM biotin to the standard DMEM growth media with 200 μg/mL G418 for 18 hours. Cells were washed 2X with ice-cold PBS and lysed on ice in 400 μL/plate of lysis buffer (50 mM Tris-HCl pH 7.4, 150 mM NaCl, 1% TritonX-100, 0.5% sodium deoxycholate, 0.1% sodium dodecyl sulfate, 0.5 mM dithiothreitol, 1 mM EDTA pH 8, 8 U/mL DNase I, and 1 protease inhibitor cocktail tablet/10 mL). Lysed cells were then scraped and combined into one 2 mL microfuge tube per construct. DNA was sheared by three sequential aspirations through a 25-gauge needle. 360 μL of chilled Tris buffer (50 mM Tris-HCl pH 7.4) was added to each tube and DNA was sheared by a further three sequential aspirations through a 27-gauge needle.

Following 15 minutes incubation on ice, lysates were centrifuged for 10 min at 4°C, 20,000xg. Separately, neutravidin-conjugated agarose beads were washed 3 times with ice-cold Tris buffer, resuspended as a 1:1 bead slurry in Tris buffer, and then aliquoted with 25 μL of slurry per 2 mL microfuge tube. Once centrifugation was complete, supernatants were transferred to the tubes containing the bead slurries and incubated on a rotating stirrer overnight at 4°C. Beads were pelleted and the supernatant was removed and discarded. Beads were resuspended at room temperature for 5 minutes in wash buffer 1 (50 mM Tris-HCl pH 7.4, 1% sodium dodecyl sulfate, and 1 complete protease tablet/10 mL). Following centrifugation, the beads were resuspended and left on ice for 5 minutes in wash buffer 2 (10 mM Tris-HCl pH 7.4, 150 mM NaCl, 0.5% sodium deoxycholate, 0.5.% TritonX-100, 1 mM EDTA pH 8, and 1 complete protease tablet/10 mL). After pelleting, the beads were again washed in wash buffer 2. The wash buffer was removed and 40 μL of 2X Laemmli sample buffer with 100 μM biotin was added to each tube to harvest the biotinylated proteins. Samples were boiled for 10 min at 100°C and the supernatants were transferred to fresh tubes using gel loading tips and stored a −20°C. The SDS-PAGE samples were then run on a short polyacrylamide gel (5% acrylamide stacking, 10% acrylamide resolving) at 180 V until the dye front had entered the resolving gel by approximately 2.5 mm. The gel was then stained with Coomassie blue for 30 min, de-stained for 4 hours and washed 3 times in distilled H2O, then left covered overnight. Gel bands for each of the conditions were carefully excised from the gel in a cell culture hood using fresh razor blades on a glass plate, leaving behind the large neutravidin band to reduce contamination by this protein. Gel slices were diced into 1–2 mm squares and transferred to low-retention Eppendorf tubes and stored at −20°C.

Quantitative mass spectrometry

In-gel digestion of the diced gel slices was performed by the Upstate Core Facility for Proteomics and Mass Spectrometry. Briefly, the small gel pieces were sequentially incubated with dithiothreitol (DTT) and iodoacetamide to reduce and alkylate cysteine residues. Proteins were digested overnight at 37°C with trypsin and the resulting peptides were extracted from the gel using an acetonitrile/trifluoroacetic acid extraction solution. Peptides were then dried using a centrifugal evaporator and stored at −20°C until ready for use. After resuspension, peptides were run on a Thermo Scientific Orbitrap Fusion Lumos Tribrid Mass Spectrometer (RRID:SCR_020562). Raw data was analyzed using Thermo Proteome Discoverer software (RRID:SCR_014477) using SequestHT as the database search algorithm and Percolator for peptide scoring, statistics, and validation. Peptides were searched against a human proteome library and the search allowed for variable modifications including methionine oxidation, N-term protein acetylation, and lysine biotinylation, as well as carbamidomethylation as a fixed modification. The search also allowed for semi-tryptic specificity with up to 4 missed cleavages.

Data Analysis

Spectral counts were calculated from the spreadsheet data generated by the Thermo Proteome Discoverer software (RRID:SCR_014477) with the following Excel formula: =round (“# peptides” / “# AAs” x 10,000). The final multiplication term is (raw peptide counts normalized to protein length) for all three groups (paxillin-, Hic-5-, and control-BioID2). These were uploaded to the CRAPome analysis pipeline (https://reprint-apms.org/), which compared the Hic-5- and paxillin-BioID2 to the control-BioID2 hits to obtain the empirical fold change (FCA), as a measure of enrichment, and the Significance Analysis of INTeractome (SAINT) probability score, as a measure of specificity (Choi et al., 2011; Mellacheruvu et al., 2013). The cutoffs were set as SAINT score ≥ 0.5, FCA ≥ 2, and abundance (average spectral counts; a measure of baseline abundance) ≥ 3.5. These scores are commonly used and assess data quality for both specificity of interaction and abundance (“amount”) of protein. Potential hits which were present in only one N were manually removed from the proximity interactor list. Cutoff determination was guided by those chosen in similar studies (Roux et al., 2018), but also involved examining the lists of output proteins and adjusting so as not to lose any well-known interactors. Lists were also filtered to remove any known common contaminants from the CRAPome database of proximity-dependent biotinylation in HeLa cells, which was deemed an acceptable substitute given that there are no currently available contaminants lists from proximity-dependent biotinylation experiments in U2OS cells. Proteins listed in the HeLa cell contaminants list with an average spectral count ≥ 3 were removed as well as any known contaminants including histones, hemoglobins, albumin, keratins, and other epithelial or secreted proteins which were likely skin contaminants. CRAPome contaminants were not removed from the final list of proximity interactors if there was literature evidence for an interaction with Hic-5 or paxillin, or if they were involved in biological processes which may put them in close proximity with Hic-5 or paxillin (Turner et al., 2010).

Gene ontology enrichment analyses were performed on the proximity interactor lists using g:Profiler (https://biit.cs.ut.ee/gprofiler/gost, RRID:SCR_006809) (Kolberg et al., 2023). g:Profiler options were as follows: organism: Homo sapiens (Human); Ordered query: Yes; and Run as multiquery: No. The lists were analyzed as ordered queries, with genes higher on the list (more enriched), being weighted more heavily in the analysis and the top 10 enriched gene ontologies were graphed.

Protein lists for network analyses were curated from literature evidence (Supplemental Data File 1) as well as from databases including as follows: MiCroKiTS (https://microkit.biocuckoo.org/, RRID:SCR_007052) (Huang et al., 2015) and Human Protein Atlas (https://www.proteinatlas.org/, RRID:SCR_006710) (Thul et al., 2017) for the centrosomal network; and Human Protein Atlas for the nuclear network. Networks were generated using Cytoscape 3.10.1 (Institute for Systems Biology, https://cytoscape.org/, RRID:SCR_003032) (Shannon et al., 2003) and the included STRING plugin (https://apps.cytoscape.org/apps/stringapp)(Doncheva et al., 2019; Szklarczyk et al., 2021). The networks included known interactions between the proximity interactors which were obtained from BioGRID interactions with a minimum evidence score of 2 (RRID:SCR_007393), downloaded through the CRAPome pipeline (Mellacheruvu et al., 2013; Oughtred et al., 2021; Stark et al., 2006), and from the STRING interactions with a confidence interval of 0.9 (Doncheva et al., 2019; Szklarczyk et al., 2021).

Supplementary Material

Tab S1

Supplemental Table 1: The identified proximity interactors of Hic-5 and paxillin.

A list of the proximity interactors of Hic-5 and Paxillin identified through mass spectrometry analysis, as well as alternative names, and the fold change and SAINT probability score for each identified protein (n = 2).

1

Supplemental Figure 1: BioID2-tagged Hic-5 and paxillin localize to the centrosome and biotinylate proteins at the centrosome.

Representative images of U2OS cells stably expressing either myc-BioID2-Hic-5, myc-BioID2-paxillin, or myc-BioID2-control were stained with the centrosomal marker, pericentrin (red), and either myc (A) or fluorescently conjugated streptavidin (SA) (B). The representative images are maximum intensity projections of the three sequential z-planes which contain pericentrin staining, in order to reduce background cytoplasmic signal. Intensity profiles of lines drawn through the centrosome show co-occurrence of pericentrin with both Hic-5 and paxillin. Arrowheads indicate the ends of the line used for the line profiles.

Data Availability Statement

The data that supports the findings of this study are available in the supplementary material of this article.