Skip to main content
NCBI home page
ACS AuthorChoice logoLink to ACS AuthorChoice
. 2024 Sep 6;23(10):4715–4728. doi: 10.1021/acs.jproteome.4c00606

Mapping Extracellular Protein–Protein Interactions Using Extracellular Proximity Labeling (ePL)

David Peeney 1,*, Sadeechya Gurung 1, Joshua A Rich 1, Sasha Coates-Park 1, Yueqin Liu 1, Jack Toor 1, Jane Jones 2, Christopher T Richie 3, Lisa M Jenkins 4, William G Stetler-Stevenson 1
PMCID: PMC11460327  PMID: 39238192

Abstract

graphic file with name pr4c00606_0007.jpg

Proximity labeling (PL) has given researchers the tools to explore protein–protein interactions (PPIs) in living systems; however, most PL studies are performed on intracellular targets. We have adapted the original PL method to investigate PPIs within the extracellular compartment, which we term extracellular PL (ePL). To demonstrate the utility of this modified technique, we investigated the interactome of the matrisome protein TIMP2. TIMPs are a family of multifunctional proteins that were initially defined by their ability to inhibit metalloproteinases, the major mediators of extracellular matrix (ECM) turnover. TIMP2 exhibits broad expression and is often abundant in both normal and diseased tissues. Understanding the functional transformation of matrisome regulators, such as TIMP2, during disease progression is essential for the development of ECM-targeted therapeutics. Using dual orientation fusion proteins of TIMP2 with BioID2/TurboID, we describe the TIMP2 proximal interactome (MassIVE MSV000095637). We also illustrate how the TIMP2 interactome changes in the presence of different stimuli, in different cell types, in unique culture conditions (2D vs 3D), and with different reaction kinetics, demonstrating the power of this technique versus classical PPI methods. We propose that screening of matrisome targets in disease models using ePL will reveal new therapeutic targets for further comprehensive studies.

Keywords: matrisome, proximity labeling, interactomics

Highlights

  • Proximity labeling (PL) can be optimized to study the interactome of secreted factors.

  • Extracellular PL is used to reveal an expanded interactome for TIMP2.

  • Candidate proximal interactors are validated by proximity ligation assay.

  • Cost-effective approach serves as a template for probing the interactomes matrisome proteins.

Introduction

The extracellular matrix (ECM) is a complex framework that supports all cellular and tissue processes and is made up of over 1000 proteins termed the matrisome. It is well established through the principles of dynamic reciprocity, which define the bidirectional interaction between cells and the extracellular space, that the ECM directly modulates cell behavior.1 These principles are mediated, in large part, through protein–protein interactions, linking the extracellular and intracellular compartments. Virtually all mammalian disease states have roots entrenched within the ECM, manifesting as abnormal abundance or dysfunctional activity of matrisomal proteins.2 Despite widespread appreciation of the significant alterations in the disease-associated matrisome, therapeutic targeting of the ECM has shown limited success across the spectrum of human diseases.

Tissue inhibitors of metalloproteinases (TIMPs) are a family of endogenous matrisome proteins that were initially defined by their ability to inhibit the enzymatic activity of metalloproteinases (MPs), the major mediators of ECM breakdown and turnover.3 In turn, this function of TIMPs makes them important regulators of the ECM structure. Since their original discovery, various additional MP-independent functions have been attributed to TIMP family members leading to their designation as multifunctional proteins with discrete functional domains.4 This can be exemplified by TIMP2, the most abundant member of the family that possesses a range of pro- and antimitogenic functions that are independent of its MP inhibitory activity.5 Due to their multifunctional capabilities, the therapeutic potential of TIMP proteins across various conditions has been proposed in recent years.6 We and others have illustrated how TIMP2 exhibits antitumor capabilities,7−10 and others have described how TIMP2 may protect against age-related neuronal dysfunction.11−13 Biochemically, TIMP2 is a promiscuous protein, binding to and inhibiting all 23 members of the matrix metalloproteinase (MMP) family as well as ADAM1214 and ADAMTS815 and targeting several membrane proteins.5 Despite the therapeutically promising capabilities of TIMP2, many of the detailed mechanisms reinforcing these functions are missing. Exploration of the TIMP2 interactome using affinity purification methods has yielded incomplete results, the limitations of which are well described.16

A powerful method for identifying PPIs is proximity labeling (PL), which utilizes target fusions with promiscuous biotin ligases such as BioID2/TurboID.17 These fusions preferentially biotinylate interacting partners that can be identified using basic enrichment steps, followed by mass spectrometry. Most studies that utilize biotin ligases target intracellular proteins. Using broadly expressed matrisome protein TIMP2, we illustrate that conditions can be optimized for proximity labeling reactions in the extracellular milieu. Using TIMP2 fusions with BioID2 and TurboID, we identify novel proximal interactors for TIMP2 across two different cell types and in a range of culture conditions, providing a new context regarding TIMP2 functions in health and disease. Furthermore, our workflow serves as a template for proximal interactome studies of other matrisome targets, representing an important technical resource for extracellular matrix biologists.

Materials and Methods

Construct Design and Retroviral Packaging

All constructs were produced in a pBABE retroviral expression (Addgene plasmid no. 80899, a gift from Kyle Roux) backbone using ligation-independent cloning (In-Fusion, Takara), and transformed into a recombination-deficient bacterial strain (NEB Stable competent cells, New England Biolabs). Insert containing clones were confirmed by restriction digest analysis and DNA sequencing. Large scale plasmid preps were performed with NucleoBond Xtra Maxi Plus purification kits (Macherey-Nagel). Detailed construct production and retroviral packaging details can be found in the Supporting Information.

Generation of Biotin-Depleted Full Media (BDFM) and Low-Serum Biotin-Depleted Media (LSBDM)

Biotin-depleted full media (BDFM) was generated by washing 2.5 mL high-capacity streptavidin agarose (ThermoFisher #) four times with 50 mL of phosphate buffered saline, centrifuging the beads at 1000G for 2 min between washes. After the final wash, all the beads were transferred to 1 L of DMEM, 10% FBS, 1% GlutaMAX, and 1% penicillin–streptomycin stored in a roller bottle. The media was then rolled at 4 °C for 16–20 h prior to filtration through a 0.22um PES filter unit. BDFM was used within 1 month of generation. Low-serum biotin-depleted media was generated by diluting BDFM 1 in 10 with Advanced DMEM (Gibco).

Extracellular Proximity Labeling Assay

Fusion protein expressing or control cells were seeded in 15 cm dishes (1.5 × 106 HT1080, 7 × 106 HS-5) in BDFM (media transfer cells; MTCs). At the same time, WT cells were seeded in 15 cm plates in BDFM at the same density (receiver cells; RCs). If performing a spheroid-based ePL experiment, WT HT1080 were seeded at 10,000 cells per well of a 96-well ultralow binding round bottomed spheroid microplate 3 days prior to seeding fusion protein-expressing or control cells (the latter grown in 2D). At 72 h post seeding, MTCs were switched to 23 mL of LSBDM, while 10 mL of media on the RCs was replaced with fresh BDFM. TurboID experiments: On day 4, 24 h after the previous step, the media from the RCs was discarded and replaced by the media on the MTCs. The remaining culture was supplemented with 1 μM biotin and 0.1 mM ATP and incubated for 1 h. BioID2 experiments: On day 4, 24–30 h after the previous step, the media from the RCs was discarded and replaced by the media on the MTCs. The remaining culture was supplemented with 1 μM biotin and 2 mM ATP and incubated for 16 h. Serum experiments: Human blood was collected through the NIH Blood Bank, clotted for 1h at room temperature, centrifuged at 2000g for 10 min at 4 °C, and frozen. In a 1 mL reaction, serum was diluted to 50% with Dulbecco’s phosphate buffered saline and supplemented with 300nM recombinant TurboID fusion protein, 1 μM biotin, 0.1 mM ATP, and incubated at 37 °C for 1h. Reactions were stopped through desalting using PD-10 columns (Cytiva) and immediately incubated with streptavidin using the protocol described below.

Lysate and Media Collection

Conditioned media was removed from the cells, supplemented with 0.1% protease inhibitor cocktail, and centrifuged at 1000g for 5 min to pellet cellular debris. The conditioned media was then concentrated to less than 1.5 mL using Amicon Ultra-15 centrifuge tubes (MilliporeSigma), prior to buffer exchange of the concentrated media into Dulbecco’s phosphate buffered saline (DPBS) using PD-10 columns (Cytiva), resulting in 3.5 mL of desalted media. Desalted media was concentrated again using Amicon Ultra-4 centrifuge tubes to a volume of 1 mL. Cell lysate collection was performed at the same time as the above steps. Cells were placed on ice and washed 3× with ice-cold 12 mL DBPS, followed by addition of 1250 μL lysis buffer (50 mM HEPES, 50 mM NaCl, 2% SDS, 1% Triton X-100, 1% NP-40, 1% sodium deoxycholate, 10 mM DTT, 1% protease inhibitor cocktail). The cells and matrix were scraped from the plate, then transferred to a 2 mL centrifuge tube and mixed by vortexing (10 × 2 s). Lysate samples were then heated to 95 °C for 10 min while shaking at 800 RPM and then cooled on ice for 3 min. Cooled samples were sonicated with 24 × 5 s bursts at 70% power and then centrifuged at 10,000g for 10 min. The supernatant was removed and combined with the desalted-concentrated conditioned media before being snap-frozen in liquid nitrogen. The final snap freezing step is highly beneficial, producing a less viscous solution that is more amenable to pulldown.

Pulldown and Sample Preparation for Mass Spectrometry

Snap frozen sample were thawed in a 37 °C water bath, then 60 μL trypsin-resistant streptavidin magnetic beads were added to the whole sample. The samples were rotated at 4 °C for 2 h. After incubation, beads were allowed to bind to the magnet for 5 min, followed by supernatant (output) removal and two washes with 1 mL wash 1 (2% SDS). This was followed by two individual 1 mL washes with wash 2 (0.1% sodium deoxycholate, 1% Triton X-100, 500 mM NaCl, 1 mM EDTA, 50 mM HEPES) and wash 3 (250 mM LiCl, 0.5% NP-40, 0.5% sodium deoxycholate, 1 mM EDTA, 10 mM Tris pH 8), followed by two washes with 1 mL wash 4 (20 mM HEPES, 150 mM NaCl). After supernatant removal, a single wash with 500 μL 100 mM triethylammonium bicarbonate (TEAB) was performed and sample transferred to a fresh lo-bind 1.5 mL tube (Eppendorf), at which point the beads were resuspended in 50 μL 100 mM TEAB, 5 mM DTT and incubated at 56 °C for 25 min to reduce the samples. Samples were then supplemented with 1.55 μL of 0.5 M iodoacetamide (15 mM working concentration) and incubated at room temperature in the dark for 30 min. The supernatant was removed, and samples were washed three times with 200 μL 100 mM TEAB, then finally resuspended in 30 μL 100 mM TEAB + 0.125 μg Trypsin-LysC mixture (Promega) and incubated for 16–18 h at 37 °C with shaking at 500 RPM. Samples were centrifuged at 500g for 20 s, then placed on the magnet, and the supernatant was transferred to a fresh 1.5 mL Lo-bind tube. For unlabeled processing: the remaining beads were washed with 30 μL of 100 mM TEAB, and then supernatants were combined. Samples were acidified with 3 μL 10% trifluoroacetic acid (TFA) and snap frozen in liquid nitrogen. For TMT-labeling: The remaining beads were washed with 15 μL of 100 mM TEAB, then supernatants combined. Peptide concentration was determined using A205, and then sample concentrations were normalized. Samples were then supplemented with 3.64 μg/μL TMT reagents and incubated in the dark for 1 h at room temperature. Each sample was supplemented with 1 μL of 15% hydroxylamine and then 1.7 μL of 10% trifluoroacetic acid. Samples were then combined in a single tube. The individual sample was then cleaned and concentrated using 100 μL C18 pipet tips (ThermoFisher) and eluted in 0.1% formic acid, 95% acetonitrile prior to snap-freezing in liquid nitrogen.

Mass Spectrometry

Before analysis, samples were dried using a SpeedVac system (Thermo Scientific) and resuspended in 16 μL of 0.1% formic acid, with 12 μL being analyzed by liquid chromatography and tandem mass spectrometry (LC-MS/MS). The LC-MS/MS analysis of tryptic peptides for each sample was performed sequentially with a blank run between each two sample runs using a Thermo Scientific Orbitrap Exploris 240 Mass Spectrometer and a Thermo Dionex UltiMate 3000 RSLCnano System. Peptides from trypsin digestion were loaded onto a peptide trap cartridge at a flow rate of 5 μL/min. The trapped peptides were eluted onto a reversed-phase Easy-Spray Column PepMap RSLC, C18, 2 μM, 100A, 75 μm × 250 mm (Thermo Scientific) using a linear gradient of acetonitrile (3–36%) in 0.1% formic acid. The elution duration was 110 min at a flow rate of 0.3 μL/min. Eluted peptides from the Easy-Spray column were ionized and sprayed into the mass spectrometer, using a Nano Easy- Spray Ion Source (Thermo Scientific) under the following settings: spray voltage, 1.6 kV; Capillary temperature, 275 °C. Other settings were empirically determined. For unlabeled samples: Raw data files were searched against human protein sequences using Proteome Discoverer 3.0 software (Thermo Scientific) based on the SEQUEST algorithm. Carbamidomethylation (+57.021 Da; cysteine) was set as the static modifications. Oxidation/hydroxylation + 15.995 Da (methionine, proline, lysine) and deamidation + 0.984 Da (asparagine, glutamine) were set as dynamic modifications. The minimum peptide length was specified as five amino acids. The precursor mass tolerance was set to 15 ppm, whereas the fragment mass tolerance was set to 0.05 Da. The maximum false peptide discovery rate was specified as 0.01. For TMT-labeled samples: The Exploris 240 instrument was operated in the data dependent mode to automatically switch between full scan MS and MS/MS acquisition. Survey full scan MS spectra (m/z 350–1800) was acquired in the Orbitrap with 35,000 resolutions (m/z 200) after an accumulation of ions to a 3 × 106 target value based on predictive automatic gain control (AGC). The maximum injection time was set to 100 ms. The 20 most intense multiply charged ions (z ≥ 2) were sequentially isolated and fragmented in the octopole collision cell by higher-energy collisional dissociation (HCD) using normalized HCD collision energy 30 with an AGC target 1 × 105 and a maxima injection time of 400 ms at 17,500 resolutions. The isolation window was set to 2 and fixed first mass was 120 m/z. The dynamic exclusion was set to 30 s. Charge state screening was enabled to reject unassigned and 1+, 7+, and >7+ ions. Raw data files were searched against human protein sequences using the Proteome Discoverer 3.0 software (Thermo Scientific) based on the SEQUEST and percolator algorithms. Carbamidomethylation (+57.021 Da; cysteine) and TMT labeling (+229.163 Da; N-terminus, lysine) was set as static modifications. Oxidation/hydroxylation + 15.995 Da (methionine, proline, lysine), and deamidation + 0.984 Da (asparagine, glutamine) were set as dynamic modifications. The minimum peptide length was specified to be five amino acids. The precursor mass tolerance was set to 15 ppm, whereas fragment mass tolerance was set to 0.05 Da. The maximum false peptide discovery rate was specified as 0.01. The resulting Proteome Discoverer Report contains all assembled proteins with peptides sequences, peptide spectrum match counts (PSM), and TMT-tag based quantification using reporter ion abundance. Raw and processed data is deposited in massive.ucsd.edu, repository number MSV000095637 (FTP download link; ftp://massive.ucsd.edu/v08/MSV000095637/).

Proximal Interactor Candidate Selection

For unlabeled experiments: We performed candidate protein selection using a series of basic calculations comparing peptide spectrum matches (PSMs) and calculated protein abundance between fusion gene and control samples (summarized in Supporting Information; calculations maintained in Supplementary Tables). For TMT-labeled experiments: total abundance of each sample is normalized (1 control, 2 fusion protein samples), prior to fold change comparison between fusion protein samples versus the control sample. Proteins with an abundance ≥ 1.5 times the control are identified as candidate proteins.

Final protein candidates are identified for filtering against the Contaminant Repository for Affinity Purification (CRAPome), with any protein candidate that is detected in over 99 (out of 716) control experiments within the CRAPome database automatically excluded.

Proximity Ligation Assay (PLA)

2 × 105 HT1080 cells were seeded in to 3-well removable chamber slides (Ibidi) and cultured overnight. Wells to be used for MMP14-related PLA assays were treated with 40 nM PMA for 16 h. Next day, all cells were treated with 100ng/mL recombinant TIMP2 (to mimic the overexpression of fusion proteins) for 1h. Media was removed from the cells prior to fixation in 4% PFA without washing. Fixed cells were permeabilized with PBS 0.1% Triton X-100, then washed 1× with PBS. The subsequent protocol was followed per manufacturer’s instructions (Duolink In Situ Red Starter Kit Goat/Rabbit, Millipore Sigma). A detailed breakdown of the full protocol is described in the Supporting Information. 40× oil images were collected by using a Zeiss Axio Observer 5 multimodal imaging microscope.

Results

Construct Design for Extracellular Proximity Labeling (ePL)

BirA is a bifunctional protein that acts as a biotin ligase and transcriptional repressor, which is well conserved in single-celled organisms.19 TurboID was developed based on directed evolution of the E. coli BirA*.20 BioID2 was developed using BirA from A. aeolicus, which is substantially smaller than E. coli BirA at 27 kDa (versus 35 kDa).21 This reduced size is largely the consequence of a truncated N-terminal DNA-binding domain that does not significantly affect the biotin ligase activity. The crystal structure of the parental forms of TurboID (PDB #1BIA22) and BioID2 (PDB #2EAY), in comparison to that of TIMP2 (AlphaFold prediction23), suggested to us that generating contrasting orientations of TIMP2:biotin ligase fusion proteins would reveal a more complete proximal interactome (Figure 1A). This is particularly important when considering that a free N-terminus of TIMP2 is required for MP inhibition and a free C-terminus of TIMP2 is essential for proMMP2 hemopexin domain binding, highlighted in Figure 1A. Constructs encoding fusion proteins of both orientations, separated by a flexible 13× GGGGS-linker (GS-linker) sequence to increase reactive radius, were produced containing an N-terminal appended secretory signal sequence to promote secretion into the extracellular compartment (Figure 1B). Retroviral-mediated fusion protein expression was analyzed by immunoblotting, confirming the successful secretion and comparable production of fusion proteins (Figure 1C). The fusions retained important functional characteristics, including the ability to inhibit MMP2 activity in the fusion orientation with a free TIMP2 N-terminus (Figure 1D). Streptavidin horseradish peroxidase staining of pull-down samples (input, output, and elution) reveals a unique pattern of biotinylated proteins detected in TIMP2-TurboID fusion protein samples, indicative of a functional proximity labeling protocol (Figure 1E). To perform a successful ePL assay, exogenous biotin and ATP are required (Figure S1).

Figure 1.

Figure 1

Open in a new tab

Expression, secretion, and function of biotin ligase fusion proteins. (A) Structure of E. coli BirA (parental TurboID protein), A. aeolicus BirA (parental BioID2 protein), and the AlphaFold predicted structure of TIMP2. (B) Basic construct design and proposed structure of one fusion protein. (C) Immunoblots assessing the expression and secretion of fusion proteins in HT1080 cells. (D) Reverse zymography (10% acrylamide) of conditioned media from fusion protein expressing cells revealing retention of MP-inhibitory activity in the fusion orientation that maintains a free TIMP2 N-terminus, which is required for MP inhibition. Endogenous/wild-type TIMP2 (22 kDa) is indicated by the green arrow. (E) Streptavidin pulldown and blotting of SDS-PAGE nitrocellulose blots with streptavidin-HRP reveals unique patterns of biotinylated proteins in fusion protein expressing samples.

ePL Reveals the TIMP2 Proximal Interactome in HT1080 Cells

HT1080 cells are an epithelial-like fibrosarcoma cell line that has been extensively utilized by researchers studying the biology of TIMPs and MPs.24−26 Considering the strong correlation between TIMP2 expression and mesenchymal cells,27 they represent an ideal cell line for preliminary ePL studies. A total of 4 experiments were performed, representing duplicates from TurboID and BioID2 variations, fully summarized in Table S1. For TurboID experiments we utilized a 1 h labeling protocol, whereas for BioID2 labeling was performed over 16 h (Figure 2A). Surprisingly, despite its higher labeling activity we found that TurboID produced a more limited proximal interactome with a smaller contribution of contaminants defined through contaminant repository for affinity purification (CRAPome) analysis (Figure 2B).28 A core of 4 proteins (CCN1, CCN2, THBS1, MMP2) were identified as proximal interactors in each experiment with at least duplicate identifications (Figure 2C). Three of these are potentially novel interactors for TIMP2, with the TIMP2-MMP2 interaction being well characterized.29 Across all experiments, at least 19 candidate interactors were identified through duplicate identification, some displaying orientation specificity, and others biotin ligase specificity. The orientation specificity of the proximal interactome can be appreciated through the TIMP2:proMMP2 interaction that requires a free C-terminal tail of TIMP2, the crystal structure of which is defined in Figure 2D.30 Initial observations suggest that both orientations display proximal interactions with MMP2. This is to be expected, considering that TIMP2 can bind to and inhibit active MMP2 via its opposing (N-)terminus. However, in unstimulated conditions MMP2 resides mostly in its pro-form, and the detected MMP2 peptides from the TIMP2-BL data exhibit poor sequence coverage that do not correspond to peptides of the prodomain of MMP2 (0, 6, 7, and 5% sequence coverage versus ≥44% for BL-TIMP2 samples) (Table S1, tab 24). Quantification of the total abundance between the opposing orientations clearly demonstrates that the TIMP2:MMP2 proximal interaction is favored by a free C-terminus of TIMP2 (Figure 2E). Unsurprisingly, most of the candidate proximal interactors are matrisome proteins including F13A1 (Coagulation Factor XIII A Chain). RNA sequencing data confirms that HT1080 cells do not express F13A1 transcripts despite its significance in TurboID experiments with both orientations (data available through GEO GSE252575). Assessment of the peptides detected from human F13A1 and alignment with analogous regions in bovine F13A1 reveals the peptides correspond with regions of identical m/z ratios (with single residue differences represented by a switch between leucine and isoleucine) (Table S1, tab 25). The observation that bovine serum contaminants are identified as proximal interactors presents as both a technical limitation and a source of intrigue, especially considering that TIMP2 is detectable in the human plasma at a concentration of approximately 110ng/mL according to the Human Protein Atlas.31 Analysis of the raw spectral files searched against the Bos taurus protein database reveals four further proximal interactors for the TurboID:TIMP2 orientation; SERPINA5, CFH, C3, and IGFBP3 (Table S1, tab 26).

Figure 2.

Figure 2

Open in a new tab

Extracellular proximity (ePL) labeling to identify the proximal interactome of TIMP2 in HT1080 cells. (A) Schematic describing the basic workflow of ePL reactions in HT1080 cells. (B) Data acquired through LC-MS/MS was scored and proximal interactors identified using a defined scoring system. (C) Proximal interactors were uploaded into the Contaminant Repository for Affinity Purification-mass spectrometry (CRAPome) and persistent contaminants were removed based on a defined threshold (identified in >99 control experiments). Duplicate experiment identified proximal interactors are illustrated in the quadrant based on the biotin ligase utilized and the fusion protein orientation. (D) Crystal structure of the TIMP2:proMMP2 complex (Protein Data Bank #1GXD) reveals a strong C-terminal interaction (metalloproteinase inhibition independent) that is (E) significantly perturbed in the presence of a C-terminal fused biotin ligase. (A+C) Created with BioRender.com.

Proximal Interactomes Are Dynamic

An important consideration when probing proximal interactomes is that protein–protein interactions are dynamic phenomena, the occurrence of which may have a few or many dependencies. These dependencies may include the expression of cofactors or induction of unique post-translational modifications. Phorbol 12-myristate 13-acetate (PMA) is a plant-derived (Croton tiglium) phorbol ester that mimics diacylglycerol, leading to a constitutive activation of protein kinase C.32 The ensuing pathway dysregulation supports an acute inflammatory response that is a hallmark of PMA’s tumor-promoting capabilities.33 Concanavalin A (ConA) is a carbohydrate binding protein (lectin) isolated from Jack bean (Canavalia ensiformis) that binds to α-mannopyranosyl and α-glucopyranosyl residues of glycoproteins and glycolipids. Interactions between lectins and glycoproteins at the plasma membrane are involved in cellular-ECM interactions that promote ECM remodeling through matrisome regulators such as MMP14 (also known as MT1-MMP), thus ConA is commonly used to stimulate the expression and activity of MMP14.34 In the same manner, the inflammatory response induced by PMA also induces the activity of MMP14,35 which can be supported by gelatin zymography analysis of HT1080 cells treated with ConA and PMA revealing enhanced MMP2 activation (Figure S2), an occurrence that strongly correlates with MMP14 activity.36 The interaction between MMP14 and TIMP2 is well-studied and has been shown to mediate the activation of MMP2 via formation of a trimolecular complex.36,37 RNA sequencing corroborates the well-established proinflammatory effects of PMA and ConA, which interestingly share common pathways and key upstream regulators despite inducing unique transcriptomic signatures (Figure 3A/3B). Analysis of chemically defined TIMP2 interactors alongside previously identified proximal interactors reveals significant transcript-level changes that could potentially rewire the TIMP2 proximal interactome (Figure 3C). Notably, these changes include a marked shift in the expression levels of multiple matrix metalloproteinases. Using TurboID as the biotin ligase, a series of ePL experiments were performed to determine the proximal interactome of TIMP2 in HT1080 cells treated with 40nM PMA or 40 μg/mL ConA, summarized in Table S2. Analysis of the unfiltered proximal interactors reveals a distinct elevation in PMA-induced TIMP2 proximal interactors that are common contaminants, as listed by the CRAPome database (Figure 3D). The contribution of CRAPome contaminants in ePL experiments treated with ConA were less pronounced. Examination of the final filtered proximal interactor candidates reveals that although several of the original interacting partners remain, others are lost or gained (summarized in Figure 3E). Of note is the identification of MMP1 as an interactor for the TIMP2-TurboID orientation in the PMA treatment only, despite MMP1 being upregulated in both cell treatments (Figure 3C/3E). Furthermore, the TIMP2:MMP14 interaction is mediated through the N-terminus of TIMP2 (Figure 3F). This interaction is predominantly detected by fusion with a free TIMP2 N-terminus. Although MMP14 is detected in all TurboID experiments as an interactor for TIMP2-TurboID, assessment of the normalized abundance reveals that the detection of MMP14 as a proximal interactor increases following treatment with ConA and, to a larger extent, PMA (Figure 3G). This observation is consistent with zymography, which illustrates that PMA is a more effective activator of MMP2 than ConA in HT1080 cells, an established functional consequence of MMP14 activity that was described earlier (Figure S2). Low levels of MMP14 are occasionally detected with the TurboID-TIMP2 orientation (free TIMP2 C-terminus). This is likely enabled by a limited retention of affinity through peripheral interactions between other regions, such as the long A-B loop of TIMP2 that exhibits extensive contact with MMP14 during binding.38 Interestingly, the same interaction is not detected by the TIMP2-BioID2 fusion (Figure 2C), revealing an important consideration with regard to reaction kinetics and the identification of proximal interactors. The TIMP2:MMP14 interaction has been reported to result in rapid internalization and, to an undefined extent, degradation of the complex,39,40 possibly explaining the inability of the slower labeling BioID2 fusions to successfully biotinylate MMP14.

Figure 3.

Figure 3

Open in a new tab

Cellular treatments can reveal proximal interactome dynamics. (A) Ingenuity Pathway Analysis canonical pathways and upstream regulators of transcriptome data from HT1080 cells treated with 40nM PMA or 40 μg/mL ConA for 24 h. (B) Venn diagram comparing the transcriptome changes in PMA versus ConA treated HT1080 cells. (C) Transcript abundance changes in defined TIMP2 interactors or previously identified proximal interactors from PMA and ConA treated HT1080 cells. (D) Quantified ePL data using TurboID fusions with TIMP2 acquired through LC-MS/MS that was processed, scored, and proximal interactors identified using a defined system. (E) Tables illustrating the identified TIMP2 proximal interactors (TurboID) in HT1080 cells treated with PMA or ConA. Gained or lost proximal interactors (versus unstimulated cells in analogous TurboID experiments) are highlighted in magenta (lost) or cyan (gained), created with BioRender.com. (F) Predicted structure of TIMP2 interacting with full length active MMP14, created by superimposing the AlphaFold MMP14 structural prediction over the TIMP2-MMP14 (catalytic domain) crystal structure (Protein Data Bank # 1BQQ). The structural domains of MMP14 are color-coded, with disordered linker regions colored gray. (G) Comparison of the normalized abundance of MMP14 across HT1080 TurboID ePL assays that were untreated or treated with ConA or PMA.

Cell culture models propagated in 2-dimensions are widely used because they are simple and economical. However, this simplicity comes with many limitations, the most important being a loss of physiological relevance. Cell cultures grown in 3-dimensions help to address this limitation by supporting extensive cell–cell and cell-ECM interactions and facilitating cell and tissue polarity.41 Spheroid models are the simplest of the 3D culture methods, with scaffold-free cell clumping induced through gravity (hanging drop method) or low-binding plastic. HT1080 cells produce robust spheroids that form within 6 days of culture, so we used this system as proof-of-principle that ePL is effective in 3D, summarized in Figure 4A. We utilized the BioID2 fusion proteins for these experiments under the assumption that the fusion proteins will require over 1 h to penetrate the spheroid fully. Furthermore, the extended processing time when harvesting 3D cultures are ill-suited to the rapid labeling that occurs in TurboID experiments. Results from the single duplicate experiment (described as n = 4 when considering both fusion protein orientations) reveals there is fair agreement within the predominant proximal interactors between 2D and 3D experiments. A total of 17 proximal interactors were identified in 3D experiments, with 4 proteins being identified in 3+ samples (MMP2, CCN1, LTBP1, LTBP3). Of the 17 proximal interactors, 8 were identified specifically in 3D ePL experiments (LTBP3, CHIA, RAC2, C3, EFEMP1, SETDB2, TF, and STC1; Figure 4B,C, summarized in Table S3).

Figure 4.

Figure 4

Open in a new tab

ePL can be performed across multiple live in vitro systems. (A) Basic workflow of HT1080 3D spheroid ePL experiments, created with BioRender.com. (B) Key illustrating the color-coded nodes in subsequent figure panels. (C) Summary of TIMP2 proximal interactors (BioID2) identified in 3D spheroid cultures, compared with previous BioID2 ePL experiments. (D) Overall summary of the TIMP2 proximal interactome in HT1080 cells, a total of 5 experiments each performed in duplicate with both fusion orientations using both TurboID and BioID2. (E) STRING analysis of the HT1080 proximal interactome for TIMP2 reveals potential interactome neighborhoods using previously reported direct protein–protein interactions and by incorporating additional potential interactions through STRING’s text mining feature. (F) TIMP2 TurboID ePL experiments in HS-5 bone marrow stromal cells reveals duplicate identified novel proximal interactors and shared interactors compared with the TIMP2 proximal interactome in HT1080 cells.

Summarizing all of the experiments from HT1080 cells, we can visualize the predominant proximal interactors for TIMP2 (Figure 4D, Table S4). Unsurprisingly, members of the matrisome represent the major interacting partners for TIMP2. Of the proximal interactors with a total of 5 hits or more, only 2 are known interactors (MMP2 and MMP14), both of which display fusion orientation predilection due to their well-characterized interaction with the C-terminus (latent MMP2) and N-terminus (active MMP14, active MMP2) of TIMP2. Upload of the candidate proximal interactome for TIMP2 into STRING (string-db.org) reveals an interaction network that may depict local protein complexes or neighborhoods within HT1080 cell cultures (Figure 4Ei). Elaborating on this analysis, the network can be expanded to include a text mining feature that indicates an interaction between proteins that are mentioned together within published scientific abstracts (Figure 4Eii).

It is recognized that protein interactions which occur in core complexes, such as those involved in essential functions, tend to be retained between cell types.42 In contrast, broader protein interactomes are dynamic and can be rewired between cell types in a manner that ultimately dictates the unique phenotypes of each cell type.42 In support of the latter point, we show that the proximal interactome for TIMP2 is rewired in HS-5 bone marrow stromal cells (Figure 4F). Candidate proximal interactors for TIMP2 in HS-5 cells are identified through at least duplicate identification across replicate experiment. A total of 29 candidate proximal interactors are identified in HS-5 cells, 6 of these shared with the compiled HT1080 proximal interactome (results summarized in Table S5). Since some of the protein candidates represent bovine serum proteins and TIMP2 is a well-recognized component of human serum, we performed a duplicate TurboID proximity labeling experiment utilizing recombinant fusion proteins and human serum. These studies revealed that TIMP2 displays a unique serum proximal interactome that, from one donor, includes F13A1 as a proximal interactor (Table S6). Other noteworthy candidates identified by these experiments include Secreted Frizzled Related Protein 1 (SFRP1), the proteinase inhibitor Pregnancy-Zone Protein (PZP), Calpain 2 (CAPN2), and Cathepsin G (CTSG).

CCN1 (CYR61) and CCN2 (CTGF) represent two of the persistent novel TIMP2 proximal interactors revealed in our study, identified in 13 and 17 samples respectively. Co-immunoprecipitation could not successfully detect either of these targets as direct TIMP2 interactors, an observation that may be influenced by the pitfalls of the technique that requires robust interactions that survive all stages of the protocol. To substantiate our ePL findings, we performed a proximity ligation assay (PLA) in HT1080 cells with TIMP2 versus CCN1, CCN2, MMP14 (positive control using PMA treated HT1080 cells), and LOXL2. Despite the similarity in name, PLA is methodologically distinct from PL and is a valuable tool for probing endogenous protein–protein proximity (within 40 nm).43 The inclusion of LOXL2 as a PLA assay target serves as a useful gauge for validating less robust candidate proximal interactors, with LOXL2 being identified in 7 samples from HT1080 cells. PLA corroborates the ePL findings, illustrating that MMP14, CCN1, and CCN2 are statistically significant proximal interactors for TIMP2 (Figure 5A–C). In each case, PLA foci occur predominantly within the perinuclear region of cells. Indeed, each of these targets have previously been identified in the perinuclear regions of cells, indicating that these targets may share secretory or endocytic pathways.44−48 Despite trending toward a proximal interaction through persistent above background signal, LOXL2 could not be convincingly validated using this method (data indicates p = 0.057 and amplitude of the signal was low). An example analysis pipeline for PLA results is outlined in Figure S3.

Figure 5.

Figure 5

Open in a new tab

Proximity ligation assay (PLA) corroborates the identification of CCN1/CCN2 as candidate proximal interactors for TIMP2. (A) Normalized quantification, (B) example control images, and (C) PLA images indicating a high abundance of PLA foci between TIMP2 coimmunostained with MMP14, CCN1, and CCN2. Co-staining between TIMP2 and LOXL2 required a high exposure to observe a surplus of PLA foci versus controls (p = 0.057). Arrows indicate regions of interest (not quantification). * = p < 0.05, Two-tailed Mann–Whitney test. Scale bar = 10 μm.

Discussion

Despite great advances in identifying and categorizing the matrisome, a considerable number of its constituents lack thorough functional characterization. These challenges are rooted in the biochemical features of matrisome proteins, in particular, members of the core matrisome, that are generally large, insoluble, and structurally linked proteins. Furthermore, members of the matrisome are predisposed to unique post-translational modifications that are uncommon in the broader proteome.49 The diversity in proteoform contributes to the overwhelming complexity of the matrisome, despite it comprising <10% of the total proteome.50 Complimentary to characterization of the matrisome in models of health and disease, knowledge of protein–protein interactions within the extracellular compartment provides vital context with regard to the biological functions of individual components. Due to the complex biochemical properties of matrisome proteins, interrogation of extracellular PPIs is best implemented in living systems.49 To this end, proximity-labeling-based methods are powerful techniques that offer great promise. Historically, these have been used to assess intracellular or plasma membrane proximal interactions, using methods that rely on BirA (BioID), APEX, or biotinylation by antibody recognition (BAR).17,51,52 The latter technique relies on the generation of truly effective antibodies, representing an additional level of technical complexity. Indeed, the overall performance and consistency of many established antibodies is questioned in a recent study.53 There have been very limited reports describing proximity labeling in the extracellular space, with a recent example utilizing an antibody-biotin ligase conjugate to investigate the EGFR proximal interactome.54 Using TIMP2 as a model, we provide evidence of the power of ePL in interrogating the proximal interactome of matrisome proteins. TIMP2 is an intriguing target for proximal interactome studies due to its well-defined functions with regards to metalloproteinase interactions, in addition to a slew of mechanistically undefined functions, reviewed extensively elsewhere.5 We identified a core of highly reproducible proximal interactors that were detected in ≥50% of the samples from HT1080 cells: CCN1, CCN2, MMP2, THBS1, and F13A1. All these proximal interactors are members of the matrisome (3 core matrisome, 2 matrisome-associated), with only 1 (MMP2) being a known interactor. To substantiate some of these findings, we performed a series of proximity ligation assay (PLA) experiments to show that TIMP2 exhibits proximal interactions with two novel candidate interactors, CCN1 (CYR61) and CCN2 (CTGF), in HT1080 cells. A third candidate interactor, LOXL2, could not be reliably corroborated using PLA. In each case, PLA foci generally localized within the intracellular compartment, specifically in perinuclear localizations. Components of the matrisome are frequently observed to localize in perinuclear regions in immunostaining protocols, possibly linked to migration, invasion, and, specifically, the formation of podosomes and invadopodia—intracellular sites for ECM attachment and degradation.55−57 Whether the perinuclear colocalization of these proximal interactions is a consequence of migratory phenotypes in cells is a question of particular interest.

That ePL proved to be an effective method for assessing the TIMP2 interactome, despite competing with high levels of endogenous TIMP2, is a testament to its power. For TIMP2, the proximal interactome is not extensively rewired across treatments (untreated, PMA, and ConA) and culture conditions (2D and 3D culture conditions). However, across cell types, the proximal interactome is largely unique, despite the cell types sharing a mesodermal lineage. HT1080 is an epithelial-like fibrosarcoma cell line, whereas HS-5 is a genetically transformed bone marrow stromal cell line.24,58 The stark differences between the TIMP2 proximal interactome in these cells serve as a testament to the idea of disease-specific PPIs that may convey functional knowledge and/or clinical diagnostic power. This illustrates the importance of choosing the most relevant models when probing proximal interactomes, a feat that will be aided through query of available transcriptomic and proteomic data sets. Recent work has highlighted the importance of selecting an appropriate model system, revealing that most tissues and cell lines express only a small fraction of the matrisome, with a significant portion remaining poorly expressed, or “dark”.59

Routinely, the proteins detected in highest abundance are the carboxylases, such as PCCA, PC, MCCC1, ACACA. These enzymes are biotin-dependent, containing a covalently bound biotin through amide linkage between a lysine and the carboxylate moiety of biotin.60 Another persistent background artifact in our analyses was the incidence of Heat Shock Protein family members detected in control samples, in particular HSP70 and HSP90 subfamily members. Negative control samples cannot reasonably detect a complete set of experimental contaminants, in large part due to inevitable variations in sample handling across the many steps prior to MS analysis. To account for this, candidate proteins were screened and filtered based on prevalence in negative control samples deposited within the CRAPome repository.28 Consequentially, HSP family members that presented as candidate proteins were ultimately excluded, including the repeat identified (n = 7) HSPA5 (HSP70 family) and (n = 5) CCT8 (Chaperonin family) (Figure 6). Furthermore, multiple instances of HSP90 proteins (HSP90AB1, HSP90B1, HSP90AA1) were detected as candidate proteins prior to CRAPome filtering, consistent with reports describing HSP90 as a biochemically defined TIMP2 interacting partner.61 The need to examine each level of the data empirically and thoroughly is further demonstrated through the likely detection of nonhuman proteins as proximal interactors, exemplified by F13A1. This matrisome-associated protein classically functions in the realms of blood coagulation, although it likely plays more diverse roles in the tissue microenvironment.62 Assessment of the TIMP2 proximal interactome in human serum samples reveals a unique serum proximal interactome, through which F13A1 is identified among other unique proximal interactors. Based on the mechanistically undefined functions of TIMP2, and reports linking its activity with antitumor and antiaging characteristics,7,11−13,63 it would be of great interest to assess whether the serum proximal interactomes of targets like TIMP2 could impart clinically relevant diagnostic information. Regardless, it is crucial that all the details concerning candidate identification (protein coverage, analysis scores, predefined protein function) be interrogated prior to further in-depth functional studies.

Figure 6.

Figure 6

Open in a new tab

CRAPome filtering can cause data loss through false-negative reporting.

A common tool for analyzing PL data is SAINT (Significance Analysis of Interactome).64 We chose to forego this method since SAINT uses only one metric (such as spectral count or peptide spectrum matches). SAINT analysis was not amenable to the use of calculated protein abundances due to the large range in abundance values. Our analysis utilizes a simple series of calculations to assign order to the candidate interactors, considering both PSMs and abundance in samples and controls (Supporting Information). To the ordered candidates, we assigned a threshold score (≥1) to determine the candidate proteins to be filtered through CRAPome analysis. To prove the utility of our method, we also performed SAINT analysis on a duplicate experiment and compared the hits with our candidate proteins. This comparison revealed that our analysis was slightly more stringent in candidate protein selection, with SAINT analysis identifying two (TIMP2-TurboID) and nine (TurboID-TIMP2) additional candidate proteins when using Bayesian false discovery rate (BDFR) ≤ 0.05 (Table S7). SAINT analysis also allows users to incorporate known interactors into their analysis platform (topology-aware probability score), adding weight to protein IDs that are known interactors. We decided to exclude inference from previously identified interactors and instead use our knowledge of known interactors to inform our determined threshold for candidate proximal interactors.

A key difference between the methods used to interrogate the proximal interactome of HS-5 and that of HT1080 is the use of TMT labeling. In TMT experiments, the pipeline skews in favor of the control sample through rounds of normalization, starting with sample normalization at TMT labeling and then during normalization of total abundance in analysis. Theoretically, fusion protein samples should produce more biotinylated protein than the controls, something we corroborate with streptavidin blots. To account for skew in favor of control samples, and considering ratio compression that occurs in isobaric tagging methods,65 we utilize a low (1.5) fold change threshold for hit identification.

Here we present proof-of-principle for the utilization of extracellular proximity labeling (ePL) to investigate the proximal interactome of secreted factors. This method effectively identifies unique proximal interactor candidates in the matrisome compartment and can be adapted downstream for improved protein coverage. One such example may be to include a deglycosylation step prior to tryptic digestion that may enhance peptide discovery and quantification. Regardless, exploration of the interactome of matrisome components such as TIMP2 will provide important context with respect to their defined functions and new research direction when exploring undiscovered functions. The protocol successfully identifies known interactors for TIMP2 (matrix metalloproteinases, HSP90 family members) and reveals novel, highly reproducible interactors for further in-depth study. The matrisome represents an untapped resource with regard to therapeutic intervention in disease. To date, there are very few therapeutics that target the extensive, well-documented changes that occur in the extracellular matrix across the spectrum of human diseases. Continued investigation of the matrisome proximal interactome in model systems such as HT1080 and HS-5 cells will uncover interactome hubs within the extracellular compartment, potentially revealing novel therapeutic targets in human disease.

Acknowledgments

This research was supported by the Intramural Research Program of the NIH, grant ZIA BC011204 (WGSS). Plasmids were constructed by the National Institute on Drug Abuse Genetic Engineering and Viral Vector Core Facility (RRID:SCR_022969). Recombinant fusion proteins were produced by the National Cancer Institute, Center for Cancer Research Protein Expression Laboratory. The authors would like to thank Harvey E. Johnston of the Signalling Programme (The Babraham Institute, Cambridge, U.K.) for stimulating discussions regarding sample cleanup prior to mass spectrometry.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.jproteome.4c00606.

  • Figure S1: Exogenous ATP and biotin are required for ePL assays; Figure S2: Gelatin zymography; Figure S3: Proximity ligation assay (PLA) analysis procedure; Raw Western Blot Images; Construct Design; Plasmid Maps; Proximal Interactor Candidate Identification; Antibodies; Proximity Ligation Assay Extended Methods (PDF)

  • Table S1: Summary of LC-MS/MS results discussed in Figure 2 (XLSX)

  • Table S2: Summary of LC-MS/MS results discussed in Figure 3 (XLSX)

  • Table S3: Summary of 3D culture LC-MS/MS results discussed in Figure 4 (XLSX)

  • Table S4: Summary of LC-MS/MS results generated from HT1080 cells (XLSX)

  • Table S5: Summary of LC-MS/MS results generated from HS-5 cells (XLSX)

  • Table S6: Summary of LC-MS/MS results generated using human serum (XLSX)

  • Table S7: Comparison of our analysis method versus SAINT analysis (XLSX)

An early draft of this study was posted at Research Square.18

The authors declare no competing financial interest.

Supplementary Material

pr4c00606_si_001.pdf (4.3MB, pdf)
pr4c00606_si_002.xlsx (561KB, xlsx)
pr4c00606_si_003.xlsx (2MB, xlsx)
pr4c00606_si_004.xlsx (293.1KB, xlsx)
pr4c00606_si_005.xlsx (58KB, xlsx)
pr4c00606_si_006.xlsx (387.8KB, xlsx)
pr4c00606_si_007.xlsx (812.1KB, xlsx)
pr4c00606_si_008.xlsx (36.5KB, xlsx)

References

  1. Bissell M. J.; Hall H. G.; Parry G. How does the extracellular matrix direct gene expression?. J. Theor. Biol. 1982, 99 (1), 31–68. 10.1016/0022-5193(82)90388-5. [DOI] [PubMed] [Google Scholar]
  2. Iozzo R. V.; Gubbiotti M. A. Extracellular matrix: The driving force of mammalian diseases. Matrix Biol 2018, 71–72, 1–9. 10.1016/j.matbio.2018.03.023. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Brew K.; Nagase H. The tissue inhibitors of metalloproteinases (TIMPs): an ancient family with structural and functional diversity. Biochim. Biophys. Acta 2010, 1803 (1), 55–71. 10.1016/j.bbamcr.2010.01.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Stetler-Stevenson W. G. Tissue inhibitors of metalloproteinases in cell signaling: metalloproteinase-independent biological activities. Sci Signal 2008, 1 (27), re6. 10.1126/scisignal.127re6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Peeney D.; Liu Y.; Lazaroff C.; Gurung S.; Stetler-Stevenson W. G. Unravelling the distinct biological functions and potential therapeutic applications of TIMP2 in cancer. Carcinogenesis 2022, 43 (5), 405–418. 10.1093/carcin/bgac037. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Coates-Park S.; Rich J. A.; Stetler-Stevenson W. G.; Peeney D. The TIMP protein family: diverse roles in pathophysiology. Am J Physiol Cell Physiol 2024, 326 (3), C917–C934. 10.1152/ajpcell.00699.2023. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Peeney D.; Jensen S. M.; Castro N. P.; Kumar S.; Noonan S.; Handler C.; Kuznetsov A.; Shih J.; Tran A. D.; Salomon D. S.; et al. TIMP-2 suppresses tumor growth and metastasis in murine model of triple-negative breast cancer. Carcinogenesis 2020, 41 (3), 313–325. 10.1093/carcin/bgz172. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Albini A.; Gallazzi M.; Palano M. T.; Carlini V.; Ricotta R.; Bruno A.; Stetler-Stevenson W. G.; Noonan D. M. TIMP1 and TIMP2 Downregulate TGFbeta Induced Decidual-like Phenotype in Natural Killer Cells. Cancers (Basel) 2021, 13 (19), 4955. 10.3390/cancers13194955. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Guedez L.; Jensen-Taubman S.; Bourboulia D.; Kwityn C. J.; Wei B.; Caterina J.; Stetler-Stevenson W. G. TIMP-2 targets tumor-associated myeloid suppressor cells with effects in cancer immune dysfunction and angiogenesis. J Immunother 2012, 35 (6), 502–512. 10.1097/CJI.0b013e3182619c8e. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Escalona R. M.; Bilandzic M.; Western P.; Kadife E.; Kannourakis G.; Findlay J. K.; Ahmed N. TIMP-2 regulates proliferation, invasion and STAT3-mediated cancer stem cell-dependent chemoresistance in ovarian cancer cells. BMC Cancer 2020, 20 (1), 960. 10.1186/s12885-020-07274-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Castellano J. M.; Mosher K. I.; Abbey R. J.; McBride A. A.; James M. L.; Berdnik D.; Shen J. C.; Zou B.; Xie X. S.; Tingle M.; et al. Human umbilical cord plasma proteins revitalize hippocampal function in aged mice. Nature 2017, 544 (7651), 488–492. 10.1038/nature22067. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Britton R.; Wasley T.; Harish R.; Holz C.; Hall J.; Yee D. C.; Melton Witt J.; Booth E. A.; Braithwaite S.; Czirr E. Noncanonical Activity of Tissue Inhibitor of Metalloproteinases 2 (TIMP2) Improves Cognition and Synapse Density in Aging. eNeuro 2023, 10.1523/ENEURO.0031-23.2023. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Ferreira A. C.; Hemmer B. M.; Philippi S. M.; Grau-Perales A. B.; Rosenstadt J. L.; Liu H.; Zhu J. D.; Kareva T.; Ahfeldt T.; Varghese M.; et al. Neuronal TIMP2 regulates hippocampus-dependent plasticity and extracellular matrix complexity. Mol Psychiatry 2023, 28 (9), 3943–3954. 10.1038/s41380-023-02296-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Kveiborg M.; Jacobsen J.; Lee M. H.; Nagase H.; Wewer U. M.; Murphy G. Selective inhibition of ADAM12 catalytic activity through engineering of tissue inhibitor of metalloproteinase 2 (TIMP-2). Biochem. J. 2010, 430 (1), 79–86. 10.1042/BJ20100649. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Santamaria S.; Martin D. R.; Dong X.; Yamamoto K.; Apte S. S.; Ahnstrom J. Post-translational regulation and proteolytic activity of the metalloproteinase ADAMTS8. J. Biol. Chem. 2021, 297 (5), 101323 10.1016/j.jbc.2021.101323. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Branon T.; Han S.; Ting A. Beyond Immunoprecipitation: Exploring New Interaction Spaces with Proximity Biotinylation. Biochemistry 2017, 56 (26), 3297–3298. 10.1021/acs.biochem.7b00466. [DOI] [PubMed] [Google Scholar]
  17. Qin W.; Cho K. F.; Cavanagh P. E.; Ting A. Y. Deciphering molecular interactions by proximity labeling. Nat Methods 2021, 18 (2), 133–143. 10.1038/s41592-020-01010-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Peeney D.; Gurung S.; Rich J. A.; Coates-Park S.; Liu Y.; Toor J.; Jones J.; Richie C. T.; Jenkins L. M.; Stetler-Stevenson W. G. Extracellular Proximity Labeling Reveals an Expanded Interactome for the Matrisome Protein TIMP2. Res. Sq 2024, 10.21203/rs.3.rs-3857263/v1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Rodionov D. A.; Mironov A. A.; Gelfand M. S. Conservation of the biotin regulon and the BirA regulatory signal in Eubacteria and Archaea. Genome Res. 2002, 12 (10), 1507–1516. 10.1101/gr.314502. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Branon T. C.; Bosch J. A.; Sanchez A. D.; Udeshi N. D.; Svinkina T.; Carr S. A.; Feldman J. L.; Perrimon N.; Ting A. Y. Efficient proximity labeling in living cells and organisms with TurboID. Nat. Biotechnol. 2018, 36 (9), 880–887. 10.1038/nbt.4201. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Kim D. I.; Jensen S. C.; Noble K. A.; Kc B.; Roux K. H.; Motamedchaboki K.; Roux K. J. An improved smaller biotin ligase for BioID proximity labeling. Mol. Biol. Cell 2016, 27 (8), 1188–1196. 10.1091/mbc.E15-12-0844. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Wilson K. P.; Shewchuk L. M.; Brennan R. G.; Otsuka A. J.; Matthews B. W. Escherichia coli biotin holoenzyme synthetase/bio repressor crystal structure delineates the biotin- and DNA-binding domains. Proc Natl Acad Sci U S A 1992, 89 (19), 9257–9261. 10.1073/pnas.89.19.9257. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Jumper J.; Evans R.; Pritzel A.; Green T.; Figurnov M.; Ronneberger O.; Tunyasuvunakool K.; Bates R.; Zidek A.; Potapenko A.; et al. Highly accurate protein structure prediction with AlphaFold. Nature 2021, 596 (7873), 583–589. 10.1038/s41586-021-03819-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Rasheed S.; Nelson-Rees W. A.; Toth E. M.; Arnstein P.; Gardner M. B. Characterization of a newly derived human sarcoma cell line (HT-1080). Cancer 1974, 33 (4), 1027–1033. . [DOI] [PubMed] [Google Scholar]
  25. Stanton H.; Gavrilovic J.; Atkinson S. J.; d’Ortho M. P.; Yamada K. M.; Zardi L.; Murphy G. The activation of ProMMP-2 (gelatinase A) by HT1080 fibrosarcoma cells is promoted by culture on a fibronectin substrate and is concomitant with an increase in processing of MT1-MMP (MMP-14) to a 45 kDa form. J. Cell Sci. 1998, 111 (18), 2789–2798. 10.1242/jcs.111.18.2789. [DOI] [PubMed] [Google Scholar]
  26. Volpert O. V.; Lawler J.; Bouck N. P. A human fibrosarcoma inhibits systemic angiogenesis and the growth of experimental metastases via thrombospondin-1. Proc Natl Acad Sci U S A 1998, 95 (11), 6343–6348. 10.1073/pnas.95.11.6343. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Peeney D.; Fan Y.; Gurung S.; Lazaroff C.; Ratnayake S.; Warner A.; Karim B.; Meerzaman D.; Stetler-Stevenson W. G. Whole organism profiling of the Timp gene family. Matrix Biol Plus 2023, 18, 100132 10.1016/j.mbplus.2023.100132. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Mellacheruvu D.; Wright Z.; Couzens A. L.; Lambert J. P.; St-Denis N. A.; Li T.; Miteva Y. V.; Hauri S.; Sardiu M. E.; Low T. Y.; et al. The CRAPome: a contaminant repository for affinity purification-mass spectrometry data. Nat Methods 2013, 10 (8), 730–736. 10.1038/nmeth.2557. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Stetler-Stevenson W. G.; Krutzsch H. C.; Liotta L. A. Tissue inhibitor of metalloproteinase (TIMP-2). A new member of the metalloproteinase inhibitor family. J. Biol. Chem. 1989, 264 (29), 17374–17378. 10.1016/S0021-9258(18)71503-2. [DOI] [PubMed] [Google Scholar]
  30. Morgunova E.; Tuuttila A.; Bergmann U.; Tryggvason K. Structural insight into the complex formation of latent matrix metalloproteinase 2 with tissue inhibitor of metalloproteinase 2. Proc Natl Acad Sci U S A 2002, 99 (11), 7414–7419. 10.1073/pnas.102185399. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Uhlen M.; Karlsson M. J.; Hober A.; Svensson A. S.; Scheffel J.; Kotol D.; Zhong W.; Tebani A.; Strandberg L.; Edfors F. The human secretome. Sci. Signal 2019, 10.1126/scisignal.aaz0274. [DOI] [Google Scholar]
  32. Nishizuka Y. The role of protein kinase C in cell surface signal transduction and tumour promotion. Nature 1984, 308 (5961), 693–698. 10.1038/308693a0. [DOI] [PubMed] [Google Scholar]
  33. Newton A. C.; Brognard J. Reversing the Paradigm: Protein Kinase C as a Tumor Suppressor. Trends Pharmacol. Sci. 2017, 38 (5), 438–447. 10.1016/j.tips.2017.02.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Gervasi D. C.; Raz A.; Dehem M.; Yang M.; Kurkinen M.; Fridman R. Carbohydrate-mediated regulation of matrix metalloproteinase-2 activation in normal human fibroblasts and fibrosarcoma cells. Biochem. Biophys. Res. Commun. 1996, 228 (2), 530–538. 10.1006/bbrc.1996.1694. [DOI] [PubMed] [Google Scholar]
  35. d’Ortho M. P.; Stanton H.; Butler M.; Atkinson S. J.; Murphy G.; Hembry R. M. MT1-MMP on the cell surface causes focal degradation of gelatin films. FEBS Lett. 1998, 421 (2), 159–164. 10.1016/S0014-5793(97)01555-X. [DOI] [PubMed] [Google Scholar]
  36. Strongin A. Y.; Collier I.; Bannikov G.; Marmer B. L.; Grant G. A.; Goldberg G. I. Mechanism of cell surface activation of 72-kDa type IV collagenase. Isolation of the activated form of the membrane metalloprotease. J. Biol. Chem. 1995, 270 (10), 5331–5338. 10.1074/jbc.270.10.5331. [DOI] [PubMed] [Google Scholar]
  37. Overall C. M.; Tam E.; McQuibban G. A.; Morrison C.; Wallon U. M.; Bigg H. F.; King A. E.; Roberts C. R. Domain interactions in the gelatinase A.TIMP-2.MT1-MMP activation complex. The ectodomain of the 44-kDa form of membrane type-1 matrix metalloproteinase does not modulate gelatinase A activation. J. Biol. Chem. 2000, 275 (50), 39497–39506. 10.1074/jbc.M005932200. [DOI] [PubMed] [Google Scholar]
  38. Fernandez-Catalan C.; Bode W.; Huber R.; Turk D.; Calvete J. J.; Lichte A.; Tschesche H.; Maskos K. Crystal structure of the complex formed by the membrane type 1-matrix metalloproteinase with the tissue inhibitor of metalloproteinases-2, the soluble progelatinase A receptor. EMBO J. 1998, 17 (17), 5238–5248. 10.1093/emboj/17.17.5238. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Maquoi E.; Frankenne F.; Baramova E.; Munaut C.; Sounni N. E.; Remacle A.; Noel A.; Murphy G.; Foidart J. M. Membrane type 1 matrix metalloproteinase-associated degradation of tissue inhibitor of metalloproteinase 2 in human tumor cell lines. J. Biol. Chem. 2000, 275 (15), 11368–11378. 10.1074/jbc.275.15.11368. [DOI] [PubMed] [Google Scholar]
  40. Zucker S.; Hymowitz M.; Conner C.; DeClerck Y.; Cao J. TIMP-2 is released as an intact molecule following binding to MT1-MMP on the cell surface. Exp. Cell Res. 2004, 293 (1), 164–174. 10.1016/j.yexcr.2003.10.007. [DOI] [PubMed] [Google Scholar]
  41. Jensen C.; Teng Y. Is It Time to Start Transitioning From 2D to 3D Cell Culture?. Front Mol Biosci 2020, 7, 33. 10.3389/fmolb.2020.00033. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Huttlin E. L.; Bruckner R. J.; Navarrete-Perea J.; Cannon J. R.; Baltier K.; Gebreab F.; Gygi M. P.; Thornock A.; Zarraga G.; Tam S.; et al. Dual proteome-scale networks reveal cell-specific remodeling of the human interactome. Cell 2021, 184 (11), 3022–3040. 10.1016/j.cell.2021.04.011. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Alam M. S. Proximity Ligation Assay (PLA). Methods Mol. Biol. 2022, 2422, 191–201. 10.1007/978-1-0716-1948-3_13. [DOI] [PubMed] [Google Scholar]
  44. Fisher K. E.; Sacharidou A.; Stratman A. N.; Mayo A. M.; Fisher S. B.; Mahan R. D.; Davis M. J.; Davis G. E. MT1-MMP- and Cdc42-dependent signaling co-regulate cell invasion and tunnel formation in 3D collagen matrices. J Cell Sci 2009, 122 (24), 4558–4569. 10.1242/jcs.050724. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Borkham-Kamphorst E.; Steffen B. T.; Van de Leur E.; Haas U.; Tihaa L.; Friedman S. L.; Weiskirchen R. CCN1/CYR61 overexpression in hepatic stellate cells induces ER stress-related apoptosis. Cell Signal 2016, 28 (1), 34–42. 10.1016/j.cellsig.2015.10.013. [DOI] [PubMed] [Google Scholar]
  46. Hirschfeld M.; zur Hausen A.; Bettendorf H.; Jager M.; Stickeler E. Alternative splicing of Cyr61 is regulated by hypoxia and significantly changed in breast cancer. Cancer Res. 2009, 69 (5), 2082–2090. 10.1158/0008-5472.CAN-08-1997. [DOI] [PubMed] [Google Scholar]
  47. Blalock T. D.; Gibson D. J.; Duncan M. R.; Tuli S. S.; Grotendorst G. R.; Schultz G. S. A connective tissue growth factor signaling receptor in corneal fibroblasts. Invest Ophthalmol Vis Sci 2012, 53 (7), 3387–3394. 10.1167/iovs.12-9425. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Chen Y.; Segarini P.; Raoufi F.; Bradham D.; Leask A. Connective tissue growth factor is secreted through the Golgi and is degraded in the endosome. Exp. Cell Res. 2001, 271 (1), 109–117. 10.1006/excr.2001.5364. [DOI] [PubMed] [Google Scholar]
  49. Naba A. Ten Years of Extracellular Matrix Proteomics: Accomplishments, Challenges, and Future Perspectives. Mol Cell Proteomics 2023, 22 (4), 100528 10.1016/j.mcpro.2023.100528. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Hynes R. O.; Naba A. Overview of the matrisome--an inventory of extracellular matrix constituents and functions. Cold Spring Harb Perspect Biol 2012, 4 (1), a004903 10.1101/cshperspect.a004903. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Kim D. I.; Roux K. J. Filling the Void: Proximity-Based Labeling of Proteins in Living Cells. Trends Cell Biol 2016, 26 (11), 804–817. 10.1016/j.tcb.2016.09.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Bar D. Z.; Atkatsh K.; Tavarez U.; Erdos M. R.; Gruenbaum Y.; Collins F. S. Biotinylation by antibody recognition-a method for proximity labeling. Nat Methods 2018, 15 (2), 127–133. 10.1038/nmeth.4533. [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Ayoubi R.; Ryan J.; Biddle M. S.; Alshafie W.; Fotouhi M.; Bolivar S. G.; Moleon V. R.; Eckmann P.; Worrall D.; McDowell I.; et al. Scaling of an antibody validation procedure enables quantification of antibody performance in major research applications. bioRxiv 2023, 10.1101/2023.06.01.543292. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Yamada K.; Shioya R.; Nishino K.; Furihata H.; Hijikata A.; Kaneko M. K.; Kato Y.; Shirai T.; Kosako H.; Sawasaki T. Proximity extracellular protein-protein interaction analysis of EGFR using AirID-conjugated fragment of antigen binding. Nat Commun 2023, 14 (1), 8301. 10.1038/s41467-023-43931-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Murphy D. A.; Courtneidge S. A. The ’ins’ and ’outs’ of podosomes and invadopodia: characteristics, formation and function. Nat Rev Mol Cell Biol 2011, 12 (7), 413–426. 10.1038/nrm3141. [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Chang J.; Chaudhuri O. Beyond proteases: Basement membrane mechanics and cancer invasion. J. Cell Biol. 2019, 218 (8), 2456–2469. 10.1083/jcb.201903066. [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Bahr J. C.; Li X. Y.; Feinberg T. Y.; Jiang L.; Weiss S. J. Divergent regulation of basement membrane trafficking by human macrophages and cancer cells. Nat Commun 2022, 13 (1), 6409. 10.1038/s41467-022-34087-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Roecklein B. A.; Torok-Storb B. Functionally distinct human marrow stromal cell lines immortalized by transduction with the human papilloma virus E6/E7 genes. Blood 1995, 85 (4), 997–1005. 10.1182/blood.V85.4.997.bloodjournal854997. [DOI] [PubMed] [Google Scholar]
  59. Rich J. A.; Fan Y.; Chen Q.; Meerzaman D.; Stetler-Stevenson W. G.; Peeney D. Analysis of cancer cell line and tissue RNA sequencing data reveals an essential and dark matrisome. Matrix Biology Plus 2024, 23, 100156 10.1016/j.mbplus.2024.100156. [DOI] [PMC free article] [PubMed] [Google Scholar]
  60. Tong L. Structure and function of biotin-dependent carboxylases. Cell. Mol. Life Sci. 2013, 70 (5), 863–891. 10.1007/s00018-012-1096-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  61. Baker-Williams A. J.; Hashmi F.; Budzynski M. A.; Woodford M. R.; Gleicher S.; Himanen S. V.; Makedon A. M.; Friedman D.; Cortes S.; Namek S.; et al. Co-chaperones TIMP2 and AHA1 Competitively Regulate Extracellular HSP90:Client MMP2 Activity and Matrix Proteolysis. Cell Rep. 2019, 28 (7), 1894–1906. 10.1016/j.celrep.2019.07.045. [DOI] [PMC free article] [PubMed] [Google Scholar]
  62. Dull K.; Fazekas F.; Torocsik D. Factor XIII-A in Diseases: Role Beyond Blood Coagulation. Int. J. Mol. Sci 2021, 22 (3), 1459. 10.3390/ijms22031459. [DOI] [PMC free article] [PubMed] [Google Scholar]
  63. Peeney D.; Kumar S.; Singh T. P.; Liu Y.; Jensen S. M.; Chowdhury A.; Coates-Park S.; Rich J.; Gurung S.; Fan Y.; et al. Timp2 loss-of-function mutation and TIMP2 treatment in murine model of NSCLC: modulation of immunosuppression and oncogenic signaling. bioRxiv 2023, 10.1101/2023.12.29.573636. [DOI] [Google Scholar]
  64. Choi H.; Larsen B.; Lin Z. Y.; Breitkreutz A.; Mellacheruvu D.; Fermin D.; Qin Z. S.; Tyers M.; Gingras A. C.; Nesvizhskii A. I. SAINT: probabilistic scoring of affinity purification-mass spectrometry data. Nat Methods 2011, 8 (1), 70–73. 10.1038/nmeth.1541. [DOI] [PMC free article] [PubMed] [Google Scholar]
  65. Ow S. Y.; Salim M.; Noirel J.; Evans C.; Wright P. C. Minimising iTRAQ ratio compression through understanding LC-MS elution dependence and high-resolution HILIC fractionation. Proteomics 2011, 11 (11), 2341–2346. 10.1002/pmic.201000752. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

pr4c00606_si_001.pdf (4.3MB, pdf)
pr4c00606_si_002.xlsx (561KB, xlsx)
pr4c00606_si_003.xlsx (2MB, xlsx)
pr4c00606_si_004.xlsx (293.1KB, xlsx)
pr4c00606_si_005.xlsx (58KB, xlsx)
pr4c00606_si_006.xlsx (387.8KB, xlsx)
pr4c00606_si_007.xlsx (812.1KB, xlsx)
pr4c00606_si_008.xlsx (36.5KB, xlsx)

Articles from Journal of Proteome Research are provided here courtesy of American Chemical Society

RESOURCES