Article | Published:

A proximity-tagging system to identify membrane protein–protein interactions

Nature Methodsvolume 15pages715722 (2018) | Download Citation


The communication between cells and between cellular organelles is often controlled by the interaction of membrane proteins. Although many methods for the detection of protein–protein interactions (PPIs) exist, membrane PPIs remain difficult to detect. Here we developed a proximity-based tagging system, PUP-IT (pupylation-based interaction tagging), to identify membrane protein interactions. In this approach, a small protein tag, Pup, is applied to proteins that interact with a PafA-fused bait, enabling transient and weak interactions to be enriched and detected by mass spectrometry. Pup does not diffuse from the enzyme, which allows high-specificity labeling. We applied this approach to CD28, a critical costimulatory receptor for T lymphocyte activation, and identified known CD28 binding partners and multiple potential interacting proteins. In addition, we demonstrated that this method can identify the interaction between a cell surface receptor and its ligand.

Access optionsAccess options

Rent or Buy article

Get time limited or full article access on ReadCube.


All prices are NET prices.

Additional information

Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.


  1. 1.

    Hung, V. et al. Proteomic mapping of the human mitochondrial intermembrane space in live cells via ratiometric APEX tagging. Mol. Cell 55, 332–341 (2014).

  2. 2.

    Lee, S. Y. et al. APEX fingerprinting reveals the subcellular localization of proteins of interest. Cell Rep. 15, 1837–1847 (2016).

  3. 3.

    Roux, K. J., Kim, D. I., Raida, M. & Burke, B. A promiscuous biotin ligase fusion protein identifies proximal and interacting proteins in mammalian cells. J. Cell Biol. 196, 801–810 (2012).

  4. 4.

    Kim, D. I., Jensen, S. C. & Roux, K. J. Identifying protein-protein associations at the nuclear envelope with BioID. Methods Mol. Biol. 1411, 133–146 (2016).

  5. 5.

    Loh, K. H. et al. Proteomic analysis of unbounded cellular compartments: synaptic clefts. Cell 166, 1295–1307 (2016).

  6. 6.

    Lobingier, B. T. et al. An approach to spatiotemporally resolve protein interaction networks in living cells. Cell 169, 350–360 (2017).

  7. 7.

    Paek, J. et al. Multidimensional tracking of GPCR signaling via peroxidase-catalyzed proximity labeling. Cell 169, 338–349 (2017).

  8. 8.

    Zhuang, M., Guan, S., Wang, H., Burlingame, A. L. & Wells, J. A. Substrates of IAP ubiquitin ligases identified with a designed orthogonal E3 ligase, the NEDDylator. Mol. Cell 49, 273–282 (2013).

  9. 9.

    Hill, Z. B., Pollock, S. B., Zhuang, M. & Wells, J. A. Direct proximity tagging of small molecule protein targets using an engineered NEDD8 ligase. J. Am. Chem. Soc. 138, 13123–13126 (2016).

  10. 10.

    Enchev, R. I., Schulman, B. A. & Peter, M. Protein neddylation: beyond cullin-RING ligases. Nat. Rev. Mol. Cell Biol. 16, 30–44 (2015).

  11. 11.

    Striebel, F. et al. Bacterial ubiquitin-like modifier Pup is deamidated and conjugated to substrates by distinct but homologous enzymes. Nat. Struct. Mol. Biol. 16, 647–651 (2009).

  12. 12.

    Özcelik, D. et al. Structures of Pup ligase PafA and depupylase Dop from the prokaryotic ubiquitin-like modification pathway. Nat. Commun. 3, 1014 (2012).

  13. 13.

    Guth, E., Thommen, M. & Weber-Ban, E. Mycobacterial ubiquitin-like protein ligase PafA follows a two-step reaction pathway with a phosphorylated pup intermediate. J. Biol. Chem. 286, 4412–4419 (2011).

  14. 14.

    Pearce, M. J., Mintseris, J., Ferreyra, J., Gygi, S. P. & Darwin, K. H. Ubiquitin-like protein involved in the proteasome pathway of Mycobacterium tuberculosis. Science 322, 1104–1107 (2008).

  15. 15.

    Burns, K. E., Liu, W. T., Boshoff, H. I., Dorrestein, P. C. & Barry, C. E. III Proteasomal protein degradation in Mycobacteria is dependent upon a prokaryotic ubiquitin-like protein. J. Biol. Chem. 284, 3069–3075 (2009).

  16. 16.

    Smirnov, D., Dhall, A., Sivanesam, K., Sharar, R. J. & Chatterjee, C. Fluorescent probes reveal a minimal ligase recognition motif in the prokaryotic ubiquitin-like protein from Mycobacterium tuberculosis. J. Am. Chem. Soc. 135, 2887–2890 (2013).

  17. 17.

    Festa, R. A. et al. Prokaryotic ubiquitin-like protein (Pup) proteome of Mycobacterium tuberculosis. PLoS One 5, e8589 (2010).

  18. 18.

    Watrous, J. et al. Expansion of the mycobacterial “PUPylome”. Mol. Biosyst. 6, 376–385 (2010).

  19. 19.

    Poulsen, C. et al. Proteome-wide identification of mycobacterial pupylation targets. Mol. Syst. Biol. 6, 386 (2010).

  20. 20.

    Tagwerker, C. et al. A tandem affinity tag for two-step purification under fully denaturing conditions: application in ubiquitin profiling and protein complex identification combined with in vivo cross-linking. Mol. Cell. Proteomics 5, 737–748 (2006).

  21. 21.

    Zhang, Q. et al. Multiple Ser/Thr-rich degrons mediate the degradation of Ci/Gli by the Cul3-HIB/SPOP E3 ubiquitin ligase. Proc. Natl. Acad. Sci. USA 106, 21191–21196 (2009).

  22. 22.

    Zhuang, M. et al. Structures of SPOP-substrate complexes: insights into molecular architectures of BTB-Cul3 ubiquitin ligases. Mol. Cell 36, 39–50 (2009).

  23. 23.

    Gossen, M. et al. Transcriptional activation by tetracyclines in mammalian cells. Science 268, 1766–1769 (1995).

  24. 24.

    Urlinger, S. et al. Exploring the sequence space for tetracycline-dependent transcriptional activators: novel mutations yield expanded range and sensitivity. Proc. Natl. Acad. Sci. USA 97, 7963–7968 (2000).

  25. 25.

    Banaszynski, L. A., Liu, C. W. & Wandless, T. J. Characterization of the FKBP·rapamycin·FRB ternary complex. J. Am. Chem. Soc. 127, 4715–4721 (2005).

  26. 26.

    Kim, D. I. et al. An improved smaller biotin ligase for BioID proximity labeling. Mol. Biol. Cell 27, 1188–1196 (2016).

  27. 27.

    Mehus, A. A., Anderson, R. H. & Roux, K. J. BioID identification of lamin-associated proteins. Methods Enzymol. 569, 3–22 (2016).

  28. 28.

    Rhee, H. W. et al. Proteomic mapping of mitochondria in living cells via spatially restricted enzymatic tagging. Science 339, 1328–1331 (2013).

  29. 29.

    Bar, D. Z. et al. Biotinylation by antibody recognition—a method for proximity labeling. Nat. Methods 15, 127–133 (2018).

  30. 30.

    Crooks, G. E., Hon, G., Chandonia, J. M. & Brenner, S. E. WebLogo: a sequence logo generator. Genome Res. 14, 1188–1190 (2004).

  31. 31.

    Liu, Q. Z. J., Wang, H. & Zhuang, M. Identification of interacting proteins using PUP-IT. Protoc. Exch. (2018).

  32. 32.

    Guo, Z. Q. et al. Small-molecule targeting of E3 ligase adaptor SPOP in kidney cancer. Cancer Cell 30, 474–484 (2016).

  33. 33.

    Elias, J. E. & Gygi, S. P. Target-decoy search strategy for increased confidence in large-scale protein identifications by mass spectrometry. Nat. Methods 4, 207–214 (2007).

  34. 34.

    Cox, J. & Mann, M. MaxQuant enables high peptide identification rates, individualized p.p.b.-range mass accuracies and proteome-wide protein quantification. Nat. Biotechnol. 26, 1367–1372 (2008).

  35. 35.

    Cox, J. et al. Andromeda: a peptide search engine integrated into the MaxQuant environment. J. Proteome Res. 10, 1794–1805 (2011).

  36. 36.

    Tian, R. et al. Combinatorial proteomic analysis of intercellular signaling applied to the CD28 T-cell costimulatory receptor. Proc. Natl. Acad. Sci. USA 112, E1594–E1603 (2015).

  37. 37.

    Vizcaíno, J. A. et al. 2016 update of the PRIDE database and its related tools. Nucleic Acids Res 44, D447–D456 (2016).

Download references


We thank Y. Han for assistance with mass spectrometry equipment, X. Li and G. Zhong for training on use of the confocal microscope, and X. Ren and Y. Xiong for flow cytometer sorting. We also thank J. Han (Xiamen University, Xiamen, China) for providing cDNAs, and A. Weiss (UCSF, San Francisco, CA, USA) and O. Acuto (University of Oxford, Oxford, UK) for providing cell lines. GST–MATH and His–MATH fusion vectors were gifts from C. Yang (Shanghai Institute of Materia Medica, Chinese Academy of Sciences, Shanghai, China). BirA* codon-optimized sequence was a gift from X. Yue (ShanghaiTech University, Shanghai, China). This work has been supported by the Shanghai Municipal Science and Technology Commission Pujiang Talents plan (15PJ1406000 to M.Z.; SMSTC 17411951800 to L.Z.), the National Natural Science Foundation of China (31570767 to M.Z.; 31670919 to H.W.), and the 1,000-Youth Elite Program of China (H.W.).

Author information

Author notes

  1. These authors contributed equally: Qiang Liu and Jun Zheng.


  1. School of Life Science and Technology, ShanghaiTech University, Shanghai, China

    • Qiang Liu
    • , Jun Zheng
    • , Weiping Sun
    • , Yinbo Huo
    • , Liye Zhang
    • , Piliang Hao
    • , Haopeng Wang
    •  & Min Zhuang
  2. Institute of Biochemistry and Cell Biology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai, China

    • Qiang Liu
    •  & Yinbo Huo
  3. University of Chinese Academy of Sciences, Beijing, China

    • Qiang Liu
    • , Jun Zheng
    • , Weiping Sun
    •  & Yinbo Huo


  1. Search for Qiang Liu in:

  2. Search for Jun Zheng in:

  3. Search for Weiping Sun in:

  4. Search for Yinbo Huo in:

  5. Search for Liye Zhang in:

  6. Search for Piliang Hao in:

  7. Search for Haopeng Wang in:

  8. Search for Min Zhuang in:


Q.L. performed all the experiments except those explicitly noted below. J.Z. generated the stable cell lines, performed mass spectrometry experiments with different bait proteins, and analyzed mass spectrometry data. W.S. purified proteins and compared enzymatic activity in vitro. Y.H. compared the activity between different substrates in vitro. L.Z. analyzed data and provided bioinformatics suggestions. P.H. guided and oversaw mass spectrometry experiments and analyzed results with J.Z. and Q.L. H.W. and M.Z. designed the research, analyzed the data, and wrote the paper. All authors reviewed and edited the paper.

Competing interests

The authors declare no competing interests.

Corresponding authors

Correspondence to Haopeng Wang or Min Zhuang.

Integrated supplementary information

  1. Supplementary Figure 1 Identification of modified lysine residues on GST and XIAP.

    (a) Protein sequences of GST-PafA, PafA-XIAP and PupE. The modified lysine residues are highlighted in red. (b) Identified peptides from mass spectrometry experiments and the peptide scores (Supplementary Table 1). (c) One peptide spectral example showing the modification site.

  2. Supplementary Figure 2 Sequence and structure of bio-Pup(E) and bio-DE28.

    (a) The sequence of bio-Pup(E). BCCP domain (blue) derived from bacteria carboxylase (Propionobacterium freednreichii, NCBI Reference Sequence: WP_013161729.1) is fused to the N terminus of full length Pup (red). BCCP has been codon optimized for mammalian cell expression (DNA sequence: GCCGGGAAGGCAGGCGAGGGAGAGATCCCCGCACCCTTGGCCGGCACGGTCAGCAAAATCCTGGTCAAGGAAGGCGACACCGTGAAGGCTGGACAGACGGTGTTGGTACTGGAGGCGATGAAGATGGAGACAGAGATCAATGCCCCGACCGATGGGAAGGTGGAGAAGGTGCTGGTTAAGGAGAGGGACGCCGTGCAGGGCGGTCAGGGACTGATCAAGATCGGCGACTACGACATCCCGACAACCGCCAGC). The lysine residue labeled in green is biotinylated in mammalian cells. This version of bio-Pup(E) was used for all intracellular studies in the paper. (b) Recombinant bio-Pup(E) purified from E. Coli. GST-bio-Pup(E) was cloned into pGEX vector. The protein was pull down by GST tag, then GST was cleaved off and the protein was further purified with gel filtration chromatography. * Residual GST left in the sample. (c) The structure and design of chemically synthesized bio-DE28. (d) Mass spectrometry analysis of confirmed peptide mass. (e) Bio-Pup(E) and bio-DE28 have similar activity in vitro. The Puplylation assay was carried out in a 20 μl reaction mix containing 200 nM GST-PafA, 10x reaction buffer (200 mM Tris, 50 mM ATP and 75 mM Mg2+, pH 8.0), and 2 μM bio-Pup(E) or bio-DE28 at 25 °C. The reactions were stopped at different time points for western blot analysis with streptavidin-HRP. The intensity of self-modified GST-PafA were quantitated and plotted (mean ± SEM on the bottom left and individual dots plot on the bottom right, n = 3 independent experiments). One representative gel from triplicate experiments was shown on the top. Original full scans of blots are shown in Supplementary Fig. 15.

  3. Supplementary Figure 3 Mass spectrometry identification of proteins that interact with CD28 cytosolic tail.

    (a) The spectral counts are combined from duplicate runs and plotted with each dot represents a protein identified. PUP-ITCD28 and PUP-ITtailless datasets are compared. The red dots are examples of known CD28 interacting proteins (Supplementary Table 2a). (b) Identification of CD28 interacting proteins by comparing the spectral counts from PUP-ITCD28 and PUP-IT5AA datasets (Supplementary Table 2b).

  4. Supplementary Figure 4 PUP-ITCD28 identified a large number of proteins involved in protein transport.

    (a) Venn diagram shows the overlapping proteins identified by different PUP-IT. Proteins identified with more than 2 unique peptides are considered in this plot. (b) 51 unique CD28 tail-binding proteins and c) 202 potential CD28 interacting candidates (Supplementary Table 2) are analyzed with gene ontology, using STRING. Top ten functional enrichment groups are presented.

  5. Supplementary Figure 5 PUP-ITCD28 maintains correct localization and function.

    (a) Immunofluorescence staining shows CD28-PafA is expressed on cell surface in the Pup inducible Jurkat used for mass spectrometry study. Top panel, CD28 in wild type Jurkat was stained. Bottom panel, CD28-PafA-myc was stained with anti-myc antibody. GFP was also expressed in iPUP Jurkat as a control for cytosolic protein. This experiment was repeated independently twice with similar results. (b) CD28-PafA can be detected on cell surface. CD28 deficient cell line (a gift from Dr. Oreste Acuto) was transfected with either CD28 or CD28-PafA, cells were then mixed with anti-CD28-PE antibody and analyzed with flow cytometer. (c) CD28-PafA mediates up regulation of CD69 upon stimulation. CD28 knockout cells were transfected with either CD28 or CD28-PafA. With the stimulation by CD28 ligand mimic antibody, the cell surface activation marker CD69 is up regulated in both CD28 and CD28-PafA transfected cells. (d) Different expression level of CD28 and CD28-PafA in transfected cells. CD28-myc has a much higher expression level than CD28-PafA-myc. Original full scans of blots are shown in Supplementary Fig. 15. (e) The experiments in (c) were repeated three times and CD28 activation index was compared between CD28 and CD28-PafA (mean ± SEM, n = 3 biological independent samples). CD28-PafA largely maintains the function of wild type CD28. CD28 activation index was defined as percentage of CD69 positive cells/CD28 expression intensity.

  6. Supplementary Figure 6 PUP-ITMUL1 and PUP-ITRNF13 identify unique interacting proteins.

    (a) Cellular localization of MUL1-PafA and RNF13-PafA. HeLa cells were transfected with either MUL1-PafA-myc or RNF13-PafA-myc, then cells were fixed and stained with anti-myc antibody. This experiment was repeated independently twice with similar results. (b) Volcano plot of PUP-ITMUL1 and PUP-ITCD28. The logarithmic ratios of protein LFQ-intensity (CD28/MUL1) was plotted against negative logarithmic P value of a two-sided two samples t-test in Perseus. Mitochondria located proteins, including RHOT1, TOM70 and HK2 are uniquely identified in PUP-ITMUL1 datasets (Supplementary Table 3b). (c) Volcano plot of PUP-ITMUL1 and PUP-ITRNF13. The logarithmic ratios of protein LFQ-intensity (RNF13/MUL1) was plotted against negative logarithmic P value of a two-sided two samples t-test in Perseus. Ubiquitin ligase RNF13 uniquely binds the ubiquitin conjugating enzyme UBE2H (Supplementary Table 3c). In (b) and (c), the green and blue dots represent significantly enriched proteins from common hits (FDR ≤ 0.05, n = 3 independent experiments).

  7. Supplementary Figure 7 Validation of potential MUL1-interacting proteins.

    (a) Pathway analysis of MUL1 interaction proteins. Each dot represents a protein identified in Supplementary Table 4 as a MUL1 specific interacting protein. These proteins are either connected in primary metabolic process or related to ubiquitin-proteasome system. The gene ontology software STRING (functional protein association networks) was used to analyze protein interactions ( The thickness of the lines indicates the strength of data support. (b) Co-immunoprecipitation of MUL1 and interacting candidates. Myc tagged MUL1 was co-transfected with different V5 tagged candidate in HeLa cells. LCK-V5 was used as the negative control. MUL1 associates with TOMM22, CYB5R1 and PEX19, but not with OCIAD1 and RHOT1. (c) TOMM22 and PEX19 interact with MUL1. The co-immunoprecipitation experiments were repeated with TOMM22 and PEX19 to further confirm the specific interaction. Experiments in (b) and (c) were repeated independently twice with similar results. Original full scans of blots are shown in Supplementary Fig. 15.

  8. Supplementary Figure 8 PUP-IT expression does not introduce more background for mass spectrometry experiments.

    Gene ontology analysis using protein lists identified in (a) iPUP cells without PafA or Pup expression (no PUP-IT transfection and no doxycycline) (n = 801 proteins) (b) iPUP cells induced with doxycycline but no PafA expression (n = 795 proteins), (c) common background proteins identified with PUP-ITCD28, PUP-ITMUL1 and PUP-ITRNF13 (n = 543 proteins). The data (Supplementary Table 3) was analyzed with DAVID ( and plotted with Graphpad Prism 6.

  9. Supplementary Figure 9 Raji cell-surface labeling by PUP-ITCD28-ex.

    (a) PUP-ITCD28-extracellular mediated Raji cells labeling. Similar experiments were set up as in Fig. 5d but the fluorescence was examined with flow cytometer. The percentage of biotin positive cells is shown in the gated window. This experiment was repeated independently twice with similar results. (b) mCherry positive Jurkat cells were transfected with FKBP-CD28 and mixed with GFP positive Jurkat cells. Cell surface Pup modification assay was performed in the same way as in (a). This experiment was repeated independently twice with similar results.

  10. Supplementary Figure 10 Recombinant protein expression and purification.

    (a) Construct design and protein sequence used to express and purify IL2-FKBP. Human IL2-FKBP fusion with GS linker and His tag was subcloned into pCDNA3.1 and transfected 293 T cells. Medium containing IL2-FKBP was collected and the protein was purified with Ni beads. FRB-PafA was subcloned into pGEX6p-1 and fused with N terminal GST (see Methods for details). (b) Purified IL2-FKBP and (c) Purified FRB-PafA are analyzed with SDS-PAGE and coomassie staining. Lane 1 in (c) is purified GST-FRB-PafA cleaved by PreScission protease and lane 2 is further purified FRB-PafA.

  11. Supplementary Figure 11 PUP-ITIL2-dependent cell-surface modification of bio-DE28.

    (a) Activated T cell surface can be modified by PUP-ITIL2. T cells are first activated then mixed with PUP-ITIL2, bio-DE28, ATP and rapamycin. The same reaction was also set up without either ATP or rapamycin. A complete assembly of PUP-ITIL2 is required for bio-DE28 labeling. This experiment was repeated independently more than three times with similar results. (b) Cell surface labeling increases with increased PUP-ITIL2. 1 × 105 Jurkat cells were first activated over night by anti-CD3 and anti-CD28. Cell surface Pup modification assays were set up the same as in Fig. 6. In 150 μl reaction volume, IL2-FKBP was added to reach the final concentration at 0.5, 1, 2, 4, and 14.8 ng/μl. With increased IL2-FKBP protein, more cell surface labeling of bio-DE28 can be observed. This experiment was repeated independently twice with similar results. (c) Representative flow cytometer data from three experimental repeats as shown in Fig. 6d. (d) The gating strategy used in (a)-(c).

  12. Supplementary Figure 12 PUP-IT is more active than BioID both in vitro and in cells.

    (a) Recombinant purified GST-BirA* (R118G, BioID) and GST-PafA (PUP-IT) were analyzed with coomassie stain. (b) In vitro modification assay of BioID and PUP-IT to compare the activity. For BioID reaction, the 40 μl reaction contains 1 μM GST-BirA* and 50 μM biotin. For PUP-IT reaction, the 40 μl reaction contains 1 μM GST-PafA and 10 μM bio-DE28. The experiments were carried out at 37 °C in the presence of ATP and Mg2+. At each time point, 10 μl reaction mixture were added to SDS loading dye to stop the reaction. Anti-GST blot shows the enzyme level while the streptavidin blot shows the product level. This experiment was repeated independently twice with similar results. (c) BirA* and PafA are fused to the C terminus of CD28 individually to generate BioIDCD28 and PUP-ITCD28 with myc tag. Jurkat cells are transiently transfected with either BioID or PUP-IT (with bio-PupE). BioID transfected cells were cultured in medium supplemented with 50 μM biotin. 24 h after transfection, cells are harvested and lysed for immune-blot analysis. This experiment was repeated independently three times with similar results. Original full scans of blots are shown in Supplementary Fig. 15.

  13. Supplementary Figure 13 Side-by-side comparison of BioIDCD28 and PUP-ITCD28 to identify CD28-interacting proteins.

    (a) Volcano plot for BioIDCD28 and BioIDCD28 tailless comparison (Supplementary Table 5a). (b) Volcano plot compare PUP-ITCD28 and PUP-ITCD28 tailless comparison (Supplementary Table 5b). In (a) and (b), the logarithmic ratios of protein LFQ-intensity was plotted against negative logarithmic P value of a two-sided two samples t-test. The green and blue dots represent significantly enriched proteins from common hits (FDR ≤ 0.05, n = 4, 2 biologically independent samples in 2 independent experiments). (c) BioID background protein analysis. The common proteins (n = 688 proteins) that are identified in BioIDCD28 WT, BioIDCD28 tailless and BioIDCD28 5AA were analyzed the same way as in supplementary Fig. 8.

  14. Supplementary Figure 14

    Unprocessed scans of original blots shown in the main text figures, and a plot with individual data points.

  15. Supplementary Figure 15

    Unprocessed scans of original blots shown in the Supplementary Figures.

Supplementary information

  1. Supplementary Text and Figures

    Supplementary Figures 1–15, Supplementary Note

  2. Reporting Summary

  3. Supplementary Protocol

    Identification of interacting proteins using PUP-IT

  4. Supplementary Table 1

    GGE modification site mapping

  5. Supplementary Table 2

    MS/MS count analysis of proteins highly enriched in PUP-IT(CD28)

  6. Supplementary Table 3

    Identification of interacting proteins using iPUP stable cell line

  7. Supplementary Table 4

    MS/MS count analysis of proteins unique to each PUP-IT

  8. Supplementary Table 5

    Direct comparison between BioID and PUP-IT

About this article

Publication history




Issue Date


Further reading