PTP-MEG2 regulates quantal size and fusion pore opening through two distinct structural bases and substrates

Abstract

Tyrosine phosphorylation of secretion machinery proteins is a crucial regulatory mechanism for exocytosis. However, the participation of protein tyrosine phosphatases (PTPs) in different exocytosis stages has not been defined. Here we demonstrate that PTP-MEG2 controls multiple steps of catecholamine secretion. Biochemical and crystallographic analyses reveal key residues that govern the interaction between PTP-MEG2 and its substrate, a peptide containing the phosphorylated NSF-pY⁸³ site, specify PTP-MEG2 substrate selectivity, and modulate the fusion of catecholamine-containing vesicles. Unexpectedly, delineation of PTP-MEG2 mutants along with the NSF binding interface reveals that PTP-MEG2 controls the fusion pore opening through NSF independent mechanisms. Utilizing bioinformatics search and biochemical and electrochemical screening approaches, we uncover that PTP-MEG2 regulates the opening and extension of the fusion pore by dephosphorylating the DYNAMIN2-pY¹²⁵ and MUNC18-1-pY¹⁴⁵ sites. Further structural and biochemical analyses confirmed the interaction of PTP-MEG2 with MUNC18-1-pY¹⁴⁵ or DYNAMIN2-pY¹²⁵ through a distinct structural basis compared with that of the NSF-pY⁸³ site. Our studies thus provide mechanistic insights in complex exocytosis processes.

Keywords: PTP-MEG2; catecholamine; exocytosis; structure; tyrosine phosphorylation.

Figure 1. Inhibition of PTP‐MEG2 reduced the spike amplitude and “foot” probability of catecholamine secretion

1. A–D

   The epinephrine (A, B) or norepinephrine (C, D) secreted from the adrenal medulla was measured by ELISA method after stimulation with high KCl (70 mM) (A, C) or angiotensin II (AngII, 100 nM) (B, D) for 1 min, with or without pre‐incubation with a specific PTP‐MEG2 inhibitor (400 nM) for 1 h.

2. E, F

   Amperometric spikes of primary mouse chromaffin cells induced by AngII (100 nM) were determined by electrochemical experiments after incubation with PTP‐MEG2 inhibitor at different concentrations.

3. G

   The distribution of the quantal size of AngII (100 nM)‐induced amperometric spikes by primary chromaffin cells after incubation with different concentrations of PTP‐MEG2 inhibitor. Histograms show the number of amperometric spikes of different quantal sizes.

4. H, I

   Secretory vesicles of primary chromaffin cells were examined by transmission electron microscopy after 100 nM AngII stimulation, with or without pre‐incubation with 400 nM Compound 7. Scale bars: (H, I), 150 nm. Red arrows indicate morphologically docked LDCVs, and green arrows stand for undocked LDCVs. PM represents plasma membrane.

5. J

   Vesicle numbers according to different distances from the plasma membrane were calculated in the presence or absence of 100 nM AngII or 400 nM Compound 7. PM represents plasma membrane.

6. K

   The percentage of vesicles with different diameters was measured after 100 nM AngII stimulation or incubation with 400 nM Compound 7. Compound 7 significantly decreased vesicle size under AngII stimulation.

7. L

   The percentage of pre‐spike foot (PSF) (left panel) and stand‐alone foot (SAF) (right panel) was calculated after incubation with the indicated concentrations of Compound 7.

Data information: Data were analyzed using one‐way ANOVA and displayed as the mean ± s.e.m. (A–D, J–L) *P < 0.05; **P < 0.01; ***P < 0.001: cells stimulated with KCl or AngII compared with un‐stimulated cells. ^# P < 0.05; ^## P < 0.01; ^### P < 0.001: cells pre‐incubated with PTP‐MEG2 inhibitors compared with control vehicles. N.S. means no significant difference. Data were from 6 (A–D, J–K) or 8 (L) independent experiments. # indicates P < 0.05 and ### indicates P < 0.001 compared with control group.

Figure 2. Interaction of PTP‐MEG2 and tyrosine‐phosphorylated NSF in the medulla and the crystal structure of the PTP‐MEG2/phospho‐NSF‐segment complex

1. A

   NSF was phosphorylated after stimulation with AngII or KCl. Adrenal medulla cells were stimulated with 100 nM AngII or 70 mM KCl for 5 min and lysed. NSF was immunoprecipitated with a specific NSF antibody coated with Protein A/G beads. A pan‐phospho‐tyrosine antibody pY²⁰ was used in Western blot to detect the tyrosine‐phosphorylated NSF in the adrenal medulla under different conditions.

2. B

   NSF was co‐localized with PTP‐MEG2 in the adrenal medulla. After stimulation with 100 nM AngII, NSF and PTP‐MEG2 in the adrenal medulla were visualized with immunofluorescence. White arrow stands for co‐localization of the PTP‐MEG2 and NSF.

3. C

   Analysis of PTP‐MEG2 and NSF fluorescence intensities by Pearson’s correlation analysis. Pearson’s correlation coefficient was 0.7.

4. D

   Phosphorylated NSF interacted with PTP‐MEG2. PC12 cells were transfected with FLAG‐NSF. After stimulation of the cells with 100nM AngII in the presence or absence of 100 μM H₂O₂ for 5 min (H₂O₂ was exploited to increase the overall tyrosine phosphorylation of the cellular proteins), the potential PTP‐MEG2 substrate in cell lysates was pulled down with GST‐PTP‐MEG2‐D⁴⁷⁰A or GST control.

5. E

   The overall structure of PTP‐MEG2‐C⁵¹⁵A/D⁴⁷⁰A in complex with the NSF‐pY⁸³ phospho‐segment.

6. F

   The 2Fo‐Fc annealing omit map (contoured at 1.0σ) around the NSF‐pY⁸³ phospho‐segment. The P‐loop was highlighted in blue.

7. G

   Plot of distance RMSDs of individual residues between the crystal structures of the PTP‐MEG2/NSF‐pY⁸³ phospho‐peptide complex (PDB: 6KZQ) and the PTP‐MEG2 native protein (PDB: 2PA5).

8. H

   Superposition of the PTP‐MEG2/NSF‐pY⁸³ phospho‐peptide complex structure (red) on the PTP‐MEG2 native protein structure (PDB: 2PA5, blue). The structural rearrangement of the WPD loop and β3‐loop‐β4 is highlighted.

9. I, J

   The closure of the WPD loop and corresponding conformational changes in the inactive state (I) and the active state (J) of PTP‐MEG2. The rotation of R⁵²¹ leads to the movement of W⁴⁶⁸ and a corresponding 7 Å movement of the WPD loop.

10. K, L

    The structural rearrangement of β3‐loop‐β4 of PTP‐MEG2 in the active state (L) compared with the inactive state (K). The disruption of the salt bridge between E⁴⁰⁶ and R⁵²¹ contributed to the new conformational state of the N‐terminal of β3‐loop‐β4.

11. M, N

    The conformational change of the P‐loop of the inactive state (M) and the active state of PTP‐MEG2 in response to NSF‐pY⁸³ segment binding (N). The disruption of the charge interaction between R⁴⁰⁹ and D³³⁵ resulted in the movement of the main chain from G⁴⁰⁸ to K⁴¹¹.

Figure 3. Molecular determinants of PTP‐MEG2 interaction with the NSF‐pY⁸³ site

1. Schematic representation of interactions between PTP‐MEG2 and the NSF‐pY⁸³ site.

2. Relative values of the decreased phosphatase activities of different PTP‐MEG2 mutants toward pNPP (p‐Nitrophenyl phosphate) compared with wild‐type PTP‐MEG2.

3. Fold change decreases in the phosphatase activities of different PTP‐MEG2 mutants toward the NSF‐pY⁸³ phospho‐segment compared with wild‐type PTP‐MEG2.

4. Sequence alignment of PTP‐MEG2 from different species and with other PTP members. Important structural motifs participating in phosphatase catalysis or the recognition of the NSF‐pY⁸³ site are shown, and key residues contributing to the substrate specificity are highlighted.

Data information: ***P < 0.001: PTP‐MEG2 mutants compared with the control group. Data were obtained from 3 independent experiments. N.S. means no significant difference. All the data were analyzed using one‐way ANOVA and displayed as the mean ± s.e.m. In (B, C), the red column indicates that the mutation sites have greater effects on the enzyme activities toward phospho‐NSF segment than the pNPP. The green column indicates that the mutation sites only have significant effects on the enzyme activity toward NSF phospho‐segment but not pNPP. The white column stands for the mutants that have similar effects on pNPP and NSF phospho‐segment.

Figure 4. The tip opening of the pY + 1 pocket is critical for PTP‐MEG2 substrate specificity

1. Surface representations of the complex structures of PTP‐MEG2‐C⁵¹⁵A/D⁴⁷⁰A/NSF‐pY⁸³, PTP1B‐C²¹⁵A/EGFR‐pY⁹⁹² (PDB: 1EEN), LYP‐C²²⁷S/SKAP‐HOM‐pY⁷⁵ (PDB: 3OMH), PTPN18‐C²²⁹S/HER2‐pY¹²⁴⁸ (PDB: 4GFU), and PTP‐STEP/pY (PDB: 2CJZ). Crystal structures of PTP‐MEG1 (PDB: 2I75), SHP1 (PDB: 4HJQ), PTPH1 (PDB: 4QUN), and SHP2 (PDB: 3B7O). The pY + 1 sites are highlighted.

2. Summary of the distance between the Cβ atoms of D³³⁵ and Q⁵⁵⁹ (corresponding to PTP‐MEG2 number) of the pY + 1 pocket in PTP‐MEG2 and other classical non‐receptor PTPs bearing the same residues at similar positions.

Figure 5. Effects of different PTP‐MEG2 mutants on properties of catecholamine secretion from primary chromaffin cells

1. A

   Primary mouse chromaffin cells were transduced with a lentivirus containing the gene for wild‐type PTP‐MEG2 or different mutants with a GFP tag at the C‐terminus. Positive transfected cells were confirmed with green fluorescence and selected for electrochemical analysis. Typical amperometric current traces evoked by AngII (100 nM for 10 s) in the control (transduced with control vector) (top left panel), WT (top right panel), G³³⁴R (bottom left panel), and D³³⁵A (bottom right panel) are shown.

2. B

   The schematic diagram shows key residues of PTP‐MEG2 defining the substrate specificity adjacent to the pY⁸³ site of NSF.

3. C

   The distribution of the quantal size of chromaffin cells transduced with lentivirus containing the genes encoding different PTP‐MEG2 mutants.

4. D

   Statistical diagram of the quantal size in Fig 5A and Appendix Fig S6A–D. The secretion amount of each group was standardized with respect to the control group.

5. E–G

   Calculated parameters of secretory dynamics, including the spike amplitude (E), PSF frequency (F), and SAF frequency (G).

Data information: (B, D–G) Green color represents the amino acids which are found to modulate foot probability based on structure analysis, and red color means amino acids which does not regulate foot probability. (D–G), * indicates PTP‐MEG2 mutant overexpression group compared with the WT overexpression group. # indicates the overexpression group compared with the control group. **P < 0.01; ***P < 0.001 and ^# P < 0.05; ^## P < 0.01; ^### P < 0.001. N.S. means no significant difference. Data were from six independent experiments. All the data were analyzed using one‐way ANOVA and showed as the mean ± s.e.m.

Figure 6. Identification of potential candidate substrates of PTP‐MEG2 that participate in fusion pore initiation and expansion

1. A

   Flowchart for the workflow to predict the candidate substrates of PTP‐MEG2 during fusion pore initiation and expansion. A total of 51 proteins were enriched with the functional protein association network STRING and the text mining tool PubTator by searching the keywords “fusion pore”, “secretory vesicle” and “tyrosine phosphorylation”. These proteins were filtered with UniProt by selecting proteins located only in the membrane or vesicle, which resulted in 28 candidates. The Human Protein Atlas database was then applied to exclude proteins with no expression in the adrenal gland. Finally, we used the post‐translational‐motif database PhosphoSitePlus to screen candidate proteins with potential phospho‐sites that matched our sequence motif prediction at the pY + 1 or pY + 2 positions.

2. B

   After the bioinformatics analysis, a total of 12 candidate PTP‐MEG2 substrates that may participate in fusion pore initiation and expansion and their potential phospho‐sites were displayed.

3. C

   The GST pull‐down assay suggested that PACSIN1, MUNC18‐1, VAMP7, SNAP25, DYNAMIN1, and DYNAMIN2 directly interact with PTP‐MEG2. PC12 cells were transfected with plasmids of candidate substrates, including SYN1, MUNC18‐3, PACSIN1, SCAMP1, MUNC18‐1, PPP3CA, STX17, VAMP7, SYT7, SYT11, SNAP25, DYNAMIN1, and DYNAMIN2 stimulated with 100 nM AngII. The tyrosine phosphorylation of these proteins was verified by specific anti‐pY antibodies (Appendix Fig S8A–E). The potential substrates of PTP‐MEG2 in cell lysates were pulled down with a GST‐PTP‐MEG2‐D⁴⁷⁰A trapping mutant and then detected by Western blot. White arrow stands for Western blot band consistent with the predicted molecular weight of the potential PTP‐MEG2 substrate.

4. D–F

   Co‐immunostaining assays of PTP‐MEG2 with potential substrates in the adrenal medulla. MUNC18‐1, VAMP7, and DYNAMIN2 all showed strong co‐localization with PTP‐MEG2 after 100 nM AngII stimulation in the adrenal medulla. White arrow stands for co‐localization of PTP‐MEG2 with MUNC18‐1, VAMP7, or DYNAMIN2. Pearson’s correlation coefficients for (D, E, and F) were 0.61, 0.65, and 0.79 respectively. The co‐immunostaining results of PTP‐MEG2 with other potential substrates are shown in Appendix Fig S7.

Figure 7. Dephosphorylations of MUNC18‐1 at the pY¹⁴⁵ site and DYNAMIN2 at the pY¹²⁵ site by PTP‐MEG2 determine the “foot” probability of catecholamine secretion from chromaffin cells

1. A

   Structural representation and location of MUNC18‐1‐Y¹⁴⁵ in the complex structure of MUNC18‐1‐SYNTAXIN1 (PDB: 3PUJ).

2. B

   Detailed structural representation of MUNC18‐1‐Y¹⁴⁵ and its interactive residues. Y¹⁴⁵ of MUNC18‐1 interacts with the main chain carboxylic oxygen of residues I⁵³⁹ and G⁵⁶⁸ and forms hydrophobic interactions with L⁶, I⁵³⁹, L¹⁴⁷, L¹⁷⁹, L¹⁸³, and L²³⁰ to tether the arc shape of the three domains of MUNC18‐1 (PDB: 3PUJ).

3. C

   Association analysis of SNPs of MUNC18‐1 with human disease.

4. D

   Interactions of the PTP‐MEG2‐trapping mutants with the MUNC18‐1‐Y¹⁴⁵ mutants and DYNAMIN2‐Y¹²⁵ mutants. PC12 cells were transfected with FLAG‐MUNC18‐1‐Y¹⁴⁵, FLAG‐DYNAMIN2‐Y¹²⁵ and different mutations of the FLAG‐MUNC18‐1‐Y¹⁴⁵A, Y¹⁴⁵H, Y¹⁴⁵E or Y¹⁴⁵F, FLAG‐DYNAMIN2‐Y¹²⁵A, Y¹²⁵E, Y¹²⁵F, 24 h before stimulation with 100 nM AngII, respectively. The cell lysates were then incubated with GST beads‐PTP‐MEG2‐D⁴⁷⁰A complex for 2 h with constant rotation. The potential PTP‐MEG2 substrates were pulled down by GST beads, and their levels were examined by the FLAG antibody with Western blot.

5. E

   Primary chromaffin cells were transduced with lentivirus containing the gene encoding wild‐type MUNC18‐1 or different mutants. These cells were stimulated with 100 nM AngII. The amperometric spikes were detected with electrochemical experiments. Typical amperometric traces are shown.

6. F, G

   The percentages of pre‐spike foot (F) and stand‐alone foot (G) for wild‐type MUNC18‐1 or different mutants were calculated.

7. H

   The MUNC18‐1‐Y¹⁴⁵ mutations decreased the interaction between MUNC18‐1 and SYNTAXIN1. PC12 cells were transfected with plasmid encoding SYNTAXIN‐1. The proteins in cell lysates were pulled down with purified GST beads‐MUNC18‐1‐Y¹⁴⁵ complex and the GST‐MUNC18‐1‐Y¹⁴⁵H/E/F, and detected with SYNTAXIN1 antibody. The right histogram shows the quantified protein levels.

8. I

   Primary chromaffin cells were transduced with lentivirus containing the gene encoding wild‐type DYNAMIN2 or different mutants. These cells were stimulated with 100 nM AngII. The amperometric spikes were detected with electrochemical experiments. Typical amperometric traces are shown.

9. J, K

   The percentages of pre‐spike foot (J) and stand‐alone foot (K) for wild‐type DYNAMIN2 or different mutants were calculated.

10. L

    Plasmids encoding DYNAMIN2 (DYNAMIN2‐WT, DYNAMIN2‐Y¹²⁵E or DYNAMIN2‐Y¹²⁵F) and LYN‐YFP, AT1aR‐C‐RLUC were co‐transferred into HEK293 cells at 1:1:1 ratio. The BRET experiment was performed to monitor the mutation effects of DYNAMIN2 on AT1aR endocytosis in response to AngII (1 μM) stimulation.

11. M

    The GTPase assay was performed to detect the GTPase activity of the DYNAMIN2‐WT, DYNAMIN2‐Y¹²⁵E, and DYNAMIN2‐Y¹²⁵F. Purified DYNAMIN2 proteins were incubated with 5 μM GTP in the presence of 500 μM DTT for 90 min, followed by addition of GTPase‐Glo™ Reagent and detection reagent. The luminescence was measured after 37‐min incubation.

12. N

    Phosphorylated Y¹²⁵ of DYNAMIN2 was identified by LC‐MS/MS. The PTP‐MEG2‐D⁴⁷⁰A trapping mutant was used to pull down potential PTP‐MEG2 substrates from the adrenal lysates after AngII stimulation. Trypsin‐digested potential PTP‐MEG2 substrates were subjected to LC‐MS/MS analysis. The doubly charged peptide with m/z 725.06 matches VYSPHVLNLTLIDLPGITK of DYNAMIN2 with Y¹²⁵ phosphorylation.

Data information: (F–H, J–M) * indicates MUNC18‐1, DYNAMIN2 mutant overexpression group compared with the WT overexpression group. # indicates MUNC18‐1, DYNAMIN2 overexpression group compared with the control group. *P < 0.05; **P < 0.01; ***P < 0.001 and ^# P < 0.05; ^## P < 0.01; ^### P < 0.001. N.S. means no significant difference. Data were from 6 (F, G, J, K, L, M) or 3 (H) independent experiments. All the data were analyzed using one‐way ANOVA, and showed as the mean ± s.e.m.

Figure 8. The structural details of the PTP‐MEG2‐MUNC18‐1‐Y¹⁴⁵ interaction and catalytic bias of PTP‐MEG2 toward NSF‐pY⁸³ and MUNC18‐1‐pY¹⁴⁵

1. The 2Fo‐Fc annealing omit map (contoured at 1.0σ) around MUNC18‐1‐pY¹⁴⁵ phospho‐segment.

2. Comparison of residues of PTP‐MEG2 interacting with NSF and MUNC18‐1. Amino acid residues of NSF and MUNC18‐1 are colored as follows: green, residues interacting with both NSF and MUNC18‐1; red, residues specifically contributing to NSF recognition; and blue, residues selectively contributing to MUNC18‐1 interaction.

3. The structural alteration of the interactions surrounding Y⁴⁷¹ of PTP‐MEG2 with the MUNC18‐1‐pY¹⁴⁵ site.

4. The structural alteration of the interactions surrounding I⁵¹⁹ of PTP‐MEG2 with the MUNC18‐1‐pY¹⁴⁵ site.

5. Relative phosphatase activities of different PTP‐MEG2 mutants toward the MUNC18‐1‐pY¹⁴⁵ phospho‐segment compared with wild‐type PTP‐MEG2.

6. Relative phosphatase activities of different PTP‐MEG2 mutants toward the DYNAMIN 2‐pY¹²⁵ phospho‐segment compared with wild‐type PTP‐MEG2.

7. The percentages of PSF and SAF for PTP‐MEG2‐R⁴⁰⁹A or R⁴¹⁰A were calculated.

8. Schematic illustration of the PTP‐MEG2‐regulated processes of vesicle fusion and secretion in chromaffin cells via the dephosphorylation of different substrates with distinct structural basis. PTP‐MEG2 regulates vesicle fusion and vesicle size during the catecholamine secretion of adrenal gland by modulating the phosphorylation state of the pY⁸³ site of NSF, which relies on the key residues G³³⁴, D³³⁵ (pY loop), Y⁴⁷¹ (WPD loop), I⁵¹⁹ (P‐loop), and Q⁵⁵⁹ (Q loop). PTP‐MEG2 regulates the fusion pore initiation and expansion procedures of catecholamine secretion by the adrenal gland (also designated as foot probability) by modulating the newly identified substrate MUNC18‐1 at its pY¹⁴⁵ site and DYNAMIN2 at its pY¹²⁵, through distinct structural basis from that of its regulation of NSF phosphorylation.

Data information: (E, F) Residues are colored according to Fig 8B. (E, F), *P < 0.05, **P < 0.01, ***P < 0.001: PTP‐MEG2 mutants compared with the control. N.S. means no significant difference. Data were obtained from 3 independent experiments. (G), * indicates PTP‐MEG2‐R⁴⁰⁹A or R⁴¹⁰A mutant overexpression group compared with the WT overexpression group. # indicates the PTP‐MEG2 overexpression group compared with the control group. **P < 0.01; ***P < 0.001 and ^## P < 0.01; ^### P < 0.001. N.S. means no significant difference. Data were obtained from 10 cells in each group of 6 independent experiments. All the data were analyzed using one‐way ANOVA and showed as the mean ± s.e.m.