Genetic, structural, and chemical insights into the dual function of GRASP55 in germ cell Golgi remodeling and JAM-C polarized localization during spermatogenesis

Abstract

Spermatogenesis is a dynamic process that is regulated by adhesive interactions between germ and Sertoli cells. Germ cells express the Junctional Adhesion Molecule-C (JAM-C, encoded by Jam3), which localizes to germ/Sertoli cell contacts. JAM-C is involved in germ cell polarity and acrosome formation. Using a proteomic approach, we demonstrated that JAM-C interacted with the Golgi reassembly stacking protein of 55 kDa (GRASP55, encoded by Gorasp2) in developing germ cells. Generation and study of Gorasp2-/- mice revealed that knock-out mice suffered from spermatogenesis defects. Acrosome formation and polarized localization of JAM-C in spermatids were altered in Gorasp2-/- mice. In addition, Golgi morphology of spermatocytes was disturbed in Gorasp2-/- mice. Crystal structures of GRASP55 in complex with JAM-C or JAM-B revealed that GRASP55 interacted via PDZ-mediated interactions with JAMs and induced a conformational change in GRASP55 with respect of its free conformation. An in silico pharmacophore approach identified a chemical compound called Graspin that inhibited PDZ-mediated interactions of GRASP55 with JAMs. Treatment of mice with Graspin hampered the polarized localization of JAM-C in spermatids, induced the premature release of spermatids and affected the Golgi morphology of meiotic spermatocytes.

Fig 1. GRASP55 interacts with junctional adhesion molecules in a PDZ-dependent manner.

(A) Schematic diagram of GRASP55 and JAM-A/B/C constructs. JAMsΔ mutants correspond to wild-type sequences lacking the last three C-terminal amino acids forming PDZ-binding motifs. GRASP55Δ lacks the first PDZ domain and GRASP55 PDZ12 corresponds to the PDZ domain tandem repeat. (B) Characterization of GRASP55/JAM interacting domains by yeast two-hybrid using pACT2 as a negative control. (C) Immunoblot for GRASP55 after peptide pull-down of testis lysates using the indicated peptides. (D) Representative curves obtained by homogenous time-resolved fluorescence (HTRF) using GST-GRASP55 FL and the indicated biotinylated peptides competed with unlabeled peptides. ΔF is calculated as the ratio of signals obtained for acceptor (665nm)/donor (620nm). ΔF0 corresponds to ΔF_max obtained in absence of competitor. (E) Representative curves obtained by isothermal titration calorimetry (ITC) of GST-GRASP55 FL with JAM-B and JAM-C.

Fig 2. Defective spermatogenesis and acrosome formation in Gorasp2 mutant mice.

(A) Sections of epididymis from control littermate and Gorasp2^-/- mice stained with hematoxylin and eosin. Note that most of the mutant lumens are empty and that some of them contain large round cells instead of differentiated spermatozoa (inserts). Scale bars: top panels, 100 μm; bottom panels, 50 μm; inserts, 10 μm. (B) Sections of testes from 35 days-old control and Gorasp2^-/- mice stained for the acrosomal protein SP56. Scale bar, 100 μm. (C) High magnification images of knock-out and control testes sections stained with anti-SP56 and peanut agglutinin (PNA). Note that SP56 staining is more restricted than PNA staining. Scale bar, 10 μm.

Fig 3. Defective polarized localization of JAM-C in spermatids of Gorasp2^-/- mice.

(A) Confocal images of GRASP55, JAM-C, peanut agglutinin (PNA) and DAPI staining of seminiferous tubule sections of adult WT mice at stage I, II-III, VIII, X and XII. Staging was performed according to Nakata. H et al [26]. Co-polarized distribution of GRASP55 and JAM-C with partial overlap is found in step 2–8 spermatids present in stage II-VIII seminiferous tubules (arrowheads). Note that JAM-C staining overlaps with PNA acrosomal staining in some spermatids. Non-polarized membrane expression of JAM-C is found in spermatogonia (asterisks). Scale bar, 20 μm. (B) Co-immunoprecipitation of GRASP55 with JAM-C from testis lysates. Western blots were probed sequentially as indicated. Arrowheads and asterisk indicate the bands corresponding respectively to JAM-C and GRASP55. (C) Left panels: Low magnification confocal images of GRASP55, JAM-C, peanut agglutinin (PNA) and DAPI staining of seminiferous tubule sections of adult Gorasp2^+/+ and Gorasp2^-/- mice at stage V and VIII. Due to changes in PNA staining, seminiferous tubule staging in knock-out animals was based on morphological criteria. Scale bar: 100 μm. Right panels: High magnification confocal images of GRASP55, JAM-C, and DAPI showing the loss of polarized localization of JAM-C in spermatids of Gorasp2^-/- adult mice as compared to controls (arrowheads). JAM-C expression in spermatogonia is present in Gorasp2^-/- mice (asterisks). Scale bar, 20 μm.

Fig 4. Genetic ablation of Gorasp2 affects Golgi morphology.

(A) Confocal images of GM130, GRASP55 and DAPI staining of seminiferous tubule sections of P35 Gorasp2^+/+ and Gorasp2^-/- mice at stage II-III, V, VIII and XII. Arrowheads indicate less compact and more diffuse Golgi in spermatocytes from Gorasp2^-/- mice as compared to littermate controls. Scale bar, 10 μm. (B) Pseudo-colored images of GM130 staining as a function of Golgi area. The threshold value was set to 5μm² based on analysis of Golgi area distribution in function of seminiferous tubule stages. The automatic pipeline for the functions was encoded in Matlab and is available upon request (arnauld.serge@inserm.fr). Representative pseudo-colored images for indicated tubule stages and genotypes are shown. Arrowheads indicate Golgi with area greater than 5 μm2. Note that Golgi complexes with large areas are essentially localized in the periphery of seminiferous tubules. Scale bar, 50 μm. (C) Graph showing the percentages of Golgi with areas greater than 5 μm2 in function of seminiferous tubule differentiation stages. Images used for quantitative analysis were obtained from three independent samples and the number of tubules analyzed was as follow: II-III, n = 27; IV, n = 8; VI-VIII, n = 37; IX, n = 16; X-XII, n = 41. ***: p <0.001. (D) Confocal images of GM130, SYCP3 and DAPI staining of seminiferous tubule sections of P35 Gorasp2^+/+ and Gorasp2^-/- mice at stage II-III. Scale bar, 50 μm. Inserts: High magnification pictures highlighting GM130 staining in pachytene spermatocytes expressing SYCP3.

Fig 5. Co-Crystal structure of GRASP55 with JAM peptides.

(A) Superimposition of the previously reported ‘ligand-free’ GRASP55 PDZ12 X-ray structure (PDB ID: 3RLE) with ‘cargo-bound’ complex, showing GRASP55 PDZ12 structure bound to JAM-B (PDB ID: 5GMI). The ‘ligand free’ published GRASP55 PDZ12 structure is displayed as yellow (PDZ1) and black (PDZ2) ribbons, while the ‘cargo-bound’ conformation is shown as yellow (PDZ1) and salmon (PDZ2); the JAM-B peptide is indicated in blue. The arrow highlights the 33-degree global rotation of the PDZ2 domain after aligning both PDZ1 domains. (B) Surface representation of the GRASP55 PDZ12/JAM-B complex structure showing the JAM-B peptide embedded in the groove formed by the GRASP55 PDZ1 (yellow) and PDZ2 (salmon) domains. (C) LigPlot+ representation of the complex pinpointing the hydrogen bonding interaction contacts (green) as well as the van der Waals and hydrophobic contacts (semicircles). Green stars (*) highlight carboxylic hydrogen bond contacts in the binding pocket, and green cross circles (¤) indicate β-sheet hydrogen bonding interactions. Conserved contact residues in the JAM-B and JAM-C complex structures are highlighted in red. (D) Detailed view of the Arg101 hydrogen bonding interaction network with the PDZ2 domain (salmon) and JAM-B (blue), as revealed by the X-rays structure. (E) LigPlot+ representation of the PDZ1 Arg101 hydrogen bonding network with JAM-B and PDZ2. The color coding is the same as that described in (C).

Fig 6. Characterization of a small compound inhibiting GRASP55 interaction with JAMs.

(A) 2D structure, PubChem Common Identifier (CID) and given name of the prioritized hit from the structure-based drug design approach. (B) Representative curves were obtained by HTRF using GST-GRASP55 FL or GST-Erbin with the indicated biotinylated peptides and competed with Graspin (IC₅₀ = 8.4 μM for GRASP55/JAM-B and 12 μM for GRASP55/JAM-C). (C) Fluorescence profiles presenting results obtained by differential scanning fluorimetry (DSF). Shifts in the melting temperatures of His-GRASP55 PDZ12 incubated with a twelve molar equivalent of ligand or DMSO as control are shown. (D) Golgi density of Gorasp2^+/+ and Gorasp2^-/- MEFs treated or not with Graspin at 50 μM during 48 h. Each circle represents one Golgi. Data are the mean ± s.e.m. of pooled results of three independent experiments (analysis of 30–90 Golgi per condition, per experiment). Student’s unpaired t-test; ***: P<0.001. (E) Left panel, immunoblot for GRASP55 and JAM-C following exposure of Gorasp2^+/+ and Gorasp2^-/- MEFs to increasing doses of Graspin. Right panel, quantification of JAM-C expression in Gorasp2^+/+ and Gorasp2^-/- MEFs treated with Graspin at 50 μM for 48 h. Data are expressed as mean of JAM-C inhibition as compared to control condition (vehicle) in five independent experiments. Student’s unpaired t-test; *: P<0.05.

Fig 7. Chemical inhibition of GRASP55 affects spermatid differentiation in vivo.

(A-B) Flow cytometry profiles of DAPI-stained germ cells isolated from testes of Gorasp2^+/+ and Gorasp2^-/- 35-days old mice (A) and 27-days old mice treated with vehicle or Graspin for two weeks (B). ES: elongated spermatids, RS: round spermatids, 2C: spermatogonia, 4C: primary spermatocytes. (C) Quantification of germ cell numbers of 27-days old mice treated with Graspin for two weeks and analyzed two days after the last Graspin injection. Vehicle, n = 10; Graspin, n = 10. Student’s unpaired t-test; *: P<0.05. (D) Confocal images of JAM-C, GRASP55 and DAPI staining of seminiferous tubule sections of 43-days old mice treated for 16 days with vehicle or Graspin. Differentiation stages of seminiferous tubules are indicated. High magnification pictures on the left panel highlight the loss of co-polarized localization of GRASP55 and JAM-C in developing round spermatids. Scale bars: left panels: 50 μm; right panels: 10 μm. (E) Quantification of co-polarized JAM-C and GRASP55 staining occurrence in control and treated mice expressed as the percentage of round spermatids showing a close apposition between GRASP55 and JAM-C stainings. Results obtained upon quantification of co-polarized staining occurrences in seminiferous tubules at stage V-VI (n = 415 counted events, n = 4 mice, 2 tissue sections/mice) and VIII (n = 498 counted events, n = 6 mice, 2 tissue sections/mice) are shown. Student’s unpaired t-test; **: P<0.01, ***: P<0.001. (F) Confocal images of SP56 and DAPI staining of seminiferous tubule sections of 43-days old mice treated for 16 days with vehicle or Graspin as indicated. High magnification pictures on the right panels highlight the loss of SP56 staining associated with elongated spermatids. Scale bars: 200 μm.

Fig 8. Graspin treatment induces germ cell release in epididymis.

(A) Quantification of epididymis content from vehicle and Graspin treated mice using flow-cytometry. Student’s unpaired t-test; *: P<0.05. (B) Cytospin of material recovered from epididymis of vehicle and Graspin treated mice and stained for DAPI and PNA. Note the increased abundance and heterogeneity of material recovered from Graspin treated animals. Scale bar, 100μm (C) Representative high magnification confocal images of spermatozoa recovered from epididymis of vehicle or Graspin treated mice and stained for DAPI (blue), JAM-C (green) and PNA (grey). Note that the co-localization of JAM-C with PNA is lost in samples from Graspin treated mice. Scale bar, 5μm

Fig 9. Graspin treatment affects Golgi morphology of germ cells.

(A) Pseudo-colored images of GM130 staining as a function of Golgi area. The threshold value was set to 5μm² based on analysis of Golgi area distribution in function of seminiferous tubule stages. Representative pseudo-colored images for indicated tubule stages and genotypes are shown. Arrowheads indicate Golgi with area greater than 5 μm2. Note that Golgi with large areas are essentially localized in the periphery of seminiferous tubules. Scale bar, 50 μm. (B) Graph showing the percentages of Golgi area greater than 5 μm2 in function of seminiferous tubule differentiation stages. Images used for quantitative analysis were obtained from three independent samples and the number of tubules analyzed was as follow: II-III, n = 30; IV, n = 12; VI-VIII, n = 47; IX, n = 16; X-XII, n = 28. *: p <0.05. (C) Confocal images of GM130, SYCP3 and DAPI staining of seminiferous tubule sections of P35 treated mice at stage II-III. Scale bar, 50 μm. Inserts: High magnification pictures highlighting GM130 staining in pachytene spermatocytes expressing SYCP3.