Figure 1.
Mossy fiber–Purkinje cell contacts in the developing mouse cerebellum.
(A) Pontine mossy fibers were selectively labeled by in utero electroporation at E14.5. Ventral view of a pup (P7) shows highly selective marking of cells in the pontine gray nucleus (see Figure S1 and Methods). The midline is indicated as a dotted line. (B) Dorsal view of cerebellum with mossy fiber axon projections in a P7 electroporated mouse. (C–E) GFP-labeled mossy fiber axons (green) projecting to the mediolateral cerebellum at P7. Purkinje cells are marked with anti-calbindin antibodies (Calb) in red. (F,G) Left panels: High-magnification views of mossy fiber–Purkinje cell contacts at P7 formed on the Purkinje cell soma (F) and the Purkinje cell axon (G). Rights panels show an orthogonal projection of a 0.45 µm thick section at the site of contact, showing the apposition of the cellular markers Calb (red) and GFP (green). (H, I) Mossy fiber–Purkinje cell contacts accumulate presynaptic markers. The EGFP-positive varicosities contain endogenous synaptic vesicle protein VAMP2 (H, blue) or synaptophysin-fluorescent protein fusion introduced by co-electroporation (I, blue) with EGFP. (J) In Thy1.2-EGFP-O mice a population of mossy fibers from multiple precerebellar nuclei is labeled at P14 (for details see Figure S1). At a subset of mossy fiber–Purkinje cell contacts the postsynaptic density marker Shank1a (blue) is concentrated. Scale bars: (C) 500 µm; (D) 50 µm; (E) 20 µm; (H,I) 5 µm; (J) 10 µm.
Figure 2.
Transient innervation of pontine mossy fibers into Purkinje cell territory during postnatal development.
(A) Schematic diagram showing position of DiI crystals insertion in the PGN (red triangle). Cerebellar hemispheres in which DiI was photoxidized forming dark precipitate are highlighted in blue. (B) Photoxidized DiI-labeled mossy fibers visualized in 50 µm coronal sections at P14 show typical mossy fiber morphologies. Red arrow in the left panel points at labeled mossy fiber axons traversing white matter (WM). Middle panel shows mossy fibers exiting the axon tract and arborizing in the IGL form multiple irregularly shaped varicosities (rosettes). Right panel outlines unlabeled cells in the cerebellar cortex, and their cellular relationships with labeled mossy fibers. Asterisks highlight examples of immature rosettes in the IGL characteristic during the second postnatal week [20]. Scale bars: middle panel 50 µm, right panel 10 µm. EGL, external granule cell layer; PCL, Purkinje cell layer; IGL, inner granule cell layer; WM, white matter; Sim, simplex lobule; mcp, medial cerebellar peduncle; v–IV, fourth ventricle; Pf, paraflocculus. (C) High magnification photograph of 7 µm semithin section from cerebellar cortex at P7, next to camera lucida drawing reconstructing mossy fiber axonal arborizations and their cellular relationships with Purkinje cell somata from 50 µm thick section of the same area. Several Purkinje cell somata are highlighted in blue. Since only a single plane of focus can be visualized in the photomicrograph, mossy fibers appear as discontinuous segments (delineated by asterisks). Reconstruction from multiple focal planes using camera lucida allows us to visualize continuous axonal arbors. Red arrowheads point at putative contacts on Purkinje cell somata. Mossy fiber segments and Purkinje cells that are not on the same focal plane as the photograph are grayed out in the camera lucida drawing. Scale bar: 10 µm. (D) Camera lucida drawings of the upper IGL and the PCL of DiI-calbindin double-labeled 50 µm sections from P0, P7, P14, and P21 time points. Purkinje cells are shown in blue. Red arrows point at putative contacts between mossy fibers and Purkinje cell somata. Scale bar: 10 µm. (E) Quantitative assessment of mossy fiber invasion into Purkinje cell territory from camera lucida drawings at P0, P7, P14, and P21 time points. Drawings encompass all of the mossy fiber segments and Purkinje cell outlines reconstructed from an area 175 µm horizontal×120 µm vertical×approximately 30 µm deep (thickness of one Purkinje cell soma), oriented parallel to the PCL, and recorded with a 100× objective. The percentage of all mossy fiber segments that enter the PCL out of all mossy fiber segments was scored (>1,000 segments from >25 areas obtained from 5–9 animals per time point, >3 50 µm sections per animal were analyzed). Statistical significance was determined by Mann-Whitney t test. Two-tailed p values were used with 95% confidence intervals, ** p<0.01, *** p<0.001.
Figure 3.
Transient mossy fiber–Purkinje cell contacts and synapses identified by correlated light-electron microscopy.
Mossy fiber–Purkinje cell contacts at P0, P7, P14, and P21 time points. Each panel consists of camera lucida drawing (top left), corresponding electron micrograph (EM), and a schematic drawn to scale highlighting major ultrastructural elements representing mossy fibers (orange), Purkinje cells (blue), glial processes (yellow), unlabeled terminals (presumably from PC collaterals, white), synaptic vesicles (small circles), and PSDs (demarcated by blue arrowheads). Contacts are highlighted by red arrowheads. Scale bars: camera lucida drawings 10 µm, electron micrographs 100 nm. (A) At P0 electron micrographs from the contact region verify direct cellular appositions between the mossy fibers and the Purkinje cells. By light microscopy mossy fibers form extensive putative contacts throughout the entire length of Purkinje cell somata. Even though mossy fiber segments apposed to Purkinje cell somata are densely filled with synaptic-like vesicles, no synaptic specializations can be seen by electron microscopy. (B) At P7 electron micrographs from the contact region verify direct cellular appositions between the mossy fibers and the Purkinje cells. These contacts occur en passant, are densely packed with synaptic vesicles, and exhibit ultrastructural characteristics of synapses: synaptic vesicles polarized towards PSD-like structures on the Purkinje cell (blue arrowheads). (C) At P14 mossy fiber segments typically do not invade more than 15 µm into PCL. By electron microscopy the diameter of mossy fibers at the sites of contact with Purkinje cell soma is smaller compared to P7. Purkinje cell somata become surrounded by thin glial process, identified by very light cytoplasm and presence of glycogen particles. (D) At P21, none of the potential mossy fiber–Purkinje cell contacts identified by LM could be verified by EM as direct contacts, even though in some cases mossy fiber terminals were positioned as close as 150 nm from Purkinje cell soma. (E) Quantitation of correlated light-EM analysis. More than 90% of putative mossy fiber–Purkinje cell contacts identified by LM are direct cellular appositions as verified by serial EM at P0 and P7. At P14, only 40% of putative contacts are direct cellular appositions, and at P21 mossy fiber–Purkinje cell contacts are completely removed. Total of 55 potential contacts analyzed by serial EM analysis (P0, n = 13; P7, n = 21; P14, n = 13; P21, n = 8; taken from 34 areas, from 12 animals). (F) The filling fraction was calculated for somatic mossy fiber–Purkinje cell contacts by dividing the number of synapses by the number of direct contacts identified in the EM analysis. Synapses were defined by the presence of clustered synaptic vesicles docked at the plasma membrane opposite a postsynaptic density in the Purkinje cell. Since no direct mossy fiber–Purkinje cell contacts are detected at P21 no filling fraction can be calculated for this developmental time point.
Figure 4.
Expression of BMP signaling molecules in the pontocerebellar system.
(A) In situ hybridization in the brainstem with BMPR1A, BMPR1B, and BMPR2 specific antisense probes. Images show an enlargement of sections through the pontine nucleus at P0 and P14. Scale bar: 200 µm. (B) In situ hybridization on cerebellar tissue (mediolateral cerebellum including SIM, Crus1, Crus2) at P0, P7, P14, P21 with a BMP4 specific antisense probe and sense control. Position of the external germinal layer (EGL), molecular layer (ML), Purkinje cell layer (PCL), and internal granular layer (IGL) are marked. Scale bar: 20 µm.
Figure 5.
Cerebellum-derived BMP signal negatively regulates growth of mossy fiber axons in vitro.
(A,B) Co-culture assay with pontine (P) and cerebellar (CB) explants. Axons are visualized with antibodies to the neuronal anti β3 tubulin subunit (Tuj1 antibody). Mossy fiber axon growth from PGN explants (P0+2 days in vitro) is reduced on the side facing cerebellar explant (panel A). Noggin (100 ng/ml) addition to the culture medium blocks the reduction in mossy fiber growth on the side facing cerebellar explant (B). Panels (A′) and (B′) show enlargements of the boxed areas in (A) and (B), respectively. (C) Collagen gel co-cultures of pontine explants (P) with BMP4-expressing HEK293 cell aggregates (BMP4-HEK) grown for 48 h. Panels in (C′) and (C″) are enlarged views of the boxed areas on the proximal and distal side with respect to the explant, respectively. (D) Quantification of axon density measured as Tuj1-positive area within fields 200 µm away from the explant. The measurements are shown as ratio of distal to proximal staining density. Growth inhibitory activity of cerebellar explants was counteracted by addition of 100 ng/ml noggin to the culture media (n = 29 explants from 3 experiments). (E) Quantification of axon density measured as Tuj1-positive area within fields 200 µm away from the explant. The measurements are shown as ratio of distal to proximal staining density. BMP4 axon growth inhibitory activity could be neutralized by addition of 100 ng/ml noggin to the culture media (n = 69 explants from 3 experiments). Statistical significance was determined by Mann-Whitney test. Two-tailed p values were used with 95% confidence intervals, ** p<0.01, *** p<0.001. Scale bars in (B) and (C): 200 µm; (B′) and (C″): 20 µm.
Figure 6.
Excessive mossy fiber–Purkinje cell contacts and synapses in the absence of Purkinje cell-derived BMP4.
(A) Camera lucida drawings of the upper IGL and the PCL of DiI-Calbindin double-labeled coronal sections from P0, P7, and P14 cerebellar hemispheres of littermate controls (BMP4fl/fl, upper panels) and BMP4 cKO mice (BMP4fl/fl::Pcp2cre/cre and BMP4fl/fl::Pcp2cre/+, lower panel). Purkinje cells are shown in blue. Red arrows point at putative contacts between mossy fibers and Purkinje cell soma. More mossy fiber segments extensively invade into the Purkinje cell territory in the BMP4 cKO tissue. (B) Quantitative analysis from camera lucida drawings of DiI-Calbindin double-labeled material showing significant increases in the percentage of mossy fiber segments entering the Purkinje cell territory and in the number of Purkinje cells receiving mossy fiber contacts (n>5,000 mossy fiber segments from 220 areas of 175 µm×120 µm×30 µm from 22 animals). Gray bars are control, black bars are cKO. ** p<0.01, *** p<0.001. (C) Elimination index for removal of mossy fibers from the PCL between P7 and P14 in control (gray) and BMP4 cKO mice (black bar). (D) Combinations of camera lucida drawings, schematic representations, and electron micrographs as in Figure 3. Two upper panels: At P7, many synaptic contacts in BMP cKO mice exhibit the same ultra-structural features as in the wild type. Lower left panel: Non-synaptic contacts between one mossy fiber segment and two neighboring Purkinje cells. Lower right panel: Some mutant contacts exhibit unusual ultra-structure with aggregated synaptic vesicles and ruffled plasma membranes at the contact site, characteristics never observed in wild-type or control mice. (E) Quantification of the density of EM-verified mossy fiber–Purkinje cell somatic contacts per 100 µm mossy fiber length in the PCL at P7 for BMP4fl/fl control mice (gray) and BMP4cKO mice quantified by light microscopy (black). The filling fraction was substantially reduced in the BMP4 cKO tissue. Based on the contact density and filling fraction, an estimated synapse density was calculated which shows no significant difference in control and cKO mice. Data were collected from >3 animals per genotype, >50 areas per animal, >12 mossy fiber segments per area. *** p<0.001.
Figure 7.
Mossy fiber defects in the mature cerebellum of BMP4 cKO mice.
Pontine neurons in BMP4 cKO (BMP4fl/fl::Pcp2cre/cre) mice or control mice (BMP4+/+::Pcp2cre/cre and BMP4fl/+::Pcp2cre/+) were marked by in utero electroporation at E14.5 and analyzed at postnatal day 21. (A) Low magnification views of coronal sections show restriction of GFP-labeled mossy fibers to the IGL and overshooting of fibers in the BMP4 cKO mice. The inverted images show the GFP-filled mossy fiber axons in black, and the bottom of the Purkinje cell layer is marked with a dashed line. (B) Quantification of overshooting mossy fiber axons in the molecular layer within 0–20 µm or >20 µm distance from the bottom of the PCL in control (gray bars) and cKO mice (black bars). Fibers overshooting more than 20 µm beyond the PCL are never observed in control cerebella (surveyed in 60 sections through Crus1, Crus2, and simplex lobules from 3 electroporated control animals in this experimental series and sections from >20 control animals in other experiments). ** p<0.01. (C) Cellular contacts of overshooting mossy fibers (green) in tissue double labeled for the Purkinje cell marker calbindin (red), the mature granule cell marker GABA α6 (magenta), or the Golgi cell marker neurogranin (magenta). The right panels show higher magnification views of Purkinje and Golgi cell contacts. Some Purkinje cell contacts identified by three-dimensional evaluation of high-resolution confocal image stacks are marked with white arrows. (D) Frequency of contacts of molecular layer mossy fiber axons in BMP4 cKO cerebella was quantified for contacts with neurogranin-positive Golgi cells, calbindin-positive Purkinje cells, and GABA alpha6-positive granule cells. Single fibers that contact Golgi as well as Purkinje cells are listed as a separate group. Note that there are neurogranin-negative Golgi cells, which might be associated with mossy fibers leading to a possible underestimation of molecular layer cell contacts (n = 93 fibers from tissue obtained from three different BMP4 cKO animals). (E) The position of mossy fiber rosettes in GFP-positive pontine axons with respect to the bottom of the Purkinje cell layer was quantified. The percentage of fibers for each 20 µm segment away from the PCL is plotted (n = 393 fibers for control and n = 638 fibers for BMP4 cKO from 3 animals per genotype). Gray bars are control, and black bars are BMP4 cKO. Scale bars: (A) left panels 100 µm, right panels 20 µm; (C) left three panels 20 µm, two right panels 10 µm. * p<0.05, ** p<0.01.