STEM tomography reveals that the canalicular system and α-granules remain separate compartments during early secretion stages in blood platelets

Abstract

ESSENTIALS: How platelets organize their α-granule cargo and use their canalicular system remains controversial. Past structural studies were limited due to small sampling volumes or decreased resolution. Our analyses revealed homogeneous granules and a closed canalicular system that opened on activation. Understanding how platelets alter their membranes during activation and secretion elucidates hemostasis.

Background: Platelets survey the vasculature for damage and, in response, activate and release a wide range of proteins from their α-granules. Alpha-granules may be biochemically and structurally heterogeneous; however, other studies suggest that they may be more homogeneous with the observed variation reflecting granule dynamics rather than fundamental differences.

Objectives: Our aim was to address how the structural organization of α-granules supports their dynamics.

Methods: To preserve the native state, we prepared platelets by high-pressure freezing and freeze-substitution; and to image nearly entire cells, we recorded tomographic data in the scanning transmission electron microscope (STEM).

Results and conclusions: In resting platelets, we observed a morphologically homogeneous α-granule population that displayed little variation in overall matrix electron density in freeze-substituted preparations (i.e., macro-homogeneity). In resting platelets, the incidence of tubular granule extensions was low, ~4%, but this increased by > 10-fold during early steps in platelet secretion. Using STEM, we observed that the initially decondensing α-granules and the canalicular system remained as separate membrane domains. Decondensing α-granules were found to fuse heterotypically with the plasma membrane via long, tubular connections or homotypically with each other. The frequency of canalicular system fusion with the plasma membrane also increased by about three-fold. Our results validate the utility of freeze-substitution and STEM tomography for characterizing platelet granule secretion and suggest a model in which fusion of platelet α-granules with the plasma membrane occurs via long tubular connections that may provide a spatially limited access route for the timed release of α-granule proteins.

Keywords: blood platelets; cytoplasmic granules; electron microscope tomography; hemostasis; platelet activation.

Figure 1. Resting platelets prepared using sodium citrate (A) or acid citrate dextrose (B) were uniformly discoid in shape and contain α-granule population that are predominantly oval in shape

Blood was collected by venipuncture into either sodium citrate or acid citrate dextrose anticoagulant. An equal volume of 6% paraformaldehyde, 0.2% glutaraldehyde in PBS was added immediately to the blood as chemical fixative. Platelets were then isolated by centrifugation, and prepared by conventional dehydration for EM. Thin sections, 50 nm, were post-stained with lead citrate and uranyl acetate, and imaged at 3,500x. A: citrate blood draw, inset 2x magnification. B: ACD blood draw, inset 2x magnification. Black arrow in A inset points to a tubular α-granule extension. White arrow points to pseudopodal bulge. C, D: Quantification of platelet shape, frequency of pseudopodal bulges, and granule tubular extensions. Error bars in C,D represent standard error of the mean. Pairwise P values for Citrate versus ACD results are C: P = 0.60, 0.67 and 0.80; D: 0.80 and 0.98 indicating that the differences are small and not statistically significant. Collecting blood into either anticoagulant maintains resting platelets as indicated by discoid shape and limited pseudopodal bulges.

Figure 2. The α-granule matrix appeared morphologically homogeneous in platelets dehydrated by freeze substitution

Blood was drawn into sodium citrate, immediately mixed with chemical fixative and fixed, resting platelets isolated by centrifugation. The platelets were then either dehydrated conventionally through a serious of graded alcohols (A) or by FS following HPF to yield non-disruptive ice (B). We refer to a chemical fixation followed by HPF/FS dehydration as a hybrid protocol. All steps before and after the hybrid HPF/FS protocol dehydration paralleled each other for both sets of samples. Thin-sections were cut at 50 nm and imaged at 9,600x. A: conventional dehydration, 2x inset to right. Nucleoids were apparent in a large portion of the granules with detail shown in the inset. Inset: Mito indicates mitochondria; * indicates putative Golgi apparatus, white arrow points to retracted granule matrix. B: Alpha-granules in hybrid FS dehydrated platelets displayed a much more morphologically homogeneous matrix with no contraction from the granule membrane, 2x inset shown to right. C: Tilt series were acquired a magnification of 9600x of hybrid HPF/FS protocol dehydrated platelets sectioned at a thickness of 300 nm. Reconstructions of sufficiently high quality were computed with the IMOD software package (University of Colorado) using the weighted back projection (WBP) algorithm rather than the simultaneous iterative reconstructive technique (SIRT), which requires a significantly longer processing time. Arrows point to nearly morphologically homogeneous α-granules shown in the XY and XZ dimensions. D: Quantification of the frequency of nucleoids in granule profiles in platelets conventionally or hybrid HPF/FS protocol dehydrated. The P value of <0.0003 indicates that the difference between the two preparations is highly significant.

Figure 3. Alpha-granule tubular extensions increased upon platelet activation

Platelets were chemically fixed at different steps in the isolation procedure to give platelet preparations at different stages of the activation process. All platelets were dehydrated by the hybrid HPF/FS protocol. In brief, blood was drawn, a portion fixed immediately, and platelets isolated (A, resting platelets, fixed at blood drawn). Other portions were fixed either at the pelleted PRP stage (not shown) or at the pelleted first wash stage (B, no incubation). A final portion was incubated for 5 min at room temperature before centrifugation and fixed at the pelleted first wash step (C, 5 min incubation. An expanded view of C (right hand frame) shows detail of the tubular extensions, arrow points to an example tubular extension, Asterisks, C, mark decondensing α-granule. D: Platelet granule profiles were quantified for tubular extensions at different stages in the isolation procedure. Error bars represent standard error of the mean. Delayed fixation, particularly with a pre-centrifugation incubation of the freshly drawn citrated blood resulted in increased frequency of tubular extensions protruding from the α-granule.

Figure 4. Tubular pipes or necks support initial granule fusion with the plasma membrane

Citrate drawn blood was incubated for 5 min to initiate activation and then PGI2 and apyrase were used as suppressors to “lock” activation state as described in Methods. Isolated platelets were “fixed” for electron microscopy by HPF and dehydrated using FS. A: Low magnification image (imaged at 3500x) shows in thin section that the secretion-active platelets are at an early step of activation with the platelets rounded, extending pseudopods and the α-granules frequently decondensing. B: A 2x expanded view showing left-to-right a pseudopod (white arrow), example individual pipes extending from decondensing α-granules and fusing with the plasma membrane (black arrows) and an example of homotypic fusion between two α-granules of different size. C: Example of a decondensing α-granule connected directly to the plasma membrane by a short pipe, a neck. The pipe appears to contain decondensing cargo. (D) A decondensing α-granule showing abundant decondensed cargo in the pipe.

Figure 5. The structural features of pipes were more readily observed by electron tomography

Secretion-active platelets were isolated as described in the legend to Figure 4 and “fixed” for electron microscopy by HPF and dehydrated by FS (see Methods). 300 nm electron tomograms were generated from samples of secretion-active platelets and rendered in three-dimensions to illustrate organelle architecture. The methodology for tomogram reconstruction is detailed in the legend to Figure 2. Yellow = OCS, blue = condensed granules (not yet activated), gold-orange = decondensing granules (activated), and green = plasma membrane. A, B, C, D, E: White arrows points to a pipe extending from a decondensing granule shown first as a tomogram slice and later as the rendered model (D) and then as the plasma membrane enclosed platelet (E). In sum, the various white arrows to the same pipe that was then traced to its connection with the plasma membrane. As shown, the 300 nm section thickness was insufficient to image the full granule depth.

Figure 6. By STEM tomography, pipes were the only detectable exit route from decondensing α-granules

Secretion-active platelets were isolated as above and “fixed” by HPF and then dehydrated by FS. STEM tomograms of 1500 nm sections were generated from samples of plastic-embedded resting (A) and activated (B) platelets. The methodology for tomogram reconstruction is detailed in the legend to Figure 2. Platelets and the included organelles were rendered in 3-dimensions. A: rendered α-granules in a resting platelet. B: morphology of an individual slice of a 1500 nm thick tomogram showing a pipe (red arrow) connecting an αgranule to the plasma membrane. C: Rendered decondensing granules (gold-orange) are overlaid on the same individual tomogram slice as B. White arrowhead points to an example of homotypic fusion between two α-granules. D: Three-dimensional rendering of condensed (blue) and decondensing granules (orange-gold) during initial secretion activation. Decondensing α-granules are located to the periphery of the cell and form pipes that are the only exit route to the plasma membrane in these fully 3-dimensionally imaged organelles. Color code: Blue = condensed granules (not yet activated), gold-orange = decondensing granules (activated).

Figure 7. Heterotypic fusion of the CS and α-granules with the plasma membrane was not accompanied by detectable fusion between the CS and decondensing α-granules

Secretion-active platelets were isolated , “fixed” by HPF and FS dehydrated as described in Methods. STEM tomograms of 1,500 nm sections were generated from samples of plastic-embedded resting (A, C, E) and activated platelets (B, D, F). The methodology for tomogram reconstruction is detailed in the legend to Figure 2. Platelets and included organelles were rendered in three-dimensions to yield near full platelet depth models. A: Rendered model of the organelles within a resting platelet including α-granules (blue), CS, (yellow), dense granules (red) and mitochondria (pink). Note that each organelle class is distinct with even the CS exhibiting limited interconnections. B: Rendered model of the organelles within a early stage, secretion-active platelet. Note that the α-granules are rendered as two subclasses, decondensing (gold-orange) and condensed (blue), based on the appearance of the granule matrix. The red arrow in B points to an example of a pipe that connects α-granules to the plasma membrane (B, best seen in Supplemental video 5). No heterotypic fusions between α-granules and the CS were detected. in the fully rendered secretion-active platelet (B). C,D: Plasma membrane surface view (green) of the resting (C) or secretion-active platelet (D). Note that only a small number of yellow CS continuities are present in the model of the resting platelet (C) while a substantial increase is apparent in the secretion active platelet (D). The red arrow in D points to an example pipe fusing with the plasma membrane. E,F: Simplified rendering of the resting platelet (E) and the secretion-active platelet showing only elements of the CS system. Here the CS elements are subclassified based on tracing within the 1,500 nm tomograms, whether there is any detectable continuity with the plasma membrane. Closed CS elements are shown in turquoise and plasma membrane connected CS elements are shown in yellow. Note that CS elements are significantly more interconnected in the secretion-active platelet (compare E,F). Many isolated, closed CS elements are found in the resting platelet and few in the early stage, secretion-active platelet. Color code: Green = plasma membrane, yellow = CS, blue = condensed granules (not yet secreting), gold-orange = decondensing α-granules (activated), red = dense granules, and pink = mitochondria.