Tubulobulbar complexes (TBCs) are elongate subcellular machines responsible for internalizing intercellular junctions during sperm release. Each complex consists of a double-membrane tubular core terminating in a clathrin-coated pit. The core is surrounded by a network of actin filaments, and a distinct swelling or bulb, which lacks an association with actin, develops in the distal third of the structure. The bulb eventually buds from the complex and enters endocytic compartments of the Sertoli cell. The relationship of the actin cuff to the formation and budding of the bulb is not known. To gain insight into this relationship, we perturbed the actin networks of TBCs with cytochalasin D. When isolated testes were perfused with a physiological buffer containing cytochalasin D, apical TBCs at stage VII of spermatogenesis were associated with lower levels of actin compared to controls. At the ultrastructural level, the actin networks in cytochalasin D-treated testes appeared patchy, and ectopic bulbs and swollen tubular regions occurred. When normal untreated samples at early stage VII were analyzed, large elongate bulbs and short tubular sections were observed. Together, these results suggest a new model for TBC vesiculation in which the actin network begins to disassemble and the tubular region begins to swell into a bulb. As actin disassembly continues, the coated pit and most of the tubular region are incorporated into the enlarging bulb. The remaining short neck of the bulb near the base of the complex undergoes scission, and the bulb is internalized.
INTRODUCTION
Intercellular junction remodeling during spermatogenesis is critical to the normal production of spermatozoa by the mammalian testis. In basal regions of the seminiferous epithelium, massive junction complexes between neighboring Sertoli cells disassemble above and reassemble below the next generation of spermatogenic cells as they move from basal to adluminal compartments of the epithelium [1]. Tight junctions within these complexes contribute to the blood-testis barrier [2]. In apical regions, junctions between Sertoli cells and mature spermatids disassemble to release the latter cells, as spermatozoa, into the duct system, while at the same time new junctions form between Sertoli cells and early elongating spermatids deeper in the epithelium. Fundamental to the process of junction remodeling in the seminiferous epithelium are structures termed tubulobulbar complexes (TBCs).
TBCs are proposed to be subcellular machines that internalize intercellular junctions during junction turnover [34–5]. They form only at intercellular junctions in the testis, and their appearance precedes sperm release [6] and spermatocyte translocation [7]. Multiple generations of complexes form as junctions disassemble [8], and the structures often occur in clusters, particularly at apical sites of attachment between Sertoli cells and spermatids [9]. In rats, as many as two dozen complexes occur in Sertoli cell regions adjacent to the convex face of each mature hook-shaped spermatid head at stage VII of spermatogenesis prior to sperm release at stage VIII [6].
Each TBC consists of an elongate tubular extension of either a Sertoli cell or a spermatid that projects into a corresponding invagination of the adjacent Sertoli cell [6]. The double-membrane core of the complex is cuffed by an actin (ACT-) filament network and the distal end of the structure is capped by a clathrin (CLT-)-coated pit [10]. A swollen region or “bulb” forms near the end of the complex. The bulb, which lacks an actin cuff and is closely related to cisternae of endoplasmic reticulum, eventually loses its association with the complex and enters endocytic compartments of the Sertoli cell [6].
TBCs have some structural and molecular similarities to podosomes that occur in other cells [11] and to membrane tubules that develop in cell-free systems [12]. They also have a protein domain profile that mirrors that of clathrin-mediated endocytosis machinery present in cells generally [3, 4, 1314–15]. In fact, except for their double-membrane cores and the presence of bulbs, TBCs resemble clathrin-coated pits with very long necks.
The function of the actin filament cuffs at TBCs is not entirely clear, although a role in the formation and maintenance of tubular regions appears likely. Knocking down cortactin (CTTN), a key regulator of actin filament/membrane dynamics in other systems [16, 17], results in shorter TBCs than in controls [18]. When the actin filament disruptor cytochalasin D is injected into the testis, typical TBCs do not occur; rather, elongate tubular regions and bulbs are completely absent and only the double-membrane coated pits occur directly at sites of Sertoli cell attachment to spermatids [19]. Both of these results are consistent with the actin networks playing a role in the formation of TBCs. Details about how the bulbs form in relationship to the actin cuffs and how the bulbs “bud” from TBCs are not known. Also not known are the fates of the coated pits and the remaining tubular elements of the complexes, although, like the bulbs, they have been assumed to undergo some form of scission or vesiculation and to be internalized by the Sertoli cell [6, 20].
Here we revisit perturbing actin filament networks in the testis with cytochalasin D, but use the ex vivo testis perfusion technique pioneered by Hoffer and coworkers [21], rather than intratesticular injection as introduced by Russell and coworkers [22], to better control dosages. Over the period of an hour, cytochalasin D significantly altered but did not eliminate the actin filament networks of TBCs at stage VII of spermatogenesis. Using fluorescence thresholding techniques as an index of the amount of actin network present in regions containing the complexes, less actin was present after cytochalasin treatment than in controls. At the ultrastructural level, the actin networks appeared patchy and concentrated into foci. Tubular regions were swollen, ectopic bulb regions were present, and the complexes were shorter and often malformed. Based on these results and on the observations of normal untreated TBCs in early stage VII of spermatogenesis, we propose a new model for the formation and budding of bulbs that involves progressive loss of the actin filament networks that enables the initiation and growth of the bulbs, and the presence of a single scission site at the base of each complex.
MATERIALS AND METHODS
Animals
Reproductively active male Sprague Dawley rats (Rattus norvegicus) obtained from Charles River were used in all experiments. Animals were housed and used in accordance with guidelines established by the Canadian Council on Animal Care and according to protocols approved by the Animal Care Committee of the University of British Columbia, and according to the Society for the Study of Reproduction specific guidelines and standards. A total of seven animals were used for this study with two (one for fluorescence and one for electron microscopy) used for each of the 10 and 20 μM experiments and three (two for fluorescence and one for electron microscopy) for the 40 μM experiments.
Organ Perfusion
The ex vivo perfusion protocol used in this study was similar to that published by Hoffer and coworkers [21]. Testes were excised from rats that were under deep anesthesia induced by isoflurane inhalation. For each experiment, one testis was used for cytochalasin D treatment, and the other testis from the same animal when possible was used as a control. The spermatic artery of each testis was cannulated with a 27-gauge syringe needle attached to the outlet port of a three-way valve. One of the delivery ports was attached by tubing to a two-channel peristaltic pump calibrated to deliver 1 ml/min of treatment or control solution. The other port of the valve was attached by tubing to a bottle suspended above the table and set to deliver fixative by gravity feed. Flasks containing treatment and control buffers were housed in a water bath set to 33°C and were aerated with 5% CO2 in O2. Fixatives also were warmed to 33°C prior to being used. Treatment and control solutions were made up in Krebs-Henseleit buffer containing 4% bovine serum albumin. Treatment solutions contained cytochalasin D (at 10, 20, or 40 μM) added from a stock concentration of 20 or 40 mM in dimethyl sulfoxide (DMSO). Control solutions contained the same volume of DMSO as added to the treatment solutions.
Testes were perfused for 1 h with treatment or control solutions followed by perfusion with fixative for 30 min.
We initially explored using latrunculin A to perturb actin at TBCs, but did not observe any obvious effects when the organ was perfused for 1 h at a concentration of 0.2 μM. We did not pursue this approach further and proceeded to use cytochalasin D, which had an observable effect.
Fluorescence Microscopy
For analysis by fluorescence microscopy, testes were perfused with a fixative containing 3% paraformaldehyde, 150 mM NaCl, 4 mM Na/PO4, and 5.0 mM KCl, and adjusted to pH 7.3. Treatment and control testes were removed from the needles and cut in half. One half of each testis was used for preparing cryosections and the other half was used for preparing epithelial fragments.
For preparing cryosections, the sample was placed in a small pool of Optimal Cutting Temperature Compound (Sakura Finetek USA) on the top of an aluminum stub and frozen using liquid nitrogen. Sections (5 μm thick) were cut on a cryomicrotome and collected onto slides coated with poly-l-lysine. The slides were immediately submerged in cold acetone (−20°C) for 5 min and air dried.
For preparing epithelial fragments, half of each testis was placed into a Petri dish containing PBS (150 mM NaCl, 4 mM Na/K PO4, 5 mM KCl, pH 7.3), the capsule was removed, and the seminiferous tubule mass was cut into small pieces using scalpels. The pieces in the PBS were transferred into 50-ml tubes and gently aspirated once through an 18-gauge needle followed by a 21-gauge needle attached to a 10-ml syringe. The fragments were transferred into a 15-ml tube and allowed to settle for 5 min. During this time, larger tubule fragments settled to the bottom of the tube and smaller epithelial fragments containing mature spermatids with associated Sertoli cell apical processes remained suspended in solution. The suspended material was transferred to another 15-ml tube and centrifuged at low speed on a tabletop centrifuge to pellet the epithelial fragments. The supernatant was removed and the pellet gently resuspended in a small volume of PBS. These fragments were then transferred to poly-l-lysine-coated slides. The fragments were allowed to settle and attach to the slide for 10 min inside a humidity chamber. Excess fluid was removed and the slides were immediately submerged in cold acetone (−20°C) for 5 min and then air dried.
Epithelial fragments and cryosections were rehydrated with PBS containing 0.05% Tween 20 (TPBS) and 0.1% BSA for 20 min at room temperature. The tissue was stained with phalloidin conjugated to Alexa Fluor 488 (Invitrogen) for 30 min in a humidity chamber followed by two washes with TPBS/0.1% BSA for 10 min each. The slides were mounted using Vectashield containing 4′,6-diamidino-2-phenylindole (DAPI; Vector Laboratories).
Sections and fragments were imaged using a conventional fluorescence microscope (Zeiss Axiophot) with a film camera. Conditions in which the experimental and control sections and fragments were photographed were kept the same to allow for comparison of the signal intensity. All of the film for one experiment and its controls were developed in the same developer and fixative together. The images were then digitally scanned at 6400 dpi for quantification. The raw image obtained was used for quantification and the images used for illustration purposes were adjusted identically in Adobe Photoshop CS6.13 using the levels function without altering the integrity of the data.
Fluorescence Quantification of Actin Staining at TBCs
Only images of intact apical processes containing stage VII spermatids with associated apical processes of Sertoli cells from fragmented tissue were used in the quantification analysis. The process used to get an index of the area occupied by the actin networks at TBC clusters was carried out using ImageJ software version 1.48 [23]. All spermatids used for quantification were rotated to have the same orientation. For each experiment, controls were used to set the appropriate threshold value that would best represent only the staining of the actin networks at TBC clusters. This threshold was then applied to all spermatids analyzed in that specific experiment. The analyze particles function was used to count the number of pixels above the threshold in a region that was defined by an oval shape. This oval shape represented the concave region of the spermatid head where apical TBCs are located. The size of the oval shape remained constant for the measurements of all experiments. The total number of pixels above threshold was used as an index of the amount of actin network present in regions containing the TBC clusters.
Statistics
The 20 μM fluorescence experiment was performed once and the 40 μM experiment was done twice. A one-tailed, unpaired t-test with Welch correction was used to determine significant differences between experimental and control groups within each experiment. The 10 μM experiment could not be quantified because of differences in the variables of the imaging process.
Electron Microscopy
Drug-treated and control testes were perfused with a fixative containing 1.5% paraformaldehyde, 0.1 M sodium cacodylate (Electron Microscopy Sciences), and 1.5% glutaraldehyde (Electron Microscopy Sciences) adjusted to pH 7.3 rather than the fixative described above for fluorescence microscopy. After fixation for 30 min by perfusion, the testes were cut into small pieces and submerged in the same fixative for another 2–3 h. The tissue was washed three times for 10 min each with 0.1 M sodium cacodylate at pH 7.3 and then postfixed for 1 h on ice in 0.1 M sodium cacodylate and 1% osmium tetroxide (Electron Microscopy Sciences). The tissue was washed three times for 10 min each with double-distilled water (ddH2O), stained for 1 h with uranyl acetate (Canemco-Marivac Inc.), washed another three times with ddH2O, and then dehydrated using a series of ascending concentration series (30%, 50%, 70%, 95%, 100%) of ethanol solutions. The ethanol was replaced with propylene oxide (two washes, 30 min each, with 100% propylene oxide) and then the tissue was left overnight in a mixture of 1:1 propylene oxide and EMbed-812 (Electron Microscopy Sciences). The following day the tissue was embedded in 100% EMbed-812 and the samples polymerized at 60°C for 48 h. The blocks were sectioned and then the sections were stained with both uranyl acetate and lead citrate.
For obtaining images of normal tissue not treated with drug or control solutions, testes were excised as described above and then immediately perfused for 30 sec with PBS to clear the blood from the testes. The testes were fixed and prepared for electron microscopy as described above.
Stained sections were imaged using a Tecnai G2 Spirit electron microscope (FEI North America NanoPort) operated at 120 kV.
Qualitative observations of the drug experiments were made while scanning sections on the microscope and on the basis of a total of 377 digital images. Electron micrographs presented in the figures are representative images. Quantitative data for the diameters of TBCs were made on an additional 153 images taken specifically for that purpose.
Quantification of TBC Proximal Tubule Diameters
Two previously unsectioned blocks of embedded tissue were sectioned for each treatment and control testis for each experiment, and images collected for quantifying the diameters of the proximal tubular regions of TBCs. Images were taken of proximal tubules from different clusters of complexes in sections from each block (total of 12 blocks, 2 each for each of the control and experimental testes from the three animals; one animal for each dose). Each TBC was photographed at 68 000× on the microscope. The diameters of tubular regions were measured using ImageJ software version 1.48. A one-tailed, unpaired t-test with Welch correction was used to test for significant differences between the diameters of the tubular regions in experimental and control groups.
RESULTS
Seminiferous Epithelial Architecture Appears Normal in Cytochalasin D-Treated Tissue
Perfusion for 1 h with 10–40 μM cytochalasin D did not disrupt the overall structure of the seminiferous epithelium. At stage VII of spermatogenesis, late spermatids remained at the apex of the epithelium and were not sloughed or released. Cells deeper in the epithelium also did not detach from the epithelium. When stained with fluorescently tagged phalloidin, labeling of actin filaments at adhesion junctions (ectoplasmic specializations) associated with late spermatids appeared less intense in sections of drug-treated tissues than in controls, but was still detectable, as was labeling of filaments in ectoplasmic specializations at basal junctions between Sertoli cells.
Cytochalasin D Reduces the Area Occupied by Filamentous Actin Associated with Apical Clusters of TBCs
The amount of filamentous actin associated with clusters of TBCs in Sertoli cell cytoplasm adjacent to late spermatids appeared reduced in drug-treated epithelium relative to controls. In drug-treated clusters, the lengths of the actin cuffs appeared progressively shorter with increasing concentrations of cytochalasin D (Figs. 1, A′–D′, and 2, A, A′, B, and B′). When analyzed using thresholding techniques, there was a significant difference in the number of pixels above threshold between control and experimental groups for the 20 μM experiment (P = 0.0019) and each of the two 40 μM experiments (P < 0.0001; P < 0.0003) (Fig. 2C).
Cytochalasin D Treatment Disrupts Tubulobulbar Actin Cuffs
When analyzed at the ultrastructural level, the actin cuffs associated with tubular regions of the membrane core were less uniform in drug-treated tissues than in controls. In cross sections of TBCs in control tissues (Fig. 3A), the actin cuffs were most dense near the membrane core and gradually became less dense peripherally. In addition, the cuffs among complexes were all of uniform diameter and appearance. In drug-treated material, the actin cuffs appeared patchy with the filaments organized into distinct foci around the membrane core (Fig. 3B). These foci also occurred in regions dissociated from the membrane cores (Fig. 3, C–E).
Actin Disruption Results in Swollen Tubular Regions, Ectopic Bulbs, Malformations, and Shorter Complexes
In tissue treated with cytochalasin D, tubular regions of the complexes appeared swollen when visualized both in cross sections (Fig. 3B) and in longitudinal sections (Fig 4, A–D) of TBCs. In control tissues, the proximal tubular regions appeared uniform in diameter and were surrounded by a continuous actin cuff along their lengths (Fig. 4A). In drug-treated material, the diameters of the tubules were enlarged (compare Fig. 4A with Fig. 4C; notice that Fig. 4A is at a higher magnification than Fig. 4C) and the actin cuffs were less obvious and patchy. When proximal tubule diameters were measured in each of the 10, 20, and 40 μM experiments, there was a significant difference in diameters between control and drug-treated testes (Fig. 5).
In addition to swollen tubular regions, ectopic bulbs were present in cytochalasin D-treated material. In control tissues, the bulbs of TBCs were positioned in their normal location at the ends on elongate tubules close to the coated pits (Fig. 6A). The bulbs were easily identified by their large diameters relative to tubular regions and their close association with cisternae of endoplasmic reticulum. In drug-treated testes, ectopic bulbs were often observed near the bases of TBCs, where they emerged from areas of attachment to spermatid heads (Fig. 6, B–D).
Malformed TBCs also were observed in actin-disrupted material. Control complexes looked normal with elongate proximal tubules, bulbs, and a single coated pit at their end (Fig. 7A). Among the misshaped complexes observed were those with multiple coated pits at their ends (Fig. 7B) and those with one or more branches along their lengths (Fig. 7, C and D).
In addition to being swollen and malformed, TBCs in drug-treated tissues appeared shorter overall than in controls (Figs. 6 and 7D).
Absence of Tubular Regions and Enlarged Bulbs
Occasionally, massively enlarged bulbs were observed in cytochalasin D-treated material that occupied virtually the entire length of the complexes (Fig. 8, A and B); in other words, elongate proximal tubules with actin cuffs were absent. In many of these cases, short distal tubular regions that normally occur between the bulbs and coated pits, and that also have actin cuffs, were absent as well (Fig. 8, A and B).
Large Bulbs and Short Tubular Regions Occur in Untreated Early Stage VII Tubules
In tubules from untreated testes that were excised from animals and immediately perfusion fixed, large elongate and folded bulbs with short tubular regions were observed (Fig. 9, A and B). These TBCs somewhat resembled those from late stage VII complexes exposed to cytochalasin D (Fig. 8, A and B).
DISCUSSION
In this study we show that disrupting the actin networks of TBCs results in increased diameters of tubular regions, the formation of ectopic bulbs, and the presence of branched complexes. The results are consistent with the conclusion that intact actin networks are necessary for maintaining the structural integrity of TBCs. Based on our observations of TBCs in cytochalasin D-treated testes, together with observations of the structures in untreated epithelium, we propose a new model for how mature TBCs may be internalized by Sertoli cells.
TBCs were originally identified and their overall structure defined in a series of classic studies by Russell and coworkers [6, 8, 9]. The structures begin as coated pits that project into Sertoli cells at sites of intercellular attachment in the mammalian seminiferous epithelium. The plasma membranes of the two cells remain attached as the pits mature, and the necks of the structures do not undergo scission to form vesicles. Rather, the necks continue to elongate, forming long double-membrane tubes, each surrounded by a cuff of actin filaments that in turn is surrounded by a cytoskeletal shell consisting of plectin (PLEC) and spectrin (SPT-) [24]. These complexes extend up to 2–3 μm into Sertoli cells in rats [6], and up to 6–8 μm in opossums [9]. As TBCs mature, swellings (bulbs) develop near the ends of the complexes. The bulbs eventually dissociate from the complexes and are degraded by Sertoli cells [8] as part of the mechanism of junction remodeling that precedes sperm release at the apex of the epithelium and translocation of spermatocytes through basal junction complexes at the base of the epithelium [4]. The precise function of the actin filament cuffs surrounding tubular regions of TBCs is not known, nor is it known how the actin networks are related to formation and scission of the bulbs.
In clathrin-associated endocytosis generally in cells, actin filament networks have been implicated in invaginating the coated pits, constricting the necks of forming vesicles, scission of the vesicles from the plasma membrane, and moving the internalized vesicles deeper into the cytoplasm [14, 25]. Actin has been found to recruit and facilitate the assembly of dynamin (DNM-) during formation of the neck and to promote dynamin-mediated vesicle scission [26]. In dynamin knockout cells, coated pits with very long, narrow necks are formed and scission does not occur [27]. In these same cells treated with latrunculin B to prevent actin polymerization by sequestering G-actin, elongate necks do not form and coated pits remain attached to the plasma membrane by wide necks. Together these results indicate that in clathrin-mediated endocytosis, an intact actin cytoskeleton participates in the formation of constricted necks and in scission of vesicles from the plasma membrane. When dynamin-mediated scission does not occur, continued actin polymerization results in abnormally long necks.
In the testis, Russell and coworkers [19] report the complete absence of normal TBCs at apical sites in rat seminiferous epithelium 3 h post-intratesticular injection of 50 μl of 500 μM cytochalasin D. What occurs instead are numerous double-membrane coated pits directly associated with attachment sites between Sertoli cells and spermatids. In other words, tubular and bulbar regions are absent. Although these results are somewhat difficult to interpret because 1) the time it takes for any single TBC to develop is not known and 2) the period during which TBCs are present at apical sites in the epithelium spans 3–4 days, the results are generally consistent with the conclusion that actin filaments are necessary for development and elongation of the tubular regions. Our results, using ex vivo testicular perfusion methods to control drug dosage, extend these earlier results and provide insight into the formation and eventual scission of the bulbs.
In the experiments reported here, perfusion with cytochalasin D did not result in disruption of overall epithelial architecture. When fluorescent phalloidin was used to label actin filaments, the characteristic pattern of staining at apical and basal intercellular junctions in the epithelium was still present, although somewhat weaker than in controls. In addition, no detachment of cells from the epithelium was detected at any of the dosages.
Although there were no dramatic changes in epithelial architecture with cytochalasin D exposure, we did observe major effects of the drug treatment on TBCs. The amount of actin network associated with apical TBC clusters was significantly different in drug-treated testes compared to controls, and individual complexes appeared shorter. When evaluated at the ultrastructural level, the diameters of tubular regions appeared larger. When quantified, there was a significant difference in proximal tubule diameters between control and drug-treated testes at each of the doses of drug tested. In addition, ectopic bulbs were present, and the complexes often appeared branched and malformed. Distal tubular regions appeared absent in many cases and the coated pits were directly associated with the bulbs or swollen regions of the complexes. We conclude that actin cuffs are necessary for maintaining the normal structure of tubular regions.
Potentially the most significant result of our study is that when the actin filament networks of TBCs are disrupted, the tubular regions swell and form bulbs.
The currently accepted model for internalization of a TBC is one in which there are multiple scission sites along the structure that result in not only internalization or budding of the bulb but also internalization of the coated pit from the end of the complex and numerous vesicles resulting from fragmentation of the proximal tubule [20] (Fig. 10A) In part, this model arose from the early ultrastructural observation that vesiculation of the inner spermatid membrane tube occurred within the proximal tubule of apical TBCs [6, 8]. If this model for internalization of TBCs is true, one might predict an abundance of small double-membrane vesicles in Sertoli cell regions where TBCs occur, and that some of these vesicles would appear to have dense cores resulting from the submembrane densities associated with the spermatid plasma membranes in the coated pits. Neither of these features is obvious in tubulobulbar-containing regions of the Sertoli cell.
Our results together with the earlier observations of Russell and coworkers [19] suggest the following alternative model for formation of the bulb and internalization of a TBC by the Sertoli cell (Fig. 10B). Polymerization of actin filaments drives the elongation of the TBC and projects it deep into the Sertoli cell. In addition to growth of the complex, the actin cuff also participates in maintaining the narrow diameter of the double-membrane tubule. Focal loss of the actin cuff near the end of each complex initiates the formation of a bulb. We predict that other components known to be present at TBCs, such as the BAR domain proteins [28] and dynamin [20, 28], also detach from this region of the tubule as the bulb forms. As the actin cuff is lost, cisternae of endoplasmic reticulum associate directly with the Sertoli cell plasma membrane of the forming bulb. Continued loss of the actin cuff and related proteins, both proximal and distal to the initiation site, expands the bulb. This expansion or elongation of the bulb eventually incorporates or resolves the coated pit into its structure distally, and reduces the length of the proximal tubule until only a small “neck” remains near the site where the complex originates from the intercellular junction. The possibility that the coated pit is resolved into the bulb may not be entirely without precedent because not all clathrin-coated structures that form generally in cells proceed to separation from the plasma membrane [13]. Scission, possibly regulated by dynamin remaining at the site, occurs at the small neck and results in the bulb being internalized into the Sertoli cell. This model is attractive because it involves only one scission site, and is consistent with the appearance of enlarged bulbs in early stage VII (see Fig. 9B) of spermatogenesis, and with the elongate and folded cigar-shaped bulbs often seen later in stage VII (see Fig. 6A). Live cell imaging of TBC formation and internalization using epithelial fragments (for apical complexes), or perhaps primary Sertoli cell culture systems (for basal complexes), could eventually be used to discriminate between the two models.
REFERENCES
Notes
[1] Financial disclosure Supported by a grant (#155397-2013) to A.W.V. from the Natural Sciences and Engineering Research Council (NSERC) of Canada Discovery Grant Program. Presented in part at the ASCB conference, December 2014, Philadelphia, Pennsylvania, and at the FASEB conference, March 2015, Boston, Massachusetts.