|Home | About | Journals | Submit | Contact Us | Français|
The recycling of internalized signaling receptors, which has direct functional consequences, is subject to multiple sequence and biochemical requirements. Why signaling receptors recycle via a specialized pathway, unlike many other proteins that recycle by bulk, is a fundamental unanswered question. Here we show that these specialized pathways allow selective control of signaling receptor recycling by heterologous signaling. Using assays to visualize receptor recycling in living cells, we show that the recycling of the beta-2 adrenergic receptor (B2AR), a prototypic signaling receptor, is regulated by Src family kinases. The target of Src is cortactin, an essential factor for B2AR sorting into specialized recycling microdomains on the endosome. Phosphorylation of a single cortactin residue, Y466, regulates the rate of fission of B2AR recycling vesicles from these microdomains, and, therefore, the rate of delivery of B2AR to the cell surface. Together, our results indicate that actin-stabilized microdomains that mediate signaling receptor recycling can serve as a functional point of convergence for crosstalk between signaling pathways.
Membrane trafficking plays a key role in determining the functional outcomes of signaling (1-3). Signaling receptor activation on the cell surface initiates several regulatory events resulting in their removal from the cell surface by endocytosis (1,4-6). Post-endocytic sorting of receptors into either the recycling pathway, which transports receptors back to the cell surface, or the degradative pathway, which transports receptors to the lysosome, determines whether cells can regain sensitivity to the signal or stay desensitized (5-8). Because sorting and recycling are rapid and continuous processes, relatively small changes in their rates can cause significant cumulative changes in the surface localization of signaling receptors (1,9). It is becoming progressively more clear that such changes in receptor trafficking and localization, especially for members of the G protein coupled receptors (GPCR) family (10), contributes to the pathophysiology of many diseases (2,11-13).
Interestingly, the recycling of many signaling receptors is restricted to a specialized “sequence-dependent” pathway that is distinct from the “bulk” recycling pathway mediating the recycling of constitutively cycling proteins like the transferrin receptor (TfR) (8,14-16). Recent studies using prototypic GPCRs, such as the Beta-2 Adrenergic Receptor (B2AR), have established that signaling receptor recycling is distinct from bulk recycling at several levels. First, unlike TfR, B2AR recycling is dependent on specific sequences on the receptor, including a C-terminal PDZ ligand (14,17-19). Second, B2AR recycling requires a highly specialized recycling machinery, the components of which include actin cytoskeletal proteins, PDZ-domain containing proteins, and the retromer complex (16,20-23). Third, B2AR recycling domains are spatially and biochemically separate from TfR recycling domains (16,18,21). While the full complement of proteins localized to these sequence-dependent recycling domains is still being identified, most proteins identified as required for B2AR recycling are localized to these domains (16,21). Fourth, B2AR sorting into these domains is actively regulated, as it requires multiple sequences and as it is controlled by the phosphorylation state of B2AR (16,18,19,24,25). Recent functional data suggest that these actin-based microdomains might represent a general recycling pathway for a considerable subset of endocytic cargo molecules including several signaling receptors and channels that contain identified PDZ ligands (26).
These observations raise the fundamental questions of why there are two distinct recycling pathways back to the cell surface, and why B2AR recycling is restricted to a specialized recycling pathway (6,15). Here, we tested the hypothesis that such restriction permits heterologous regulation of B2AR recycling by distinct signaling pathways, allowing for signaling crosstalk. Using assays to directly visualize B2AR endosomal sorting and surface delivery, we show that B2AR recycling is positively modulated by the activity of Src family kinases (Src). Src acts via phosphorylation of cortactin, an essential regulator of the endosomal actin microdomains that mediate B2AR recycling. Cortactin phosphorylation increases the dynamics of scission of recycling tubules, without apparent effects on the rate or extent of actin assembly on endosomal microdomains. Our results show that endosomal actin-based microdomains can act as a control point for signaling crosstalk, and suggest a physiological reason for restricting signaling receptor recycling to a specialized actin-dependent pathway.
To test whether B2AR recycling was regulated, we first used a sensitive assay that we have previously developed to detect B2AR recycling (25,27). We tagged the N-terminal domain of B2AR with a pH-sensitive GFP (SpH) (28), which is quenched in intracellular vesicles due to their acidic environment, but becomes fluorescent when exposed to neutral extracellular medium. To better detect recycling, we imaged receptor dynamics at the cell surface using rapid live cell imaging using Total Internal Reflection Fluorescence microscopy (TIR-FM). Treatment with isoproterenol (iso), a B2AR agonist, induced rapid clustering and endocytosis of the SpH-tagged B2AR (SpH-B2AR) that corresponded to a loss of surface fluorescence (Fig 1A-B, MovieS1). Washout of agonist induced recovery of surface fluorescence, indicating that the expressed receptor trafficked normally as established previously (MovieS2) (19,25). Serial acquisition of images, at 10 Hz, after 5 minutes of iso captured transient bursts of Sph-B2AR fluorescence in the plasma membrane, appearing abruptly and lasting approximately 100-300 ms (Fig 1C and D). We have previously ascertained that these events, spatially and temporally distinct from endocytic sites (Fig 1D and MovieS3), denote individual events mediating B2AR recycling at the cell surface (16,18,25,27). Importantly, this provided a sensitive assay to detect changes in the rate of B2AR recycling in response to acute manipulations of the signaling pathway in the same cells, thus avoiding the cell to cell variability and potential indirect effects of changes in endocytic rates seen with traditional assays measuring ensemble changes in surface fluorescence.
We first focused on the Src family kinases (Src) as a potential factor that cross-regulates B2AR recycling, as Src has been implicated in regulating the actin cytoskeleton and trafficking (29-31), and as endosomal actin is critical for B2AR recycling (16,20). To test whether Src activity acutely modulates B2AR recycling, we observed the effect of Src inhibitors on the number of B2AR recycling events per unit time. Considering that Src family kinases play roles in other cellular functions including trafficking, acute pharmacological inhibition, as we use here, was necessary to avoid compensatory changes and indirect effects on other steps of trafficking such as endocytosis (29,32,33). SpH-B2AR cells were exposed to iso as above, and imaged for 1 minute, then treated for 5 minutes with 10μM PP2, potent Src family kinase inhibitor (34), before being imaged a second time. In the same cells, PP2 treatment decreased the number of recycling events by ~25% (Fig 1E). This decrease was independent of initial event rate for the cells and was highly statistically significant (Fig 1F, n = 23 cells, p = 5E-05), suggesting that Src signaling enhanced B2AR recycling. Similar results were seen when 10 μM Src Inhibitor 1 (Src-I1), an independent inhibitor of Src (35), was used (Fig 1E and F, n = 16 cells, p = 0.004). To control for the effects of inhibitor addition and duration of the experiment, treatment with DMSO (vehicle) and an inactive variant of PP2 (PP3) were also tested and showed no significant change in recycling (Fig 1E and 1F, PP3: n = 14 cells, p = 0.665; vehicle: n = 11 cells, p = 0.346). To test whether the effect of Src inhibitors was specific to B2AR (i.e., sequence-dependent recycling) or was a general effect on recycling, we performed an analogous experiment using the transferrin receptor (TfR), a prototypic cargo that recycles by bulk membrane flow. In contrast to B2AR, PP2 treatment did not change the number of SpH-TfR recycling events per unit time (Fig 1E and 1F, 8 cells, p = 0.56). Together, our results suggest that Src family kinases specifically regulate sequence-dependent B2AR recycling and not bulk recycling.
We next attempted to define the step of B2AR recycling affected by Src family kinase inhibitors. We and others have established that the surface recycling of B2AR is mediated by specific tubular microdomains that are present on the endosome, which undergo fission and deliver receptor-containing membranes to the cell surface (16,18,21). Considering this, we hypothesized that Src activity could increase surface delivery of B2AR vesicles by regulating three steps: 1) the number of endosomes from which vesicles were generated, 2) the number of B2AR tubules generated from each endosome, and/or 3) the rate at which vesicles were generated from each tubule on an endosome.
To address these possibilities, we used spinning disk confocal microscopy to directly visualize the behavior of endosomal tubules mediating the recycling of B2AR in live cells, using an N-terminally FLAG-tagged B2AR. This tagging method has been used extensively to study B2AR trafficking, and we and others have previously established that it does not interfere with the function or trafficking of B2AR (16,18,19,36). When HEK293 cells expressing this receptor were labeled with a fluorescent anti-FLAG antibody, as described previously (16,18), we readily detected surface expressed receptors. Activation of B2AR with iso induced robust internalization and transport of B2AR to endosomes (Fig 2A). As expected, endosomal B2AR was observed in tubular domains that underwent scission and generated vesicles (Fig 2B, E).
To test the first possibility, i.e., whether the number or morphology of endosomes were regulated by Src inhibitors, we acquired rapid 3D image stacks of cells exposed to iso, before and after PP2 treatment (Fig 2A and B). Quantitative image analysis showed no difference in the number of endosomes per cell before vs. after PP2 addition (Fig 2C, n= 593 and 565 for before and after, p = 0.45). When compared pairwise and normalized, the endosome number remained essentially unchanged after inhibitor treatment (99.6 ± 6% of cells before PP2). Further, the size distribution (Fig 2D) of endosomes was also unchanged (p = 0.62), indicating that the Src inhibitor PP2 had no significant effect on the number or general morphology of B2AR endosomes. To test the second possibility, we analyzed the number of tubules observed on individual endosomes before vs. after Src inhibitor addition. When endosomes were binned according to the number of tubules generated by them, there were no differences in the population distribution of endosomes showing 0, 1, 2, 3, or more than 3 tubules after PP2 treatment (Fig 2E; p > 0.5 for all pairwise t-tests for each bin). To test the third possibility, we directly analyzed the lifetimes of individual tubules, defined as the time from appearance to scission and generation of a vesicle (Fig 2F). Treatment with PP2 caused a significant increase (~20%, p = .0002, n = 89 and 69 for before and after) in the lifetimes of endosomal tubules, consistent with a decrease in the rate of surface delivery of B2AR (Fig 2G). Together, our results suggest that Src family kinase activity regulates the rate of scission of B2AR recycling tubules without affecting general endosome morphology or function.
Actin microdomains on endosomes define the physical sites of B2AR recycling tubules, and actin association in these domains was required and sufficient for B2AR recycling (16). Therefore, we next tested whether Src modifies endosomal actin to regulate B2AR recycling. When F-actin was visualized using a variety of markers, including coronin-1 (Fig 3A and B), Arp3 (Fig 3C), cortactin (Fig 3C), and LifeAct (Fig 3E), actin was highly localized to the B2AR recycling domains as expected. Treatment with PP2 did not change the localization of actin at these domains (Fig 3A, C). When endosomes were binned according to the number of coronin spots (Fig 3D), there was no difference in the population distribution of endosomes after PP2 addition. No difference was seen also when Arp3 was used as a marker for actin domains (77±4.7 vs. 82±5.4 for percentage of endosomes with one Arp3 spot, p= 0.67). This further confirms our interpretation that Src did not regulate the number of B2AR recycling domains on endosomes (Fig 2E). Therefore, we next tested whether the dynamics of actin in recycling domains was changed after treatment with a Src inhibitor. To do this, we directly measured actin turnover using Fluorescence Recovery After Photobleaching (FRAP) before and after PP2. When a small region of the cell containing an individual actin-spot on the endosome (marked by B2AR) was bleached, actin fluorescence recovered rapidly after bleaching on both surfaces, as expected for rapid turnover of actin on endosomal recycling microdomains (Fig 3E and F). As a control, regions of the cortical actin also recovered at broadly similar timescales. Importantly, quantitative analysis of the recovery curves showed no significant difference before vs. after PP2 addition (Fig 3G). When points from both graphs were fit to single exponential fits, the preferred fit was a single curve for both plots (as opposed to two separate curves, p = 0.49), suggesting that Src inhibition does not induce gross changes in endosomal actin assembly or localization.
Considering that with the Src inhibitor PP2 did not modify actin assembly at an overall level, and that actin and associated proteins have been implicated in providing the driving force for fission during vesicle scission at the cell surface (37-39), we hypothesized that Src might regulate specific components or subsets of components in the actin machinery. Therefore, we next sought to identify a specific target of Src family kinase in regulating B2AR recycling. Cortactin was a highly attractive candidate, as it is endosomally localized, is required for the generation of actin-mediated recycling domains, is phosphorylated by Src, and can form a complex with vesicle scission proteins like dynamin (16,29-31). Src phosphorylates cortactin at three sites - Tyr 421, Tyr 466, and Tyr 482 - and can modify its function (40-42). To ask whether Src regulated B2AR surface recycling via phosphorylation of cortactin, we asked if these phosphorylation sites were required for Src-mediated regulation of B2AR recycling, by mutating these residues to phenylalanines, either together (Y3F) or individually (Y421F, Y466F, or Y482F). All these mutants localized to B2AR recycling domains on the endosome comparable to wild-type cortactin (Fig 4A, Y3F and Y466F shown as examples). Further, PP2 treatment did not change the localization or number of cortactin domains on the endosome (Fig 4B). When co-expressed with SpH-B2AR, Y3F and Y466F abolished the decrease in B2AR surface insertion events induced by PP2 (Fig 4C and D, Movie S4). As controls, PP2 still induced a decrease in recycling when wild-type cortactin or the other two mutants were expressed (Fig 4C and D). This effect on B2AR recycling is likely specific to Src-mediated phosphorylation of cortactin, as inhibition of Abl kinase or MEK, two other kinases known to phosphorylate cortactin, under identical conditions, did not change the rate of B2AR recycling (Fig S1). Consistent with this, treatment of cells with PP2 significantly reduced the phosphorylation of cortactin at Y466, as shown by immunoblotting using a phospho-specific antibody directed against this site (Fig 4E and F, p = 0.01), also directly confirming that PP2 inhibited Src activity under our conditions. Together, our results indicate that phosphorylation of cortactin by Src at Y466 is required for Src-mediated regulation of B2AR recycling.
Previous studies have established that, while internalized signaling receptors, such as B2AR, and constitutively cycling proteins, such as TfR, are transported to the same endosomal compartments, they recycle via biochemically and physically distinct pathways (15,16,18,19). This study addresses two fundamental questions raised by these observations. Why are there two recycling pathways to the cell surface from the same endosome, and why is B2AR recycling restricted to a specialized “sequence-dependent” pathway? One possibility is that these additional biochemical requirements provide opportunities for finer control over recycling of signaling receptors. We provide direct evidence in support of this, by showing that Src-mediated signaling can specifically regulate the sequence-dependent recycling pathway without affecting bulk recycling.
Our previous results suggest that the sorting of B2AR into sequence-dependent recycling microdomains is regulated by a homeostatic extracellular signaling axis, consistent with a hierarchical model for receptor sorting in the endosome (18,25,27). The first step excludes receptors from accessing the bulk recycling tubules, while the second step concentrates receptors in sequence-dependent tubules. For B2AR, exclusion from bulk recycling requires Protein Kinase A-dependent phosphorylation of serine 345/346 on the C-terminal tail of the receptor. Our current results therefore suggest that both homologous and heterologous regulation can control B2AR recycling at the level of partitioning into endosomal microdomains, with homologous PKA signaling restricting B2AR to actin-based microdomains, and heterologous signaling (e.g., Src) regulating the rate of recycling from these microdomains.
How does Src-mediated cortactin phosphorylation regulate endosomal scission? Src has been reported to dock to B2AR interacting proteins, such as AKAP (23), and this might serve to recruit Src specifically to the B2AR endosomes to phosphorylate cortactin. Cortactin can promote Arp2/3-nucleated actin filament assembly, at least in vitro (43), partly through N-WASP (41,44). However, canonical WASP and WAVE components are not localized to the endosome (16,45,46). Src-mediated phosphorylation of cortactin has also been implicated in recruitment of the adapter Nck and WIP, thus modifying actin nucleation (30,42). However, our results suggest that the effect of Src signaling on actin dynamics, at a gross level, is minimal, as Arp2/3 localization and actin turnover appear unchanged (Fig 3). Interestingly, cortactin can also interact with dynamin (29,47,48), the canonical GTPase for membrane scission. During endocytosis, cargo initiates a Src signaling cascade that phosphorylates both cortactin and dynamin to stimulate endocytosis (29), although the individual roles played by cortactin and dynamin is not clear in this case. Consistent with this, Src inhibition has been reported to decrease endocytic rates of cargo proteins including GPCRs (29,32,33). However, the changes we observe are unlikely to be due to potential effects of Src inhibitors on B2AR endocytosis. Our assay directly monitors discrete recycling events and not ensemble surface fluorescence, thus resolving endocytic defects from recycling defects. Additionally, we measure acute effects of pharmacological inhibition within minutes of Src inhibition (Fig 1E), making it unlikely that the effect is an indirect consequence of longer-term inhibition of Src as might be observed with Src depletion. Further, a cortactin mutant that cannot be phosphorylated at Y466 abolished the Src-mediated regulation of B2AR recycling (Fig 4D), indicating that the primary target of Src is cortactin and not dynamin. It is possible, however, that a similar cooperative effect between cortactin and dynamin, or an as yet unidentified scission protein, might be the basis for regulation of fission of endosomal recycling tubules by cortactin phosphorylation.
In view of the established physiological consequences of recycling for receptor signaling (1-3,6), endosomal actin microdomains might serve as highly versatile control points for crosstalk between receptor signaling pathways. Further, recent data suggest that, in addition to directing receptors into recycling vesicles, actin-based microdomains might serve as scaffolds for spatially and temporally distinct phases of signaling from the endosome (49). Modulating the lifetimes of receptor association with these domains could, therefore, regulate receptor function at multiple levels.
C-terminal PDZ ligands have been identified in many signaling receptors and channels, and are increasingly being implicated in recycling (50). Further, internal PDZ ligands, which have more disparate and complex identities, have also been identified in several receptors. In the case of B2AR, actin association, either through PDZ domains or by direct linkage, is sufficient to localize receptors to the sequence-dependent recycling pathway (16,20). Considering this, it is likely that the specialized recycling pathway used by B2AR is also used by a substantial fraction of signaling receptors in cells. Further, many of these receptors can generate sustained signaling from the endosome. The heterologous control of B2AR recycling by Src via cortactin phosphorylation, described in this study, could therefore be a general mechanism for regulating signaling receptor recycling and function. While our experiments suggest that, in our system, Src family kinases are major regulators of B2AR recycling at the endosome, it is possible that multiple signaling pathways may regulate endosomal actin and signaling receptor recycling under different physiological circumstances. Cortactin may also be phosphorylated by several other kinases, including FAK (51), ACK (52), Abl kinase (53) and MEK/ERK (41,54), suggesting that many signaling pathways might converge to heterologously regulate B2AR recycling. Further, in addition to cortactin, the actin network includes several classes of proteins that can modulate its nucleation, elongation, branching, and disassembly (55,56). Many of these factors, such as the WASH complex, can modify tubule formation, and vesicle scission (29,37,45,46,57,58). The exact details of actin assembly and turnover on the endosome are still being defined. As more examples of such regulation of receptor recycling are defined, future studies will determine whether other signaling pathways can regulate distinct components of the actin cytoskeleton for combinatorial control of receptor recycling and signaling.
HEK293 cells were purchased from ATCC and maintained in high-glucose DMEM (Thermo) with 10% FBS (Gibco). DNA transfections were performed with Effectene (Qiagen) according to manufacturer's instructions. Stable cell lines were generated by maintaining transfected cells in Zeocin (Invitrogen) and/or Geneticin (Gibco) for two weeks or more. Clonal populations were isolated using standard procedures.
Plasmid constructs for expression of B2AR, cortactin, coronin, arp3 and lifeAct-RFP constructs have been previously described (16,18). Mutant versions of cortactin were generated by QuikChange (Stratagene) site-directed mutagenesis according to manufacturer's instructions. Isoproterenol hydrochloride (Sigma) was made to 10 mM stocks in water. PP2 (Cayman Chemicals) was stored as a 10 mM stock in DMSO (Invitrogen).
Confocal imaging was performed on an Andor Revolution XD Spinning disk system with a Nikon Eclipse Ti inverted microscope equipped with a temperature controlled chamber, using a 100x or 60x 1.45 NA TIRF objective (Nikon). Cells were imaged live at 37°C in Opti-MEM (Invitrogen) with 10% FBS and 40 mM HEPES pH 7.4. Light sources were solid state 488 nm, 561 nm, or 647 nm lasers. A cooled iXon+ 897 EM-CCD camera was used for image capture. For visualizing receptors in the endosome, FLAG-tagged receptors were labeled with M1 antibodies (Sigma) conjugated with Alexa-488, 568, or 647 (Invitrogen) as described (18). Cells were imaged 5-15 minutes after addition of Iso with 5 minutes PP2 or DMSO (vehicle) treatment. Confocal stacks and time-lapse images were collected as tiff images and analyzed with ImageJ. The numbers of tubules or actin spots were counted manually. Object area was determined on thresholded images using the ImageJ area metric. Tubule lifetimes were estimated manually, in a double blind manner, using timelapse confocal images collected at 5 second intervals. FRAP assays were done as previously described (18). Fluorescence of actin was measured by drawing a 6 pixel circular ROI manually around the actin spot. Total internal reflection fluorescence microscopy (TIR-FM) to detect individual insertion events was carried on the same microscope system as described previously (18,59). Cells expressing SpH-TfR or SpH-B2 were imaged after 5 minutes of iso, at 10 frames per second. Cells were then treated with 10 μM PP2, Src-I1, PP3 or DMSO for 5 minutes and imaged again. Images were collected as tiff images, and exocytic events were counted manually. While the images shown have been scaled to present the visual data clearly, all fluorescence measurements and quantitations were performed on images acquired directly from the camera without any adjustments. To minimize bias, all assays were done in a double blind manner by scrambling file names using the Scrambler.py file name randomization script (https://gist.github.com/SavinaRoja/1629319). Simple statistical tests were performed using Microsoft Excel, and graph generation and curve fitting were performed using Graphpad Prism.
Cells expressing SSF-B2 were grown to 80% confluence in 10 cm dishes the treated with either iso or water (no treatment) for 5 minutes followed by either PP2 or equivalent amount of DMSO for 5 minutes. After three washes in cold DPBS 1mM Ca Mg cells were scraped in lysis buffer (150 mM NaCl, 10 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid [HEPES],1 mM ethylene glycol tetra acetic acid [EGTA], 0.1 mM MgCl2 and 0.5% Triton X-100 supplemented with 1 mM PMSF, 1 mM DTT (dithiothreitol), PhosSTOP (Roche), 10μM Pepstatin A and 50 μM Leupeptin (Sigma)). Scraped cells were then rotated at 4°C for 15 minutes and lysate was separated by centrifugation at 4°C. Samples were brought to equal concentration in lysis buffer then reduced and boiled at 95 °C for 10 min with 4x BioRad Sample Buffer. Samples were run in 4% to 10% Tris Acrylamide gels and protein was transferred to nitrocellulose for blotting. Membranes were blocked with 4% BSA in TBST. Phosphorylated cortactin was detected with polyclonal Anti-Cortactin (phospho Y466) antibody (AbCam) at 1:1000 in 4% BSA TBST, total cortactin was detected with monocolonal anti-cortactin (Santa Cruz Biotechnology) at 1:5000 and actin was detected with HRP conjugated anti actin (Cell Signaling) at 1:10,000. Primary antibodies were detected with HRP conjugated secondary and visualized with Luminata HRP substrate (Millipore). Phospho Y466 antibody was stripped using Restore™ Western Blot Stripping Buffer (Thermo Scientific) according to manufacture instruction. The same membrane was re-probed for total cortactin. Densitometry measurements were made in ImageJ, and phospho-Y466 was normalized to total cortactin for statistical measurements.
Why are signaling receptors recycled via a specialized “sequence-dependent” pathway, when many proteins can recycle by “bulk” without apparent requirements? This study shows that sequence-dependent recycling provides control points for heterologous signaling pathways to regulate receptor recycling and resensitization. Specifically, we show that Src kinase regulates GPCR recycling without affecting bulk recycling, by phosphorylating cortactin and controlling the dynamics of specialized sequence-dependent endosomal microdomains. Such a mechanism might be generally applicable for signaling crosstalk.
We thank Paul J. Barton for writing Scrambler.py, used for randomly renaming files for objective blind analysis. We thank Daniel Shiwarski, Amanda Soohoo, Shanna Bowersox, and Isaac Shamie for essential technical help and comments, Tim Jarvela and Collin Bachert for technical help and advice, and Drs. Adam Linstedt, Nathan Urban, Tina Lee, Victor Faundez, Mark von Zastrow, Peter Friedman, Guillermo Romero, Alessandro Bisello, and Jean-Pierre Vilardaga for reagents, comments, and helpful discussions. R.V. was supported by NIH/NIDA Grant T90-DA023420, and M.A.P. was supported by DA024698 and DA036086 from the NIH and 13BGIA17020019 from the American Heart Association.