Author response: Low-affinity integrin states have faster ligand-binding kinetics than the high-affinity state
Year: 2021
Type: peer-review
Abstract: Article Figures and data Abstract Editor's evaluation Introduction Results Discussion Materials and methods Data availability References Decision letter Author response Article and author information Metrics Abstract Integrin conformational ensembles contain two low-affinity states, bent-closed and extended-closed, and an active, high-affinity, extended-open state. It is widely thought that integrins must be activated before they bind ligand; however, one model holds that activation follows ligand binding. As ligand-binding kinetics are not only rate limiting for cell adhesion but also have important implications for the mechanism of activation, we measure them here for integrins α4β1 and α5β1 and show that the low-affinity states bind substantially faster than the high-affinity state. On- and off-rates are similar for integrins on cell surfaces and as ectodomain fragments. Although the extended-open conformation's on-rate is ~20-fold slower, its off-rate is ~25,000-fold slower, resulting in a large affinity increase. The tighter ligand-binding pocket in the open state may slow its on-rate. Low-affinity integrin states not only bind ligand more rapidly, but are also more populous on the cell surface than high-affinity states. Thus, our results suggest that integrin binding to ligand may precede, rather than follow, activation by 'inside-out signaling.' Editor's evaluation This manuscript describes a detailed measurement and calculation of integrin ligand-binding kinetics, which are very important for the understanding of integrin activation. The data clearly indicate that low-affinity-binding states with closed conformation bind ligand with the mode of 'fast on, fast off,' while the high-affinity-binding state with the open conformation shows very slow off-rate. The kinetics measurements were well designed and a lot of work was done in this study. https://doi.org/10.7554/eLife.73359.sa0 Decision letter Reviews on Sciety eLife's review process Introduction Integrins are a family of receptors that in all adherent cells in the body mechanically integrate the intracellular and extracellular environments and mediate cell migration and adhesion. Their α- and β-subunits associate noncovalently to form an extracellular ligand-binding head and multidomain 'legs' that connect to single-pass transmembrane and cytoplasmic domains with binding sites for cytoskeletal adaptor or inhibitory proteins (Figure 1A). Integrins populate an ensemble with three overall conformational states: the low-affinity bent-closed (BC) and extended-closed (EC) conformations and the high-affinity extended-open (EO) conformation (Figure 1A). The equilibrium between these conformational states is allosterically regulated by extracellular ligand binding, intracellular adaptor/inhibitor binding (Bouvard et al., 2013; Iwamoto and Calderwood, 2015), and tensile force applied by the actin cytoskeleton on the integrin β-subunit that is resisted by ligand embedded in the extracellular matrix or on cell surfaces (Kim et al., 2011; Legate and Fässler, 2009; Li and Springer, 2017; Nordenfelt et al., 2016; Park and Goda, 2016; Sun et al., 2016; Zhu et al., 2008; Figure 1A). The EO conformation has ~1000-fold higher binding affinity for ligand than the two closed conformations and is the final competent state to mediate cell adhesion and migration (Li and Springer, 2018; Li et al., 2017; Schürpf and Springer, 2011). Many previous studies have emphasized the importance of force in regulating integrin adhesiveness (Alon and Dustin, 2007; Astrof et al., 2006; Li and Springer, 2017; Nordenfelt et al., 2016; Nordenfelt et al., 2017; Sun et al., 2019; Zhu et al., 2008). Recent measurements of the intrinsic ligand-binding affinity of each conformational state and the equilibria linking them enabled a thermodynamic comparison of integrin activation models (Li and Springer, 2017; Li et al., 2017). Remarkably, only the combination of adaptor binding and cytoskeletal force can activate integrins in an ultra-sensitive, switch-like manner over a narrow range of signal input that is the sine qua non of cellular signaling (Kuriyan et al., 2012; Li and Springer, 2017). The large increase in length between the bent and extended conformations (Figure 1A) is indispensable for switch-like integrin activation. Figure 1 Download asset Open asset Ligand-interaction kinetics of integrin ensembles. (A) Three overall integrin conformational states (Luo et al., 2007). Individual domains are labeled next to the extended-open (EO) state. The structural motifs that move during opening (α1-helix, α7-helix, and β6-α7 loop) are labeled in the βI domain of the extended-closed (EC) and EO state. F represents tensile force exerted across ligand–integrin–adaptor complexes by the cytoskeleton and resisted by immobilized ligand. (B) Reaction scheme showing the apparent 1 vs. 1 kinetics of integrin and ligand binding (left) and the scheme taking into account conformational change (right). (C) Fabs utilized in this study, the integrin domains they bind, and their conformational specificities. Despite these advances, thermodynamics cannot describe the sequence of events in a multistep transition; furthermore, energy-driven processes such as cytoskeleton movements occur under nonequilibrium conditions. To quantitatively relate the steps involved in signal transmission across the plasma membrane in integrins, ligand-binding on- and off-rates of each conformational state are of key importance. These parameters also determine whether integrin encounter of ligand is timely and whether the ligand remains bound for a sufficiently long time for the integrin to exert its function in the presence of force. Previous representative measurements (Dong et al., 2018; Kokkoli et al., 2004; Mould et al., 2014; Takagi et al., 2003) on integrin interaction with ligand have yielded kinetics on mixtures of conformational states, that is, apparent on- and off-rates averaged over conformational states (Figure 1B, left). However, the ligand-binding kinetics of individual integrin conformational states (Figure 1B, right) remain unknown. These kinetics must be determined before we can understand how integrin function is regulated and how integrins work in concert with the cytoskeleton to provide traction for cell migration and firm adhesion for tissue integrity (Figure 1A). For two classes of force-regulated adhesion molecules, each of which have a single low-affinity state and a single high-affinity state, selectins (Phan et al., 2006) and FimH (Yakovenko et al., 2015), it has been postulated that the kinetics are faster for the lower-affinity state. If after binding to the low-affinity state subsequent conformational change to the high-affinity state is rapid, fast ligand-binding kinetics to the low-affinity state efficiently couples ligand binding to stabilization by applied force of the high-affinity state, which has a long lifetime (Yakovenko et al., 2015). However, the ligand-binding kinetics of the states of these receptors have not yet been measured. Work from our group on integrin αVβ6 showed that removal of the hybrid domain in the αVβ6 head resulted in a 50-fold increase in affinity for ligand yet decreased the apparent on-rate of ligand binding (Dong et al., 2018). However, whether this is related to an increase in the population of the open conformation could not be determined due to the lack of tools to stabilize specific conformational states. There is a similar lack of measurements of state-specific ligand-binding kinetics on other type I single-pass transmembrane receptors, many of which have both inactive and active conformations. In contrast, the field is more advanced for multipass receptors, such as the β2-adrenergic receptor in the G protein-coupled receptor family, which is stabilized by binding of intracellular G proteins in a high-affinity state that has slower on- and off-rates (DeVree et al., 2016). In this study, we utilized well-characterized conformation-specific Fabs (Li and Springer, 2018; Li et al., 2017; Su et al., 2016; Figure 1C) at saturating concentrations to stabilize integrins α4β1 and α5β1 into defined ensembles containing only one or two of the three integrin conformational states and measured the ligand-binding kinetics of each defined ensemble. Together with previously determined intrinsic ligand-binding affinities and populations of conformational states (Li and Springer, 2018; Li et al., 2017), our measurements enable us to define ligand-binding kinetics intrinsic to each conformational state. For each integrin, the two closed states have indistinguishable on- and off-rates for ligands. Remarkably, the on-rate for ligand of the low-affinity closed integrin conformations is ~40-fold (α4β1) or ~5-fold (α5β1) higher than that for the high-affinity EO conformation. The ~1000-fold higher affinity of the EO conformation than the closed conformations is achieved by the ~25,000-fold lower off-rate of the EO conformation for both α4β1 and α5β1 integrins. These findings suggest for two representative β1 integrins that most ligand binding occurs to the BC and/or EC states. These measurements have important implications for the order of the steps in integrin activation. Results Ligand-binding kinetics of intact α4β1 and α5β1 on cell surfaces We measured binding kinetics of intact α4β1 on Jurkat cells to two fluorescently labeled ligands, a phenylureide derivative of Leu-Asp-Val-Pro (FITC-LDVP) and a fragment of vascular cell adhesion molecule (VCAM) containing its first two domains (Alexa488-VCAM D1D2) (Figure 2). Before adding ligands, cells were equilibrated with saturating concentrations of Fabs for 30 min at 22°C to stabilize specific conformational states (Li and Springer, 2018). Integrin extension, that is, the EC and EO states, was stabilized with 8µM 9EG7 Fab, which binds to the β1-subunit knee (Figure 2B). The EO conformation was stabilized with a combination of 4 µM 9EG7 Fab and 2 µM HUTS4 Fab; the latter binds near the interface between the βI andhybrid domains and stabilizes the EO conformation (Figure 2C). Ligand-binding was monitored as mean fluorescence intensity (MFI) by flow cytometry without washing (Figure 2). Beginning at about 10 min, a 500-fold higher concentration of unlabeled ligand was added to measure the kinetics of dissociation. Background MFI at each fluorescent ligand concentration, measured under identical conditions except in the presence of 10 mM EDTA, showed no significant difference at different time points during the association and dissociation measurements (Figure 2—figure supplement 1) and was averaged across different time points and subtracted to obtain specific binding. Figure 2 with 2 supplements see all Download asset Open asset Binding kinetics of ligands to intact α4β1 on Jurkat cells. (A–E) Binding and dissociation of FITC-LDVP (A–C) and Alexa488-VCAM D1D2 (D, E) to α4β1 on Jurkat cells measured by flow cytometry. Cartoons in panels A–C show the schemes for measuring kon in the association phase and koff in the dissociation phase in the basal ensemble (A), the extended ensemble stabilized with Fab 9EG7 (8 μM) (B, D), and the open ensemble stabilized with Fabs 9EG7 (4 μM) and HUTS4 (2 μM) (C, E). Specific mean fluorescence intensity (MFI) with the MFI in EDTA (Figure 2—figure supplement 1) subtracted is shown as open (association) or filled (dissociation) symbols; fits are shown as thin lines. (F) Tabulation of results. konapp and koffapp are from global fits of data at all ligand concentrations. koffapp/konapp is also shown and compared to the equilibrium Kd measurements shown in Figure 4A and D of our previous work (Li and Springer, 2018), except for Alexa488-VCAM D1D2 binding to extended states, which was measured here (Figure 2—figure supplement 2). Errors for Kd values are SD from three independent experiments. Kinetic constants were determined multiple times. Kinetic constants determined in a preliminary experiment were used to determine the range of concentrations and times used in the experiment shown. Each of the three concentrations in the experiment shown is an independent experiment capable of determining the reported kinetics. By globally fitting kinetic constants to all three analyte concentrations, we obtained single on- and off-rates that combine the measurements in the three independent determinations and at the same time report errors in concentrations used and model assumptions about concentration dependence. Errors for konapp and koffapp values are standard error (SE) from global nonlinear least-square fitting of data at all ligand concentrations. From SE, the 95% confidence interval for the fitting parameter can be computed as [fitting value – 2 * SE, fitting value + 2 * SE]. Errors for koffapp/konapp are propagated from errors of konapp and koffapp . Figure 2—source data 1 Source data for Figure 2A. https://cdn.elifesciences.org/articles/73359/elife-73359-fig2-data1-v2.txt Download elife-73359-fig2-data1-v2.txt Figure 2—source data 2 Source data for Figure 2B. https://cdn.elifesciences.org/articles/73359/elife-73359-fig2-data2-v2.txt Download elife-73359-fig2-data2-v2.txt Figure 2—source data 3 Source data for Figure 2C. https://cdn.elifesciences.org/articles/73359/elife-73359-fig2-data3-v2.txt Download elife-73359-fig2-data3-v2.txt Figure 2—source data 4 Source data for Figure 2D. https://cdn.elifesciences.org/articles/73359/elife-73359-fig2-data4-v2.txt Download elife-73359-fig2-data4-v2.txt Figure 2—source data 5 Source data for Figure 2E. https://cdn.elifesciences.org/articles/73359/elife-73359-fig2-data5-v2.txt Download elife-73359-fig2-data5-v2.txt Under basal condition with all three integrin states present in the ensemble, binding of FITC-LDVP to Jurkat cells reached equilibrium within 3 min (Figure 2A). Upon addition of a 500-fold excess of LDVP, dissociation of FITC-LDVP was rapid and was 99.7% complete by 5 min (Figure 2A). In contrast, both binding and dissociation of FITC-LDVP were slower when only the extended conformations (EC and EO) were present on Jurkat cells (Figure 2B). Reaching steady state required ~5 min after addition of 20 nM FITC-LDVP, ~10 min with 10 nM ligand, and was not reached after 10 min with 5 nM ligand. After 10 min of dissociation, only 19.4% of ligand had dissociated (Figure 2B). Association and dissociation were even slower when only the EO conformation was present (Figure 2C). After 15 min of association, much less ligand had bound (Figure 2C) than when both EC and EO conformations were present (Figure 2B). Dissociation was also slower, with only 1.2% of bound ligand dissociating after 10 min (Figure 2C). VCAM D1D2 binds with ~100-fold lower affinity than LDVP to α4β1 (Figure 5 in Li and Springer, 2018). As a result, binding to the basal ensemble was too low to measure over the noise from unbound ligand; however, we were able to measure binding kinetics to intact α4β1 stabilized in the extended (EC + EO) and EO states (Figure 2D and E). When the two extended conformations (EC and EO) were present, binding of all three concentrations of Alexa488-VCAM D1D2 (10 nM, 20 nM, and 30 nM) reached equilibria within 2 min. Upon addition of a large excess of LDVP, dissociation of Alexa488-VCAM D1D2 was also fast; 100% dissociated by 5 min (Figure 2D). Association and dissociation both became markedly slower when only the EO conformation of α4β1 was present (Figure 2E). To address the generality of these results, we studied another integrin and cell type by measuring binding of a fluorescently labeled two-domain fragment of fibronectin (Alexa488-Fn39-10) to intact α5β1 integrin on K562 cells (Figure 3). The BC conformation of α5β1 integrin on K562 cells is more stable than that of α4β1 integrin on Jurkat cells (Li and Springer, 2018). Therefore, to assure that the extended states (EC + EO) were saturably populated, they were stabilized with a combination of two Fabs, 6 µM 9EG7 Fab and 2 µM SNAKA51 Fab (Figure 3A, schematic). The EO state of α5β1 was stabilized with the same combination of Fabs as used for α4β1 (Figure 3B, schematic). Although binding affinity was too low to measure kinetics of the basal ensemble (Li et al., 2017), we were able to measure Alexa488-Fn39-10 kinetics with the EC + EO and EO ensembles of intact α5β1 (Figure 3). When α5β1 was stabilized in the EO conformation, Alexa488-Fn39-10 bound and dissociated significantly more slowly than when both the EC and EO states of α5β1 were present in the ensemble (Figure 3A and B). Faster binding and dissociation of Alexa488-Fn39-10 from the EC + EO ensemble than EO showed that the EC state of α5β1 binds and dissociates faster than the EO state, just as found for α4β1. Figure 3 Download asset Open asset Binding kinetics of Alexa488-Fn39-10 to α5β1 on K562 cells. (A, B) Binding of Alexa488-Fn39-10 to α5β1 on K562 cells measured by flow cytometry. Measurements were on integrins in extended ensembles (EC + EO ) in the presence of Fabs 9EG7 (6 μM) and SNAKA51 (2 μM) (A) or in the open (EO state) in the presence of Fabs 9EG7 (6 μM) and HUTS4 (2 μM) (B), as illustrated in the cartoons. Mean fluorescence intensity (MFI) with background in EDTA subtracted (Figure 2—figure supplement 1) is shown as symbols and fits are shown as lines; the association phase has open symbols and solid lines, and the dissociation phase has filled symbols and dashed lines. (C) Tabulation of results. Details about experimental repeats and kinetic parameter uncertainty, including errors and confidence intervals, are as described in Figure 2 legend. koffapp/konapp is also shown with propagated error and compared to equilibrium Kd measurements shown in Figure 7B of our previous study (Li et al., 2017). Figure 3—source data 1 Source data for Figure 3A. https://cdn.elifesciences.org/articles/73359/elife-73359-fig3-data1-v2.txt Download elife-73359-fig3-data1-v2.txt Figure 3—source data 2 Source data for Figure 3B. https://cdn.elifesciences.org/articles/73359/elife-73359-fig3-data2-v2.txt Download elife-73359-fig3-data2-v2.txt To quantify the binding kinetics of intact α4β1 and α5β1 under each condition, we globally fit specific binding in both association and dissociation phases at each concentration of fluorescently labeled ligand to the 1 vs. 1 Langmuir binding model (lines in Figure 2A-E and Figure 3A, B). Apparent on- and off-rates, konapp and koffapp, were well fit, with low fitting errors relative to the measured values (Figure 2F and Figure 3 C). The ratio of the apparent off- and on-rates, koffapp/konapp , agrees reasonably well with the previously determined equilibrium dissociation constants shown as Kd in Figure 2F and Figure 3C (Li and Springer, 2018; Li et al., 2017). These agreements suggest that the 1 vs. 1 Langmuir binding model can reasonably fit the kinetic data; that is, the kinetics of conformational change between states is sufficiently fast relative to the kinetics of ligand binding to not perturb ligand-binding kinetics. Thus, the relative populations of unbound states in the ensemble were not perturbed by removal by binding to ligand. Overall, these results show that ligand binds to and dissociates from the EO conformation more slowly than from the BC and EC conformations. The kinetics measured here for the basal and EC + EO ensembles are apparent because they include contributions from distinct conformational states present in these ensembles. In contrast, EO state kinetics are measured exactly because EO is the only state present in the EO ensemble. In the final section of 'Results,' we will use previous measurements of the populations of the states in each ensemble to calculate the on- and off-rates for conformations within mixtures of states. Binding kinetics of soluble α5β1 ectodomain for Fn39-10 We utilized biolayer interferometry (BLI) (Wallner et al., 2013) to measure the kinetics of binding of an ectodomain fragment of α5β1 to the biotin-labeled Fn39-10 fragment of fibronectin immobilized on streptavidin biosensors (Figure 4). The ectodomain was truncated just prior to the transmembrane domains of the α5 and β1 subunits and was expressed in a cell line containing a glycan processing mutation so that it had high-mannose rather than complex-type N-glycans. Truncation of α5β1 and high mannose glycoforms substantially raises the basal population of the EO relative to that of the EO state of intact α5β1 on cell surfaces ( Figures 4 and 7E in Li et al., 2017), enabling measurement here of basal ensemble Fn39-10 binding kinetics. Figure 4 Download asset Open asset Binding kinetics of α5β1 ectodomain to Fn39-10. (A–D) Binding of unclasped high-mannose α5β1 ectodomain measured with biolayer interferometry (BLI). Schemes for measuring ligand binding and dissociation in the association phase and dissociation phase are shown in each panel. α5β1 ectodomain (analyte) at the indicated concentrations in nM was bound to biotin-Fn39-10 immobilized on streptavidin biosensors without Fab (A), with 2 μM Fab 8E3 (B), with 2 μM 9EG7 and 5 μM HUTS4 Fabs (C), or with 1 μM Fab 12G10 (D). Arrows mark the start of the dissociation phase. Response curves are in gray and fitting curves in black. (E) The equilibrium binding (response) was calculated from konapp and koffapp values at each α5β1 ectodomain concentration and fit to a dose–response curve to calculate Kd values. These values serve as a check on the koffapp/konapp values in (F). (F) konapp and koffapp values from nonlinear least-square fit of data in panels (A–D) and Kd values from equilibrium response analysis in panel (E) with 1 vs. 1 Langmuir binding model, and koffapp/konapp values. Errors with * are difference from the mean of two measurements at the same concentrations of α5β1 ectodomain from two independent purifications. Errors without * for koffapp , koffapp, and Kd are SE from nonlinear least-square fits with single koffapp and koffapp as global parameters fitting to all analyte concentrations. Details about experimental repeats and uncertainties for konapp and koffapp , including errors and confidence intervals, are as described in Figure 2 legend. Errors for koffapp/konapp are propagated from errors of konapp and koffapp . Figure 4—source data 1 Source data for Figure 4A. https://cdn.elifesciences.org/articles/73359/elife-73359-fig4-data1-v2.txt Download elife-73359-fig4-data1-v2.txt Figure 4—source data 2 Source data for Figure 4B. https://cdn.elifesciences.org/articles/73359/elife-73359-fig4-data2-v2.txt Download elife-73359-fig4-data2-v2.txt Figure 4—source data 3 Source data for Figure 4C. https://cdn.elifesciences.org/articles/73359/elife-73359-fig4-data3-v2.txt Download elife-73359-fig4-data3-v2.txt Figure 4—source data 4 Source data for Figure 4D. Download Binding kinetics were measured by Fn39-10 biosensors to containing the α5β1 ectodomain in the or presence of Dissociation kinetics were measured by of to the integrin but containing identical Fab concentrations (Figure Kd values were previously shown to be independent of the Fab used to stabilize a state (Figure 3 in Li et al., 2017). However, we were that binding of Fabs, that bind near ligand-binding slow kinetics and this by the Fabs used to stabilize the EO state. The kinetic curves showed that the α5β1 ectodomain EO state more slowly than the mixtures with the closed states and also dissociated more slowly (Figure Overall, these ensembles found for the EC + EO ensemble and EO state of intact α5β1 on K562 cells and extended measurements to the basal α5β1 ensemble. The on- and off-rates of the EO state for Fn39-10 determined in the presence of 12G10 Fab were and than determined in the presence of Fabs (Figure D and As 12G10 Fab binds to the ligand-binding in the β1 domain (Figure we use koffapp and konapp kinetics determined with the Fabs, which bind from the ligand-binding for and kinetic for each state in the final section of The off-rates of the closed states to the low affinities of the closed states was too binding to measure kon or koff in the presence of saturating We used another We first ligand binding to integrins to steady state in the of a We added different concentrations of Fab and measured dissociation kinetics (Figure 5 and Figure Dissociation of the ligand from the EO state is very slow as shown and is in our experimental time high Fab concentrations, when the EO state to or here as Fab binds and to (Figure and B). After saturating concentrations of Fab are added to basal or EC ensembles with ligand, the off-rate is by two the conformational change from to and the dissociation of ligand from (Figure and B). Thus, the off-rate at saturating concentration of Fab is by the of both steps and the of the lower of . Figure 5 Download asset Open asset Dissociation of FITC-LDVP from α4β1 on Jurkat cells in the presence of (A, B) FITC-LDVP dissociation from basal or extended ensembles of intact α4β1 on Jurkat cells measured flow cytometry. FITC-LDVP nM) was with Jurkat cells in the (A) or in the presence of Fab 9EG7 (4 μM) (B) for 10 min to steady state. 10 μM unlabeled LDVP, with indicated concentrations of Fab, were mean fluorescence intensity values as a function of time at indicated Fab concentrations were globally to , with MFI at the start of dissociation and background MFI as parameters and koffapp as the individual fitting parameter at each Fab (C) of koffapp on Fab koffapp at each Fab concentration in panels (A) and (B) were to dose–response curves to determine the off-rate at saturating Fab concentration, , and the Fab concentration when the off-rate of the . in panels (A) and (B) were first with and 4 μM Fab for and with the conditions with similar koffapp values in each errors are SE from nonlinear least-square Figure data 1 Source data for Figure Download Figure data 2 Source data for Figure Download Figure 6 Download asset Open asset Dissociation of α5β1 ectodomain from biotin-Fn39-10 in the presence of (A, B) high-mannose α5β1 ectodomain dissociation from biotin-Fn39-10 immobilized on streptavidin biosensors was monitored by biolayer interferometry (BLI). Reaction schemes are illustrated in each panel. nM α5β1 ectodomain was with biotin-Fn39-10 biosensors for 10 min to binding in the (A) or the presence of 2 μM 9EG7 Fab were into the α5β1 ectodomain in the presence or of 9EG7 Fab as before and also containing the indicated concentrations of Fab for measurement of dissociation. The response at each Fab concentration as a function of time was to the single , for the response at the start of dissociation and koffapp . (C) of at saturating Fab koffapp was fit to Fab concentration a dose–response curve for the off-rate at saturating Fab concentration to determine . The Fab concentration when the off-rate of the was also in panels (A) and (B) were first with and 10 Fab for and with the conditions with similar koffapp values in each Errors are SE from nonlinear least-square Figure data 1 Source data for Figure Download Figure data 2 Source data for Figure Download We measured FITC-LDVP dissociation from basal or extended ensembles of α4β1 on Jurkat cells after addition of a range of Fab concentrations (Figure and B). binding of Fab to cell surface was from the to a of (Figure The
Source: eLife
Authors: Jing Li, Jiabin Yan, Timothy A Springer
Institutions: Boston Children's Hospital, Harvard University
Language: English
Subfield: Immunology and Allergy
Field: Medicine
Domain: Health Sciences
Open Access status: gold
Estimated APC paid: $2,500
OpenAlex ID: W4206341910