The role of the alveolar epithelial glycocalyx in acute respiratory distress syndrome

Abstract

The alveolar epithelial glycocalyx is a dense anionic layer of glycosaminoglycans (GAGs) and proteoglycans that lines the apical surface of the alveolar epithelium. In contrast to the pulmonary endothelial glycocalyx, which has well-established roles in vascular homeostasis and septic organ dysfunction, the alveolar epithelial glycocalyx is less understood. Recent preclinical studies demonstrated that the epithelial glycocalyx is degraded in multiple murine models of acute respiratory distress syndrome (ARDS), particularly those that result from inhaled insults (so-called “direct” lung injury), leading to shedding of GAGs into the alveolar airspaces. Epithelial glycocalyx degradation also occurs in humans with respiratory failure, as quantified by analysis of airspace fluid obtained from ventilator heat moisture exchange (HME) filters. In patients with ARDS, GAG shedding correlates with the severity of hypoxemia and is predictive of the duration of respiratory failure. These effects may be mediated by surfactant dysfunction, as targeted degradation of the epithelial glycocalyx in mice was sufficient to cause increased alveolar surface tension, diffuse microatelectasis, and impaired lung compliance. In this review, we describe the structure of the alveolar epithelial glycocalyx and the mechanisms underlying its degradation during ARDS. We additionally review the current state of knowledge regarding the attributable effect of epithelial glycocalyx degradation in lung injury pathogenesis. Finally, we address glycocalyx degradation as a potential mediator of ARDS heterogeneity, and the subsequent value of point-of-care quantification of GAG shedding to potentially identify patients who are most likely to respond to pharmacological agents aimed at attenuating glycocalyx degradation.

INTRODUCTION

Acute respiratory distress syndrome (ARDS) is a devastating critical illness that affects over 200,000 patients in the United States per year and is associated with substantial morbidity and a mortality rate of 30%–35% (1–3). This complex syndrome is characterized by pulmonary vascular leak, surfactant dysfunction, impaired lung compliance, and hypoxemic respiratory failure (4, 5). Unfortunately, despite decades of preclinical research, clinical trials in ARDS have largely failed to identify pharmacological treatments that improve patient outcomes (6). The inability of promising preclinical data to translate to successful clinical trials is increasingly attributed to the substantial heterogeneity that exists within ARDS. Consequently, randomized studies have generally neglected to match treatment allocation to the mechanisms driving an individual patient’s lung injury (7). The development and personalized implementation of efficacious ARDS therapeutics demands both an understanding of these distinct mechanisms of lung injury and the identification of biomarkers that can determine the predominant driver of injury in each individual with ARDS (8–10).

One potentially targetable mechanism of ARDS pathogenesis is the degradation of the alveolar epithelial glycocalyx, a mesh-like layer composed of glycosaminoglycans (GAGs), proteoglycans, and glycoproteins lining the alveolar surface. In animal models of lung injury (as well as in patients with ARDS), the epithelial glycocalyx is degraded, shedding GAG fragments into the alveolar airspace (11, 12). In humans, the magnitude of this shedding is highly heterogeneous, with some patients with ARDS having little evidence of epithelial glycocalyx degradation (13). However, other patients with ARDS demonstrate substantial airspace GAG shedding, which is predictive of prolonged duration of mechanical ventilation and hospital course (13). GAG shedding can be easily and rapidly identified at the bedside, using a point-of-care assay of noninvasively collected airspace fluid samples. Such assays are capable of rapidly identifying potentially targetable mechanisms of lung injury (so-called “treatable traits”) and may facilitate the implementation of personalized treatment strategies for ARDS (14).

In this review, we will describe the structure of the alveolar epithelial glycocalyx. We then review the existing literature on the mechanisms of epithelial glycocalyx degradation during lung injury and the attributable effect of glycocalyx degradation on ARDS pathogenesis. Finally, we explore the potential role for point-of-care quantification of airspace GAG shedding as a novel, and potentially causal, biomarker in ARDS, which may identify patients most likely to benefit from pharmacological agents aimed at either preventing epithelial glycocalyx degradation or promoting its regeneration.

ALVEOLAR EPITHELIAL GLYCOCALYX STRUCTURE

The presence of a continuous layer lining the alveolar epithelium has been appreciated since electron microscopy studies from the 1960s. These foundational studies by Drs. Weibel and Gil noted the presence of a superficial surface film, now known to be comprised of pulmonary surfactant, overlying an alveolar hypophase or “dense basal layer” hypothesized to be composed of mucopolysaccharides (i.e., GAGs) and proteins (15–17). Following these original anatomic descriptions, there have been decades of intense study on the structure and function of surfactant, which is now known to have critical biophysical and immunomodulatory functions (18, 19). In comparison, the alveolar hypophase, now recognized to be the alveolar epithelial glycocalyx, has been understudied (20). Emerging studies now demonstrate that the alveolar epithelial glycocalyx is a critical part of the epithelial surface and that its degradation contributes to lung injury pathogenesis following multiple different pulmonary insults.

The alveolar epithelial glycocalyx, like the glycocalyces of other endothelial and epithelial cells throughout the body, is predominantly composed of long unbranched GAGs that are anchored to the apical cell membrane by transmembrane or glycosylphosphatidylinositol-anchored proteoglycans (21). The main GAG species that make up this structure are heparan sulfate (HS, a linear polysaccharide composed of repeating units of N-acetyl glucosamine and a hexuronic acid—either glucuronic acid or its epimer, iduronic acid), chondroitin sulfate (CS, a linear polysaccharide composed of repeating units of N-acetyl galactosamine and a hexuronic acid), and hyaluronic acid (HA, an unsulfated linear polysaccharide composed of N-acetyl galactosamine and glucuronic acid) (22, 23) (Fig. 1). HS and CS are synthesized on a protein backbone in the Golgi apparatus by a complex series of steps including chain polymerization (mediated by exostosins-1 and -2), sulfation (mediated by sulfotransferases such as N-deacetylase/N-sulfotransferase-1 and -2), and epimerases (22). These sulfated GAG chains and their attached protein backbones, which include HS proteoglycans (HSPGs) or CS proteoglycans (CSPGs), such as syndecans, are then transported to the cellular surface, where they form the glycocalyx (24). Once extracellular, glycocalyx HS may undergoing additional modification by extracellular sulfatases such as sulf-1 and -2; this postsynthetic control of HS sulfation suggests homeostatic importance of glycocalyx sulfation. Indeed, sulfation of HS and CS at specific sites imparts geographic domains of negative charge across these molecules, allowing electrostatic binding to cognate positively charged residues of proteins. GAG binding modifies protein function, allowing GAGs to regulate downstream cellular signaling (25). HA, in contrast to HS and CS, is unsulfated and does not bind to proteoglycans (26). HA is synthesized by hyaluronan synthases, which are enzymes with multiple transmembrane domains that are located on the inner surface of the plasma membrane. During synthesis, the growing HA polymer is extruded through the membrane to the extracellular space, where it then anchors to the cell surface by binding to HA synthase enzymes or cell surface receptors such as CD44 (27). The extracellular HA also forms complexes with other GAGs, which enhances the structural stability of the glycocalyx (28). Together this mesh-like anionic layer supports many homeostatic functions of both endothelial and epithelial cells throughout the body. A detailed schematic of the structure of the epithelial glycocalyx is provided in an excellent review by Ochs et al. (20).

Figure 1.

Structure of the alveolar epithelial glycocalyx. The epithelial glycocalyx is a layer of glycosaminoglycans (GAGs) that line the epithelial surface. Heparan sulfate (HS) is a linear polysaccharide composed of repeating units of glucosamine and a hexuronic acid (either glucosamine or its epimer, iduronic acid). Chondroitin sulfate (CS) is a linear polysaccharide composed of repeating units of galactosamine and a hexuronic acid. Both HS and CS are attached to a proteoglycan backbone and can be sulfated at specific sites as indicated. Hyaluronic acid (HA) is a linear polysaccharide that is composed of galactosamine and glucuronic acid. It binds to cluster differentiation 44 (CD44) and is not covalently linked to proteoglycans of the epithelial surface and is not sulfated. Gal, galactosamine; GalNAc, N-acetyl galactosamine; GlcA, glucuronic acid; GlcNAc, N-acetyl glucosamine; IdoA, iduronic acid; Xyl, xylulose. The schematic figure was created using BioRender.

MECHANISMS OF ALVEOLAR EPITHELIAL GLYCOCALYX DEGRADATION

Recent studies have demonstrated that the alveolar epithelial glycocalyx is degraded in multiple murine models of lung injury, suggesting that it may play an important role in the pathogenesis of ARDS (11–13). Haeger et al. (12) found that GAGs, including HS and CS, are shed into the airspaces, but not the circulation, of mice following the administration of intratracheal lipopolysaccharide (LPS). These results are supported by in vitro experiments in which LPS induced HS shedding in cultured A549 epithelial cells (29). Interestingly, these findings are not unique to LPS, as HS is also shed into the bronchoalveolar lavage (BAL) fluid in murine models of bleomycin-induced lung injury, where full-length and highly sulfated HS was observed to persist in the airspaces for 3 wk after the initial injury (11). Furthermore, similar epithelial glycocalyx degradation occurs after influenza infection, which induces a profound and prolonged GAG shedding into the airspaces (30). In contrast to the substantial airspace GAG shedding observed in these models characterized by initial epithelial injury (so-called “direct” lung injury), we observed little evidence of alveolar epithelial glycocalyx degradation following cecal ligation and puncture (CLP), a model characterized primarily by pulmonary endothelial dysfunction (so-called “indirect” lung injury) (Fig. 2). Taken together, these preclinical findings suggest that direct forms of lung injury induce epithelial glycocalyx degradation, with minimal effect on the pulmonary endothelial glycocalyx. Conversely, indirect forms of lung injury induce prominent endothelial glycocalyx degradation with minimal damage to the epithelial glycocalyx. This compartmentalization of glycocalyx degradation in direct versus indirect lung injury is supported by observational studies of critically ill humans (31, 32).

Figure 2.

The epithelial and endothelial glycocalyces of the lung during health and lung injury. A: during homeostasis the lung has both a pulmonary endothelial glycocalyx and an alveolar epithelial glycocalyx. Lung injury from inhaled insults (“direct lung injury”) is associated with degradation of the alveolar epithelial glycocalyx. Conversely, lung injury from systemic insults (“indirect lung injury”) is largely characterized by degradation of the pulmonary endothelial glycocalyx. Notably, patients with acute respiratory distress syndrome (ARDS) often have multiple simultaneous insults, which can lead to degradation of both the endothelial and epithelial glycocalyces. B: direct pulmonary insults in mice (lipopolysaccharide, influenza, and bleomycin) induces shedding of heparan sulfate (HS) into the alveolar lining fluid (bronchoalveolar lavage, corrected for urea dilution) (11, 12, 30). C: indirect lung injury in mice (cecal ligation and puncture, CLP) does not induce shedding of HS into the alveolar lining fluid. CLP airspace samples pooled (n = 3/group). D: CLP does induce shedding of HS from the endothelial glycocalyx into the plasma (34). For B and D, n > 5 mice/group. Bars = standard error of the mean. Individual comparisons by t test; multiple comparisons by ANOVA. *P < 0.05. FFU, focus forming unit; IN, intranasal; IT, intratracheal; LPS, lipopolysaccharide. The schematic figure (A) was created using BioRender.

Consistent with this compartmentalization of glycocalyx shedding, the mechanisms driving endothelial and epithelial glycocalyx degradation are distinct. In sepsis-associated endothelial glycocalyx degradation, GAG shedding is mediated by activation of heparanase (an HS-specific glucuronidase), which directly degrades HS and releases small octasaccharide fragments into the plasma (33, 34). This contrasts with direct lung injury-associated alveolar epithelial glycocalyx degradation, during which the shed GAGs are largely intact (>20 saccharides in length) and highly sulfated (11, 12). The presence of full-length GAGs within the alveolar space suggests that epithelial glycocalyx degradation is mediated by enzymes that cleave the proteoglycans anchoring GAGs to the epithelial surface, rather than glucuronidases (such as heparanase) or other glycan-targeted enzymes that directly cleaves GAGs into small fragments (22). Indeed, inducible deletion of heparanase in mice after intratracheal bleomycin had no effect on persistent airspace GAG shedding after bleomycin (11). Similarly, constitutive deletion of heparanase did not prevent the magnitude of lung injury after intratracheal LPS (35). Notably, these results are in contradiction to the findings of Li et al. (29), who reported that pharmacological inhibition of heparanase attenuated LPS-induced damage to tight junctions in both A549 cells and murine LPS-induced lung injury. In addition, although much of the published literature has focused on mechanisms of HS shedding, it is important to note that CS and HA are also shed into the airspaces during lung injury (13). CS shedding could also be a consequence of syndecan shedding, which can have CS side chains (36). Alternatively, shedding of CS and HA may also be a consequence of shedding of versican, a large CS-containing proteoglycan that interacts with HA via its core protein (27, 37).

Recent work highlights that epithelial glycocalyx degradation during direct lung injury is likely mediated by multiple redundant proteases. For example, LPS and bleomycin both induce expression of matrix metalloproteinases (MMPs), which are proteases capable of cleaving proteoglycans (such as syndecan-1) anchoring GAGs to the epithelial surface (11, 12, 38–40). LPS is associated with an induction of both MMP 9 and (to a lesser extent) MMP 2; intriguingly, genetic knockout of MMP 9 does not attenuate epithelial glycocalyx shedding, perhaps due to compensatory upregulation of MMP 2 (12). Similarly, bleomycin induced an early activation of alveolar MMP 2, followed by later induction of MMP 9 (11). In addition, work by Gill et al. (40) demonstrated that MMP7 is an important protease that promotes shedding of syndecan-1 from the epithelial glycocalyx. Together, these findings suggest that broad MMP inhibition may be necessary to attenuate glycocalyx shedding in direct lung injury. Indeed, doxycycline, a broad inhibitor of MMPs and related proteases, attenuates the severity of epithelial glycocalyx shedding and/or lung injury in LPS, bleomycin, or influenza infection (11, 12, 41). Other proteases, such as a disintegrin-like metalloproteinases (ADAMs), likely also contribute to the extent of epithelial glycocalyx degradation during lung injury. Pruessmeyer et al. (42) demonstrated that ADAM17 mediates shedding of syndecan-1 and -4 from the epithelial cell surface in response to tumor necrosis factor (TNF)α and interferon (IFN)γ. Together, this body of work suggests that a complex system in which a redundant induction of proteases, including MMPs and ADAMs, mediate epithelial glycocalyx degradation following direct lung injury.

CONSEQUENCES OF ALVEOLAR EPITHELIAL GLYCOCALYX DEGRADATION DURING ARDS

Epithelial Barrier Dysfunction

To determine the attributable effect of epithelial glycocalyx degradation in lung injury, our group developed a murine model of targeted, enzymatic (bacterial heparinase-induced) degradation of the epithelial glycocalyx. Using this model, we observed that targeted epithelial glycocalyx degradation was sufficient to induce an increase in BAL protein content. This increased alveolar permeability to protein surprisingly occurred in the absence of alveolar inflammation or the development of pulmonary edema, suggestive that epithelial heparan sulfate degradation was not simply causing nonspecific injury to the alveolar septum (12). Interestingly, these findings were specific to HS degradation, as similar targeted degradation of epithelial CS with chondroitinases did not impair epithelial barrier function in mice (12). Although the precise mechanisms by which epithelial glycocalyx shedding causes epithelial barrier dysfunction are unknown, work by Brauer et al. (43) demonstrated that cell membrane-associated syndecan-1 regulates the response to influenza by facilitating cytoprotective signaling that limits bronchial epithelial apoptosis and attenuates lung injury. These findings suggest that epithelial glycocalyx degradation may impair barrier function by increasing epithelial apoptosis, which could explain the findings of increased barrier permeability in the absence of a significant inflammatory response that was reported during targeted epithelial glycocalyx degradation (12). In addition, it is critical to note that the importance of the epithelial glycocalyx to barrier integrity is not limited to the lung, as work by multiple groups has demonstrated the importance of the glycocalyx to barrier regulation in other epithelial interfaces (including the gut and the eye) (44, 45).

Surfactant Dysfunction

Surfactant is a complex mixture of lipids (90%) and proteins (10%) secreted by alveolar type II epithelial cells into the alveolar space, where it serves to reduce surface tension at the air-liquid interface to prevent alveolar collapse during exhalation (46). The alveolar epithelial glycocalyx is interposed between epithelial cells and surfactant; however, technical challenges in visualization of the epithelial glycocalyx structures have limited investigation into the potential interactions between these two structures (20). We recently reported that targeted degradation of HS from the alveolar epithelial glycocalyx impairs surfactant function, leading to microatelectasis and impaired lung compliance (12, 13). Interestingly, no loss of compliance was induced by the addition of exogenous GAG fragments (HS or CS) to uninjured mice, suggesting that surfactant dysfunction arises from injury to the native glycocalyx, rather than the presence of GAG fragments in the airspaces.

The clinical relevance of these findings is supported by an observational human study in which airspace GAG fragments were predictive of duration of respiratory failure (13). In addition, patients with lysosomal storage diseases, such as mucopolysaccharidosis type IIIA, exhibit impaired surfactant metabolism and decreased lung function (47). Although the precise mechanisms by which the epithelial glycocalyx supports surfactant function are uncertain at this time, we observed that heparin, a highly sulfated form of HS, can directly bind to surfactant proteins A, B, and D (13). Surfactant proteins A and D both belong to the collectin family of proteins, which contain a carbohydrate recognition domain capable of binding to glycans (48). We speculate that an intact epithelial glycocalyx could serve to tether surfactant to the epithelium, ensuring optimal surfactant distribution necessary for the maintenance of alveolar recruitment.

Impaired Epithelial Repair

In addition to these acute effects on pulmonary physiology, alveolar epithelial glycocalyx degradation impedes recovery from injury. During bleomycin-induced lung injury, a model in which lung injury is followed by a prolonged period of recovery, HS fragments persist in the air spaces for 3 wk after the initial insult. Treatment with doxycycline, a broad MMP inhibitor, starting 7 days after bleomycin administration attenuated persistent GAG shedding and improved lung function, suggestive that ongoing glycocalyx degradation may impede the process of epithelial repair (11). Although the exact mechanisms underlying these findings are unknown, work by Altemeier et al. (49) demonstrates that syndecan-1 is important for lung epithelial migration and adhesion, which suggests that shedding of syndecan-1 may contribute to impaired epithelial repair. In addition, work by Parimon et al. (50, 51) demonstrates that syndecan-1 shedding influences the progression or resolution of lung injury in both lung fibrosis, by regulating epithelial reprogramming through extracellular vesicles, and lung tumorigenesis, by regulating exosomal miRNAs. Furthermore, given that the epithelial glycocalyx is necessary for barrier function, this prolonged GAG shedding may delay recovery of normal alveolar-capillary integrity (12). Finally, shed GAG fragments are also biologically active and capable of binding to and sequestering lung-reparative growth factors, such as fibroblast growth factor (FGF) and hepatocyte growth factor (HGF), thus providing another possible mechanism by which persistent glycocalyx degradation impairs lung recovery (11).

Increased Severity of Secondary Bacterial Pneumonia

In addition to the direct effects of alveolar epithelial glycocalyx degradation on lung injury and repair, epithelial glycocalyx degradation may influence host-pathogen interactions. Although it is well known that viral pneumonias place the lung at increased susceptibility for severe secondary bacterial pneumonias, the mechanisms by which this occurs are incompletely understood (52). Our group recently reported that the influenza-injured lung microenvironment induces methicillin-resistant staph aureus (MRSA) to upregulate the pore forming cytotoxin leucocidin A/B (LukAB), which contributes to the increased severity of secondary MRSA pneumonia. Interestingly, alveolar epithelial HS that is shed during influenza infection binds and activates LukAB, contributing to the severity of postinfluenza MRSA pneumonia (30). Studies of syndecan-1 shedding in corneal infections suggest that epithelial glycocalyx shedding may promote secondary bacterial infection via additional mechanisms, such as promoting Streptococcal adhesion to cryptic fibronectin sites on the epithelial surface (53) or sequestration of host-protective cationic antimicrobial peptides (54).

THE ROLE OF HYALURONIC ACID IN THE EPITHELIAL GLYCOCALYX

As described earlier, HA differs from HS and CS in that it is both unsulfated and not covalently bound to the epithelial surface. Given these structural differences, it is not surprising that HA also has distinct functions within the epithelial glycocalyx (27, 55). Like other GAGs, epithelial HA is shed in multiple forms of lung injury, including inflammatory lung disease and pulmonary fibrosis, via the action of hyaluronidases and inflammatory stimuli, such as reactive oxygen species (13, 55, 56). Furthermore, in the aftermath of injury, the lung generates a provisional HA glycocalyx which can have both protective and deleterious effects in a context-dependent manner (27, 55, 57–59). For example, increased HA production and/or impaired HA clearance causes an ongoing innate inflammatory response that is mediated by HA-binding proteins including tumor necrosis factor (TNF)-stimulated gene 6 (57, 60). Here, increased HA leads to the generation of heavy-chain-HA complexes, which can be degraded into smaller LMW fragments that can engage cell receptors, including CD44, TLR4, and TLR2, and impact downstream biological effects including inflammation and fibrosis (58). In addition, recent work has also demonstrated that imbalances in HA impact the resolution of influenza in mice, during which HA complexes with the HA-binding protein inter-α-inhibitor, ultimately leading to ongoing inflammation and decreased lung function, an effect that can be abrogated with administration of exogenous hyaluronidases (59). The clinical importance of epithelial HA in inflammatory lung disease is also highlighted by recent work by Kratochvil et al. (61), which demonstrated that HA is a major component of the respiratory secretions of patients with COVID-19. Finally, in addition to these inflammatory consequences of HA shedding during lung injury, HA fragments and associated proteins, such as inter-alpha-inhibitor, have been shown to influence epithelial sodium channel activity, which could impact alveolar fluid clearance and ARDS pathogenesis (62, 63).

ALVEOLAR EPITHELIAL GLYCOCALYX DEGRADATION AS A MEDIATOR OF ARDS HETEROGENEITY

The failure of strong preclinical data to translate into efficacious ARDS therapies is increasingly attributed to the mechanistic heterogeneity of this complex illness (3, 7, 64). Calfee et al. (65) identified two distinct phenotypes of ARDS (8), each of which may experience different (and possibly opposing) treatment responses to therapeutics. This work inspired interest in precision medicine approaches to ARDS care, in which rapid, biomarker-based identification of the specific mechanisms driving lung injury would allow for personalization of targeted therapies (10, 14). Epithelial glycocalyx degradation may be one such heterogeneous and potentially targetable mechanism that contributes to lung injury in patients with ARDS. Interestingly, the heterogeneity that has been noted in GAG shedding in ARDS depends on both injury-specific factors (direct vs. indirect) as well as unexpected patient-level factors (such as sex) (13).

The ability to rapidly identify a subgroup of patients in whom lung injury is driven by epithelial glycocalyx degradation would be highly useful, as it would enable predictive and/or prognostic enrichment of clinical trials aimed at determining the efficacy of therapies preventing glycocalyx degradation. GAGs are a promising biomarker in part due to their durability: they are unaffected by proteases and are durable in freeze-thaw cycles (66). Although mass spectrometry with multiple reaction monitoring is the gold standard for GAG quantification, this approach is expensive and relies on a high level of technical expertise. Our group has recently optimized a simple and inexpensive ($2 per sample) colorimetric assay (dimethylmethylene blue) of GAGs that is accurate in both urine (where it reflects excreted circulating GAGs and thus endothelial glycocalyx degradation) and HME fluid (where it reflects alveolar epithelial glycocalyx degradation) (13, 67). Assays such as these could be useful in identifying a subgroup of patients most likely to benefit from therapeutic strategies aimed at preventing GAG shedding, such as MMP inhibitors or MMP-blocking monoclonal antibodies (68, 69).

CONCLUSIONS

The alveolar epithelial glycocalyx is a complex and understudied structure with importance to both lung homeostasis and lung injury pathogenesis. During injury, multiple inflammatory stimuli activate MMPs and other (likely redundant) proteases, cleaving HSPGs and CSPGs that anchor the glycocalyx to the epithelial surface. The subsequent shedding of the epithelial glycocalyx has important consequences that contribute to the onset and progression of lung injury, including alveolar hyperpermeability, disruption of surfactant function, increased bacterial virulence, and impaired epithelial cell repair. Additional observational studies in humans are needed to better understand the contribution of epithelial glycocalyx degradation to lung injury heterogeneity, potentially allowing the epithelial glycocalyx to serve as a therapeutic target in precision medicine approaches for ARDS care.

GRANTS

This work was supported by the National Heart, Lung, and Blood Institute Grants R01 HL125371 (to E. P. Schmidt) and F32 HL162230 (to A. N. Rizzo) and the Congressionally Directed Medical Research Program Grant W81XWH-18-1-0682 (to E. P. Schmidt).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

A.N.R. and E.P.S. conceived and designed research; A.N.R. performed experiments; A.N.R. and E.P.S. analyzed data; A.N.R. and E.P.S. interpreted results of experiments; A.N.R. prepared figures; A.N.R. and E.P.S. drafted manuscript; A.N.R. and E.P.S. edited and revised manuscript; A.N.R. and E.P.S. approved final version of manuscript.