Exosomes, which are nano-sized vesicles ranging between small diameters of about 30–150 nm, are derived from the endosomal system. The formation and secretion of exosomes begin with the invagination of the plasma membrane (endocytosis) to form a cup-like structure harboring cell-surface proteins and extracellular deposits (Kalluri and LeBleu, 2020). This invagination forms an early endosome which then mature to form late endosomes. Subsequently, intraluminal vesicles (ILVs) are formed by the progressive, inward budding of the late endosomal membrane. The accumulation of ILVs in the late endosome generates multivesicular bodies (MVBs) (Kalluri and LeBleu, 2020; Yue et al., 2020). At this stage, MVBs can fuse with autophagosomes or lysosomes for the degradation of their endosomal contents. Degraded components can then be recycled by the cell. Alternatively, MVBs (containing mature or late endosomes) that do not traffic to lysosomes move toward the cytoplasmic side of the plasma membrane with the aid of the cellular cytoskeletal and microtubule network (Kalluri and LeBleu, 2020). Here, MVBs are able to dock via MVB-docking proteins and fuse with the plasma membrane as exocytic MVBs. Exocytic MVBs release their lipid bi-layered ILV contents to the extracellular space as exosomes (Kalra et al., 2016) (Figure 1).

It still remains to be completely understood, the processes that govern the formation of ILVs in MVBs and their further release as exosomes. However, a proposed mechanism implicated in this process involves the Endosomal Sorting Complex Required for Transport (ESCRT). ESCRTs comprise of several proteins that assemble into four ESCRT complexes: namely ESCRT-0, -I, -II, and -III, in association with others like ALIX and vacuolar protein sorting-associated proteins (VPS4, VTA 1) (Colombo et al., 2013). Initiating this process is the organization of endosomal membranes into specialized units that are highly enriched for the tetraspanin class of membrane proteins. Tetraspanins (e.g., CD63, CD9, and CD81) function to cluster the required proteins for ILV formation and are considered common markers for the identification of exosomal vesicles (Pols and Klumperman, 2009). ESCRTs are then recruited to the site of ILV formation where ESCRT-0 recognizes and binds to ubiquitinated proteins present on the outside of the endosomal membrane. ESCRTs-I and -II are then assembled to the cytosolic area to initiate and drive intraluminal membrane budding. The assembling of ESCRT-I/II is reportedly stimulated by some factors, such as the abundance of phosphatidylinositol 3-phosphate (PIP3) in the membrane early endosomes, ESCRT-0 proteins (e.g., hepatocyte growth factor-regulated tyrosine kinase substrate-HRS) and the ubiquitination of the cytosolic tail of endocytosed proteins and/or their curved membrane topology (Henne et al., 2013; Abels and Breakefield, 2016). Upon activation of ESCRT-II, ESCRT-III is recruited through an associated protein, ALIX. ALIX binds to the tumor susceptibility gene 101 (TSG101) component of ESCRT-I complex and also to the charged MVB protein 4A (CHMP4A) components of ESCRT-III, thus, serving as an intermediate between ESCRT-III and ESCRT-I association (McCullough et al., 2008). ESCRT-III in conjunction with deubiquitinating enzymes then finalize the processes that involve vesicle closure and the detachment of ILVs from the membrane. ESCRT-III forms structures that constrict budding neck followed by scissoring of the membrane (driven by accessory proteins, such as VPS4) and finally, the dissociation and recycling of the ESCRT machinery (Kalra et al., 2016; Schöneberg et al., 2017).

Recent studies identify an alternate ESCRT-independent pathway for ILV formation which involves clustering of sphingomyelin in lipid rafts where it is converted to ceramide by sphingomyelinases. The accumulation of ceramide subsequently induces the merging of microdomain, thus, triggering ILV formation (Stuffers et al., 2009; Kalra et al., 2016). As such, exosomes presumed to be formed through this pathway may lack the expression of common markers, such as ALIX and TSG 101 involved in the endosomal pathway. Small RAB GTPases facilitate the docking, fusion and secretion of exosomes. For example, RAB27A promotes docking of MVBs and fusion to the plasma membrane, whiles RAB27B is involved in vesicle transfer from the Golgi to MVBs, and mobilization of MVBs to the actin-rich cortex under the plasma membrane (Ostrowski et al., 2010; Kalra et al., 2016).

Vesicles that arise through direct outward budding and fission of the plasma membrane are known as microvesicles (MVs) or ectosomes (Figure 1). These range between relatively larger diameters of about 50–2,000 nm (Akers et al., 2013). Ectosome biogenesis involves plasma membrane phospholipid re-arrangements/redistribution, as well as, contractions of cytoskeletal proteins.

Enzymes (collectively described as aminophospholipid translocases) involved in the exchange of lipids between the inner and outer leaflet of the cell membrane for asymmetric maintenance are activated to induce changes within the bilayer to allow for budding and membrane abscission. Flippases are translocases required for the transfer of phospholipids from the outer leaflet to the inner leaflet and floppases, on the other hand, transfer phospholipids from within out. Following this, actin and myosin interactions within the cytosol induce the contraction of cytoskeletal structures; a mechanism that underlies bud formation and subsequent release of ectosome portions. Microvesicles, like exosomes, can be highly enriched in sets of proteins and are particularly identified with markers, such as vesicle-associated membrane protein 3 (VAMP3) and ADP-ribosylation factor 6 (ARF6) (Hugel et al., 2005; Muralidharan-Chari et al., 2009).

During programmed cell death, apoptotic cells undergo several stages. These stages begin with condensation of the nuclear chromatin and then membrane blebbing. Cellular components that undergo disintegration are then processed into distinct membrane-enclosed vesicles. These vesicles comprise of a type of EV known as apoptotic bodies that range in sizes between 500 and 4,000 nm. Apoptotic bodies pack organelles or cytosolic components and are capable of transferring genetic information from cell to cell via their uptake (Bergsmedh et al., 2001; Cocucci et al., 2009; Kalra et al., 2016).

The regulated migration of lymphocytes is necessary for their effective function. This regulation phenomenon termed, “lymphocyte homing” enables for the trafficking of lymphocytes to specific sites of the body. Lymphocyte homing has been particularly demonstrated in gut immunology. Antigen-primed naïve T lymphocytes in the gut lymph nodes are imprinted to allow them home back into gut tissues as effector or memory T lymphocytes: a necessity for efficient adaptive immunity in the gut mucosa (Myint et al., 2020). High endothelial venules (HEV) of the gut associated lymphoid tissues (GALT) highly express mucosal addressin cell adhesion molecule-1 (MAdCAM-1) that engages the receptor integrin α4β7 expressed on activated lymphocytes. This interaction enables for the homing of antigen-specific B and T lymphocytes into gut tissues (Streeter et al., 1988; Berlin et al., 1995). The inability of β7 integrin deficient mice to mount antigen-mediated humoral responses in the gut affirms the critical role of integrin α4β7 in gut mucosa B-cell immune responses (Schippers et al., 2012). Additionally, HEVs in GALT chemo-attract chemokine receptor 7 (CCR7)-bearing naïve and central memory T lymphocytes via CCL21 (Okada et al., 2002). Gut-tropic lymphocytes expressing chemokine receptor 9 (CCR9) are signaled by CCL25 and integrin α4β7/MADCAM-1 expressing epithelial cells of HEV to facilitate their migration and homing to the gut (Myint et al., 2020). Lymphocyte homing niche in the gut is physiologically regulated. The expression of MAdCAM-1 in the gut is stabilized in limited levels to maintain homeostasis (Pabst et al., 2000). Elevated signal from inflammatory mediators as seen in gut inflammatory diseases upregulate MAdCAM-1 expression and promote the accumulation of activated T lymphocytes that abundantly express integrin α4β7 (Myint et al., 2020).

Remodeling of homing niches to regulate the tissue-specific migration and recruitment of activated lymphocytes has been demonstrated in recent studies. One mechanism by which the homing niche is remodeled appears to be through gut-tropic derived exosomal secretions, demonstrated by Park et al. (2019b). Activated gut-tropic T lymphocytes secrete exosomes that highly express functional integrin α4β7 with the ability to bind MAdCAM-1 and preferentially home to the gut mucosa. These α4β7 expressing exosomes function to suppress MAdCAM-1 through the cargo and delivery of microRNA milieu that target the transcriptional factor, Nirenberg-Kim (NK) 2 homeobox 3 (NKX2.3) necessary for MAdCAM-1 regulation. Consequently, other homing niche factors, such as the chemokines CCL25 and CCL28 were suppressed by gut-tropic lymphocyte-derived exosomes (Park et al., 2019b). It is therefore not far-fetched to consider exosomes of lymphocytes and other host cells as key remodeling regulators of homing niches to maintain homeostasis.

EV-mediated regulation of homing niches (by modifying the expression of adhesion molecules and chemokines) may operate in stem cell trafficking, and warrants further research. Here we explain the overview of how integrins and chemokines govern stem cell trafficking. Physiologically, stem cells home to specific tissue areas via chemoattraction. For example, hematopoietic stem cell precursors home to the bone marrow after selectin- mediated braking (E-selectins, P-selectins, and L-selectins) that facilitates migration on adhesion ligands expressed by vascular endothelium (Liesveld et al., 2020). Following braking, intracellular signaling cascades are activated through chemokine- chemokine receptor interactions (e.g., CXCL12/CXCR4) to trigger the activation states of integrins like α_Lβ₂ (LFA-1) and α₄β₁ (VLA-4). This facilitates the arrest of cells and their subsequent migration via integrin/cell adhesion molecule (CAM) interaction (Schweitzer et al., 1996; Huttenlocher and Horwitz, 2011; Sahin and Buitenhuis, 2012). Increased CXCL12/CXCR4 signaling is implicated in the retention of leukocytes, such as mature neutrophils at inflammatory sites (demonstrated in tissue-damaging inflammatory diseases) (Yamada et al., 2011; Isles et al., 2019), or in the bone marrow (demonstrated in Warts, Hypogammaglobulinemia, Infections, and Myelokathexis (WHIM) syndrome-associated neutropenia) (Kawai and Malech, 2009). Additionally, FoxP3⁺regulatory T cells trafficking and homing to hematopoietic stem cell niche in the bone marrow involves the signaling of CXCR4 (Zou et al., 2004; Hirata et al., 2018). It has been demonstrated that, loss of adhesion molecules like integrins (e.g., β7-integrins: α4β7, αEβ7) and selectins (e.g., P-selectin) influences stem cell and lymphocyte trafficking (Frenette et al., 1996; Schön et al., 1999; Lucas, 2019).

The subventricular zone (SVZ) of the brain is another example of a host site that serves as a niche for neural stem cells (NSC), with inflammation playing a key role in the homing and recruitment of NSC for CNS regeneration. Increased expression of stromal derived factor 1α (SDF-1α or CXCL12) by astrocytes and endothelial cells of ischemic areas in the brain, as well as their constitutive expression and activation of CXCR4 (receptor of CXCL12), facilitate NSC migration toward ischemic brain explants (Imitola et al., 2004). LeX-positive NSCs (denoting a subpopulation of SVZ adult stem cells that express the extracellular matrix (ECM)-associated carbohydrate, Lewis X) potentially differentiate into cells with neuronal phenotypes following transplantation. These cells acquire enhanced migratory and proliferative characteristics upon exposure to SDF-1α (Corti et al., 2005). Other soluble factors involved in the migration and recruitment of NSCs include vascular endothelial growth factor (VEGF), interleukin-6, hepatocyte growth factor (HGF) and platelet derived growth factor (PDGF) (Garzón-Muvdi and Quiñones-Hinojosa, 2009). Just as with the effect of soluble growth factors on NSC homing, the expression of integrins, such as α6β1 in SVZ neurogenic niche play a vital role in NSC recruitment, demonstrated in a rodent model of cerebral ischemia (Prestoz et al., 2001).

Both myoblasts and non-proliferating myotube forms of skeletal tissue are capable of secreting EVs, in vitro (Forterre et al., 2014a; Choi et al., 2016). Skeletal muscle-derived EVs have been demonstrated to be key players in muscle physiology and systemic homeostasis via paracrine actions (Rome et al., 2019). The release of small EVs into circulation is particularly enhanced by physical exercise, inflammation/stress, and several muscle-related conditions (Frühbeis et al., 2015; Barone et al., 2016). Skeletal muscle is a major contributor of exercise-induced secreted molecules. Some of these secreted molecules were identified in EVs harvested from conditioned medium (CM) of myotube cultures (Forterre et al., 2014b; Deshmukh et al., 2015). Skeletal muscle-derived EVs play vital roles in the differentiation and regeneration of muscle tissue by triggering cues for myogenic processes and myofiber regeneration. Exosomes derived from human skeletal myoblasts were found to contain various myogenic factors, such as insulin-like growth factors (IGFs), fibroblast growth factor-2 (FGF2), and hepatocyte growth factor (HGF) that enhanced terminal myogenic differentiation of stem cells, in vitro (Choi et al., 2016). Conversely, skeletal muscle-derived exosomes may be implicated in suppressing myogenic events. During inflammatory conditions or tissue damage, inflammatory cells and the milieu of pro-inflammatory mediators, such as chemokine ligand 2 (CCL2) induce the production of muscle EVs which package more myostatin (negative regulator of myoblast proliferation and differentiation) and less decorin (myostatin antagonist) (Tidball, 2017; Kim S. et al., 2018).

Crosstalk between the skeletal system and the peripheral nervous system have been demonstrated to be at least in part, mediated by skeletal muscle-derived exosomes. Motor neuron regeneration and survival has been found to be positively impacted by EVs released from muscle cells, in vitro (Madison et al., 2014). Additionally, skeletal muscle denervation induced the release of muscle EVs that preferentially support motor neuron regeneration accuracy. Muscle-derived EVs (containing muscle specific markers, such as α-sarcoglycan) were taken up by denervated, but not naïve nerve tissue around the neuro-muscular junction and significantly induced the exclusive projection of motor neurons to muscle branch, thus, ensuring anatomically accurate motor neuron regeneration in rats (Madison and Robinson, 2019). However, mechanisms underlying exactly how muscle-derived EVs are taken up and localized within the denervated nerve remain a knowledge gap incompletely bridged.

MicroRNAs, carried by muscle EVs have been implicated in the crosstalk between muscle and other cell types (Jalabert et al., 2016; Nie et al., 2019), particularly between muscle and bone in the events of bone remodeling. Skeletal muscle-derived exosomes enriched in miR-27a-3p positively impacted osteogenesis by delivering and increasing miR-27a-3p to target adenomatous polyposis coli (APC), thereby, activating the Wnt/β-catenin pathway to promote osteogenic differentiation of MC3T3-E1 pre-osteoblasts (Xu Q. et al., 2018). Another area of evidence for muscle-bone crosstalk mediated by muscle EVs is osteoclast biogenesis in osteoporosis. Mouse muscle cell line (C2C12 myoblasts)-derived EVs suppressed osteoclast formation and mitochondrial energy metabolism in bone marrow cells (Takafuji et al., 2020a). This group demonstrated that Myo-EVs were successfully taken up by bone cells in culture and suppressed the expression of RANKL-induced osteogenic factors, such as cathepsin K (CTSK), nuclear factor of activated T-cells 1 (NFATc1) and dendrocyte expressed seven transmembrane protein (DCSTAMP). Evidently, Myo-EVs suppressed the oxygen consumption rate and expressions of peroxisome proliferator-activated receptor co-activator 1 beta (PGC1β), NADH-ubiquinone oxidoreductase chain 4 (ND4) and cytochrome c in mouse bone cells (Takafuji et al., 2020a), possibly by an miRNA-mediated inhibition of CREB/PGC1β pathway in osteoclast precursors (Takafuji et al., 2020b). Oxidative stress has been shown to modify miRNA contents of Myo-EVs; miR-34a is elevated in C2C12 derived EVs following oxidative stress. Muscle EVs enriched in miR-34a home to the bone and induce senescence of bone marrow-derived stem cells by decreasing Sirtuin1 expression on primary bone marrow cells (Fulzele et al., 2019).

In metabolic abnormalities like diabetes and insulin resistance, paracrine and/or endocrine effects of Myo-EVs have been reported. A study revealed that the pancreas takes up muscle EVs in vivo. Such EVs transferred muscle specific miRNAs to pancreatic beta cells and islets, thereby, modulating gene expression and proliferation (demonstrated in vitro with MIN6B1 pancreatic cell line). In this study, muscle EVs derived from insulin resistant skeletal muscle were particularly enriched in miR-16 cargo that could be transferred to pancreatic cells and regulate PTCH1, potentially contributing to the pathogenesis of type-2 diabetes (Jalabert et al., 2016).

As positive regulators of metabolic function, exercise-induced skeletal muscle exosomes potentially promote adipocyte lipolysis. IL-15 is known to be highly expressed in skeletal muscle, and acts to decrease fat depots in adipose tissue (Quinn et al., 2005). Additionally, myonectin, a recently established myokine involved in lipid metabolism has been identified as a link between skeletal muscle and systemic lipid metabolism (Seldin and Wong, 2012). Considering the abundance of IL-15 mRNA in the contracting skeletal muscle (Nielsen et al., 2007) and the exercise-induced release of muscle IL-6 in vesicular structures (Lauritzen et al., 2013), it is not far-fetched to speculate the packaging and cargo of this lipolysis-influencing myokine in skeletal muscle derived exosomes. Alterations in skeletal muscle (e.g., muscle activity and injury) have been shown to influence the release of key microRNAs (e.g., miR-21, miR-148b, and miR-486 in tissues, plasma and exosomes) that regulate insulin responsiveness (D'Souza et al., 2018).

Although not classically described as typical secretory cells, cardiac muscle cells can be induced to release EVs, in vitro (Gupta and Knowlton, 2007; Chistiakov et al., 2016). Cardiomyocytes, which constitute the prime contractile cell type in cardiac tissue, display augmented secretions of EVs under several cellular stress conditions like inflammation, injury, hypoxia, alcohol exposure and glucose starvation (Garcia et al., 2015; Yu and Wang, 2019). Similar to skeletal muscles, cardiac muscle EVs cargo nucleic acids, including miRNAs that can regulate gene expression in recipient cells (Waldenström et al., 2012).

Oxygen and nutrient supply are imperative for the efficient function of the myocardium. As such, blood vessels feeding the myocardium play a key role in cardiac function. The crosstalk between cardiac myocytes and intra-cardiac endothelial cells has been established. This crosstalk is vital to meet the metabolic needs for efficient heart function (Garcia et al., 2016) and has been found to be mediated by paracrine signal molecules, direct cell-cell communication, and EVs (Colliva et al., 2020). In the light of cardiomyocyte-endothelial cell communication, cardiomyocyte-derived EVs effect both anti-angiogenic and pro-angiogenic regulatory functions (Ribeiro et al., 2013; Wang et al., 2014; Chistiakov et al., 2016). Heat shock proteins (Hsps) well-known to protect cardiomyocytes from stresses (Willis and Patterson, 2010; Edwards et al., 2011; Yu and Wang, 2019; Yu et al., 2019) are highly expressed in muscle cells and secreted from cardiomyocytes via exosomes. Particularly, cardiomyocyte Hsp20 released through exosomes by the non-classical endoplasmic pathway promote angiogenesis; enhanced proliferation, migration, and tube formation of endothelial cells by Hsp20 via the activation of vascular endothelial growth factor receptor (VEGFR) signaling cascade (Zhang et al., 2012). The crucial role of cardiac microvascular endothelial (CME) cells in protecting against myocardial injury by way of activating endothelial nitric oxide synthase (eNOS) has been evidently linked to signals from cardiomyocytes. A study reported that, cardiomyocytes regulated CME cells via EV transfer of long intergenic non-protein coding RNA, Regulator of Reprogramming (Linc-ROR) to target miR-145-5p, ultimately resulting in the activation of eNOS pathway (Chen et al., 2020). Conversely, anti-angiogenic pathways are mediated by cardiomyocyte-derived EVs and contribute to endothelial dysfunction, insufficient myocardial angiogenesis and cardiovascular complications (Wang et al., 2014). Myocyte-derived exosomes from type-2 diabetic rats but not wild type inhibited proliferation, migration, and tube formation of cardiac endothelial cells in culture through miRNA regulation. It was observed that diabetes induced the release and transfer of exosomes harboring high levels of miR-320 (while being less enriched in miR-126) that target angiogenic factors (Hsp20 and insulin-like growth factor-1) in endothelial cells (Wang et al., 2009, 2014), potentially contributing to diabetes-related vascular dysfunction.

Cardiac remodeling is crucial in the pathophysiology of cardiac injury and hence, should be appropriately regulated to prevent irreversible changes to the structure and function of the myocardium (Tracy et al., 2020). Cardiac muscle exosomes play a key role in regulating myocardial remodeling by the transfer of encapsulated miRNAs to target fibrotic factors (Chaturvedi et al., 2015; Chistiakov et al., 2016). A study revealed that miR-133a from cardiomyocyte exosomes targeted type IA1 collagen (Col1A1) and connective tissue growth factor, resulting in decreased myocardial fibrosis in hypertensive rats (Castoldi et al., 2012). Other cardiomyocyte-derived exosome miRNAs that have been identified to promote fibrosis and fibroblast differentiation in animal models include miR-208a (Yang et al., 2018), miR-92a (Wang et al., 2020) and miR-195 (Morelli et al., 2019). While a number of exosome-derived miRNAs contribute to myocardial fibrosis, others, such as miR-378 tend to exhibit a protective effect by targeting mitogen-activated protein kinase kinase 6 (MKK6) to suppress p38 mitogen-activated protein kinase phosphorylation in cardiac fibroblasts (Yuan et al., 2018). Other miRNAs implicated in regulating the fibrotic cascades in this regard include miR-29b (Chaturvedi et al., 2015) and miR-373 (Xuan et al., 2019).

Recently, the roles of cardiomyocyte-derived exosomes have expanded to encompass a role in mechanisms that underlie the therapeutic efficacy of stem cells in myocardial infarction (Hu et al., 2018). Cardiomyocyte exosomes obtained from culture conditioned medium under stress microenvironment (peroxide-induced oxidative stress) hastened stress-induced injury of bone marrow-derived stem cells (Hu et al., 2018).

Vascular smooth muscles in a synthetic or proliferative, non-contractile state exhibit an increased production of extracellular vesicles (Kapustin and Shanahan, 2016; Schurgers et al., 2018), demonstrated by the inverse relationship between the expression of contractile vascular smooth muscle cell markers and exosome secretion (Kapustin et al., 2015). Inflammatory cytokines and growth factors like tumor necrosis factor-α (TNF-α) and platelet-derived growth factor (PDGF), respectively stimulate the phenotypic transition and exosome secretion by vascular smooth muscle cells (Kapustin and Shanahan, 2016). The activation of vascular smooth muscle cells (SMC) form one of the key events in atherogenesis and the onset of vascular complications, such as stroke and myocardial infarction (Bennett et al., 2016). Overexpression of Krüppel-like factor 5 (KLF5), a transcription factor requisite for mediating SMC proliferation and migration in vascular remodeling events, mediates the secretion of exosomes enriched on miR-155 from human smooth muscle cells. The transfer and uptake of SMC-derived exosomal miR-155 by endothelial cells have been shown to impair endothelial integrity by suppressing their proliferation, migration, tube formation (angiogenesis), and expression of tight junction proteins (Zheng et al., 2017).

Recently, smooth muscles of the pulmonary vasculature have been shown to release EVs that package a myriad of RNA transcripts transferrable to pulmonary arterial endothelial cells (de la Cuesta et al., 2019). Migration and apoptosis of pulmonary arterial smooth muscle cells account for the modulation of miR-143. As a result, exosomes derived from these cells selectively pack abundant miR-143-3p that exhibit pro-angiogenic and pro-migratory effects on recipient pulmonary arterial epithelial cells (Deng et al., 2015). In another recent finding, pulmonary arterial epithelial cells cultured with exosomes derived from platelet-derived growth factor-stimulated smooth muscle cells showed an enhanced migratory but not proliferative ability via the effect of exosomal miRNAs. It was observed that exosomes from PDGF-stimulated smooth muscle cells were deficient in miR-182, miR-1246, and miR-486 and this alteration proved critical for the enhanced migratory phenotype of endothelial cells in pathologic states (Heo et al., 2020).

The microenvironment or matrix of the vasculature is to a large extent regulated by microvesicles secreted from muscle cells, endothelial cells, and infiltrating immune cells of blood vessels (Reynolds et al., 2004; New et al., 2013). Small EVs secreted by vascular smooth muscle cells normally cargo various factors like the high affinity calcium ion-binding matrix Gla protein (MGP) and fetuin-A that function to inhibit calcification, thus protecting against vascular mineralization and the formation of atherosclerotic plagues (Reynolds et al., 2005). However, inflammation and mineral imbalance influence this protective role by inducing the release of EVs enriched in calcification promoters and depleted of inhibitors, thus, conditioning an environment conducive for vascular mineralization and stiffening (Reynolds et al., 2005; Kapustin et al., 2011).