Skin-Associated B Cells in Health and Inflammation

The skin is an essential barrier organ exposed to a variety of daily insults that range from simple physical injuries like cuts and UV damage to infections with sophisticated skin pathogens such as HSV or Staphylococcus aureus. In addition to these external threats, the skin is the frequent target of allergy, autoimmunity, and cancer. The skin is colonized with microbes and consists of protective layers that encompass physical and cellular components as well as antimicrobial proteins and microbial metabolites (reviewed in Ref. 1). The outermost layer of the skin is the keratinized epidermis, and the dermis and s.c. adipose tissue lay below it, and leukocytes populate all of these anatomical skin sites (reviewed in Ref. 2). To enter the skin from blood, leukocytes migrate through postcapillary venules in the dermis from where they can access microenvironments within the dermis or traverse further into the epidermis (reviewed in detail in Ref. 3). To egress from skin, leukocytes have to reach dermal afferent lymph vessels, which will transport them via larger collecting lymph vessels to the draining lymph node (reviewed in Ref. 4).

In the past decades, knowledge about the skin immune system has been rapidly expanding by defining the protective and pathological roles of different skin-associated leukocyte subsets (e.g., T cells) as well as other skin cells, such as keratinocytes (reviewed in Ref. 5). In contrast, the existence of skin-localized B cell functions remained largely unknown, despite a recognized systemic importance of B cells for skin immunity and inflammation. Specifically, B cells that differentiate into Ab-secreting cells give rise to systemic Ab titers that enforce defense against cutaneous infections but also support cutaneous allergies and autoimmune diseases. Today, there is growing evidence for a localized function of B cells within skin in driving disease through multiple mechanisms. Subsets of skin-associated B cells have also emerged as critical negative regulators of skin inflammation. In addition, there is evidence that skin-associated B cells are involved in skin homeostasis and repair by regulating wound healing and the cutaneous microbiome. In this article, we review skin-associated B cells as novel players in cutaneous homeostasis and inflammation as illustrated in Figs. 1 and 2, respectively. A role for B cells in skin cancers has been reviewed elsewhere (6).

Both recirculating (mobile) and resident T cell subsets exist in skin; T cells resident for prolonged periods of time are termed “tissue-resident memory T cells” and have distinct surface markers (i.e., CD103 and CD69 on CD8⁺ T cells) as well as differentiation programming (reviewed in Ref. 7). Although it is clear that subsets of B cells egress from skin by entering afferent lymph vessels after some time of residency (8), it is currently unknown how long B cells dwell in skin. In this review, we use the term “skin-resident” for B cells that have left the vascular space and reside in the skin and “skin-associated” for B cells that are either skin recirculating (present in skin-draining lymph) or skin resident.

Although initial histological studies were unable to detect B cells in healthy human skin (9, 10), cannulation of afferent lymph vessels draining healthy skin in sheep (8, 11) and humans (12, 13) demonstrated the existence of skin-recirculating B cells. More recent studies that employ flow cytometry identified and characterized B cells residing in healthy skin of mice and humans (14, 15). Skin-associated B cells and Ig-secreting cells are also conserved in the skin of lower vertebrates, such as teleost fish (16). In mammalian skin, B cells localize to the dermis where they are sparsely scattered as individual cells during homeostasis and often as cell clusters or more-organized lymphoid structures during inflammation (14, 17–20)

Skin-associated B cells are a heterogenous population with both conventional (B-2) and innate-like (B-1–like) B cells (8, 14). Relative to their blood-borne counterparts, IgG⁺ skin-infiltrating human B-2 B cells have a lower representation of the IgG1 subclass and show a preferred usage of certain Vh genes, suggesting the existence of skin-specific B cell subsets (15). Whereas the majority of skin B cells resemble conventional B-2 B cells, innate-like (B-1–like) B cells are enriched in skin compared with blood in humans and mice (14). Also, a population of ovine skin-recirculating B cells in afferent lymph expresses high levels of IgM and myeloid markers such as CD11b, indicative of B-1–like B cells (8). Innate-like B cells often recognize conserved pathogen structures and produce natural Abs without previous antigenic exposure as well as mount rapid T cell–independent Ab responses early postinfection (reviewed in Refs. 21 and 22). Their often polyreactive Abs (mostly IgM and IgA) are, therefore, an effective first line of defense against infection (21, 22). In mice, B-1 B cells have a high propensity to migrate into skin (14) and other barrier sites such as the respiratory tract or the small intestinal mucosa, where they limit early infections by secreting IgA and IgM and innate-stimulatory cytokines such as GM-CSF (23–25). Thus, innate-like B cell localization to mammalian barrier organs represents a form of barrier enforcement and first line of defense against invading pathogens following barrier breach.

Whereas most Abs act systemically by blood distribution throughout the body and diffusion into tissues, tissue titers can be enhanced by localized Ab production. For example, tissue-resident plasma cells secrete multimeric IgA and IgM, which are transcytosed through epithelial barriers from abluminal to luminal side via polymeric IgR (reviewed in Ref. 26). As a result, they reach glandular excretions and coat mucosal surfaces as secretory Igs (26). The localized role of IgA, and to some degree, that of IgM, is well appreciated as a first line of defense against toxins and pathogen invasion of mucosal barriers (26–28). In contrast, studies analyzing Ab production in mammalian skin are limited. To this end, Ab secretion in healthy skin has been demonstrated for humans and sheep (8, 29). In human skin, IgA is secreted in eccrine sweat glands and appears in sweat and sebum, and Abs of IgG and IgM isotype reach the skin surface by undefined mechanisms (29, 30). Our own studies found that the isotype predominantly produced by skin-resident plasma cells of naive mice is IgM (G.F. Debes and S.E. McGettigan, submitted for publication), which may reflect the relative lack of eccrine sweat glands in mouse skin. More extensive studies are necessary to determine how each Ig isotype reaches distinct skin compartments in different mammalian species.

During inflammation, the presence of Ab-secreting plasma cells in lesional human skin has been well documented for decades. Moreover, plasma cells, identified histologically, facilitate dermatopathological diagnosis of several inflammatory entities (e.g., necrobiosis lipoidica and acne keloidalis nuchae) (31). The isotype (i.e., subclasses of IgG, IgA, IgM, and/or IgE) produced by plasma cells in lesional skin varies depending on the skin site and type of inflammation (for examples, see Refs. 32–34). Ig-secreting cell accumulation in inflamed skin leads to higher Ig titers in lymph draining the skin site, reflecting an increase in tissue titers (8). Although localized Ab production is likely an effective, or sometimes desperate, attempt to control skin pathogens, it is destructive in the case of pathogenic autoantibody production to cutaneous autoantigens [e.g., in pemphigus (18); further discussed below].

Diverse microbes colonize throughout the skin, including epidermal surfaces and appendages such as hair follicles (35), and colonization reportedly reaches as deep as the dermis and dermal (s.c.) adipose tissue (36). The microbial communities of the skin are critically involved in orchestrating local cutaneous physiological and pathophysiologic processes, including wound healing, inflammation, immune responses to skin pathogens (37–39), and even susceptibility to skin cancer (40). In contrast, skin colonization with pathogenic bacteria, such as methicillin-resistant S. aureus, can cause serious disease upon skin barrier breach (41). In the intestine, Abs (in particular, thymus-independent secretory IgM and IgA) bind to commensal bacteria and facilitate symbiotic host–microbe interactions (42–44), and the question arises of whether there is a similar role for Abs in shaping skin microbiomes. A study in RAG-1–deficient mice, which lack all B and T cells as well as Abs, was unable to detect alterations in the ear skin surface microbiome using 16S rRNA sequencing (45). However, electron microscopy studies using immunogold labeling of human skin samples showed that skin microbes are coated with IgA, IgG, and IgM (46), supporting a role for cutaneous Abs in regulating microbial colonization of skin microenvironments. In addition, IgM in human sweat from healthy skin can bind S. aureus (46). Thus, it is possible that, akin to their role in the intestine, secretory Igs influence microbial communities in the skin. Future studies are warranted that test the role of Abs for their ability to modulate microbial colonization of distinct skin compartments, including that of deeper skin layers.

In a number of autoimmune diseases with cutaneous manifestations, the role of B cells and Abs has generally been considered systemic (i.e., pathogenic autoantibodies produced in lymphoid tissue enter blood and diffuse into skin where they cause disease). Recently, B cells are increasingly recognized as an important part of the infiltrating immune compartment in autoimmune and inflammatory skin diseases. B cell numbers are elevated in lesional skin relative to control skin specimens in a number of inflammatory diseases, including, but not limited to, psoriasis (47), pemphigus (18), lupus profundus (19, 20), systemic sclerosis (scleroderma) (48), discoid lupus erythematosus (49, 50), Sjögren’s syndrome (51), IgG4-related skin diseases (33, 34), atopic dermatitis (52), and allergic contact dermatitis (53). Although an increase in skin-infiltrating B cells already suggested a role for B cells in cutaneous inflammation, the therapeutic success of systemic B cell depletion with rituximab confirmed a pathogenic role for B cells in a number of these diseases (54–56). In pemphigus, atopic dermatitis, and scleroderma, disease severity and progression are positively correlated with the number of skin-infiltrating B cells (18, 48, 52). Specifically, in scleroderma, a chronic connective tissue disorder that often progresses to extensive cutaneous and vascular fibrosis, only those patients with skin-infiltrating B cells advanced in disease severity over a period of 12 mo (48). Furthermore, a pathogenic role for B cells in the initiation phase of skin fibrosis was confirmed using B cell depletion therapy in scleroderma patients and mouse models of the disease (57–59). Collectively, the recent reports of skin-infiltrating B lineage cells in skin pathologies raise the question of whether B cells directly contribute to cutaneous inflammation through local effector functions (as discussed in the sections below). Of note, skin-infiltrating B cells are not always pathogenic (e.g., psoriasis does not usually improve with B cell depletion) (60), and rituximab treatment can induce psoriasis in some individuals (61), suggesting a protective role of B cells (discussed below). Czarnowicki et al. (52) investigated B cell subsets in two skin diseases in which B cells appear to have opposing roles: atopic dermatitis and psoriasis, which are driven by Th2 and Th17/Th1 cell responses, respectively (62). The authors suggested that a higher B cell activation status reflected accumulative Ag exposure and ability to contribute to T cell activation and other pathogenic mechanisms, such as Ab production and IgE switching (52). Thus, it is likely that BCR specificity for skin-associated Ags (autoantigens, allergens, or microbiota) in combination with B cell effector functions (discussed below) dictates the distinct roles of skin-associated B cells in different types of cutaneous inflammation.

A prominent example in which skin-infiltrating B cells may drive disease is pemphigus vulgaris, a severe autoimmune blistering disease that affects mucosal membranes and skin. Pemphigus is characterized by circulating anti-desmoglein 1/3 (Dsg1/3) autoantibodies, which recognize desmosomal adhesion proteins on epidermal keratinocytes. In humans, placental transfer of the anti-Dsg autoantibody from diseased mothers induces pemphigus symptoms in their newborns (reviewed in Ref. 63). Likewise, passive transfer of the autoantibodies into newborn mice is sufficient to induce pemphigus-like disease, confirming that the pathogenic autoantibody is the main etiology of the disease (63). The vast majority of pemphigus patients accumulate anti-Dsg1/3 autoantibodies in skin; however, circulating Dsg1/3 autoantibodies can be found in patients without skin lesions and in some healthy controls, indicating that in addition to systemic autoantibodies, other factors determine disease (reviewed in Ref. 64). The pathogenic potential of anti-Dsg1/3 Abs after transfer into mice correlates with serum titers (65). An increase in cutaneous autoantibody titers could be due to enhanced localized production and/or diffusion from systemic Abs, ultimately boosting localized binding to cutaneous autoantigens. In support of this notion, Yuan et al. (18) recently showed that autoreactive Dsg1- and Dsg3-specific Ab-secreting cells and B cells localize to lesional skin relative to healthy skin, suggesting a pathogenic role in situ by these autoreactive cells. Akin to pemphigus, the majority of scleroderma patients present with both systemic and local autoantibodies as well as skin lesion–infiltrating B cells and plasma cells (48). Finally, IgG4-related skin diseases, which are fibroinflammatory disorders affecting multiple organs, are induced by accumulation of terminally differentiated IgG4⁺ plasma cells and Ab deposition in affected skin (33, 34). Taken together, localized Ab production is a feature of several skin diseases, including those in which Ab binding to cutaneous autoantigens is a known driver of disease.

Although in some instances, such as pemphigus vulgaris, dissociation of tissue due to autoantibody binding alone causes disease (63), many times, complement fixation and/or Fc-mediated Ab effector functions are necessary for tissue destruction. For example, in cutaneous manifestations of the autoimmune disease lupus erythematosus, the lupus band test on skin biopsies reveals the deposition of autoantibody–Ag complexes as well as complement proteins along the dermoepidermal junction (66). In cutaneous vasculitis, Igs and complement proteins are deposited around dermal vessel walls (67). Complement activation in the skin has the potential to exacerbate inflammatory damage by direct cell lysis and/or via recruitment and activation of innate immune cells, such as macrophages and neutrophils. Thus, autoantibody production in skin is a powerful means for localized destruction and inflammation through complement and/or other downstream mechanisms.

Formation of ectopic lymphoid tissue or tertiary lymphoid organs (TLOs) in the skin may provide a local microenvironment for skin-associated B cells to promote skin inflammation. TLOs are considered outposts of secondary lymphoid organs because they possess similar characteristics, such as organized B cell and T cell areas with Ag-presenting dendritic cells on a network of fibroblasts as well as high endothelial venules, lymph vessels, and germinal center (GC)-like structures (68). TLOs are found in or near sites of chronic inflammation (e.g., during autoimmunity, infection, or transplant rejection). In extracutaneous lesions of autoimmunity, such as in systemic lupus erythematosus (SLE) and rheumatoid arthritis, TLOs are considered pathogenic because they perpetuate local inflammation by enhancing autoantibody production and autoreactive T cell activation (reviewed in Refs. 69 and 70). Even though TLO-like structures that support interactions between skin B and T cells exist in certain inflammatory skin diseases, little is known about the role of TLOs in the pathogenesis of cutaneous pathologies. In lupus erythematosus panniculitis (lupus profundus; a form of cutaneous lupus that is typified by inflammation of the s.c. adipose tissue), aggregates of B cells and plasma cells or lymphoid follicles containing distinct GCs are found in the majority of cutaneous lesions (19, 20). Kogame et al. (20) showed that additional components of a TLO are present in these lupus profundus lesions, including CXCL13⁺ cells, peripheral node addressin⁺ (PNAd⁺) high endothelial venules, and podoplanin⁺ lymphatic vessels. These findings confirm that TLO structures develop in lupus profundus. In pemphigus skin lesions, B cells and IL-21⁺ CD4⁺ T cells interact closely in aggregate structures that resemble developing TLOs (18). Aggregates of B cells and T cells have also been noted in skin manifestations of Sjögren’s syndrome (51), an autoimmune disorder of exocrine glands. Skin-localized GC-like structures likely enhance pathogenic Ig affinity maturation and class switching to different IgG subclasses, as well as Ab production, thereby augmenting humoral responses to skin-associated Ags. Localized activation of effector T cells and Ab production in lymphoid aggregates and TLOs could help explain why in one individual with systemic autoantibodies that target skin, some skin areas are affected, whereas other areas are spared from disease symptoms. Notably, B cells are drivers of lymphoid neogenesis, including TLOs, by provision of lymphotoxin (70). Therefore, when skin-resident B cells initiate formation of a local TLO, this may be the tipping point that accelerates disease in many cases of chronic skin inflammation.

B cells are increasingly recognized as key effector cells in inflammation and infection by secretion of inflammatory cytokines such as IL-6, GM-CSF, IFN-γ, and IL-4 (71), and B cell cytokine secretion in skin is a new area of investigation. Recently, Matsushita et al. (59) showed that uninflamed skin of mice harbors IL-6⁺ B cells, which drastically increase in inflamed skin in the bleomycin-induced scleroderma model. Importantly, the authors found that skin pathology, including skin fibrosis and thickening, were significantly reduced in mice whose B cells lack the ability to express IL-6, demonstrating a pathogenic role for IL-6⁺ B cells in fibrosing skin inflammation (59). Furthermore, blockade of the cytokine BAFF [a B cell–targeting approach used to treat SLE (72)] inhibited expansion of IL-6⁺ but not that of anti-inflammatory IL-10⁺ B cells, leading to reduced pathological changes in bleomycin-induced skin fibrosis (59). These findings stress the importance to further define skin-associated B cells and their functions because this may reveal novel tools to selectively target distinct B cell subsets.

Ag presentation to effector T cells is another Ab-independent effector function of B cells that is increasingly recognized to drive inflammation. Specifically, effector T cells that recognize Ags on B cell MHC class II (MHCII) release potent inflammatory cytokines and subsequently expand. For example, mice with B cells that cannot present Ags to T cells are protected from neuroinflammation in a mouse model of multiple sclerosis because of reduced pathogenic T cell responses (73). Studies by our laboratory showed that skin-resident and skin-recirculating B cells express high levels of MHCII and costimulatory molecules (CD80/86 and CD1) and are therefore well equipped for T cell activation (8). Although studies of B cells as APCs in skin inflammation are limited, Young et al. (74) showed a pathogenic role for B cells in a mouse model of chronic graft-versus-host disease, a serious complication of allogeneic hematopoietic stem cell transplantation. Specifically, donor B cells upregulated MHCII and costimulatory molecules and augmented donor CD4⁺ T cell expansion, alloreactivity, and survival, which in turn promoted chronic graft-versus-host disease in skin and lung (74). Although the authors concluded that B cell–T cell cross-talk occurred in secondary lymphoid tissues (74), it is possible that such interactions were also present directly in skin and/or skin-associated TLOs. Effector T cell activation by skin-associated B cells and downstream inflammation caused by T cell cytokines, such as IFN-γ, IL-17, and IL-4, have the potential to exacerbate many inflammatory skin disorders. Therefore, additional studies into the APC function of skin-associated B cells are highly warranted.

B cells with the capacity to suppress immune responses, termed regulatory B cells (Bregs), have recently received much attention because they limit inflammatory and autoimmune processes in many organ systems and diseases (reviewed in Refs. 75 and 76). Notably, B cells with the ability to suppress skin inflammation were already described more than 40 y ago (77, 78). The suppressive mechanisms employed by Bregs are varied and include expansion and maintenance of regulatory T cells, adenosine generation, and secretion of the suppressive cytokines IL-35 and IL-10 (reviewed in Refs. 79 and 80). In skin, B cells that produce IL-10 are critical to limiting inflammation, and the significance of other Breg functions has not been explored to date.

In mouse models, IL-10⁺ Bregs (also known as B10 cells) limit inflammation in cutaneous hypersensitivity (81, 82), scleroderma (59), and psoriasis-like inflammation (83, 84). Earlier studies suggested that splenic Bregs suppress T cell priming and/or polarization in the initiation phase and that innate-like Bregs (e.g., peritoneal B-1 B cells) are important in the remission phase of inflammation, with all Bregs acting within lymphoid tissues (81, 82). Our laboratory recently showed that IL-10⁺ B cells reside in normal mouse and human skin, and analysis of IL-10 reporter mice revealed that B cells actively transcribe IL-10 in the skin even in the absence of inflammation (14). Moreover, IL-10⁺ B-1 B cells, which are potent suppressors of inflammation in both cutaneous and intestinal inflammation (82, 85), are recruited into inflamed skin (14). Matsushita et al. (59) also showed that IL-10⁺ Bregs reside in skin and are critical in limiting inflammation and fibrosis in a scleroderma model. Therefore, IL-10⁺ Bregs are well positioned to suppress inflammation in the skin itself in addition to acting in lymphoid tissue. Breg presence and active IL-10 transcription in uninflamed skin (14) suggest that skin Bregs fulfill a steady-state function by suppressing aberrant responses to physical insults and/or skin commensal microbiota.

A potential pathway to induce skin Bregs is through cutaneous UV exposure. Specifically, moderate UV light exposure of skin leads to enrichment of skin lymph node–resident B cells with the capacity to inhibit dendritic cell activation of T cells in cutaneous hypersensitivity models in an IL-10–dependent manner (86, 87) as well as IL-10–negative Bregs that suppress experimental autoimmune encephalitis (88). It will be interesting to further investigate this pathway as well as potential differences between individual mammalian species with inherent differences in sunlight (UV) exposure (e.g., caused by hair density and nocturnal versus diurnal chronotypes).

In humans, B cell depleting therapy (i.e., rituximab, anti-CD20) is able to induce or exacerbate the inflammatory skin disease psoriasis (reviewed in Ref. 61). This apparent paradoxical effect of an immunosuppressive therapy supports a protective role for B cells in this largely Th1 and Th17 cell–mediated disease. Interestingly, psoriasis is associated with low levels of dermal IL-10 and is clinically responsive to locally administered IL-10 (89, 90), possibly reflecting a niche for regulation by IL-10⁺ cell types such as IL-10⁺ Bregs. Extending the mouse studies that demonstrate a critical role for IL-10⁺ B cells in suppressing psoriasiform inflammation (83, 84), individuals with psoriasis have reduced numbers of IL-10⁺ B cells in their blood (91, 92). Other cutaneous inflammatory diseases, such as the skin-blistering disease pemphigus and scleroderma, can also be associated with reduced numbers of circulating IL-10⁺ Bregs during active disease (93–96) and an increase in disease remission (97). Notably, IL-10⁺ B cells localize to the inflamed skin in SLE (98). It is therefore possible that skin-associated Bregs limit skin inflammation even in B cell– and autoantibody-mediated inflammation.

Collectively, there is accumulating evidence that IL-10⁺ B cells suppress skin inflammation not only in mouse models but also in humans. Additional studies are needed to dissect the Breg subsets at cutaneous and extracutaneous sites and IL-10–dependent and –independent mechanisms by which they suppress different types and phases (i.e., acute versus chronic; initiation versus remission) of skin inflammation. In particular, in B cell–mediated autoimmune diseases, it would be advantageous to be able to deplete disease-driving B cells selectively, while sparing Bregs. Conversely, tools to selectively target Bregs that impede effective cutaneous immune responses, based on differential surface marker expression, for example, could potentially be desirable when treating cutaneous tumors or chronic infections. To complicate matters, Bregs that make IL-10 do not appear to be a separate B cell lineage and are rather an activation or differentiation state of several traditional B cells subsets (99). Similarly, in T cells, IL-10 is produced not only by bona fide Foxp3⁺ regulatory T cells but also by effector T cells and represents an essential mechanism for limiting immunopathology at effector sites (100, 101). For example, during influenza virus infection, IFN-γ⁺ CD4 and CD8 effector T cells acquire the ability to secrete IL-10, which is critical for host survival because it limits tissue destruction and pulmonary inflammation (102). Therefore, further studies are needed to decode the signals that induce IL-10 or other anti-inflammatory properties in distinct skin-associated B cell subsets, allowing for selective therapeutic manipulation.

As a barrier organ with various physical insults that happen on a daily basis, there is a vast need for repair of small and larger skin wounds. B cells localize to cutaneous wounds (103, 104) and are potentially involved in wound healing (105). Specifically, mice that overexpress BCR coreceptor CD19 (CD19-transgenic mice) exhibit accelerated, whereas CD19-deficent mice have delayed, wound healing (105). The authors of the study suggested a TLR4-initiated increase in cytokines, such as IL-6, IL-10, TGF-β, platelet-derived growth factor, and basic fibroblast growth factor, as the mechanism of B cell–enhanced wound healing (105). However, CD19-deficient and -transgenic mice have greatly altered B cell subsets with a relative lack or increase of innate-like B cells (i.e., MZ and B-1 B cells), respectively (106–108). As B-1–like B cells, which are CD19^hi, localize to skin (14), it is tempting to speculate that these cells promote wound healing. In further support of a role of B cells in wound healing is an observation that exogenous application of splenic B cells to wounds accelerates the healing process in wild-type and diabetic mice (109). Future studies are needed to dissect the mechanism by which different subsets of skin-associated B cells may contribute to the wound healing process and how this may be targeted for therapeutic purposes.

With the relatively recent discovery that B cells are part of the skin immune system, we are just beginning to understand the capacity of skin-associated B cells in modulating cutaneous immune responses. The newly uncovered pathogenic and protective functions of B cells in skin and those of B cells at other extralymphoid sites (e.g., the joints and synovia, the CNS, and intestinal mucosa) offer a glimpse of the power of skin-associated B cells in disease settings. Thus, further illuminating the functions of skin B cell subsets provides the opportunity to develop novel approaches to modulate cutaneous immune responses in homeostasis, inflammation, infection, and cancer. We anticipate that exciting new developments in cutaneous B cell biology are on the horizon.

We thank Neda Nikbakht, Gregg Silverman, and members of the Debes Lab for critical comments on the manuscript and stimulating discussions. We also thank John Stanley for advice and discussion of pemphigus pathogenesis. The authors are indebted to Paul Schiffmacher and Tim Flanagan for figure illustrations.

The authors have no financial conflicts of interest.