Microbial Interplay in Skin and Chronic Wounds

Microbial infections in chronic wounds can often lead to lower-limb amputation, decrease in quality of life, and increase in mortality rate, and there is an unmet need to distinguish between pathogens and colonisers in these chronic wounds. Hence, identifying the composition of healthy skin microbiota, microbes associated with chronic wound and healing processes, and microbial interactions and host response in healing wounds vs. non-healing wounds can help us in formulating innovative individual-centric treatment protocols. This review highlights various metabolites and biomarkers produced by microbes that have been identified to modulate these interactions, particularly those involved in host–microbe and microbe–microbe communication. Further, considering that many skin commensals demonstrate contextual pathogenicity, we provide insights into promising initiatives in the wound microbiome research. The skin microbiome is highly diverse and variable, and considering its importance remains to be a hotspot of medical investigations and research to enable us to prevent and treat skin disorders and chronic wound infections. This is especially relevant now considering that non-healing and chronic wounds are highly prevalent, generally affecting lower extremities as seen in diabetic foot ulcers, venous leg ulcers, and pressure ulcers. Pathogenic bacteria are purported to have a key role in deferring healing of wounds. However, the role of skin microflora in wound progression has been a subject of debate. In this review, we discuss biomarkers associated with chronic wound microenvironment along with the relevance of skin microflora and their metabolites in determining the chronicity of wounds.


Introduction
Human skin protects us from physical, chemical, and immunological risks from the outside world and also hosts a major ecosystem that harbours skin's indigenous microbiota or the skin microbiome [1]. In all likelihood, the most significant role of this multifunctional organ is to safeguard against infections from invading microbes. Although the skin serves to prevent pathogenic microbes from gaining entry into the host, the presence of hair follicles and other appendages provides the human skin with a surface area of approximately 30 m 2 which allows interaction with a diverse array of microbes [2]. The moist, sebaceous, and dry microenvironments of the skin harbour a variety of microbial populations, including bacteria, fungi, viruses, and microeukaryotes [1].
For well over a century, scientists have debated whether microbes on the skin play a definitive role in causing disease or are merely colonisers [3]. The microbial populations of skin protect against foreign microbes via both direct and indirect processes, like the production of antimicrobial compounds (direct) [4] and competitive exclusion (indirect) [5]. However, the role of skin-dwelling commensals varies depending on the species, with certain species alternating between healthy and diseased states. An altered microbiome, together with a break in skin's protective barrier, increases the risk of infection, progressing to skin-related disorders.
Infection has been identified as one of the most important contributors to the development and maintenance of chronic wounds [6].
Chronic wounds are a serious healthcare issue, with diabetic foot ulcers (DFU), pressure ulcers (decubitus ulcers (DU)), venous leg ulcers (VLU), and non-healing surgical wounds being the most prevalent. Chronic wounds are more common in elderly people who have latent medical problems like diabetes, vascular disease, pulmonary disorders, kidney malfunctions, and obesity [7]. Impaired wound healing has also been linked to a weakened immune system, poor nutrition, and prolonged mechanical stress [8]. Chronic wounds are attributed to disturbingly high mortality rates: the 5-year mortality rate of diabetic foot ulcers is 30.5%, which is comparable to the mortality rate linked with cancer (31%) [7]. Chronic non-healing wounds are also accompanied by significant treatment expenses reaching $95 billion in annual healthcare expenses [9]. Unfortunately, effective therapies are still missing, despite the rising frequency and high costs of care [7].
A characteristic aspect of these non-healing wounds is the prevalence of skin commensals in them, associated with varied bacterial communities nestled in an altered microenvironment. Recent investigations have taken recourse to evaluation of markers and indicators such as enzyme activity, volatile molecules, or other metabolites along with sensors for changes in pH, temperature, and odour, in addition to microbiological investigations, to better understand the complex dynamics of chronic wound healing better [10][11][12][13]. This review examines the importance of the human skin microbiome and how these important natural communities are altered in chronic wounds. We also reflect upon the gamut of microbe-host and microbe-microbe interactions that affect the skin and its robustness, including wound healing modulation along with a discussion on a variety of microbiome-derived metabolites that have been identified as important intermediaries in skin microbial populations.

Skin Microbiome
The indigenous microflora of skin as a whole has been linked to the proper establishment of an intact boundary over a lifetime. The interactions between the diverse skin commensals that inhabit this vast plane are either neutral or mutually profitable. For instance, during brief developmental phase in early life, skin commensals and hair follicles work in accordance to facilitate tolerance to skin commensal microbes and maintain skin immune cell homeostasis [14]. The dynamic environment of the skin and its associated microbiome is the most reliable predictors of temporal age, outperforming hallmarks of the gut and oral microbiome [15][16][17]. These commensal communities are not only just considered hitchhikers, but contribute actively to keeping the skin barrier intact. For instance, skin microbiota has been linked to crucial barrier function processes such as regulation of the skin inflammatory response, epidermal differentiation, and augmentation of wound healing [18,19].
The healthy skin microbiome is heavily influenced by several host-related factors including age, anatomical site, and others ( Fig. 1). Studies have shown that the skin microflora of an individual is developed intra-partum, where the mode of maternal delivery has a significant impact on microbial makeup [20,21]. The colonisation of the skin by microbes is also heavily influenced by anatomical site, as evidenced by diverse microbial populations dwelling in various epidermal topographical niches [22]. A 16S rRNA gene-based study of a healthy human skin microbiome revealed the presence of minimum 19 phyla along with more than 1000 species of bacteria from 20 different skin sites [23 •]. Phyla Actinobacteria (52%), Firmicutes (24%), Proteobacteria (17%), and Bacteroidetes (7%) made up the most of epidermal microflora, while Corynebacteria (Actinobacteria), Propionibacteria (Actinobacteria), and Staphylococci (Firmicutes) were the most prevalent genera [23•, 24]. Moisture content and anatomical site both play a role in determining microbial composition. For instance, sebaceous sites like occiput, glabella, alar crease, external auditory canal, retro-auricular crease, and back have the highest microbial burden [23•] and are favoured by Propionibacterium and Staphylococcus, while sites with high moisture content are dominated by Corynebacterium and Staphylococcus. Even though dry regions (like volar forearm, hypothenar palm, and buttock) comprised a higher proportion of beta-proteobacteria, Flavobacteriales, and other Gram negative organisms, it showed high microbial richness and variance overall [24]. Corresponding to epidermal colonisation, recent research shows that the cutaneous microbiome transcends into the sub-epidermal layers of the skin, accompanied by greater abundance but a lower proportion of Proteobacteria (Burkholderiales and Pseudomonadales) and Actinobacteria [25] While the relevance of anatomical locations and moisture content in determining the human skin microbiome has been clearly established, studies have also shown that genetics and environmental variables including climate play a role in defining the 'normal' microbiome [31,32]. For instance, skin commensals obtained from forearms of Venezuelan subjects (dominance of Staphylococcus and Proteobacteria) varied significantly from those of Americans (primarily Propionibacterium) [33]. Furthermore, besides gender and age, the cutaneous microbiome can also be reliable predictors of whether residents reside in urban or rural location amidst the same metropolitan region [26]. In a similar study, Hospodsky et al. [34] found that hands of women in Tanzania contained more soil-associated microorganisms including members of Rhodobacteraceae and Nocardioidaceae than women in the USA. In another study by Wang et al. [35], the pan-microbiome concept suggests that the microbial assemblage of healthy skin differed between nations and displayed considerable variation in Chinese subjects as opposed to another ethnic group, Pakistanis. The ethnic and environmental variances highlight the need to broaden our existing knowledge on skin microbiome diversity to include a wider range of geographic and cultural communities since variations in skin commensals might significantly contribute to wound healing and therapy.

Skin and Microbial Interactions
The stratum corneum, the outermost layer of the skin, is made up of densely packed, keratinocytes supported by cornified envelopes providing a powerful shield to the host from external influences. Moreover, stratum corneum's low water content along with hydrolysis of epidermal phospholipids into free fatty acids (FFA) decreases the pH of the skin surface, making it unsuitable for pathogen colonisation [36]. The human skin is also known to release several antimicrobial peptides (AMP) which are effective against a wide range of pathogens, like bacteria, viruses, and various parasites. Primary skin-derived AMPs are cathelicidins and beta-defensins, while several other proteins such as ribonucleases and peptidoglycan recognition proteins with potent antimicrobial activity have also been reported from skin tissue [37].
Rather than living off the host, skin microbial community plays an active part in antimicrobial defence via both indirect and direct methods. Skin microflora, indirectly, works as a competitive barrier against potential infections by colonising the epidermal niches and harnessing the available resources [16]. Furthermore, microbial metabolism products may indirectly boost the skin's antibacterial potential. Lipase activity, for example, enables the production of FFAs by hydrolysis of sebum lipids in a variety of skin commensals, including Corynebacterium acnes and Staphylococcus epidermidis [38]. Similarly, the nasal commensal Corynebacterium accolens was reported to hinder pathogenic Streptococcus pneumoniae via conversion of cutaneous triacylglycerols to FFAs [39], suggesting that FFAs, synthesised by skin commensals, may have a direct antibacterial effect in addition to lowering skin pH. Through the synthesis of antimicrobial compounds, the local commensal microbial communities also contribute to colonisation resistance. Coagulase negative Staphylococci have shown to suppress the growth of closely-related bacterial species by producing bacteriocins -extremely potent compounds which are heat-resistant  [20, 21, 23 • , 26-30]. Source: the authors and synthesised in ribosomes [40][41][42]. Bacteriocins are reported to be synthesised by various skin commensals. For instance, S. epidermidis are reported to generate a variety of bacteriocins and phenol-soluble modulins that selectively eradicate pathogenic bacteria however inactive towards S. epidermidis [43]. The antibacterial arsenal of S. epidermidis includes the serine protease Esp that have been reported to deter Staphylococcus aureus in human nasal carriage [43]. Likewise, Staphylococcus lugdunensis produces lugdunin, a non-ribosomal peptide, which was shown to be efficacious against a variety of skin infections and particularly effective in decreasing the nasal carriage of S. aureus [44] (Fig. 2).
The production of antibiotics by skin commensals may also be considered as one of the major defence strategies adopted against infectious bacterial pathogens. Mupirocin, a topical antibiotic that is used to treat staphylococcal or streptococcal skin infections [48], is produced by Pseudomonas fluorescens, a bacterium which is occasionally discovered among cutaneous commensal populations [49]. Interestingly, mupirocin resistance is inherent in certain skin microbes, like Micrococcus spp. and Corynebacterium spp. [50]. This beneficial property of mupirocin enables the targeted treatment of harmful pathogens while conserving the local microbiota. The commensal microbes that live in the human body not only strive for niche possession, but also defend against invasive pathogens and hence have developed multiple mechanisms to overcome microbial competition, most of which remain largely unexplored.

Chronic Wounds
Venous, diabetic, and pressure ulcers are some of the most frequent chronic wounds [7]. Venous leg ulcers (VLUs) develop when there is venous dysfunction. VLUs are most common among chronic non-healing wounds contributing 60-80% of cases and are caused by a variety of factors. Inept veins or valves, as well as reduced muscle performance, can result in inadequate calf muscle pump activity, which can cause ambulatory venous hypertension leading to local venous dilation and pooling [51]. This results in the entrapment of leucocytes which may release proteolytic enzymes damaging tissues. Venous pooling also causes inter-endothelial pore expansion and the deposition of fibrin and other macromolecules, which trap growth factors and render them useless for wound healing. Bed-bound, wheelchair-bound, and neurologically disabled individuals are more likely to develop arterial, diabetic, and pressure ulcers, which is often followed by necrosis of the affected tissue. Lower-extremity ulcers, which affect roughly 23% of people with autoimmune illness, can be a late debilitating consequence of connective tissue disorders [52]. The pathogenic processes that create arterial, diabetic, and pressure ulcers are mostly ischemic in nature, resulting from several factors such as lymphatic obstruction, reperfusion, and cell deformation. Tissue inflammation, in all its manifestations, is a critical process implicated in the aetiology of ulcer development.
Wound healing and repair are a complicated and systematic process that is closely regulated by various types of cells which are, in turn, regulated by a number of growth factors, cytokines, and chemokines. Wounds become chronic if this process is hampered in any way and does not allow skin barrier to reseal the open wound. Chronic wounds of all forms have been reported to show fibrosis along with increased rate of keratinocyte proliferation and lack of migration. In contrast to the typical wound repair, chronic wounds have been reported to show hampered angiogenesis, recruitment and impeded activation of stem cells as well as extracellular matrix remodelling, with persistent inflammation [45 •, 53-55]. Wounds allow bacteria from the skin microflora and those from the environment, to get access to the underlying cells and tissues to colonise and develop in optimum circumstances [56]. During the normal cutaneous wound healing process, commensal bacteria interact with skin cells, which help in modulating the innate immune response [57]. Chronic, hard-to-heal wounds generally harbour polymicrobial biofilms that encourage pathogenic expansion in the wound bed and impeded wound healing [58•]. Biofilm formation has been reported in 60% of chronic wounds as opposed to just 6% of acute wounds in a clinical examination of wound specimens (n = 50) from adult patients [58•]. Furthermore, Wolcott et al. [58•] determined that Pseudomonas was the most dominant genus in chronic wound biofilms as well as the most prevalent bacterial genus observed in monomicrobial biofilms. Microbial clustering on wound surface results in biofilms, which are enclosed in an exopolymeric material made up of polysaccharides, lipids, and protein [59]. Under these conditions, microbes tend to suitably alter their reproductive rate and metabolic activities and utilise quorum sensing to relay changes in dynamics of their population density via organic signalling molecules synthesised by them. Biofilms trigger the host immune responses on a constant basis, thereby delaying the wound from entering the proliferative phase, hindering the healing of the wound [59].
The composition of the human skin microbiome changes over time, while the presence and number of bacteria in wounded skin vary according to wound type. The three major phyla, namely, Firmicutes, Proteobacteria, and Actinobacteria, which are found in healthy skin are also found in pressure ulcers [60]. In their clinical observation study of chronic wounds from 2963 subjects, Wolcott et al. [58•] found that Staphylococcus was the most common and these chronic wounds (DFU, DU, VLU, and non-healing surgical wounds) were predominated by S. aureus and S. epidermidis. Furthermore, despite the fact that microbial heterogeneity was unaffected by wound type, the most prevalent genera were noted to be S. epidermidis, whereas Pseudomonas aeruginosa was found to be relatively abundant in the examined wounds, indicating biofilm development. This investigation corroborated the previous findings of James et al. [61] who using culture and molecular techniques established Grampositive cocci to be the most predominant bacterial population in chronic wound specimens, whereas Gram-negatives were shown to be involved in the production of biofilms in chronic wounds with the predominance of Staphylococcus and Pseudomonas in chronic wounds. Despite the fact that chronic wounds are normally exposed to high amounts of oxygenation, anaerobic bacteria have been found to be more prevalent in chronic wounds than acute wounds, with Finegoldia, Prevotella, Peptoniphilus, Peptostreptococcus, and Anaerococcus as regular components of chronic wound microbiome (Table 1).
With respect to species diversity, DFUs were found to be considerably less diverse than unwounded skin samples from diabetic subjects as well as control skin samples, in terms of three different alpha diversity estimates, namely, observed species richness, Chao 1 estimator, and Shannon index [70]. Furthermore, permutational multivariate analysis of variance demonstrated that the beta diversity of bacterial populations in control skin was considerably different from that of DFUs. While Corynebacterium was reported to be the most frequent genus in diabetic foot associated ulcers and osteomyelitis (n = 20) by Johani et al. [71], Gardiner et al. [70] reported Staphylococcus as the most dominant genus in chronic DFUs, followed by Acinetobacter and Corynebacterium. However, using NGS approach, Kalan et al. [72] reported that certain S. aureus strains were exclusively present in unhealed wounds, while few other generalist strains were more broadly distributed across wound types suggesting an association between strain type and adverse clinical outcome of the wound in DFU subjects. Further, a similar result was obtained in the type 2 diabetic mouse model, with wounds inoculated with S. aureus strains showing poor outcomes with slow healing rate and poor re-epithelialisation [72]. Moreover, microbial community transition dynamics can significantly affect wound healing rates. For example, Loesche et al. [62] classified the microbial communities associated with DFUs into four major community types, and the temporal transitions between these types dictated the wound healing rates. Further, treatment with systemic antibiotics could only destabilise the microbiota but did not significantly alter the relative abundance of specific bacterial taxa.
Chronic wounds are polymicrobial in nature, and hence, understanding the dynamic and complex interplay between pathogens and skin commensals, rather than just the presence or absence of particular bacteria, is probably more helpful in terms of understanding the wound dynamics and progression [73]. The pathogenic impact of anaerobes, for instance, might be exacerbated by the presence of aerobes because they utilise oxygen, causing tissue hypoxia and favouring anaerobe proliferation [74]. This may be considered a symbiotic connection, where two or more species cooperate to enhance virulence and hinder healing [75]. It is now well established that the two of the most prevalent pathogens in infections of chronic wounds are P. aeruginosa and S. aureus. They are usually detected together, and combined infections are more harmful than their monomicrobial infections [76,77]. In addition, P. aeruginosa and S. aureus have been demonstrated to have higher antibiotic resistance when cultured together in a wound model [76]. Also, Bacteroides fragilis has been identified as the most common anaerobic bacterium isolated in DFUs in various investigations [78], and it plays an essential role in the composition of microbial populations and biological interactions. In an interaction study, Mastropaolo et al. [79] compared the synergistic interactions among partners in polymicrobial wound infections such as B. fragilis, C. perfringens, and E. coli in an obese diabetic mouse model and showed strong synergism between B. fragilis and E. coli but not C. perfringens.
Overall, there is mounting evidence that polymicrobial interactions may promote the pathogenic capacity of other microbes or reduce their virulence and hence have a considerable influence on the degree of severity as well as the progression of wound infection. Hence, it is critical to investigate beyond the mere presence or absence of microbes in such non-healing wounds but a further attempt to understand possible microbial interconnections.

Microbial Interactions in Chronic Wounds
Chronic wounds are usually polymicrobial in nature. Such infections are accompanied by persistent biofilms and show higher resistance to antibiotic therapy in comparison to monomicrobial infections [80]. In one of the earliest attempts to demonstrate polymicrobial infections, guinea pigs were co-infected with E. coli and B. fragilis, and after 7-day post-infection, each species showed an increase in bacterial CFU by more than 100 times in comparison to monomicrobial infection, as well as increase in inflammation and purulence, which was reported suggestive of poor healing and indicated pathogenic synergy [81]. Likewise, co-infection of ulcers with P. aeruginosa and S. aureus has also been linked to the persistent non-healing condition [77]. However, certain investigations on the co-infection of P. aeruginosa and S. aureus have also revealed that P. aeruginosa inhibits the proliferation of S. aureus [46,82]. A study involving mouse wound excisional model revealed that detection of peptidoglycan by P. aeruginosa is critical for its competitive edge in the vicinity of other Gram-positive taxa. The presence of N-acetylglucosamine or peptidoglycan fragments induces secretion of elastase and pyocyanin by P. aeruginosa through the proposed two-component response regulator PA0601 [46]. When P. aeruginosa and S. aureus were co-infected at the same time in the wound, P. aeruginosa exceeded S. aureus by more than 100-fold at 4 dpi. The results showed that sensing of S. aureus peptidoglycan by P. aeruginosa triggered the release of lytic virulence factors allowing P. aeruginosa to overtake S. aureus in co-infected wounds, although P. aeruginosa deletion mutants of PA0601 (involved in peptidoglycan sensing) were unable to outnumber S. aureus in the same way [46]. During co-infection with P. aeruginosa in porcine wounds, virulence factor protein A of S. aureus was drastically under-expressed by more than threefold at 2 and 4 dpi, whereas alpha-hemolysin and Panton-Valentine leukocidin (PVL) that cause necrosis in Table 1 Major genera in chronic wounds [60,[62][63][64][65][66][67][68][69] [67][68][69] wounds were significantly upregulated at 4 dpi. Furthermore, compared to their monomicrobial infection, P. aeruginosa and S. aureus co-infection caused downregulation of pro-inflammatory cytokines, including IL-1 alpha, IL-1 beta, IL-6, and IL-8 [45•]. These findings reveal that negative interactions between microbes can lead to changes in the expression and activity of multiple virulence factors along with host immune response regulators, giving a competitive advantage to a few select species over others in a polymicrobial infection (Fig. 2). Polymicrobial infections compromise the integrity of skin's surface and can exacerbate the chronicity of wounds with impaired healing. In a porcine burn wound model, coinfected with P. aeruginosa and Acinetobacter baumannii, the mammalian tight junction proteins zona-occludens-1 and zona-occludens-2 (ZO-1, ZO-2) were significantly downregulated as opposed to non-infected controls, leading to a functionally compromised epidermis [83]. In another study of porcine wound infection model, upon P. aeruginosa and S. aureus coinfections, there was a marked decrease in wound re-epithelisation due to inhibition of keratinocyte growth factor 1, compared to monomicrobial infections [45 •]. These findings also show that polymicrobial infections with biofilms hinder wound closure and may make the host more susceptible to other opportunistic infections. The microorganisms in chronic wounds and the neighbouring healthy skin are often not quite easily distinguishable [70,84].
Commensals have traditionally been overlooked in acute wounds; however, there is increasing support that they may play a bigger part in the pathogenesis and prognosis of chronic wounds. Corynebacterium spp., for instance, releases a secretory component that suppresses the agr regulatory system in the pathogen S. aureus and its virulence factors [85], thereby preventing colonisation and infection on healthy skin. However, C. striatum has been identified as an emerging multidrug-resistant wound pathogen [86], causing a high proliferation rate in the epidermal layer in diabetic wound model in mice [72], where competition between C. striatum and S. aureus may result in additional damage to wound tissue. S. epidermidis, an identified skin commensal, is also known to play a role in wound worsening; yet, certain strains of S. epidermidis enhance the ability of immune system to speed up the healing of the wound [87]. In a comparative examination of diabetic mouse models, db/db diabetic mice wounds had a larger proportion of Staphylococci than wounds of db/ + diabetic mice, suggesting a negative correlation to wound healing ability [88]. Furthermore, in diabetic mice wounds with the predominance of Staphylococci, genes involved with an acute inflammatory response were significantly upregulated, and the persistent inflammation was attributed to the chronicity of the wounds [89].
Our present knowledge of how the local skin microbiome impacts the predisposition for wound infections is limited.
Future research aimed at studying the potential of local microbiota to manipulate the host response and how their interactions with other opportunistic pathogen(s) during infection would affect wound healing efficacy can provide significant clues on managing chronic wounds. Simply put, commensal activity is crucial in a wound environment, but its ecological importance is complicated and should not be overlooked. Identifying the secreted components generated by these organisms would be a significant step forward in understanding how they build up microbial populations, communicate with other microbial communities, and regulate dialogue with the host cell.

Conclusion
The skin is a unique and important part of the human body that shields and protects us from our surroundings. It also offers a variety of habitats with unique microenvironments that contribute to a diverse array of microbial communities taking residing there. While tremendous strides have been made in recognising the significance of the skin microbiome in the establishment of host immune responses and microbial resistance, there is still much more to learn. The majority of current research has been concentrated on prominent skin microbes belonging to the genera Staphylococcus and Cutibacterium, which include both helpful and harmful species. However, other major genera such as Corynebacterium, Kocuria, Micrococcus, and Brevibacterium (belonging to the phylum Actinobacteria which is also a predominant group associated with human skin) have not been extensively researched. All of the mentioned genera are known to cause infection, but they are mainly understudied, and their effects on the skin and other skin microbiota are unclear. Even well-studied species like S. epidermidis are less explored in the context of molecular processes that drive microbe-microbe interactions and immune responses. Even though it has been well established that fungal colonisers contribute heavily to skin microbial populations and in regulating skin health, their interactions and immunomodulatory roles are a little-known component of the vast skin microbiome. The potential of microbes to adapt to changes in the wound microenvironment leads to virulence, exacerbation of wound, and delay in wound healing. To provide effective therapies for chronic wounds, it is critical to examine the function of microbes at the cellular and molecular levels, not only focusing on bacteria but also on other natural flora such as viruses and fungus. A better knowledge of the interactions between skin cells, normal flora, and their environment can help us in distinguishing between normal and pathological healing and aid researchers in developing better therapies or strategies for effectively eliminating the pathogenic microorganisms found in chronic wounds. Understanding the molecular markers and defence mechanisms that regulate the fine line between the commensal and pathogenic nature of microbes will help develop innovative therapies and even direct towards engineering a 'healthy' skin microbiome. Determining critical microbial elements and metabolomic patterns that could be employed as diagnostic biomarkers to determine clinical populations at risk of wound infection can lead to non-invasive diagnostic and therapeutic procedures that are more precise.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.