What’s in a name? Characteristics of clinical biofilms

Abstract In vitro biofilms are communities of microbes with unique features compared to individual cells. Biofilms are commonly characterized by physical traits like size, adhesion, and a matrix made of extracellular substances. They display distinct phenotypic features, such as metabolic activity and antibiotic tolerance. However, the relative importance of these traits depends on the environment and bacterial species. Various mechanisms enable biofilm-associated bacteria to withstand antibiotics, including physical barriers, physiological adaptations, and changes in gene expression. Gene expression profiles in biofilms differ from individual cells but, there is little consensus among studies and so far, a ‘biofilm signature transcriptome’ has not been recognized. Additionally, the spatial and temporal variability within biofilms varies greatly depending on the system or environment. Despite all these variable conditions, which produce very diverse structures, they are all noted as biofilms. We discuss that clinical biofilms may differ from those grown in laboratories and found in the environment and discuss whether the characteristics that are commonly used to define and characterize biofilms have been shown in infectious biofilms. We emphasize that there is a need for a comprehensive understanding of the specific traits that are used to define bacteria in infections as clinical biofilms.


Introduction
Historicall y, biofilms hav e been c har acterized by v arious featur es that distinguish them from planktonic populations .T he first descriptions of biofilms were based on their morphological properties, as tools for visible observations were the only methods available until a few decades ago.The term ' biofilm ' was used for the first time in a publication from 1981 (McCoy et al. 1981 ).Before introducing the term 'biofilm', several studies have described the phenomenon of bacteria making clumps or small microcolonies.In the 1930s, some of the first detailed descriptions of microbial attachment to glass surfaces submerged in water were published.They observed growing cells on the surface forming microcolonies increasing in size and described the organisms to grow in 'a fairly uniform film' (Henrici 1933 , Zobell andAllen 1935 ).The first reported clinical observation of what we today recognize as biofilms wer e pr esented in 1977.A Gr am-stained smear of a sputum sample from a cystic fibrosis (CF) patient revealed 'heaps of bacteria ' (Hoib y 1977 ).Ho w e v er, clumps of bacteria had already been reported in the 1650s by van Leeuwenhoek and were also mentioned in a publication from 1883 that described how bacteria grow on a surface to form clumps , i.e .'biofilms' (Weismann et al. 2007 ).The visual inspection of biofilms entered a new era after the introduction of confocal laser scanning microscopy, which allo w ed inv estigation of the formation of in vitro gr own bacterial communities in greater detail (Lawrence et al. 1991 ).The subsequent introduction of various molecular methods allo w ed a more holistic approach to characterize biofilms.For example, staining of specific exopol ysacc harides has r e v ealed the existence of self-produced matrix components (Co w an et al. 2000, Sohm et al. 2011 ).Further, the introduction of various system le v el a ppr oac hes made it possible to c har acterize genomic, tr anscriptomic , proteomic , and metabolomic differences between microorganisms living different lifestyles.As we continue to investigate environmental and clinical systems, it will be important to know if small clusters or groups of cells (that are commonly observed in these samples) are exhibiting biofilm-like properties and physiology.T here ha ve been many descriptions and discussions regarding the definition of biofilms (Costerton et al. 1999, Sauer et al. 2022 ) and the main goal of this r e vie w is not to establish a new definition of a biofilm, but rather to discuss the various characteristics, alone or in combination, that can be used to define a biofilm.

How to diagnose a biofilm?
Biofilms can exhibit great diversity depending on their species, composition, and local environment.Factors such as nutrient av ailability, pH, temper atur e, and the presence of multiple organisms all have an impact on the structure and composition of a biofilm within a single species.As a result, the characteristics of a biofilm can vary greatly between different ecosystems .T hus , under certain conditions biofilms are intricate and highly dynamic communities of microorganisms that interact with each other and with their surr oundings, ada pting to cr eate complex structures (Flemming and Wuertz 2019 ).These comm unities ar e extr emel y r esilient and can endur e harsh conditions.As a conse-quence, biofilms can survive and thrive in a variety of environments and can range in size from a fe w micr ons to se v er al millimeters in thickness (Reysenbach andCady 2001 , Bjarnsholt et al. 2013 ) and can be enclosed in a matrix consisting of extracellular polymeric substance (EPS) that can promote individual cells to stick together, adhere to surfaces, and provide protection from envir onmental str essors (Wingender et al. 1999 ).
Not sur prisingl y, ther e ar e ob vious differ ences between in vivo biofilms (i.e.those occurring in a clinical setting and in the natural or man-made environment), and in vitro biofilms grown in the laboratory (Hall-Stoodley et al. 2004 ), but e v en in vitro biofilms can be diverse in terms of phenotype and architecture (Pamp andTolker-Nielsen 2007 , McBain 2009 ).This diversity highlights the importance of a discussion on which characteristics can be considered hallmarks of a biofilm.The various characteristics shown in Fig. 1 will serve as the foundation for this r e vie w.We divided the central aspects that can be used to c har acterize a biofilm into the following four features, (i) physical, (ii) chemical, (iii) phenotypic, and (iv) gene expr ession pr ofiles .T his r e vie w will include findings of biofilms from various settings, but the primary focus will be on clinical biofilms.

Physical features
The physical features of bacterial biofilms are complex and varied and can play an important role in the survival and persistence of bacteria in diverse en vironments .T he three k e y physical features tr aditionall y used to describe a biofilm are (i) a community of cells in close proximity, (ii) adhesion/attachment of cells to a biotic or abiotic surface, and (iii) a ggr egates encased in a self-produced or externall y pr o vided matrix.T hese c har acteristics v ary gr eatl y depending on the environment and the micr oor ganisms involv ed, resulting in a diverse range of biofilm sizes and structures.If one were to take this approach to define a biofilm, an obvious question is whether a cellular a ggr egate m ust be a certain size or contain a minimum number of cells before it can be c har acterized as a biofilm?For example, can we define two cells embedded in a matrix as a biofilm?Or conv ersel y, can we define thousands-millions of cells in close contact as a biofilm e v en if they are not embedded in an obvious matrix?

Physical size
Biofilms have been found in a broad size range in infections ranging from large multicellular aggregates to small clusters of only a few μm in diameter (Bjarnsholt et al. 2013 ).The physical dimensions of a ggr egates is incr easingl y r ecognized as an important factor as it influences the de v elopment of physiological heterogeneity within biofilms (Stewart and Franklin 2008 ).The dynamics underl ying the observ ed size distribution ar e not clear, but they are influenced by multiple parameters such as access to metabolic substr ates, gr azing by predators or immune cells, antimicrobial compounds , physical constraints , and so on.In theory, there is no upper limit of biofilm size but in many soft tissue infections, biofilms are typically found in the range from 5 to 200 μm (Bjarnsholt et al. 2013 ).In the environment, biofilms such as the photosynthetic mats commonly found in hot springs can easily be observed by the naked eye.While there is no consensus on a specific threshold for the number of cells required to form a biofilm, man y articles hav e used a lo w er diameter of 5 μm to distinguish biofilms from single cells (Bay et al. 2018, Kolpen et al. 2022 ).In a recent study, it w as further sho wn that successful phagoc ytosis of bacterial a ggr egates by pol ymor phonuclear leucocytes (PMNs) dr amaticall y decr eased with a ggr egate diameters of > 5 μm (Alhede et al. 2020b, Pettygr ov e et al. 2021 ).
T hus , a selectiv e pr essur e may act on biofilms to attain a certain size to resist such competition.Ho w ever, such selective pressur es ar e complex and can v ary depending on the specific envir onmental conditions and species involved.It thus appears that the size and structure of biofilms can be influenced by a combination of genetic factors , en vironmental cues , and microbial interactions within the community and with their surroundings.

Matrix
Self-produced EPS, or the biofilm matrix, remains one of the most common c har acteristics used in definitions of biofilms .T he bacterial EPS consists of a range of different biopolymers such as pol ysacc harides, pr oteins, and DNA and its function, composition, and div ersity hav e been thor oughl y r e vie wed else wher e (Flemming et al. 2023 ).
Pr oduction of self-pr oduced matrix has been demonstrated in e .g. CF sputum (J ennings et al. 2021 ) and in c hr onic wounds (Kirketerp et al. 2008 ).Ho w ever, is a self-produced matrix a necessity to define an a ggr egate of cells as a biofilm?It has , e .g. been shown that se v er al species r el y on the matrix production of other species to form biofilm (Chenicheri et al. 2017 ).In airway infections, bacteria are found in aggregates embedded in host mucus and it has been shown that host-derived eDNA surrounds aggregates of bacteria effectiv el y shielding them from their surroundings (Alhede et al. 2020a ).In wounds, bacteria can be immobilized in necrotic tissue and wound slough (Kirketerp-Moller et al. 2008 ) and in many other soft tissue infections, host secretions have been found to contain bacteria (Bjarnsholt et al. 2013 ).
Inter estingl y, a r ecent study sho w ed a multitude of single, spatiall y separ ated, bacterial cells in secr etions fr om a r ange of acute and c hr onic pulmonary infections (Kolpen et al. 2022 ).The study sho w ed the presence of polysaccharides within the biofilms by Alcian Blue staining but the finding questions whether the conventional 'biofilm' mode of growth and self-produced EPS is necessary for surviving a persistent inflammatory r esponse.Finall y, Jennings et al. ( 2021 ) r ecentl y demonstr ated that self-pr oduced matrix is produced and surrounds Pseudomonas aeruginosa aggregates from CF sputum.T he infectious microen vironment is often characterized by being high in nutrients, but oxygen depleted, thus creating a limit for the metabolic rate (Bjarnsholt et al. 2022, Lic htenber g et al. 2022a ), which may hinder production of EPS as this production is associated with ele v ated metabolic expenditur e (Lic htenber g et al. 2022a ).

Aggregation and adhesion mechanisms
A multitude of different mechanisms of biofilm formation have been elucidated through decades of r esearc h and an expansion of the biofilm life cycle was r ecentl y pr oposed to include both attac hed and nonattac hed biofilms (Sauer et al. 2022 ).Initial surface colonization by bacteria is ac hie v ed thr ough activ e adhesion (via e.g.type IV pili) and follo w ed b y clonal gro wth and potential recruitment of other bacteria that can 'stick' to the matrix.For nonattac hed biofilms, thr ee mec hanisms ar e curr entl y known: (i) restricted motility whereby clonal expansion will create aggregated bacteria, (ii) bridging a ggr egation wher e bacteria stic k to each other by production of EPS, and (iii) depletion aggregation wher e a ggr egates can be enclosed by pol ymers by entr opic forces in certain environments (Kragh et al. 2023 ).
These mechanisms describe how attached or nonattached clusters of bacteria can form.If the mechanism can be identi-Figure 1. Characteristics that are commonly used to define biofilms include the number of cells attached to a surface and/or present in aggregates, attac hment factors, pr esence of (heter ogeneous) (sub)populations and physicoc hemical gr adients, toler ance to antibiotics and external str essors, cell-to-cell comm unication, alter ed gene expr ession, metabolicall y distinct phenotypes, and the pr esence of self-pr oduced extr acellular matrix.fied from a given cluster of cells, this can be used to infer other information about the bacterial community, e.g.certain gene expr ession patterns ar e corr elated with some of the mechanisms.For example, in Gr am-negativ e species an increased levels of c-di-GMP is associated with matrix production (Andersen et al. 2021 ).Ho w e v er, none of these mechanisms can infer extensive information on the behaviour or phenotypical traits of the bacteria in the biofilm, other than that they, at some point, formed a biofilm.Additionall y, these mec hanisms do not explain the occurr ence of slo w gro wing, spatially separated, single cells in inflamed host secretions (Kolpen et al. 2022 ).

Gene expression profiles
There is an ever-growing number of studies in which gene expression is compared between planktonic (suspended) microbial cells and biofilm-associated cells (Whiteley et al. 2001, Schembri et al. 2003, Dotsch et al. 2012, 2015, Alio et al. 2020, Zheng et al. 2022, Wang et al. 2022b, Toliopoulos and Giaouris 2023 ).In almost all of these studies differences in expression levels are observed for a smaller or lar ger fr action of genes, although comparisons between differ ent studies ar e difficult at best, due to differ ences in experimental conditions (different model systems for biofilm and planktonic gr owth, temper atur e, gr owth media, dur ation of biofilm formation, and so on) and as a consequence there is very little overlap between genes identified as up-or downregulated in biofilms in different studies (Coenye 2010 ).In addition, many studies are limited by the low accuracy of the laboratory models used, and the transcriptomic profiles obtained from in vitro or nonhuman in vivo models , ma y differ substantiall y fr om the transcriptome during human infection, as was, e.g.shown for P. aeruginosa (Cornforth et al. 2018, 2020, Harrington et al. 2022, Lewin et al. 2023 ) and Staphylococcus aureus (Xu et al. 2016, Ibberson and Whiteley 2019, Le Masters et al. 2021 ).
In addition, microbial biofilms are not homogeneous populations (Lenz et al. 2008 ) and as a consequence gene expression data obtained from such populations by definition present an 'aver a ge pictur e', that may not necessaril y r eflect meaningful biological signals.Early studies on heterogeneity in biofilms required gener ating m utants in whic h gene expr ession could be monitor ed micr oscopicall y (e.g. by cr eating GFP tr anscriptional fusions) (Ito et al. 2009 ), a combination of isolating single cells and qPCR (Perez-Osorio et al. 2010 ), or isolating subpopulations, follo w ed b y tr anscriptional pr ofiling with micr oarr ays (Williamson et al. 2012, Heacock-Kang et al. 2017 ).More recently, probe hybridizationbased a ppr oac hes hav e been used to ma p spatial differ ences in bacterial biofilms (Dar et al. 2021, Livingston et al. 2022 ).While these a ppr oac hes differ in r esolution and thr oughput, they all confirm that spatially resolved heterogeneity is the norm, not the exception, highlighting the importance of the (physicochemical) micr oenvir onment in sha ping the micr obial tr anscriptome and phenotype (Dar et al. 2021, Bjarnsholt et al. 2022, Lic htenber g et al. 2022a ).This heterogeneity is not only observed in biofilms, but also in planktonic cultures (Lenz et al. 2008, Ryall et al. 2012 ).For example, it was found that up to 90% of the biomass of P. aeruginosa 'planktonic' cultures consists of cellular aggregates with a diameter of 10-400 μm (Sc hlehec k et al. 2009 ).Recent technological adv ances hav e made it possible to perform single-cell RNA (scRNA) sequencing on bacterial cells and pioneering scRNA-seq studies have confirmed heterogeneity in various planktonic bacterial populations, including Bacillus subtilis, Salmonella enterica, Esc heric hia coli , and Clostridium perfringens grown in v arious ric h media (Brennan and Rosenthal 2021, Kuchina et al. 2021, Homberger et al. 2023, McNulty et al. 2023 ).Various innov ativ e scRNA-seq a ppr oac hes hold gr eat pr omise for the futur e inv estigation of heterogeneity of microbial populations, both sessile and planktonic, and especiall y a ppr oac hes that allow to link specific expression profiles with spatial information and/or information about the physicoc hemical micr oenvir onment will yield nov el insights (Wang et al. 2023 ).
Variability between strains from one species or multiple closely related species should also be considered.In P. aeruginosa , variability in transcriptional profiles between 77 clinical strains was higher when these were grown as biofilms than when they were gr own planktonicall y, suggesting the impact of the genetic background of individual strains on which genes are expressed in biofilms is bigger than the impact on which genes are expressed in planktonic cultures (Thoming et al. 2020 ).The core biofilm transcriptome (i.e.genes differentially expressed between planktonic and sessile cultures in all 77 clinical P. aeruginosa isolates) consisted of only 143 genes, 103 that were commonly upregulated in biofilms and 30 commonly downregulated compared to planktonic cultures.Among the upregulated genes were several genes required for p y over dine biosynthesis, heme assimilation, and central carbon metabolism, as well as genes encoding super oxide dism utase and fumar ate hydr atase.Downr egulated core genes include genes involved in denitrification and aerobic arginine catabolism (Thoming et al. 2020 ).Among the top 250 biofilm-expressed genes in seven Stenotrophomonas maltophilia isolates, 106 genes were commonly expressed in all isolates, while 142 of the 250 most str ongl y expr essed genes were only expressed in one of se v en isolates (Alio et al. 2020 ).Notabl y, the expr ession of the majority of these 250 genes str ongl y expr essed in S. maltophilia biofilms is not biofilm-specific, as they are also highly expressed in planktonic cultures.In S. aureus , profound differences were observed in biofilm-associated gene expression in repr esentativ es of three important MRSA clones (Vlaeminck et al. 2022 ).When comparing expression differences between planktonic and sessile populations at the KEGG pathway le v el, the number of pathways v aried fr om 11 ( S. aureus ST239), ov er 27 ( S. aureus USA300) to 58 ( S. aureus HEMRSA-15).Mor eov er, onl y a single common differ entiall y expr essed gene was identified acr oss these three S. aureus clones , i.e .clfA , encoding clumping factor A (Vlaeminc k et al. 2022 ).Interstr ain heter ogeneity in gene expression was also observed in Salmonella Typhimurium (Zheng et al. 2022 ) and Listeria monocytogenes (Toliopoulos and Giaouris 2023 ) biofilms.
While most studies have focused on differences between planktonic and sessile cultures, it is worth mentioning that based on transcriptomic analyses, dispersed P. aeruginosa cells (i.e. cells r eleased fr om a biofilm) ar e differ ent fr om both planktonic and sessile cells, and that the mode of dispersion has a profound influence on gene expression in dispersed cells (Chua et al. 2014, Wille et al. 2020 ).
The curr entl y av ailable data seem to indicate that ther e is no such thing as a universal 'biofilm transcriptome', nor is there any evidence for a universal 'planktonic transcriptome' or 'dispersed cell transcriptome'.An important reason for this is the heterogeneity commonly found in microbial populations; these populations more resemble a collection of subpopulations with distinct pr operties, r ather than a collection of cells with identical properties.With further technical advances in transcriptome analysis and imaging, it will likely become feasible to determine spatial differences in gene expression in microbial biofilms at the singlecell le v el.T his ma y shed mor e light on the r elationship between the micr oenvir onment, local differ ences in gene expr ession, and phenotype.

Chemical features
From a spatial perspective, the distribution of e.g.metabolites may be used to c har acterize biofilms.In planktonic cultures, a homogenous distribution will be expected whereas biofilms will produce heter ogeneous landsca pes of metabolite concentr ation due to reaction-diffusion processes (Stewart 2003, Pabst et al. 2016, Stewart et al. 2016, 2019, Kirketerp-Møller et al. 2020 ).
Are certain metabolic products always present in biofilms?Often e.g.active denitrification or fermentation is used to exemplify that oxygen has been consumed by dense biofilm structures (Pabst et al. 2016 ).Ho w ever, the expression of anaerobic metabolic pathways is not biofilm specific.
Ther e ar e onl y fe w studies inv estigating the pr oteome of biofilms by proteomics and/or metabolomics.A recent study used targeted and untargeted metabolomics to compare the metabolism of biofilm and planktonic cultures of the clinical uropathogenic E. coli UTI 89 strain.A metabolic reprogramming was found to be involved in biofilm formation by increasing metabolites, such as amino acids, sugars, lipids, uridines, and organic acids that are essential for EPS synthesis (Lu et al. 2019 ).The metals Fe 3 + , Mn 2 + , and Mg 2 + have been reported to regulate biofilm formation by regulation of functional metabolism in E. coli (Guo and Lu 2020, Guo et al. 2021, Wang et al. 2022a ).
The nucleotide second messengers cAMP and bis-(3 -5 )-cyclic dimeric GMP (c-di-GMP) ar e involv ed in biofilm formation.High intr acellular le v els of c-di-GMP ar e associated with formation of a biofilm, while low le v els ar e associated with the planktonic lifestyle (Hengge 2009, Dahlstrom and O'Toole 2017, Collins et al. 2020, Martinez-Mendez et al. 2021 ).In gener al, the expr ession and/or activity of flagella is reduced by high le v els of c-di-GMP whereas the expression of adhesins and biofilm-associated exopol ysacc harides is upregulated.In P. aeruginosa, c-di-GMP positiv el y r egulates the pr oduction of se v er al matrix components (alginate, CdrA adhesin, Cup fimbriae, and Pel/Psl pol ysacc harides) (Borlee et al. 2010, Baraquet and Harwood 2013, Fazli et al. 2014 ).Opposite to c-di-GMP, the global transcription factor cAMP r eceptor pr otein (CRP) can both pr omote and inhibit biofilm formation.As an example, CRP promote biofilm formation in E. coli and P. aeruginosa , whereas it inhibits biofilm formation in Serratia marcescens and Vibrio cholerae (Liu et al. 2020 ).In addition, it modulates biofilm maintenance in Shewanella putrefaciens by interaction with the c-di-GMP effector, BpfD (Liu et al. 2022 ).There is compelling evidence that these secondary messengers are k e y biofilm modulators.During biofilm formation, a high le v el of intercellular c-di-GMP forces the cells to use a large amount of energy for the production of exopolysaccharides that can subsequently lead to resource depletion and a low cellular metabolic state (Lic htenber g et al. 2022b ).The le v el of c-di-GMP is supposedly a good indicator of the presence of biofilms.The challenge is whether it can be measured direct in clinical biofilms and furthermore, can we expect continuous high levels of c-di-GMP in biofilm cells after prolonged embedment in human tissue?

Phenotypic features
The phenotypic features of biofilms have been studied extensively to gain insights into how a biofilm functions in different environments .T hey are crucial for the survival and persistence of biofilms in different harsh en vironments .All the characteristics presented in this r e vie w influence the phenotype of a biofilm.Biofilms often exhibit a high degree of heterogeneity, meaning that different regions within the biofilm can have different populations of bacteria with distinct phenotypes .T he phenotypic variations of the individual bacterial cells can be attributed to genetic differences (Hallet 2001 ), epigenetic modifications (Guespin- Michel 2001, Smits et al. 2006 ), or envir onmental cues (Spr att and Lane 2022 ).This phenotypic heterogeneity enables some bacteria to adopt specialized roles within the biofilm, such as metabolically active cells in surface layers or dormant cells in deeper regions which forms distinctiv e micr oenvir onments in the spatial organization of a biofilm (Pamp et al. 2008 ).The heterogeneity of biofilms has pr edominantl y been studied in vitro , and it is unclear whether the same spatial differences occur in clinical biofilms; likewise, it is unclear how this differs acr oss v arious infection sites, bacterial species, and infection durations.

The metabolic state of a biofilm
Biofilms per se are often characterized as inactive/dormant as well as hypoxic or anaerobic.Ho w ever, this is a dynamic process, as O 2 is consumed because they have high metabolism during growth; when O 2 is then depleted, growth will decrease.In the absence of external oxygen sinks, O 2 will then build up again by diffusion and growth can resume until a steady state is reached, Ho w ever, in vivo , other O 2 consumers will be pr esent suc h as PMNs that use O 2 for their oxygen r adical pr oduction.This will lead to persistent hypoxic conditions surrounding the biofilms.On the scale of a single biofilm or a ggr egate, heter ogenic metabolic states can develop in very small aggregates (Wessel et al. 2014 ) where the outer layers of the biofilm are supplied with substr ate, whic h is then depleted to w ar ds the inner parts of the biofilm.This can lead to subpopulations displaying different susceptibilities to antibiotics that are influenced by metabolic state (Lic htenber g et al. 2022c ).The metabolic state can be manipulated by increasing the suppl y of substr ate, whic h has been demonstr ated by a ppl ying hyperbaric oxygen treatment to biofilms which resensitized the biofilm to antibiotics that tar get activ el y gr owing bacteria (Kolpen et al. 2016, 2017, Lerc he et al. 2017 ).A r ecent publication suggested that single-celled bacteria also displayed low metabolic rates in infections of the lo w er r espir atory tr act (Kolpen et al. 2022 ).T hus , the inactive state is dictated by the environment and may give insight into the phenotype of the bacteria but cannot be used as a defining factor of biofilms.

Biofilm tolerance
Biofilms possess various mechanisms to increase tolerance to antibiotics and to e v ade and persist the host immune system.The mechanisms of tolerance to w ar ds antibiotics have been thoroughl y r e vie wed else wher e (see e.g.Van Ac ker et al. 2014, Ciofu and Tolker-Nielsen 2019, Ciofu et al. 2022 ), but briefly it can be subdivided into different categories; (i) the physical tolerance , i.e .ac hie v ed when penetration is restricted and the antibiotic does not r eac h all bacteria in the biofilm.(ii) The physiological tolerance , where e .g. slo w gro wth renders the antibiotic target inactive (e.g. protein synthesis).(iii) The transcriptional tolerance, where expression of specific (sets of) genes confers tolerance .T his has been argued to include e.g.elevated c-di-GMP levels that lead to upregulation of efflux pumps (Gupta et al. 2014 ).
To withstand and persist despite a highly activated immune defense some pathogenic bacteria produce various compounds causing necrotic killing of PMNs (Jensen et al. 2007, Löffler et al. 2010 ).In addition, it has been reported that the size of bacterial a ggr egates significantl y affects the outcome of phagocytosis of S. aureus , E. coli , P. aeruginosa , and S. epidermidis .Aggregates with a diameter size of 5 um or smaller were successfully phagocytosed by PMNs, while lar ger a ggr egates wer e less likel y to be pha gocytosed (Alhede et al. 2020b ).
The subject of biofilm tolerance is still widely debated but many of the tolerance mechanisms are associated with bacteria residing in dense biofilms while tolerance also occurs in cells not associated with a biofilm.The tolerance of biofilms must be considered the most crucial characteristic in relation to infections.

Characteristics of clinical biofilms-where ar e w e?
All the c har acteristics and mec hanisms described abov e, hav e been shown to contribute to the 'biofilm' lifestyle in environmental and in vitro gro wn biofilms.Ho w e v er, the r elativ e importance of each factor is unknown for clinical biofilms .T he question is whether they are all present and required to define a clinical biofilm.Micr oscopy ima ges of tissue sections from patients reveal that clinical biofilms can be organized in very small aggregates consisting of less than 100 cells (Kolpen et al. 2022 ), but it is unknown whether these small microcolonies show the same c har acteristics as lar ger colonies in terms of metabolic state and incr eased toler ance-c har acteristics, whic h ar e normall y used to distinguish biofilms from single cells.
The self-produced EPS matrix has been shown to confer incr eased toler ance in some settings (Goltermann and Tolker-Nielsen 2017 ) but on the other hand, the metabolic state of the micr oor ganisms has also been shown to be of major importance (Lopatkin et al. 2019 ).T hus , an incr eased antibiotic toler ance may be acquired independently of EPS production.Biofilm infections often have a long-time span with a potential change in c har acteristics that are not well understood (Cao et al. 2023 ).Such longitudinal changes are, thus still very difficult to investigate using laboratory-and animal experiments .New technologies , such as MALDI imaging (MALDI mass spectr ometry ima ging) and scRNAseq are starting to emerge and being used on clinical samples making it possible to investigate spatial differences in proteomics, metabolomics, and gene expression in and around bacterial com-m unities dir ectl y in the infection site .T his will undoubtedly yield more knowledge of the clinical biofilm characteristics in the future.
The term 'biofilm' can be associated with all the factors described in this r e vie w (and more), but despite all the c har acteristics that have been used to describe biofilms, very few are omnipr esent, if an y.We ar e still dependent on visualizing bacteria in the infection to determine if the cells are situated in a biofilm, but e v en then, the role of nongrowing single cells may be neglected.This questions whether the classification of bacteria according to arc hitectur e pr omotes a better understanding of infections and we argue that for infections , it ma y be more appropriate to classify bacteria according to treatment response.