Enteroaggregative E. coli (EAEC) is a pathotype of diarrheagenic E. coli (DEC) characterized by a stacked-brick adherence (Nataro et al., 1987; Vial et al., 1988). EAEC is a cause of acute and chronic diarrhea worldwide (Flores and Okhuysen, 2009; Hebbelstrup Jensen et al., 2014, 2016), and a common cause of travelers’ diarrhea (Adachi et al., 2001; Hebbelstrup Jensen et al., 2014; Porter et al., 2017; Barrett and Brown, 2018; Guiral et al., 2019; Walters et al., 2020). EAEC is associated with acute diarrhea in children (Nataro et al., 2006; Hebbelstrup Jensen et al., 2014, 2016; Modgil et al., 2021; Kabir et al., 2022). In developing countries EAEC infection is associated with failure-to-thrive (FTT) (Steiner et al., 1998; Hebbelstrup Jensen et al., 2016; Rogawski et al., 2017; Das et al., 2021). FTT is characterized by a rate of weight gain below normal (Jaffe, 2011). Beyond stunting adult height, FTT can lead to learning difficulties and reduced adult earning potential (Jaffe, 2011).

Currently, there is no prevention for EAEC infection (Flores and Okhuysen, 2009), though travelers’ diarrhea is treated with azithromycin or rifaximin (DuPont et al., 2001; Infante et al., 2004; Connor et al., 2012; Riddle et al., 2017; Tribble, 2017). Perhaps due to widespread antibiotic use, EAEC worldwide have high rates of antibiotic resistance (Mendez Arancibia et al., 2009; Aslani et al., 2011; Guiral et al., 2019; Elhadi et al., 2020; Eltai et al., 2020).

EAEC is characterized by its aggregative adherence and ability to create a biofilm (Nataro et al., 1987, 1992; Albert et al., 1993; Czeczulin et al., 1997). The aggregative adherence fimbriae (AAF) are important for both adherence to epithelial cells and other surfaces (Nataro et al., 1987; Nagy et al., 2016; Petro et al., 2020; Schiller et al., 2021), and biofilm formation (Nataro et al., 1992; Czeczulin et al., 1997; Boisen et al., 2012). One of the five genetically different AAF types (AAF1 to AAF5, also called AAF/I to AAF/V) are found in all “typical” EAEC (Savarino et al., 1994; Boisen et al., 2008). AAF1 was shown to be important for colonization and biofilm formation on spinach and abiotic surfaces (Nagy et al., 2016). AAF2 and AAF4 were significantly associated more with diarrhea cases than asymptomatic children in India (Modgil et al., 2021). A study on EAEC-mediated diarrhea in Iranian children found that isolates positive for agg4A (AAF4), pic, and sepA formed a stronger biofilm in vitro than strains without the three genes (Nezarieh et al., 2015). Biofilm formation on the intestinal mucosal is thought to be important for EAEC to cause illness (Vial et al., 1988; Tzipori et al., 1992; Nataro et al., 1996; Nataro and Kaper, 1998; Hebbelstrup Jensen et al., 2014).

As part of a biofilm, many pathogens have an extracellular matrix made up of proteins, carbohydrates, and extracellular DNA (eDNA) (Sanchez-Torres et al., 2010; Flemming et al., 2023). These extracellular components have important roles for antibiotic tolerance and transmission of antibiotic resistance, inflammation, and immune evasion (Tetz et al., 2009; Tetz and Tetz, 2010; Zhao et al., 2013; Gunn et al., 2016; Sharma et al., 2016). Targeting the extracellular components of a biofilm can improve treatment effectiveness (Jiang et al., 2020). The prototype EAEC strain 042 (AAF2) is known to make a polysaccharide-rich coat (Borgersen et al., 2018), but the overall composition of the EAEC biofilm is unknown.

In this work, we characterized the extracellular matrix of biofilms formed by recently isolated EAEC strains. We selected two strains for each of the five genetically distinct AAF types to provide a cross section of different EAEC strains. We defined EAEC for this work as having both aggR [EAEC virulence gene regulator (Nataro et al., 1994; Morin et al., 2010, 2013; Boisen et al., 2019)] and the genes for production of an AAF. We hypothesized that the selected EAEC would have an extracellular matrix with different combinations of protein, eDNA, and carbohydrate that would correlate to AAF type and/or to level of biofilm formation.

In this study, we found that quantitative biofilm staining did not correlate with AAF type, that overall, the EAEC biofilm appears similar in images, and that common treatments to target biofilms had limited effects. We explored the composition of the extracellular biofilm matrix of EAEC by several methods. Fluorescent stains were used to localize protein, eDNA, and glycoprotein within the biofilm. The results of those staining studies showed that EAEC strains have similar biofilm composition, and that all of the targets (protein, eDNA, glycoprotein) are present in the EAEC extracellular biofilm matrix. Overall, glycoprotein staining was less prevalent than eDNA and protein staining. The lower level of glycoprotein staining we observed might be due to the selective nature of the WGA conjugate we used. We did note that two strains, E3V1C (AAF1) and 55989 (AAF3), which had a high level of quantitative biofilm staining, showed more compact biofilm in the images. We do not know the reason for the distinction, but suspect that the nature of those two biofilms may allow for higher levels of crystal violet staining. Overall though, we were surprised that we did not observe more differences among the EAEC strains because of the differences in biofilm staining levels among the strains ( Figure 1 ) and heterogeneity in the virulence factors such as adhesins ( Table 1 ). We found the extracellular matrix of EAEC to be primarily associated with the individual cells, and that the matrix did not appear to extend out very far from the bacteria. We had hoped to observe a complex 3D structure of biofilm with confocal imaging. Instead, we found a mostly flat biofilm of one to two bacterial cell layers thick on the glass disks. This result could be due to the static growth conditions, which may allow weaker cell-to-cell attachment than a biofilm formed in the human gut which is subject to a strong flow.

We added treatments that target protein, eDNA, or carbohydrate into the medium with a growing biofilm to determine if such treatments would reduce biofilm staining. We expected to find that treatments that target protein, eDNA, or carbohydrate would reduce the staining of the extracellular component they targeted, i.e. that DNase would reduce the amount of TOTO-1 iodide staining, reflecting a reduction in eDNA. However, we observed no apparent reduction in matrix components in biofilm images when those treatments were used in the growth medium. In contrast, we did see changes in the arrangement of the EAEC cells in some images. For example, the E19V1A (AAF2) and 55989 (AAF3) biofilms treated with proteinase K had apparently fewer cells than the PBS control-treated biofilms in images ( Figure 3 ). The reason for the apparent discrepancy may be due to the proteinase K detaching cells from the biofilm off of the glass disks rather than altering the biofilm matrix. For the growing biofilm, it may also be that if a treatment targets one component of the extracellular matrix, the bacteria are able to compensate by overproducing other matrix components. For the quantification of biofilms grown with the treatments, we found that 6/10 strains showed a reduction in biofilm formation with proteinase K treatment. This result demonstrates that protein does play a role for the developing biofilm for most, though not all, strains. The effect of proteinase K treatment was also reflected in the crystal violet-stained images of the strains ( Figure 5 ). However, the crystal violet-stained images revealed that even though sodium metaperiodate treatment did not alter quantitative biofilm staining for most strains, that treatment with that carbohydrate-cleaving agent caused rearrangement of the cells on the glass disks. Although some of the differences between the images and the quantitative data may be due to the different methodology used in the studies, taken together, the results indicate that to assess the EAEC biofilm, both quantitative and microscopic analyses are necessary.

We also asked if treatment with proteinase K, DNase, and sodium metaperiodate could affect a fully-formed EAEC biofilm. However, with those single treatments, we did not generally measure a reduction in quantitative staining, despite the abundant presence of those components as detected by fluorescent staining ( Figure 6 ). These results strongly suggest that the protein, eDNA, and carbohydrates are protected within the biofilm. The only exceptions were that treatment with proteinase K reduced quantitative biofilm staining for two of EAEC strains tested (K5V4 AAF3 and K261 AAF4), while DNase treatment increased biofilm staining for two strains (P73V1 and K5V4). For P73V1 (AAF1), sodium metaperiodate treatment also increased quantitative biofilm staining ( Figure 6 ).

We expected to find that mature, fully-formed biofilms would be more resistant than a developing biofilm to the treatments tested. However, we found that some biofilms showed greater resistance while growing (P73V1 AAF1, K261 AAF4), and others were more resistant as a mature biofilm (E14V1D AAF2, D5613 AAF5). These findings indicate that the treatments have different effects depending on the state of the biofilm. It is possible that the permeability of the biofilm matrix may differ for a growing biofilm than for a mature biofilm. For example, D5613 (AAF5), which has high biofilm staining, was not impacted by any post-treatment but with during-growth treatment, proteinase K reduced biofilm staining ( Figure 4 ). Finally, we found that although DNase treatment increased biofilm staining for two of five strains (P73V1 AAF1 and K5V4 AAF3) with fully mature biofilms ( Figure 6 ), the biofilm from those same strains did not show a change when DNase was added to the developing biofilm ( Figure 4 ). In the image of the DNase-treated mature biofilms from P73V1 (AAF1) and K5V4 (AAF3), it is possible that the change in the biofilm structure after DNase treatment led to increased retention of crystal violet stain, since the quantitative assays measure total bound crystal violet ( Figure 7 ).

Assessing the treatments as a whole (summarized in Figure 8 and Table 2 ), we found that proteinase K was the most likely treatment to cause a statistically significant change in biofilm staining. In contrast, DNase added during biofilm growth and on a mature biofilm on average did not alter staining in a statistically significant way.

Sodium metaperiodate treatment caused mostly non-significant effects as demonstrated by quantitative data, although the images clearly demonstrated a rearrangement of the EAEC within the biofilm. We initially hypothesized that the differences observed after sodium metaperiodate treatment might be due to an increase of simple sugars freed from cleaved polysaccharides, providing more energy for biofilm growth. However, the arguments against that hypothesis are that 1) Dulbecco’s Modified Eagle Medium (DMEM) media is already a high glucose medium with 4.5 g/L of d-glucose, 2) the post-growth treatment was only for 1 hour, a time period that would only allow for a few doublings, and, 3) sodium metaperiodate treatment did not alter the CFU/mL for planktonic cells, as tested under identical conditions ( Supplementary Figure 2 ). Instead, we posit that the sodium metaperiodate catalyzes changes to the polysaccharides which leads to increased spacing around each cell without reducing the total number of cells, and allowing for increased uptake of crystal violet. These observations support the hypothesis that the sodium metaperiodate interacts with the biofilm by breaking down polysaccharide.

Overall, we found that crystal violet was a better stain than the fluorescent dyes for capturing visible differences in structure among the different biofilms, perhaps due to the relatively two-dimensional nature of the biofilms in these in vitro assays. The crystal violet-stained biofilms exhibited a notable lacy architecture. We are not sure whether the lacy pattern reflects the natural growth pattern of the biofilm or was partly influenced by the drying process that occurs prior to staining.

Our biofilm treatment results are similar to those published by Lim et al. for an O157:H7 enterohemorrhagic E. coli strain, for which DNase had minimal impact and proteinase K caused a significant reduction of the biofilm (Lim et al., 2019). In contrast, other non-EAEC E. coli biofilms are impacted by DNase and proteinase K, including visual changes to the biofilm (Tetz et al., 2009; Tetz and Tetz, 2010). DNase is also a strong inhibitor of biofilm formation for the unrelated bacteria Pseudomonas aeruginosa, Streptococcus pneumoniae, and Staphylococcus aureus (Whitchurch et al., 2002; Hall-Stoodley et al., 2008; Deng et al., 2022). Proteinase K causes a significant reduction for Staphylococcus aureus and Helicobacter pylori biofilms, but unlike for EAEC, sodium metaperiodate caused no visual alteration to the biofilm (Beltrame et al., 2015; Shukla and Rao, 2017; Windham et al., 2018).

Our findings indicate that the EAEC biofilm is complex and contains protein, DNA, and carbohydrate, and further, suggest that a single treatment is unlikely to eliminate an EAEC biofilm. In addition, our results indicate that different EAEC may require different treatments for successful biofilm elimination, though we acknowledge that these findings will need to be assessed under in vivo conditions. Finally, in this study we demonstrate that AAF type does not dictate biofilm staining level, and that the finding of low biofilm staining does not predict higher susceptibility to biofilm treatments.

The original contributions presented in the study are included in the article/ Supplementary Material . Further inquiries can be directed to the corresponding author.

VV: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Visualization, Writing – original draft. AM: Conceptualization, Data curation, Funding acquisition, Project administration, Supervision, Writing – review & editing.