Using Lactobacilli to Fight Escherichia coli and Staphylococcus aureus Biofilms on Urinary Tract Devices

The low efficacy of conventional treatments and the interest in finding natural-based approaches to counteract biofilm development on urinary tract devices have promoted the research on probiotics. This work evaluated the ability of two probiotic strains, Lactobacillus plantarum and Lactobacillus rhamnosus, in displacing pre-formed biofilms of Escherichia coli and Staphylococcus aureus from medical-grade silicone. Single-species biofilms of 24 h were placed in contact with each probiotic suspension for 6 h and 24 h, and the reductions in biofilm cell culturability and total biomass were monitored by counting colony-forming units and crystal violet assay, respectively. Both probiotics significantly reduced the culturability of E. coli and S. aureus biofilms, mainly after 24 h of exposure, with reduction percentages of 70% and 77% for L. plantarum and 76% and 63% for L. rhamnosus, respectively. Additionally, the amount of E. coli biofilm determined by CV staining was maintained approximately constant after 6 h of probiotic contact and significantly reduced up to 67% after 24 h. For S. aureus, only L. rhamnosus caused a significant effect on biofilm amount after 6 h of treatment. Hence, this study demonstrated the potential of lactobacilli to control the development of pre-established uropathogenic biofilms.


Introduction
Urinary tract infections (UTIs) are the most common type of healthcare-associated infections reported by the Centers for Disease Control [1][2][3], with an estimated annual worldwide incidence of 250 million cases [4]. The high incidence of these infections results in considerable treatment costs, increased length of hospital stays, and high mortality rates, posing a huge financial burden on healthcare systems [5][6][7][8]. Device-associated UTIs, caused by the insertion of urological devices (UDs), such as urinary catheters or ureteral stents, contribute to about 97% of UTIs [3,5,9]. Despite the efforts to maintain sterility, the contamination of UDs is almost inevitable since they work as a bridge connecting the nonsterile external environment and the patient's body [10,11]. The most common microorganisms contributing to device-associated UTIs are Escherichia coli, Klebsiella pneumoniae, Pseudomonas aeruginosa, Staphylococcus aureus, Candida spp., Enterococcus faecalis, and Proteus mirabilis [12][13][14][15].
Device-associated UTIs mostly originate from the formation of microbial pathogenic biofilms on the device's surface [16]. Biofilms are present in about 80% of human microbial infections [12], and once established, they are extremely hard to eliminate [17,18]. Biofilms are defined as a consortium of microorganisms surrounded by a self-synthesized matrix of extracellular polymeric substances [19,20], which protects the embedded bacteria against host defenses and antimicrobial agents [12]. The conditions present in the urinary tract are particularly favorable to microbial adhesion and biofilm development due to the diversity of shear forces prevailing throughout the urinary tract [5,21], the presence of The culturability of probiotics in the sessile state was simultaneously evaluated ( Figure 2). It can be observed that the E. coli culturability reduction (Figure 1a) was accompanied by the presence of viable probiotic cells in the biofilms (Figure 2a), except for L. rhamnosus, which lost its biofilm culturability after 24 h of interaction with E. coli. On the contrary, the reduction in S. aureus biofilm culturability (Figure 1b) was not followed by the presence of viable L. plantarum cells in biofilms (no colonies were detected at 6 h and 24 h; Figure 2a). L. rhamnosus was present in sessile conditions, but its The data present the mean and standard deviation (SD) of at least three independent experiments. Statistically significant differences between the control and treatment were considered for p-values < 0.05 (*** indicates p < 0.001).
Both Lactobacillus strains were able to reduce pre-formed biofilms. As regards E. coli biofilms (Figure 1a), the number of sessile culturable cells was significantly reduced when exposed to probiotics in comparison to the negative control sample (p < 0.001). The highest reductions were obtained after 24 h of exposure, with reductions of 70% for L. plantarum and 76% for L. rhamnosus (p < 0.001). Additionally, reductions of 69% and 61% were obtained after 6 h of biofilm treatment with L. plantarum and L. rhamnosus, respectively (p < 0.001). Regarding S. aureus biofilms (Figure 1b), the same tendency was observed. There was a significant reduction in S. aureus culturability at both experimental times (p < 0.001), being the most significant decrease after 24 h of contact (77% for L. plantarum and 63% for L. rhamnosus); after 6 h, reductions of 57% and 59% were obtained for L. plantarum and L. rhamnosus, respectively. For both pathogens, the antimicrobial activity of probiotics increased from 6 h to 24 h of exposure, demonstrating a progressive effect over time.
The culturability of probiotics in the sessile state was simultaneously evaluated ( Figure 2). It can be observed that the E. coli culturability reduction (Figure 1a) was accompanied by the presence of viable probiotic cells in the biofilms (Figure 2a), except for L. rhamnosus, which lost its biofilm culturability after 24 h of interaction with E. coli. On the contrary, the reduction in S. aureus biofilm culturability (Figure 1b) was not followed by the presence of viable L. plantarum cells in biofilms (no colonies were detected at 6 h and 24 h; Figure 2a). L. rhamnosus was present in sessile conditions, but its culturability decreased by 85% between 6 h and 24 h of interaction with S. aureus (Figure 2b). Additionally, looking at Figures 1 and 2, there is about 1-3 log CFU·cm −2 difference between the populations of E. coli and S. aureus and probiotics within the biofilms. culturability decreased by 85% between 6 h and 24 h of interaction with S. aureus ( Figure  2b). Additionally, looking at Figures 1 and 2, there is about 1-3 log CFU·cm −2 difference between the populations of E. coli and S. aureus and probiotics within the biofilms. Complementary assays were performed to analyze the presence of probiotic cells in the planktonic fraction in order to explain the decrease in pathogens culturability, even in the absence of Lactobacillus cells in the biofilm (data not shown). For example, L. rhamnosus cell densities of 2.5 × 10 5 ± 4.3 × 10 3 and 1.6 × 10 4 ± 4.2 × 10 3 CFU·mL −1 were detected in the suspension after 24 h of contact with E. coli and S. aureus biofilms, respectively, suggesting that probiotics may act on the pathogenic sessile cells through the release of harmful substances to the extracellular medium. The antimicrobial activity of lactobacilli was also evaluated against S. aureus by the disk diffusion method (Supplementary Material). The cell suspensions and cell-free supernatants of both viable and lysed probiotics evidenced clear inhibition zones on the swabbed S. aureus on Luria-Bertani agar plates when compared with the negative control (Table S1, Supplementary Material), indicating that these probiotic strains inhibited the growth of pathogens through the production and secretion of antimicrobial substances into their surroundings. Figure 3 presents the results of biofilm quantification using CV staining. Concerning E. coli biofilms (Figure 3a), both probiotics maintained the biofilm amount at 6 h of contact in comparison with control (p > 0.05). However, after 24 h of exposure, L. plantarum and L. rhamnosus significantly reduced the biofilm mass by 51% (p = 0.04) and 67% (p = 0.011), respectively, demonstrating their capacity to disrupt the pre-formed biofilms. In the case of S. aureus biofilms, the opposite behavior was observed. After 6 h of contact, both probiotics reduced biofilm mass, although only L. rhamnosus exhibited statistical difference when compared to control (reduction of 42% for L. plantarum (p = 0.053) and 20% for L. rhamnosus (p = 0.049)); at 24 h of contact, the biofilm amount remained nearly constant. Therefore, there was a poor correlation between the CV staining method and cell culturability. Complementary assays were performed to analyze the presence of probiotic cells in the planktonic fraction in order to explain the decrease in pathogens culturability, even in the absence of Lactobacillus cells in the biofilm (data not shown). For example, L. rhamnosus cell densities of 2.5 × 10 5 ± 4.3 × 10 3 and 1.6 × 10 4 ± 4.2 × 10 3 CFU·mL −1 were detected in the suspension after 24 h of contact with E. coli and S. aureus biofilms, respectively, suggesting that probiotics may act on the pathogenic sessile cells through the release of harmful substances to the extracellular medium. The antimicrobial activity of lactobacilli was also evaluated against S. aureus by the disk diffusion method (Supplementary Material). The cell suspensions and cell-free supernatants of both viable and lysed probiotics evidenced clear inhibition zones on the swabbed S. aureus on Luria-Bertani agar plates when compared with the negative control (Table S1, Supplementary Material), indicating that these probiotic strains inhibited the growth of pathogens through the production and secretion of antimicrobial substances into their surroundings. Figure 3 presents the results of biofilm quantification using CV staining. Concerning E. coli biofilms (Figure 3a), both probiotics maintained the biofilm amount at 6 h of contact in comparison with control (p > 0.05). However, after 24 h of exposure, L. plantarum and L. rhamnosus significantly reduced the biofilm mass by 51% (p = 0.04) and 67% (p = 0.011), respectively, demonstrating their capacity to disrupt the pre-formed biofilms. In the case of S. aureus biofilms, the opposite behavior was observed. After 6 h of contact, both probiotics reduced biofilm mass, although only L. rhamnosus exhibited statistical difference when compared to control (reduction of 42% for L. plantarum (p = 0.053) and 20% for L. rhamnosus (p = 0.049)); at 24 h of contact, the biofilm amount remained nearly constant. Therefore, there was a poor correlation between the CV staining method and cell culturability.  The data present the mean and SD of at least three independent experiments. Statistically significant differences between the control and treatment were considered for p-values < 0.05 (* and ** indicate p < 0.05 and p < 0.01, respectively).

Discussion
Device-associated urinary tract infections are a critical problem caused by the high propensity of medical devices to microbial colonization. Previous studies have proposed the use of probiotics as a useful strategy to control pathogenic biofilms, demonstrating that probiotics cells and metabolites can displace adhering uropathogens from urinary devices materials and block bacterial adhesion to uroepithelial cells [47,48]. Probiotics can exert their antibiofilm activity by adopting different strategies: displacement, exclusion, and competition [47]. Recently, our research group evaluated the ability of L. plantarum biofilms to prevent E. coli adhesion and biofilm formation on silicone rubber, following an exclusion strategy [42]. In the present work, the potential of two probiotic strains (L. plantarum and L. rhamnosus) to disperse pre-formed biofilms of E. coli and S. aureus under physiologically relevant conditions was assessed by adopting a displacement strategy [48].
Regarding the antibiofilm activity of probiotics, they significantly inhibited the proliferation of E. coli and S. aureus biofilms by reducing their cell culturability and biomass amount after 24 h of treatment. Furthermore, both Lactobacillus strains caused similar reductions in the culturability of both model pathogens. The activity of probiotics can be related to interfacial cell-cell interactions and to the production and release of antagonizing metabolites that are able to destabilize the biofilm organization, as demonstrated in previous studies [54][55][56]. Although the incorporation of probiotics within the biofilms suggests that those adhered Lactobacillus cells may contribute to pathogen inhibition, the absence of biofilm culturable cells of probiotics at some time points (L. rhamnosus after 24 h of contact with E. coli and L. plantarum after 6 h and 24 h of contact with S. aureus) suggests that, beyond the antibiofilm action by integration and contact with the biofilm, lactobacilli may act through the release of antimicrobial substances from the planktonic cells. Moreover, the differences found between the culturability of E. coli and S. aureus and the culturability of probiotics mean that there was not a direct exchange of pathogenic cells by probiotic cells during treatment. This suggests that pathogen inactivation occurred essentially by the secretion of antimicrobial substances into the surrounding environment. The data present the mean and SD of at least three independent experiments. Statistically significant differences between the control and treatment were considered for p-values < 0.05 (* and ** indicate p < 0.05 and p < 0.01, respectively).

Discussion
Device-associated urinary tract infections are a critical problem caused by the high propensity of medical devices to microbial colonization. Previous studies have proposed the use of probiotics as a useful strategy to control pathogenic biofilms, demonstrating that probiotics cells and metabolites can displace adhering uropathogens from urinary devices materials and block bacterial adhesion to uroepithelial cells [47,48]. Probiotics can exert their antibiofilm activity by adopting different strategies: displacement, exclusion, and competition [47]. Recently, our research group evaluated the ability of L. plantarum biofilms to prevent E. coli adhesion and biofilm formation on silicone rubber, following an exclusion strategy [42]. In the present work, the potential of two probiotic strains (L. plantarum and L. rhamnosus) to disperse pre-formed biofilms of E. coli and S. aureus under physiologically relevant conditions was assessed by adopting a displacement strategy [48].
Regarding the antibiofilm activity of probiotics, they significantly inhibited the proliferation of E. coli and S. aureus biofilms by reducing their cell culturability and biomass amount after 24 h of treatment. Furthermore, both Lactobacillus strains caused similar reductions in the culturability of both model pathogens. The activity of probiotics can be related to interfacial cell-cell interactions and to the production and release of antagonizing metabolites that are able to destabilize the biofilm organization, as demonstrated in previous studies [54][55][56]. Although the incorporation of probiotics within the biofilms suggests that those adhered Lactobacillus cells may contribute to pathogen inhibition, the absence of biofilm culturable cells of probiotics at some time points (L. rhamnosus after 24 h of contact with E. coli and L. plantarum after 6 h and 24 h of contact with S. aureus) suggests that, beyond the antibiofilm action by integration and contact with the biofilm, lactobacilli may act through the release of antimicrobial substances from the planktonic cells. Moreover, the differences found between the culturability of E. coli and S. aureus and the culturability of probiotics mean that there was not a direct exchange of pathogenic cells by probiotic cells during treatment. This suggests that pathogen inactivation occurred essentially by the secretion of antimicrobial substances into the surrounding environment. The presence of culturable cells of probiotics in the planktonic fraction and the existence of inhibition zones caused by probiotics cell-free supernatants reinforce this hypothesis.
The poor correlation between the crystal violet (CV) method and cell culturability results can be attributed to the ability of CV to non-specifically bind to some components Antibiotics 2021, 10, 1525 6 of 12 of the biofilm matrix (such as DNA, exopolysaccharides (EPS), and proteinaceous material) and the peptidoglycan wall of both live and dead cells [57,58].
The antimicrobial substances released by probiotics may penetrate the extracellular matrix of biofilms, interfering with their integrity and cell culturability. Several authors reported the ability of Lactobacillus strains to disrupt mature biofilms. Jaffar et al. [58] described that L. plantarum significantly disrupted the pre-formed biofilm of Aggregatibacter actinomycetemcomitans by approximately 61% on polystyrene after 24 h. Conversely, Song et al. [59] reported that L. rhamnosus was capable of disrupting biofilms of Candida albicans by 99.9% due to the production of lactic acid and antimicrobial peptides, which presumably disrupted the cytoplasmatic membrane and inactivated cytoplasm molecules. This mechanism was supported by Fayol-Messaoudi et al. [60], who suggested that the bactericidal activity of Lactobacillus strains may be due to the synergistic action of lactic acid and secreted bacteriocins. Some studies have shown that bacteriocins produced by L. plantarum effectively suppressed the growth and biofilm formation of several microorganisms [39,61,62], including S. aureus [63]. McMillan et al. [64] found that L. rhamnosus was incorporated into uropathogenic biofilms, including E. coli, and caused significant E. coli killing supposedly due to the production of bacteriocins or biosurfactantlike substances. In agreement, Cadieux et al. [65] described that L. rhamnosus strongly inhibited the growth of uropathogenic E. coli by producing bacteriocins, hydrogen peroxide, and lactic acid. Similarly, Otero et al. [45] evidenced an inhibitory effect of Lactobacillus strains in S. aureus growth after 6 h of co-culture, possibly due to the combined effect of hydrogen peroxide and lactic acid. In addition, some authors suggested that L. rhamnosus can produce biosurfactants with antibiofilm activity against E. coli and S. aureus [43,64]. Previous studies have demonstrated the ability of lactic acid bacteria biosurfactants to inhibit biofilm development and induce its dispersion from surgical implant materials [66][67][68]. Furthermore, an assorted number of studies attributed the disruption of the architecture of pathogenic biofilms by L. rhamnosus to the secretion of molecules that downregulate the genes involved in biofilm development, DNA replication, translation, glycolysis, and gluconeogenesis [55,56]. Song et al. [69] reported that L. rhamnosus significantly disrupted the architecture of E. coli biofilms by 82% by decreasing the transcriptional activity of transcriptional activators (luxS, lsrK, and lsrR) of the quorum-sensing in E. coli. In addition, Ahn et al. [70] and Kim et al. [71] described that lipoteichoic acid produced by L. plantarum disrupted pre-formed biofilms of single and multispecies pathogens by interfering with EPS production.
In this work, the lack of glucose in the growth medium hindered probiotics fermentation since high glucose concentrations enhance medium acidification [72,73]. This was confirmed by comparing the initial and final pH values of artificial urine medium (AUM), where no differences were found. Additionally, Todorov et al. [74] studied the effect of the pH on the production of bacteriocins of L. plantarum and found that the inhibitory activity of bacteriocins was detected at pH values between 5.0 and 6.5. Moreover, the presence of yeast extract, potassium dihydrogen phosphate, and dipotassium hydrogen phosphate in culture medium as nitrogen and phosphorus sources, respectively, enhanced the production of bacteriocins by L. plantarum [74]. In the present work, since the initial pH of AUM is 6.5, and yeast extract, potassium dihydrogen phosphate, and dipotassium hydrogen phosphate are present in AUM [75], this may have contributed to the production of bacteriocins by this probiotic strain. Regarding biosurfactants, despite the bactericidal potential of biosurfactants, their action is more frequently associated with their capacity to affect microbial adhesion by interfering with surface tension and hydrophobicity [76,77], so biosurfactants are unlikely to be effective on mature biofilms. Thus, the disruptive activity of probiotics is most likely explained by the production and secretion of antimicrobial exometabolites, such as hydrogen peroxide, bacteriocins, and biosurfactants, rather than by a possible pH change of the culture media. Overall, these results suggest that the ability of probiotics to displace pre-formed uropathogenic biofilms can be attributed to the production of exometabolites that inactivate sessile cells and destabilize the biofilm structure.

Preparation of Silicone Surfaces
Biofilm formation experiments were performed on silicone coupons (1 × 1 cm; Neves & Neves, Lda, Maia, Portugal) with the intention of mimicking the most common material of urinary catheters [78]. The surfaces were prepared as previously described by Carvalho et al. [42]. Briefly, the coupons were washed with 70% (v/v) ethanol (VWR, Radnor, PA, USA), air-dried, and then sterilized through ultra-violet (UV) radiation for 30 min. The sterile coupons were fixed to the bottom of 12-well polystyrene plates (VWR, USA) using double-sided adhesive tape, which was sterilized beforehand for 30 min through UV radiation.

Bacterial Strains and Culture Conditions
The probiotic strains used in this study were Lactobacillus plantarum and Lactobacillus rhamnosus (1 × 10 11 CFU·g −1 ; Biomodics ApS, Rødovre, Denmark). These bacteria were preserved at −80 • C in De Man, Rogosa and Sharpe (MRS) broth (Merck KGaA, Madrid, Spain) with 30% (v/v) glycerol, streaked on MRS agar (Scharlab, S.L., Barcelona, Spain) plates, and incubated for 48 h at 37 • C. Lactobacillus inocula were prepared by collecting bacterial colonies from the MRS agar plate into 250 mL of MRS broth and incubating overnight at 37 • C in an orbital shaker at 120 rpm (Agitorb 200, Aralab, Rio de Mouro, Portugal). MRS is the culture media routinely used for lactobacilli growth [79].
Escherichia coli CECT 434 and Staphylococcus aureus CECT 976 were chosen as model microorganisms of biofilm-based urinary tract infections. These bacterial strains were preserved at −80 • C in Luria-Bertani (LB) broth (Thermo Fisher Scientific, Waltham, MA, USA) containing 30% (v/v) glycerol, streaked on LB agar plates, and incubated for 24 h at 37 • C. The starting cultures were prepared by collecting single colonies from LB agar plates to 250 mL of artificial urine medium (AUM) [75] and incubating overnight at 37 • C and 120 rpm. AUM was used to simulate the nutrient composition of human urine [75].

Influence of Probiotics on Pre-Formed Biofilms
The antibiofilm assays followed a displacement strategy [47], which consisted of the formation of E. coli and S. aureus biofilms, after which the biofilms were inoculated with the Lactobacillus strains separately in order to evaluate their ability to disperse pre-formed biofilms. The bacterial cultures grown overnight were harvested by centrifugation at 3202× g for 10 min at 25 • C (Eppendorf Centrifuge 5810R, Hamburg, Germany), and the final cell concentration was adjusted in fresh AUM to an optical density at 610 nm of 0.15 for E. coli, 0.20 for S. aureus, and 0.70 for both Lactobacillus strains, equivalent to approximately 10 8 CFU·mL −1 , the recommended bacterial density to be used in urological experiments [5]. This was confirmed by colony-forming unit (CFU) counts. The uropathogenic biofilms were formed on silicone coupons placed inside 12-well plates where each well was filled with 3 mL of the respective pathogenic suspension. The plates were incubated for 24 h at 37 • C under shaking conditions in order to generate shear stresses similar to those found inside urinary catheters [42,80]. Afterwards, cell suspensions were removed, and each well was loaded with 3 mL of the respective probiotic suspension for periods of contact of 6 h and 24 h under the same growth conditions. A negative control was prepared by adding sterile AUM to pathogenic biofilms. At the end of each experimental period, biofilms were analyzed as previously described [42]. Briefly, cell suspensions were removed from the wells, the nonadherent cells were washed with a sodium chloride solution (8.5 g·L −1 NaCl), and the biofilm amount and culturability were determined by crystal violet (CV) staining and CFU counts, respectively. All experiments included at least three independent biological replicates with two technical replicates each. The number of biofilm culturable cells per cm 2 of silicone was determined as indicated by Carvalho et al. [42]. Briefly, the coupons were transferred to Falcon tubes with 2 mL of saline solution and the biofilm cells were detached from the coupons by vortexing (ZX4, Velp Scientifica, Usmate Velate, Italy) for 2 min at full power. Then, serial dilutions of the obtained biofilm cell suspensions were performed in saline solution, plated on LB agar (selective media for E. coli and S. aureus) and MRS agar (selective media for L. plantarum and L. rhamnosus), and incubated at 37 • C for 24 h and 48 h, respectively.
The percentages of CFU reduction in pathogens were estimated as follows: where CFU control corresponds to the number of culturable cells of pathogens in the negative control samples (biofilms not exposed to probiotic cell suspensions), and CFU biofilm is the number of culturable cells in biofilms treated with probiotics.

Biofilm Amount Determination
The total mass of biofilms was quantified using the CV staining method [42]. Briefly, after washing the non-adherent cells, silicone coupons were transferred to 24-well polystyrene plates (Thermo Fisher Scientific, USA), and biofilms were fixed with 1 mL of 100% ethanol (VWR, USA) for 15 min. Then, the wells were air-dried, and biofilms were stained with 1 mL of 1% (v/v) CV (Merck, Germany) solution for 5 min. The dye bounded to the biofilm was solubilized by adding 1 mL of 33% (v/v) acetic acid (VWR, USA) solution. Finally, 200 µL of each well was transferred to a 96-well polystyrene plate (VWR, USA), and the biofilm mass was determined through absorbance measurement at 570 nm (Abs 570 nm ) in a microtiter plate reader (SpectroStar Nano, Biogen Cientifica S. L., Madrid, Spain). When the absorbance values exceeded 1, samples were diluted in 33% (v/v) acetic acid. The biofilm amount was expressed as Abs 570 nm values.

Statistical Analysis
Statistical analysis was performed using the IBM SPSS Statistics version 26 for Windows (IBM SPSS, Inc., Chicago, IL, USA). Descriptive statistics were used to calculate the mean and standard deviation (SD) for the number of culturable cells and biofilm mass. The homogeneity of variances and normality of data were verified for all response variables tested using the Kolmogorov-Smirnov and Shapiro-Wilk tests. Since the response variables were not normally distributed, a nonparametric analysis using the Kruskal-Wallis test was performed to assess whether there were statistically significant differences among groups, and the differences between those groups were determined by the Mann-Whitney test. Statistically significant differences were considered for p-values < 0.05, corresponding to a confidence level of 95% (*, **, and *** indicate p < 0.05, p < 0.01 and p < 0.001, respectively). All reported data are presented as mean ± SD from at least three experiments with duplicates.

Conclusions
L. plantarum and L. rhamnosus were able to displace pre-established biofilms of E. coli and S. aureus at similar levels. The antibiofilm activity of these probiotic strains may be primarily linked to the release of self-produced substances, which reduced the number of culturable pathogens in biofilms. Additionally, the integration of probiotic cells into the biofilm may have contributed to the destabilization of the biofilm organization.
This proof-of-principle study supports the potential of Lactobacillus strains to be used as biocontrol agents against pathogenic biofilms developed on urinary tract devices. It will pave the way for more experiments on the topic in an effort to elucidate the mechanisms underlying the lactobacilli activity.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author. The data are not publicly available yet as some data sets are being used for additional publications.