Enhancement of rhizocompetence in pathogenic bacteria removal of a constructed wetland system

For several decades, constructed wetlands (CWs) have been used as an ecological and green technology to treat wastewaters (Wu et al. 2015). CWs offer a low-energy, land-intensive, and less operational-requirements alternative to conventional treatment systems, especially for small communities (Ghrabi et al. 2011; Shen et al. 2015; Tee et al. 2016). By using natural functions of wetland vegetation, soils and their microbial populations, these engineered systems are designed to treat contaminants in surface water, groundwater or waste streams (Vymazal 2014). CWs have a great potential for the treatment of wastewater of different origin (Zaytsev et al. 2011; Wang et al. 2015) such as domestic, agricultural and industrial wastewater (Liu et al. 2015). CWs have been successfully used to remove a wide variety of pollutants such as organic compounds, suspended solids, pathogens and metals (Zhang et al. 2014; Sánchez 2017). During wastewater treatment in CWs, pollutants are removed through an integrated combination of biological, physical and chemical interactions between the plants, the substrate and the inherent microbial community (Farrar et al. 2014). According to the wetland hydrology, CWs are typically classified into two types: free water surface (FWS) CWs and subsurface flow (SSF) CWs. FWS systems are similar to natural wetlands, with a shallow flow of wastewater over a saturated substrate. On SSF systems, wastewater flows through the substrate, which supports the growth of plants. In addition, based on the flow direction, SSF CWs could be further divided into vertical flow and horizontal flow CWs. A combination of various wetland systems, or a hybrid CW, was also introduced for the treatment of wastewater (Wu et al. 2015).

Microorganisms play a vital role in degradation of different pollutants in CWs. It has been recognized that the removal of most of them in CWs is due primarily to microbial activity (Meng et al. 2014). Each pollutant can be associated with a specific microbial functional group; therefore the employment of design and operational methodologies that enhance the activity of that group will better optimize CWs efficiency (Eastman et al. 2009; Faulwetter et al. 2009). Thus, it has long been thought that many naturally occurring rhizosphere bacteria and fungi may offer a viable substitute for the use of chemicals and are antagonistic towards crop pathogens. For instance, plant-growth-promoting rhizobacteria (PGPR) have been shown to be beneficial to plant growth and health by exhibiting an active role in nitrogen fixation, the production of phytohormones and antifungal compounds, and induced systemic resistance. (Sindhu et al. 2009; Dilfuza et al. 2015).

Based on the importance of rhizosphere competence or root colonization in beneficial plant–microbe interactions (Ben Saad et al. 2016), the main goal of the present study was to enhance the inactivation of pathogenic bacteria in a horizontal subsurface flow CW system planted with Phragmites australis using antagonistic bacteria. This work aimed to demonstrate the beneficial application of biotechnology to confer higher rhizosphere competence in the removal of the model pathogenic bacterium Salmonella typhi ATCC 560 by means of environment-friendly biological approaches.

Samplings were conducted at CWs of the Technological Demonstration Center (TDC) located at the Agronomic Institute of Tunisia. The demonstration pilot systems has been developed and erected in the framework of the project ‘Sustainable Concepts towards a Zero Outflow Municipality (Zer0-M) (Regelsberger et al. 2007). The TDC treated the sewage discharged from the student house. The black water was introduced to a storage tank and was treated in a septic tank followed by horizontal subsurface flow (HSSF) and vertical subsurface flow (VSSF) CWs. The treated water was stored in a tank for green area irrigation. Sampling was conducted at different compartments in the CWs. Water samples were taken from the septic tank, the input of the VSSF and the input and the output of the HSSF. P. australis samples (roots and leaves) were collected from each CW. Roots were taken from both HSSF and VSSF at the entrance, middle, and exit at a depth of approximately 30 cm under the gravel surface. Other P. australis roots and soil samples were collected from the riverside located at Technopark Borj Cedria (northeast Tunisia). To isolate bacteria from the rhizosphere, the roots were initially separated from the rhizomes, and then small pieces of roots were immersed in sterile saline solution (0.85 g/L NaCl) and vortexed for 15 min in order to release attached bacteria. The same protocol was carried out to isolate bacteria from the sets of reeds. Concerning the wastewater samples, these samples underwent decimal dilutions in sterile saline solution and were spread over selective medium King's B agar (KB) (Ben Saad et al. 2016).

The identification of selected bacteria was based on the phenotypical aspect of colonies, the microscopic examination, and the use of standard microbiological and biochemical tests. Siderophores were detected by the method of Jalal & van der Helm (1990) using a spectrophotometric assay where a peak at 495 nm on the addition of 2% aqueous solution of FeCl₃ to 1 mL of supernatant indicated the presence of siderophores.

An antagonism test was performed among the isolated bacterial strains to avoid negative interaction between them after their bioinoculation. On the other hand, another antagonism test was performed between bacterial strains and the model pathogenic bacterium Salmonella thyphi ATCC 560. The Petri dish surface was seeded with an indicative strain and then blank discs with the putative antagonist strain were deposited on the culture medium. Blank discs were prepared by drenching with a filtered supernatant of a liquid culture of the putative antagonist strain, collected after centrifugation at 4,000 rpm for 15 min. The diffusion of the antimicrobial agents was enhanced by incubation at 37 °C for 24 h. Antagonist activity was revealed by the appearance of an inhibition zone around the discs (Ben Saad et al. 2016).

The different types of motility (swimming, swarming and twitching) were determined by the method of Reimmann et al. (2002). The biofilm production of bacterial isolates was detected by two methods. The first described by Freeman et al. (1989), which consisted of plating the test strains on solid medium containing brain heart infusion broth (37 g/L), sucrose (50 g/L), agar (10 g/L) and Congo Red indicator (8 g/L) (Sujatha & Ammani 2013). After incubation at 30 °C for 24 h, biofilm production was indicated by the appearance of black colonies. After this qualitative study, a quantitative study described by O'Toole & Kolter (1998) was carried out, using the dye crystal violet (CV), and estimating biofilm production spectrophotometrically at optical density of 600 nm.

Bacterial DNA from the strain which showed the highest antagonist activity against S. typhi (PFH₁) was extracted and purified using the v-DNA reagent (GenIUL) according to the manufacturer's instructions. The concentration of the extracted DNA was measured using a spectrophotometer at 260 nm. DNA purity was estimated from the A260/A280 ratio. The complete 16S rRNA gene was amplified using universal bacterial primers 27F (5′-TAC GGY TAC CTT GTT AYG ACT T-3′) and 1492Rmod (5′-AGR GTT TGA TCM TGG CTC AG-3′). Each polymerase chain reaction (PCR) reaction with a final volume of 25 μL contained: 2 μL of template DNA, 0.5 μL of each deoxynucleoside triphosphate at a concentration of 10 μM, 0.75 μL of MgCl₂ 1.5 mM, 0.5 μL of each primer at a concentration of 10 μM, 0.125 μL of Taq DNA polymerase (Invitrogen), 2.5 μL of PCR buffer supplied by the manufacturer (Invitrogen, Paisley, UK) and Milli-Q water up to the final volume. Reactions were carried out in a Bio-Rad thermocycler using the following program: initial denaturation at 94 °C for 5 min, followed by 30 cycles of 1 min at 94 °C, 1 min at 55 °C and 2 min at 72 °C, and a final extension step of 10 min at 72 °C. PCR products were verified and quantified by agarose gel electrophoresis with standard low DNA mass ladder (Invitrogen). Purification and one-shot Sanger sequencing of 16S rRNA gene products were performed by Genoscreen (Lille, France) with primers 27F and 1492Rmod on ABI3730XL to obtain the complete sequence. The gene sequence was deposited in GenBank under accession number MG770377 and it was subject to a BLAST search (Altschul et al. 1997) to obtain an indication of the phylogenetic identification.

Two identical HSSF CWs were used to perform the experiment. Both basins were filled with gravel and planted with P. australis. One was used as a control (T) and the other one served for the different bio-assays (F). The size of each CW bed was 0.3 m × 0.44 m × 0.28 m (Ben Saad et al. 2016).

For optimization purposes, the kinetic growth of the selected strain PFH₁ was characterized using a spectrophotometer to estimate absorbance of cell suspensions (OD₆₀₀). Different growth curves were obtained to determine the lag time for the microorganism to adapt to the new conditions. It was cultivated in two different media: autoclaved wastewater and nutrient broth at room temperature.

Afterward, inocula were prepared by suspending bacteria from nutrient agar plates in nutrient broth to obtain populations of about 10⁸ UFC/mL (Bennett & Lynch 1981). The antagonist bacterium was inoculated using a sterile fine pipe directly connected to the rhizosphere. Treatment with contaminated water resulted in the following designations: F, a CW with P. inoculated with PFH₁; T, a CW with P. but no bacterial inoculation. The water in the basin was kept stagnant for 2 days in order to let microbes settle and attach before starting the experiment. After the adaptation period, 10⁶ UFC/mL of the indicator pathogenic bacterium S. typhi was added to both pilot-scale CWs (F and T).

Based on growth kinetic parameters of the bacterium of interest (PFH₁), sequential bio-injections were also performed in the pilot-scale CW F to test the spatio-temporal dynamics and microbial ecological processes of root colonization by an antagonist and to explore the impact of the accumulation effect of sequential bioinoculation of antagonist bacteria to promote inhibition of pathogenic bacteria.

The monitoring of pathogenic bacteria removal after bioinoculation in both types of experiments was determined by culture on selective medium (Salmonella, Shigella agar).

After sampling, isolation and purification stages, 19 bacterial strains were isolated from different ecological niches. The bacterial colonies, cultivated on KB medium, were round, smooth and cream-white. All strains studied were oxydase+, catalase+ and Gram-negative.

The isolated strains were selected and screened for general functional properties of plant-growth-promoting rhizobacteria; namely, siderophore production and antagonist activity against pathogenic bacteria, in addition to bacterial motility (swimming, swarming and twitching motilities) and biofilm production (Table 1).

Bacterial biofilm formation is important for root colonization. Indeed, root-associated bacteria have been extensively studied, and many of these promote the growth of host plants or are used as biocontrol agents (Dekkers et al. 1998; Alizadeh 2011). Plant-growth-promoting bacteria have been reported to discontinuously colonize the root surface, developing as small biofilms along epidermal fissures. Among the isolates, PFH₁ strain could be a possible candidate to be inoculated into the rhizospheric zone to enhance the reduction of pathogenic bacteria. Furthermore, the antagonism test revealed this strain as the most antagonistic bacterium against S. typhi. Certainly, the ability of microbes to produce a wide range of antimicrobial compounds, including lytic agents, antibiotics, bacteriocins, protein exotoxins and other secondary metabolites, is critical to their success in antagonistic activities (Liu et al. 2013).

The 16S rRNA sequence analysis of bacterium PFH₁ revealed that this microorganism had 99% of similarity with Enterobacter cloacae, a Gram-negative proteobacterium belonging to the Enterobacteriaceae family. The PFH₁ strain was classified under Enterobacter sp. and its sequencing data were submitted to GenBank NCBI under the accession number MG770377 (https://www.ncbi.nlm.nih.gov/nucleotide/MG770377.1?report=genbank&log$=nucltop&blast_rank=18&RID=FZRT9BXG01R).

Enterobacter is grouped in a subclade with Klebsiella, with which it is closely related. The E. cloacae species comprise an extremely diverse group of bacteria that have been found in diverse environments, ranging from plants to soil and humans (Liu et al. 2013). Enterobacter species have been reported as important engineering and plant-growth-promoting bacteria (Nie et al. 2002). Some Enterobacter strains may play important roles in plant–microbe interactions and hence in biocontrol mechanisms (Taghavi et al. 2010). Several studies have shown the relevant ability of Enterobacter sp. to remove pollutant from wastewater; for example Lu et al. (2006) have demonstrated a good metal uptake capacity and high resistance to various heavy metals of a local bacterial isolate Enterobacter sp. J1. In this sense, the selected bacterium has been used to control the pathogenic density and has proved its application to improve the water treatment in CWs.

To optimize our experiments and before inoculation into the rhizosphere, the kinetic growth of PFH₁ strain was investigated in nutrient broth and autoclaved wastewater at room temperature in order to determine the specific growth characteristics of PFH_1, namely the lag time (ʎ_t), the maximum specific growth rate (μ_max), and the bacterium generation time (t_G) (Figure 1).

The in situ experiment was done as follows. Phase I started with the sowing of the bacterium of interest into the rhizosphere environment. The main events were activation of the antagonist inoculum and establishment of an antagonist population in the plant rhizosphere. Phase II is the process by which the introduced antagonist and native root-associated microbes establish a population density and persist in the rhizoplane, rhizosphere or inside the root.

The analysis of bacterial inactivation curves showed an increase in bacterial reduction along time in both pilot-scale CWs with a difference in the bacteria inhibition rate. In turn, the injection of selected bacteria into the rhizosphere of P. australis improved the kinetics of S. typhi inactivation by approximately 1 U-Log₁₀ (N/N₀) compared to the control (T). The enhancement of bacteria inactivation in the inoculated CW (F) during a short retention time is probably related to a good colonization ability of the rhizosphere and to the antagonist activity of the selected inoculated strain (Albright & Ode 2011).

Several studies have demonstrated the effectiveness of macrophyte systems in the elimination of pathogenic bacteria (Hill & Sobsey 2001), as well as the reduction of the pathogenic bacterium S. typhi by 2.3 U-Log₁₀ for the treatment of primary sewage in small communities and rural areas using gravel during a retention time of 23 to 52 hours (Hench et al. 2003).

The exploitation of the results of bacterial reduction by the modified kinetic model of Chick–Watson has allowed us to determine different kinetics parameters, such as the coefficient of inactivation (k) and the bacterial reduction rate in contact with autochthonous rhizobacteria with and without bioinoculation (A). The analysis of those parameters showed an increase in k, represented by the slope of the inactivation curve, which was more pronounced in the pilot-scale CW with an antagonistic bacteria (PFH₁) (Table 2). The increase of this coefficient confirmed the effectiveness of inoculated bacteria in CW F to strengthen the rhizospheric effect and increase the reduction of pathogenic bacteria.

Concerning the bacterial reduction rate, this parameter showed a small decrease in CW F compared to the control (T). It revealed the inactivation of target bacteria at the first contact with autochthonous rhizospheric biomass with and without bioinoculation. The stabilization of this parameter indicated directly the need of the inoculated bacteria for a period in which to become acclimatized to the in situ environment. Therefore, the first inactivation effect was governed by autochthonous biomass by various interactions such as antibiosis, biological antagonism, the competition for nutrients and parasitism (Di Francesco et al. 2016).

In summary, by a single inoculation of antagonistic bacteria PFH₁, the pathogenic bacteria removal has been increased by 1 U-Log₁₀ of initial indicator bacteria. This result affirms well the use of bioinoculation for biocontrol.

The enhancement of the rhizobacterium potential in the pilot-scale CW planted with P. australis is strongly related to antagonist bacteria growth parameters, namely, ʎ_t, μ_max and t_G.

To strengthen the rhizocompetence in pathogen removal bacteria, sequential injections of the selected bacteria were performed at time 0, 60, 120 and 180 min. The choice of the injection time was based on the bacterium lag time, which equals 60 min (Figure 3).

By a series of multi-bioinoculations, the enhancement of pathogenic bacteria reduction compared to the inhibition rate in pilot-scale control (T) was of 2 U-Log₁₀ (N/N₀).

The accumulative effect of sequential bio-injection and the keeping of the exponential growth phase of the bacteria (based on the growth curve of PFH₁ strain) allow an increase k to 1.33 min⁻¹, determined after three bio-injections in CW (F) versus a value of k equal to 0.35 min⁻¹ determined in the control CW (T) without bio-injection.

Moreover, the increase of inactivation rate determined for inoculated pilot-scale CW (F) compared to the control one (Table 3) can be observed.

The difference in kinetic parameters (k and A) determined for both CWs T and F is proportional to bacterial growth factors, the maximum specific growth rate, and the bacterium generation time. The enhancement of pathogen inactivation rate was positively correlated with the growth bioinoculum factors (A′ and k′) and the number of injections.

In summary, after the accumulation effect of three sequential bio-injections into a rhizosphere environment of P. australis, the rhizocompetence in bacterial removal was increased and evidenced by k′ (1.33 min⁻¹) and A′ (30.4) relative to the antagonist activity of the bacteria with a reduction in contact time. The rhizosphere is a complicated biological system and too many factors affect bacterial inoculum survival (Strigul & Kravchenko 2006).

Constructed wetlands have proven to be an effective treatment for the inactivation and removal of pathogens in wastewaters. Several mechanisms are involved in pathogen treatment in CWs such as sedimentation, natural die-off, inactivation or death related to temperature, oxidation, predation, biofilm interaction, mechanical filtration, exposure to biocides and UV radiation (Weber & Legge 2008). Numerous studies have indicated that inoculated microorganisms could enhance the removal of various types of pollutants in CWs, in particular nitrogen and phosphorous (Zaytsev et al. 2011; Shao et al. 2014; Pei et al. 2016).

From the present research, it has been proven that application of bioinoculation has a clear potential to enhance the pathogenic bacteria removal process. Indeed, the preliminary results showed the beneficial effect of the bioinoculated strain (PFH₁) in the rhizosphere to increase remarkably the efficiency of the water treatment system for the reduction of pathogenic bacteria with a reduction in contact time. The single bioinoculation of selected bacteria into the rhizosphere of P. australis improved the kinetics of S. typhi inactivation by approximately 1 U-Log₁₀ (N/N₀) compared with the control. By a series of multi-bioinoculations, the enhancement of pathogenic bacteria reduction compared to the inhibition rate in the pilot-scale control (T) was of 2 U-Log₁₀(N/N₀).

The bioinoculation of antagonist showed positive results for most of the evaluated traits (single and multisequential injections), demonstrating the great potential of this practice to increase water quality. Furthermore, the results of the present study confirm the possibility of developing a commercial bioinoculant to be applied in biological water treatment processes to improve the treated water quality.

Screening for strains characterized by a high removal efficiency is a major problem to be solved before bioinoculation is performed in CWs. Therefore, it is better to isolate and screen microbial communities from the polluted environment (Meng et al. 2014). Moreover, the effect of intensified treatment of pollutants could only be maintained for a short time because the selected strains often lose the capability that they possess at laboratory pure-culture conditions (Chen et al. 2017).

This study has contributed to an eco-friendly strategy to improve the water treatment process by a CW, and highlighted the fact that better pathogen removal efficiency can be obtained through bioinoculation. As a future perspective of this work, the application of this strategy in field conditions, with multi-inoculation of antagonist substances protected by natural polymers to inactivate pathogenic bacteria in treated water without chemical additive, extension in the retention time or addition of complementary water treatment stages, should be taken into consideration.

This work is supported by CERTE contract programs funded by the Ministry of Higher Education and Scientific Research of Tunisia.