Flagellar Phenotypes Impact on Bacterial Transport and Deposition Behavior in Porous Media: Case of Salmonella enterica Serovar Typhimurium

Abstract

Bacterial contamination of groundwater has always been an ecological problem worthy of attention. In this study, Salmonella enterica serovar Typhimurium with different flagellar phenotypes mainly characterized during host-pathogen interaction were analyzed for their transport and deposition behavior in porous media. Column transport experiments and a modified mobile-immobile model were applicated on different strains with flagellar motility (wild-type) or without motility (ΔmotAB), without flagella (ΔflgKL), methylated and unmethylated flagellin (ΔfliB), and different flagella phases (fliC^ON, fljB^ON). Results showed that flagella motility could promote bacterial transport and deposition due to their biological advantages of moving and attaching to surfaces. We also found that the presence of non-motile flagella improved bacterial adhesion according to a higher retention rate of the ΔmotAB strain compared to the ΔflgKL strain. This indicated that bacteria flagella and motility both had promoting effects on bacterial deposition in sandy porous media. Flagella phases influenced the bacterial movement; the fliC^ON strain went faster through the column than the fljB^ON strain. Moreover, flagella methylation was found to favor bacterial transport and deposition. Overall, flagellar modifications affect Salmonella enterica serovar Typhimurium transport and deposition behavior in different ways in environmental conditions.

1. Introduction

Groundwater is an important part of water resources and plays a vital role in maintaining ecosystems around the world. However, 80% of wastewater is released into the environment without adequate treatment [1]. Wastewater contains many severe multi-pollutant sources, such as bacteria [2], and some of them are problematic to plants, animals, or human health. These microorganisms reach the groundwater through different layers of the subsurface region [3]. In the environment, a pathogenic bacterium like Salmonella enterica is widely disseminated, and it is among the most prevalent food-borne bacterial pathogens [4], causing gastroenteritis worldwide [5]. Water plays an important role in the spreading of this organism among plants, animals, or humans [6]. To take measures to reduce bacterial contamination of water, it is particularly important to understand the transport and adhesion mechanism of bacteria in porous media to protect soil and groundwater from contamination.

To achieve these goals, considerable efforts have been made to decipher the impact of bacterial transport in porous media. Thus, diverse parameters have been investigated: physical properties of porous media such as particle shape [7], soil structure [8,9], pore size [10], porosity [11], hydrodynamic and hydraulic properties of porous media such as flow rate [12], saturation degree [13], and chemical and physicochemical properties related to the liquid phase in which bacteria are transported such as pH and ionic strength [14]. Different properties concerning the bacteria-like cell shape [15], hydrophobicity [16], and motility [17] have been considered. Bacterial motility is the ability of the bacteria to move independently and is considered a virulence factor [18]. Flagella are motility appendages rotating by a reversible motor to ensure their movement toward nutrients and away from toxins [19] in many bacteria. Bacteria have two modes during the swimming motions driven by flagella; runs and tumbles. Runs are long straight swimming, and during tumbles, bacteria stop runs and change their orientation quickly [20]. Escherichia coli flagella have been described recently to affect transport and deposition; the flagellated bacteria could cause higher deposition onto quartz sand/silica surfaces [21]. Likewise, Haznedaroglu et al. [22] found that several Salmonella enterica serovars which were differentially flagellated displayed altered deposition. Flagellated bacteria had more deposition than non-flagellated strains onto quartz. However, the flagellar phenotype was not the only difference between the serovars with distinct genotypes that were employed in this work.

The multiprotein complex forming the flagellar filament contains mainly one globular protein, the flagellin, but many S. enterica serovars express one of two flagellins, FliC or FljB, through a phase variation mechanism [23]. FliC-expressing bacteria are more efficient in colonizing intestinal epithelia than FljB-expressing bacteria [24]. These two flagellins can be methylated on their lysine residues [25] by the methylase FliB, which increases flagellin surface hydrophobicity without affecting bacterial motility [16]. In order to synthesize the flagellar filament, many proteins are needed, including the hook-filament junction protein FlgK and the second hook-filament junction FlgL [26]. Likewise, when the genes motAB are inactivated, the bacteria are still flagellated but not motile.

Much research has been carried out to understand the impact of bacterial motility on bacterial transport and deposition in porous media. Recently, Bai et al. [10,27,28] performed bacteria transport and deposition experiments at a laboratory column scale to investigate the simultaneous influence of soil physical and cell properties on the mechanisms governing transport and deposition processes. Different species of motile and non-motile bacteria were used for this work. They found that the transport of non-motile species increased with the increasing heterogeneity of the porous media. This transport reduced non-motile bacteria retention in the porous medium by reducing the contact between bacteria and retention sites. Cell motility could allow the bacteria to swim upstream, leading to longer retention and, subsequently, cell deposition in the explored regions. Furthermore, flagella have also been reported to act as an adhesin on abiotic [29] and on biotic surfaces via lipids [30], in both cases with a better affinity on hydrophobic surfaces. So, the flagella could directly impact bacterial transport through porous media [31,32].

Even though bacterial motility has been widely investigated in the existing literature [33,34,35], there is a lack of understanding about the role of flagellar properties on bacterial transport and deposition behavior in porous media. For this purpose, previous works from Bai et al. were extended here by using one bacterial strain in porous media but with different flagellar phenotypes via mutations. Transport and deposition experiments of Salmonella enterica serovar Typhimurium (S. Typhimurium) mutants (six different flagellar phenotypes: without flagella, motility, and methylation or expressing one type of flagellin) have been performed on a homogenous sandy medium at column scale under saturated flow conditions. The breakthrough curves (BTCs) and retention profiles (RPs) of S. Typhimurium were measured and numerically modeled to decipher the flagellar impact on transport and retention mechanisms.

2. Results

2.1. Flagella and Their Motility Impact Transport and Deposition Behavior of S. Typhimurium Strain in Sandy Porous Media

Several S. Typhimurium strains were used to investigate the importance of motile flagella on bacterial transport and deposition in porous media (

Table 1. Cell properties for S. Typhimurium strains.

Flagellar Phenotypes	Wild Type	ΔflgKL	ΔmotAB	fljBON	fliCON	ΔfliB
	EM774	EM4969	EM4939	EM1013	EM1012	EM3734
Cell motility	Motile	Non-motile	Non-motile	Motile	Motile	Motile
Flagella	On	Off	On	On	On	On

): motile wild-type (WT) strain, non-motile with flagella (ΔmotAB) and non-motile without flagella (ΔflgKL). The ΔmotAB mutant is a non-motile strain without a stator, which is needed for flagella rotation [36]. Bacterial transport and deposition experiments were performed in triplicate, and a good repeatability of breakthrough curves (BTCs), as well as retention profiles (RPs), were obtained for each strain (Figure 1(a1–b3)).

The observed BTCs obtained from WT (Figure 1(a1)) and ΔmotAB (Figure 1(a3)) exhibited an asymmetrical shape with an obvious tailing, conversely to the BTC symmetrical shape of the strain without flagella (ΔflgKL, Figure 1(a2)). However, the flagellated non-motile strain, ΔmotAB, showed a more obvious tailing compared to WT. Meanwhile, some slight differences were observed between WT and ΔmotAB: an average higher peak for the WT (before 0.75 V/V₀, 0.6 C/C₀, Figure 1(a1)) compared to those of ΔmotAB (around 0.75 V/V₀, 0.5 C/C₀, Figure 1(a3)). The average mass percentage of ΔmotAB recovered, M_eff, was 36.78% (

Table 2. Experimental conditions and mass balance information of S. Typhimurium transport experiments under saturated conditions.

Bacterial Strains	Replicate	Saturated Degree (%)	C Initial (CFU/mL)	Porosity (%)	Bulk Density (g/cm3)	Pulse Time (min)	Darcy Velocity (cm/min)	Retardation Factor	Recovery (%)
									Meff	Mretained	Mtotal
WT	1	98.2	7.60 × 108	37.88	1.65	2.40	0.906	0.94	51.97	34.64	86.61
	2	98.5	1.40 × 109	36.41	1.69	2.33	0.936	0.79	46.49	26.99	73.48
	3	98.1	6.40 × 108	37.98	1.64	2.37	0.939	0.86	42.83	27.35	70.18
								(0.86) a	(47.1)	(29.66)	(76.76)
ΔflgKL	1	98.4	1.27 × 109	38.92	1.62	2.20	1.010	0.66	44.41	17.69	62.10
	2	98.9	1.23 × 109	38.26	1.64	1.85	1.191	0.66	44.83	22.41	67.24
	3	99.1	1.57 × 109	37.91	1.65	2.23	0.991	0.65	44.87	29.69	74.56
								(0.66)	(44.7)	(23.26)	(67.97)
									0.42 b	0.16	0.18
ΔmotAB	1	96.4	1.30 × 109	37.68	1.65	2.32	0.951	0.98	38.78	31.20	69.98
	2	97.9	1.73 × 109	37.17	1.66	2.44	0.903	0.99	38.45	25.37	63.82
	3	96.6	1.93 × 109	37.77	1.65	2.67	0.826	0.75	33.12	22.81	55.93
								(0.91)	(36.78)	(26.46)	(63.24)
									0.03	0.42	0.10
ΔfliB	1	98.4	9.00 × 108	37.92	1.65	2.36	0.934	1.01	44.8	20.93	65.73
	2	97.3	1.60 × 109	35.12	1.72	2.11	1.045	1.04	46.25	8.36	54.61
	3	99.5	9.33 × 108	38.06	1.64	2.22	0.994	0.75	42.89	13.48	56.37
								(0.93)	(44.65)	(14.26)	(58.9)
									0.44	0.03	0.04
fljBON	1	98.6	1.17 × 109	36.84	1.67	2.07	1.066	0.79	40.44	33.58	74.02
	2	99.2	9.00 × 108	36.79	1.67	2.34	0.941	0.70	38.34	17.76	56.10
	3	98.8	1.00 × 109	36.91	1.67	2.36	0.935	0.84	38.98	14.46	53.44
								(0.77)	(39.25)	(21.93)	(61.18)
									0.04	0.29	0.13
fliCON	1	93.5	1.57 × 109	36.46	1.68	2.15	1.025	0.72	51.02	24.54	75.56
	2	96.0	1.03 × 109	36.09	1.69	2.33	0.945	0.65	46.28	17.53	63.81
	3	100.0	1.10 × 109	36.38	1.69	2.27	0.972	0.86	46.90	13.70	60.60
								(0.74)	(48.07)	(18.59)	(66.66)
									0.77	0.05	0.2

^a The values given were the mean value of the corresponding parameters, ^b the values given parallel were the p-values for analyzing significance with WT using t-test.

), lower than WT (47.1%), indicating that bacterial motility could improve bacteria transport through sandy media. This result was consistent with the slightly higher retardation factors observed for ΔmotAB (average retardation factor, 0.91, Table 2) compared to the WT (0.86, Table 2). The overall bacteria mass recovery (M_total) was subsequently determined as the sum of M_eff and M_retained. However, different bacteria retention rates in the sand, M_retained, were obtained for WT (29.7%) and ΔmotAB (26.5%) strains, and higher M_total were obtained for WT (76.8%) in comparison to ΔmotAB (63.2%) (Table 2) due to the higher recovery of WT strain in the effluent. The observed RPs of ΔmotAB showed a different behavior compared to WT, especially the bacteria attached to the column inlet (Figure 1(b1,b3)). All replicates of ΔmotAB showed a non-monotonic distribution profile behavior with a higher number of attached bacteria (S/C₀) at the column inlet (layer 0–2 cm). Conversely, the monotonic behavior of WT exhibited no significant difference in retention rate among all layers.

The BTCs of WT and ΔmotAB were well fitted (Figure 1(a1,a3)) with R² > 0.98 (

Table 3. Fitted parameters (replicates) for S. Typhimurium through saturated columns using the MIM model in HYDRUS-1D code.

Bacterial Strains	Replicate	λ (cm)	θm/θ	Katt (min−1)	Kd (min−1)	Kstr (min−1)	R2
WT	1	0.69	0.54	3.12 × 10−2	3.93 × 10−1	7.87 × 10−6	0.9963
		(0.06) *	(0.004)	(1.58 × 10−2)	(2.10 × 10−1)	(6.56 × 10−3)
	2	0.48	0.51	1.46 × 10−1	6.07 × 10−1	8.35 × 10−2	0.9958
		(0.07)	(0.008)	(1.88 × 10−2)	(6.09 × 10−2)	(1.25 × 10−1)
	3	0.29	0.52	8.59 × 10−2	4.29 × 10−1	1.85 × 10−1	0.9963
		(0.03)	(0.002)	(1.36 × 10−2)	(8.16 × 10−2)	(2.49 × 10−2)
Mean values		0.49	0.53	0.09	0.48	0.09	0.996
		(0.2001) **	0.0166	0.0574	0.1147	0.0929	0.0003
ΔflgKL	1	0.35	0.62	3.25 × 10−2	3.38 × 10−1	1.76 × 10−1	0.9987
		(0.03)	(0.002)	(5.00 × 10−3)	(1.17 × 10−1)	(6.23 × 10−2)
	2	0.21	0.58	1.23 × 10−1	8.76 × 10−1	1.71 × 10−1	0.9968
		(0.04)	(0.004)	(1.82 × 10−2)	(1.71 × 10−1)	(8.46 × 10−2)
	3	0.29	0.65	2.68 × 10−2	4.99 × 10−1	1.71 × 10−1	0.9948
		(0.05)	(0.003)	(2.52 × 10−2)	(9.80 × 10−1)	(6.95 × 10−2)
Mean values		0.28	0.62	0.06	0.57	0.17	0.997
		0.0699	0.0375	0.0539	0.2761	0.0029	0.0020
ΔmotAB	1	0.39	0.52	1.35 × 10−1	3.80 × 10−1	2.18 × 10−1	0.9793
		(0.14)	(0.03)	(3.95 × 10−2)	(7.83 × 10−2)	(3.14 × 10−1)
	2	0.55	0.53	9.22 × 10−2	2.63 × 10−1	2.03 × 10−1	0.9668
		(0.20)	(0.04)	(4.15 × 10−2)	(1.37 × 10−1)	(4.68 × 10−1)
	3	0.21	0.64	4.93 × 10−2	5.72 × 10−1	4.63 × 10−1	0.9969
		(0.02)	(0.002)	(5.48 × 10−3)	(1.37 × 10−1)	(5.45 × 10−2)
Mean values		0.38	0.56	0.09	0.40	0.29	0.981
		0.1703	0.0654	0.0428	0.1560	0.1459	0.0151
ΔfliB	1	0.14	0.67	3.37 × 10−1	5.63 × 10−1	1.03 × 10−1	0.9823
		(0.04)	(0.02)	(1.83 × 10−2)	(5.74 × 10−2)	(1.12 × 10−1)
	2	0.30	0.51	1.36 × 10−1	5.74 × 10−1	5.82 × 10−2	0.9852
		(0.13)	(0.02)	(5.11 × 10−2)	(1.33 × 10−1)	(3.21 × 10−1)
	3	0.34	0.56	7.50 × 10−2	4.42 × 10−1	2.00 × 10−1	0.9989
		(0.03)	(0.003)	(4.58 × 10−3)	(4.35 × 10−2)	(5.62 × 10−2)
Mean values		0.26	0.58	0.18	0.53	0.12	0.989
		0.1042	0.0852	0.1368	0.0732	0.0726	0.0088
fliCON	1	0.32	0.65	2.99 × 10−2	6.92 × 10−1	1.69 × 10−4	0.9945
		(0.06)	(0.005)	(3.91 × 10−2)	(9.65 × 10−1)	(4.69 × 10−2)
	2	0.59	0.40	2.87 × 10−1	8.25 × 10−1	1.57 × 10−1	0.9958
		(0.16)	(0.008)	(1.09 × 10−1)	(1.78 × 10−1)	(3.92 × 10−2)
	3	0.35	0.59	6.01 × 10−2	5.37 × 10−1	5.01 × 10−2	0.9988
		(0.02)	(0.002)	(1.17 × 10−2)	(1.06 × 10−1)	(1.47 × 10−2)
Mean values		0.42	0.54	0.13	0.68	0.07	0.996
		0.1496	0.1295	0.1407	0.1441	0.0804	0.0022
fljBON	1	0.22	0.51	1.64 × 10−1	7.61 × 10−1	3.60 × 10−1	0.9985
		(0.02)	(0.002)	(2.10 × 10−2)	(8.43 × 10−2)	(2.04 × 10−2)
	2	0.76	0.54	1.14 × 10−1	5.29 × 10−1	3.86 × 10−1	0.9947
		(0.24)	(0.02)	(6.74 × 10−2)	(2.03 × 10−1)	(2.07 × 10−1)
	3	0.33	0.59	7.71 × 10−2	6.00 × 10−1	2.68 × 10−1	0.9972
		(0.04)	(0.005)	(6.61 × 10−3)	(9.98 × 10−2)	(1.00 × 10−1)
Mean values		0.44	0.54	0.12	0.63	0.34	0.997
		0.2885	0.0416	0.0436	0.1185	0.0621	0.0019

* The values given were the standard error coefficients (S.E.Coeff) obtained from HYDRUS-1D code, ** The values given parallel were the standard deviation.

) using the MIM model. The dispersivity λ and mobile water fraction θ_m/θ could reflect the flow patterns in porous media; lower λ and higher θ_m/θ show a more uniform, thus less preferential flow. The dispersivity of WT was 0.49 cm while for ΔmotAB was 0.38 cm, indicating that WT had more dispersive flow pathways. Slightly lower mobile water fractions θ_m/θ were obtained for WT (53%) compared to those of ΔmotAB (56%) (Table 3), indicating that lower pore water volumes were required for WT transport compared to those of ΔmotAB. These results showed a different flow pattern between flagellated motile and non-motile strains; the non-motile strain tended to have more homogeneous transport in porous media than the motile strain. The same order of magnitude of the attachment coefficient k_att was obtained for WT and ΔmotAB on the sandy columns (Table 3). The same tendency was also observed for the detachment coefficient, k_d. The k_d values obtained for both mutants were greater than k_att values (Table 3), indicating that attachment was approaching linear equilibrium sorption conditions.

To examine the importance of flagella presence in transport behavior, two non-motile bacteria were compared: ΔflgKL (without flagella) and ΔmotAB (with flagella). The observed BTCs obtained from ΔflgKL occurred earlier than for ΔmotAB (Figure 1(a2,a3)). Average earlier breakthrough (0.4 V/V₀) for ΔflgKL was obtained compared to ΔmotAB (0.5 V/V₀). An average higher peak (around 0.75 C/C₀) for ΔflgKL was observed compared to ΔmotAB (around 0.6 C/C₀). Earlier breakthroughs and higher peaks for ΔflgKL were consistent with lower average retardation factors of this strain (0.66, Table 2) compared to ΔmotAB (0.91).

Similar overall bacteria mass recovery (M_total) was obtained for both mutants with (63.2% for ΔmotAB) or without flagella (68.0% for ΔflgKL). However, the M_eff recovery in the effluent of ΔflgKL was higher than for ΔmotAB (44.7% vs. 36.8%), indicating that the flagella presence may reduce bacterial transport. In addition, ΔflgKL exhibited a lower number of bacteria retained in the sand compared to ΔmotAB (M_retained of 23.3% for ΔflgKL and of 26.5% for ΔmotAB, Table 2). The observed RPs of both ΔflgKL and ΔmotAB strains showed similar non-monotonic retention profile behavior (Figure 1b2,b3). Thus, ΔflgKL and ΔmotAB strains showed a higher number of attached bacteria at the column inlet (Figure 1(b2,b3), layer 0–2 cm). This phenomenon indicated that the bacteria were retained mostly in the first layer of the column and displayed the most important step for blocking bacteria contamination.

ΔflgKL had lower mean dispersivity λ (0.28 cm) but higher mobile water fraction θ_m/θ (62%) compared with ΔmotAB (56%). This phenomenon could be attributed to the difference in flagella presence, meaning that the bacteria with flagella could have a less uniform and more preferential transport in porous media. The mean value of the attachment coefficient k_att reached 0.06 min⁻¹ for ΔflgKL, slightly lower than ΔmotAB (0.09 min⁻¹). It showed that ΔflgKL is less adhering to sand than ΔmotAB, a result in agreement with experimental observations. ΔflgKL also had a higher detachment coefficient k_d (0.57 min⁻¹) than ΔmotAB (0.40 min⁻¹), indicating a higher possibility of being reversibly detached from the solid grains.

2.2. Flagella Phases Influence Transport and Deposition Behavior of Bacteria in Porous Media

The effect of different flagellins of S. typhimurium, here with fliC^ON and fljB^ON strains, was investigated to observe their role in transport and deposition in porous media. We used two mutants that are locked in FliC or FljB, respectively, with different positions and orientations of flagellins. When the flagellum is composed of FliC flagellin, it plays an important role during Salmonella infection in mice and humans [37].

The observed BTCs of fliC^ON and fljB^ON strains (Figure 2(a2,a3)) showed similar arrival breakthroughs for both strains. However, slightly higher average C/C₀ peaks were observed for fliC^ON compared to those of fljB^ON. Thus, the average peak C/C₀ value of the fliC^ON strain was about 0.7, while the fljB^ON strain was around 0.6. The baseline for fliC^ON and fljB^ON strains was achieved in V/V₀ 1.5 and 2.0, respectively. These results showed that fliC^ON goes faster through the column than fljB^ON.

The overall bacteria mass recovery (M_total) reached 61.2% for fljB^ON and 66.6% for the fliC^ON strain. Even though both strains exhibited slightly similar retardation factors (0.78 for fljB^ON and 0.74 for fliC^ON), the fljB^ON strain had a lower percentage of recovery from the effluent than the fliC^ON strain (39.3% vs. 48.1%, respectively, Table 2, Figure 3). In addition, the fljB^ON strain was more retained on the sandy grains (M_retained of 21.9% vs. 18.6%, respectively).

Thus, higher recovery in the effluent and lower retention on the sand for the fliC^ON compared to the fljB^ON strain were in agreement with faster transport of the fliC^ON strain observed from BTCs data. Based on the RPs (Figure 2(b2,b3)), the two mutants also showed that the bacteria were more easily attached to the column inlet but with a greater efficacity for fljB^ON. For the fliC^ON and fljB^ON strains, the fliC^ON strain had a little lower dispersivity λ but the same mobile water fraction θ_m/θ of 54% as the fljB^ON strain. It seemed that there was no obvious difference between these two strains for the transport pattern. The fliC^ON strain had lower K_str than the fljB^ON strain.

2.3. Effects of Flagella Methylation on Transport and Deposition of Bacteria in Porous Media

The posttranslational methylation of flagellin via fliB plays an important role in the invasion of host cells [16]. It should be noted that the BTC behavior of the ΔfliB strain replicates (Figure 2(a1)), and the average peaks, as well as BTCs arrival, were different from those obtained for WT (Figure 1(a1)).

The average retardation factor for the ΔfliB strain was 0.93 and 0.86 for WT (Table 2), indicating a delay for BTCs of ΔfliB. Lower bacteria effluent recovery (M_eff) was obtained for ΔfliB (44.7%) compared to WT (47.1%). Meanwhile, ΔfliB presented a much lower retention rate (14.3%) in the sand in comparison to WT (29.7%) (Table 2).

According to the RPs (Figure 1(b1) and Figure 2(b1)), the ΔfliB strain was more easily attached to the first layer of the column inlet than WT. The ΔfliB strain had lower λ (dispersivity of 0.26 cm, Table 3) and higher mobile water fraction θ_m/θ (58%) compared to WT (53%), fliC^ON (54%), and fljB^ON strains (54%). Based on these findings, bacteria without flagellin methylation have more uniform and less preferential transport. ΔfliB had a higher attachment coefficient k_att (0.18) than WT (0.09), fliC^ON (0.13), and fljB^ON (0.12), which showed that the ΔfliB strain was more likely to be captured by physicochemical mechanisms in the sand than by physical straining. It should be noted that parameters obtained from numerical simulations remained mostly in the same order of magnitude; however, their comparison between various mutants of the same strain allows us to observe some tendencies related to their specific biological properties.

3. Discussion

The prokaryotic flagellum is an extraordinary multi-subunit organelle, complex in its regulation and assembly, best known as a motility organelle responsible for bacterial movement, necessary for chemotaxis, and involved in biofilm formation [31,38,39]. When bacteria are individual cells in liquid, the rotation of their flagella generates a swimming motion. Conversely, bacteria, when immobilized on surfaces, form a biofilm, leading to higher resistance to antimicrobials and long-term colonization [40].

In our results, it should be noted that the total recovery rate of all bacteria was lower than 100%, suggesting that S. Typhimurium was strongly attached to the sand or the column and was starting biofilm formation. Furthermore, bacteria could reach crevices formed by the grains of sand via their flagella, leading to a better attachment [41] and to a lower recovery rate from the column. Similar results were obtained at a low flow rate by others [42]. They found that non-motile E. coli strains had higher retention time than motile E. coli strains in homogeneously packed sand column experiments. It showed that bacterial motility could be a factor in preventing bacterial attachment to solid surfaces. However, the BTCs of non-motile Pseudomonas fluorescens strains were found to be the first to emerge compared to motile bacteria at low water velocity [43]. Unfortunately, no data was available concerning the nature of the mutation present in the non-motile strain of this work.

A lower average retardation factor for the ΔflgKL mutant was found compared to the ΔmotAB mutant (Table 2). This may be due to the absence of flagella for ΔflgKL. It may have reduced the possibility of interaction with sandy grains, thus promoting the preferential transport of this strain through the porous media. The non-motile ΔmotAB mutant has flagella, and they could be attached to the surface of the sand more easily. It is harder for the strain without flagella to be captured by the sandy grains. This result was in agreement with low retardation factors of ΔflgKL, indicating preferential transport and low retention in the porous media. Similar results were observed for Azotobacter vinelandii [44]. Flagellated strain had a higher deposition rate than a non-flagellated one on a quartz surface in a radial stagnation point flow cell setup [44]. Combined with the result of this study, it could be seen that flagella promote bacteria adsorption to different surfaces on different experimental scales, and flagellated bacteria have more probability of sticking on porous media.

WT had higher recovery from effluent than the ΔmotAB mutant. WT had a more preferential transport in porous media than the ΔmotAB mutant, which may explain the high mass recovery in the effluent observed. These results agree with the existing literature, suggesting a linear relationship between preferential transport and mass recovery in the effluent [10]. Due to its motility, WT is not only highly recovered in the effluent but also more retained in the sand compared to ΔmotAB when it is in contact with the sandy grains (Table 2), in agreement with what has been reported in the literature. Thus, Zhang et al. (2021) [21] observed motile and non-motile E. coli ΔfliC strains and found that the motile strain was deposited more easily onto a quartz sand surface, grain-to-grain contacts, and narrow flow channels. It showed that the deposition of non-motile bacteria would be reduced in sand and water environments. The motile bacteria could be transported faster than bacteria without motility by providing a driving force for bacterial transport, which emphasized the importance of motility to bacteria during their movement in porous media. Different conclusions have been reported regarding the impact of cell motility on bacteria transport. Some authors suggested that the ability of the cell to swim is an important factor that enhances transport [45]. Others reported lower mass recovery of the motile strains compared to the non-motile ones. They suggested that small pores of the porous medium increase the possibility of the non-motile bacteria being trapped, causing a low recovery rate [10]. However, it should be noted that these results have been observed for different species: the motile E. coli and the non-flagellated Klebsiella sp., which renders it difficult for comparison, while in this work, different mutants of the same strain were investigated, giving a more accurate comparison between different experiments.

WT and ΔmotAB mutants showed different behaviors in the observed RPs. The difference between the RPs of these two strains might be explained by bacterial near-surface swimming. When bacteria swim next to surfaces, it increases the possibility of being retained onto them. Therefore, S. Typhimurium flagellar motility is needed for reaching the host cell surface but is influenced by physical forces during “near-surface swimming” by increasing local bacterial density [46]. Furthermore, E. coli is known to have near-surface swimming on solid surfaces and to have three swimming modes: landing, near-surface swimming, and swimming away; the three processes require different dynamic conditions [47]. Thus, in this work, S. Typhimurium WT via near-surface swimming on the sand surface could keep an equilibrium among these three modes to maintain stable bacterial retention in each layer, while the non-motile strains are mostly captured by the sand surface of the first layer without flagella dynamic to swim away. Likewise, the ΔmotAB mutant showed a delay in BTCs compared with WT. In this case, the non-motile flagella can add a physical constraint because they are most likely spreading to all directions around the bacterial body, as observed during the tumbling phase. Thus, the bacterial cells are occupying more space, leading to a potential increase in the residence time in the column.

The fliC^ON mutant had faster transport in BTCs, higher recovery in the effluent, and a lower retention rate compared to the fljB^ON strain. Furthermore, experiments performed by others have highlighted structural differences between fljB^ON and fliC^ON strains [48]. They found that FliC had a higher density in the D3 domain compared to FljB. This allows FljB to be more flexible and motile under high-viscosity conditions. Also, the fljB^ON mutant has been found to stay longer because of near-surface swimming than the fliC^ON mutant [24]. This observation can also explain why FliC promotes bacteria transport in porous media. It should be noted that the replicates BTCs of fljB^ON showed better repeatability than fliC^ON, probably reflecting the motility advantage of fliC^ON compared to fljB^ON described in previous studies. FliC flagella modulate the swimming behavior on the surface to facilitate the search of invasion sites which can be influenced by the column packing [24].

The ΔfliB mutant presented around half of the retention rate in the RPs in comparison to WT. It may probably be because the ΔfliB filament without methylation has lower hydrophobicity, causing lower attachment to sand. Hydrophobicity has been found to be a driving force in bacteria retention in porous media. Thus, a study involving Campylobacter jejuni and E. coli cells has demonstrated that the more hydrophobic C. jejuni had more attachment to porous media than E. coli. However, it should be noted that hydrophobicity’s impact on bacteria transport and deposition has mainly been studied between two very different bacterial species, making it difficult to reach an accurate conclusion [49]. Recently, the flagella methylation of S. Typhimurium has been found to adhere to hydrophobic surfaces of epithelial cells [16]. As observed in this work, when the hydrophobicity is high, the retention is higher.

The findings of this work could be useful to improve the risk assessment process of bacterial contamination in soil and groundwater. The pathogen concentrations and the weather conditions have already been used with hydrodynamic modeling to estimate the risk for public health [50,51,52]. Here, we observe that biological cell motility plays a clear role in the capacity of pathogens to reach groundwater. For that reason, we suggest that microbial motility should be studied on putative soil contaminants and added to risk assessment models.

4. Materials and Methods

4.1. Bacterial Strains

The bacteria used in this study were all in S. Typhimurium SL1344 background and obtained as described here [16,24]. The different flagellar phenotypes used were: wild-type strain (WT), non-motile without flagella (ΔflgKL) [26], non-motile with flagella (ΔmotAB) [36], motile with flagella unmethylated on lysine residues (ΔfliB), motile expressing only the flagellin FliC (fliC^ON) and motile expressing one flagellin FljB (fljB^ON) [16] (Table 1) FljB and FliC have different positions and orientations of flagellin in filament and FljB showed higher motility than bacteria expressing FliC under high viscosity [48]. S. Typhimurium have peritrichous flagella distributed all over the cells [36]. The structure of flagella includes the basal body, the hook, and the filament (Figure 4).

Bacteria were inoculated on a lysogeny broth (LB) agar (35 g/L) Petri dish overnight at 37 °C, then bacteria strains were grown at 37 °C, 85 rpm overnight in LB (20 g/L). The growth curves of bacteria were measured by absorbance at OD₆₀₀. Bacterial suspension harvested from the stationary phase of bacterial culture was used in this work. Batch experiments [53] were conducted to understand whether there was a change in bacterial concentration during column transport experiments in the absence of hydrodynamic flow conditions. Next, 250 mL conical flasks with 30 g Fontainebleau sand, 8 mL 0.1 mmol/L NaCl solution, and 2.5 mL bacteria suspensions were put on a shaker (85 rpm, 21 °C) for 45 min. The time for shaking was the same as the duration of the column transport experiments. The initial and final bacterial concentrations were determined by the number of colony-forming units (CFU) using the plating method [54] on LB agar. The result showed that there was not an obvious change in bacteria amount during the experiment. Accession numbers of FljB and FliC flagellin were PDB: 6 RGV and 1 IO1, respectively.

4.2. Column Transport Experiments

Transport experiments were performed in Plexiglass columns with an inner diameter of 3.4 cm and a length of 16 cm, packed with Fontainebleau sand. The particle size of Fontainebleau sand ranged from 0.16 to 0.79 mm, with a median diameter size of 0.36 mm. Two filter papers were used at the inlet and outlet of the column to make sure the solution came out without fine sand particles during the transport experiment. Before every experiment, the pump, tubes, and other column components were sterilized with bleach and 15 min of UV. Other materials were sterilized using an autoclave. The empty column with pipes, two caps, and two filter papers was weighted, and the mass was noted as M₀ (g). The sand was inserted in successive layers into the column to assure homogenous distribution, and both column and sand were weighted and noted as M₁ (g). The mass of sand was estimated as:

$$M_s=M_1-M_0$$

(1)

and the sand bulk density (ρ_b) was calculated by dividing the dry mass of the sand(M_s) by the volume of the column (V_c):

$${\rho }_b=M_S/V_c$$

(2)

The porosity ($\epsilon$) was then estimated by the sand bulk density as follows:

$$\epsilon =\left(1-{\rho }_b/2.65\right)\times 100\%$$

(3)

The column was flushed upward using a background solution of NaCl (0.1 mmol/L), then the column was rinsed using the background solution reversed, and at the same time, the conductivity of the effluent solution was continuously measured at the column outlet until this parameter reached a stable value. Then, the column fully saturated with the background solution was weighed, and this mass was noted as M₂ (g). Assuming that water density is 1 g/cm³, the volume occupied by water (V_w) was estimated as:

$$V_w=M_2-M_1$$

(4)

The saturation degree (S) could then be described as follows:

$$S=V_w/V_p$$

(5)

here V_p is the volume of the pores, calculated as

$$V_p=V_c\times \epsilon$$

(6)

For bacteria injection experiments, 20 mL of bacterial suspension was injected into each column at the initial concentration given in Table 2 for each replicate (Figure 5). The breakthrough curve of each strain was obtained by plotting bacterial concentration as a function of pore volume V/V₀. Bacteria concentration was deduced by measuring the absorbance of effluent at OD₆₀₀ at the column outlet. The bacterial suspension was flushed with a background solution (0.1 mmol/L NaCl solution) until the absorbance of the effluent returned to the baseline level. The ratio of bacteria mass recovered from the effluent (M_eff) and retardation factor (R) was then calculated [55,56]. The retardation factor was calculated by the residence time of bacteria to the theoretical residence time of water in the column. M_eff corresponds to the ratio of the bacteria mass recovered at the column outlet to their mass injected at the column inlet. The retardation factor represents the ratio of the resident time of bacteria to the theoretical water resident time.

After bacterial transport experiments, the sandy porous media inside each column was divided into 8 layers to explore the spatial distribution of the bacteria. First, a sterilized spoon was used to remove the sand into 8 flasks containing 0.1 mmol/L NaCl solution. The same solution was also used to wash the remaining sand off of the spoon. Then, we shook the flasks filled with sandy porous media for 15 min using an orbital shaker to isolate the retained bacteria from the sand. The Plating method count [54] was employed to determine the bacterial concentration in each layer of sand. The mass of bacteria recovered from all layers was regarded as M_retained, which was calculated by the total number of CFU retained in the sand divided by the CFU of the initial bacterial suspension injected into the column. The M_total is the sum of M_eff and M_retained.

4.3. Simulation

The same modified mobile-immobile (MIM) model under two kinetic deposition sites, as described by Bai et al. (2016) [27], was used in this work. The two-region MIM model assumes that bacteria transport is limited to the mobile water region and that water in the immobile region is stagnant. Briefly, the one-dimensional bacterial transport model was described as [9]:

$$\frac{\partial \mathsf{\theta }\mathrm{C}}{\partial \mathrm{t}}+\rho \frac{\partial s_1}{\partial t} +\rho \frac{\partial s_2}{\partial t}=\frac{\partial }{\partial \mathrm{x}}\left({\theta }_{m}D\frac{\partial C}{\partial x}\right)-\frac{\partial qC}{\partial \mathrm{x}}$$

(7)

where θ is the total water content, C is the bacterial concentration in the liquid phase (CFU/cm³), t is time (min), ρ is the bulk density of the porous media (g/cm³), s₁ is the concentration of bacteria in the solid phase accounting for attachment or detachment, s₂ is the concentration of bacteria in the solid phase related to irreversible straining, x is the distance (L), θ_m is the mobile water content, D is the dispersion coefficient (cm²/min), and q is the Darcy velocity (cm/min). It was assumed that there was no bacterial exchange between the mobile and immobile regions. Then the bacterial mass transfer between solution and s₁, s₂ sites can be described as [57]:

$$\rho \frac{\partial \mathrm{S}}{\partial \mathrm{t}}=\rho \frac{\partial \left(s_1+s_2\right)}{\partial \mathrm{t}}={\theta }_m{\psi }_t{k}_{att}C-k_d\rho s_1+{\theta }_m{k}_{str}{\psi }_{x}C$$

(8)

where k_att, k_d, k_str are the first-order attachment, detachment, and straining coefficients, respectively (1/min); ψ_t and ψ_x describe the deposition of bacteria in the solid retention under time- and depth-dependent conditions, respectively. Another hydrodynamic parameter is λ (cm) which describes the dispersivity of the medium and can be estimated by [56]:

$$\lambda =\frac{D_m\ast {\theta }_m}{q}$$

(9)

where D_m is the dispersion coefficient of the mobile region (cm²/min).

Transport (θ_m, λ) and deposition parameters (k_att, k_d, k_str) were obtained by fitting of theoretical MIM model, implemented in Hydrus 1D code (software version: Hydrus-1D_4.17, free in https://www.pc-progress.com/en/Default.aspx?Downloads (accessed on 7 June 2021)) to the experimental breakthrough curves of bacteria.

4.4. Statistical Analysis

To determine whether flagellar phenotypes affect bacterial transport and deposition in porous media, the replicates for each strain were analyzed with WT for significance using Student’s t-test.

5. Conclusions

In this study, the effects of S. Typhimurium flagellar phenotypes on bacterial transport and deposition in sandy porous media were explored.

Flagella affected S. Typhimurium transport and deposition in porous media. S. Typhimurium strain with non-motile flagella reduced bacterial transport and increased bacterial retention in comparison to the strain without flagella. Flagella motility promotes bacterial transport, and this was confirmed by the higher retardation factor of the BTCs and the higher recovery of effluent of motile flagellated strains. As observed during host-pathogen interaction studies, flagella phases of S. Typhimurium have a different influence on bacterial movement in porous media. The fliC^ON strain showed a faster transport and higher recovery in the effluent than the fljB^ON strain, but the FljB flagellin promoted bacterial adhesion. Flagella methylation influenced S. Typhimurium transport and deposition in porous media, probably because of the higher hydrophobicity of the flagella filament’s outer surface.

Our results will lead to more robust transport models and could help to develop new environmental strategies through a more accurate risk assessment process.