Biofilm Producing Rhizobacteria With Multiple Plant Growth-Promoting Traits Promote Growth of Tomato Under Water-Deficit Stress

Plant growth-promoting rhizobacteria (PGPR) not only enhance plant growth but also control phytopathogens and mitigate abiotic stresses, including water-deficit stress. In this study, 21 (26.9%) rhizobacterial strains isolated from drought-prone ecosystems of Bangladesh were able to form air–liquid (AL) biofilms in the glass test tubes containing salt-optimized broth plus glycerol (SOBG) medium. Based on 16S rRNA gene sequencing, Pseudomonas chlororaphis (ESR3 and ESR15), P. azotoformans ESR4, P. poae ESR6, P. fluorescens (ESR7 and ESR25), P. gessardii ESR9, P. cedrina (ESR12, ESR16, and ESR23), P. veronii (ESR13 and ESR21), P. parafulva ESB18, Stenotrophomonas maltophilia ESR20, Bacillus cereus (ESD3, ESD21, and ESB22), B. horikoshii ESD16, B. aryabhattai ESB6, B. megaterium ESB9, and Staphylococcus saprophyticus ESD8 were identified. Fourier transform infrared spectroscopy studies showed that the biofilm matrices contain proteins, polysaccharides, nucleic acids, and lipids. Congo red binding results indicated that these bacteria produced curli fimbriae and nanocellulose-rich polysaccharides. Expression of nanocellulose was also confirmed by Calcofluor binding assays and scanning electron microscopy. In vitro studies revealed that all these rhizobacterial strains expressed multiple plant growth-promoting traits including N2 fixation, production of indole-3-acetic acid, solubilization of nutrients (P, K, and Zn), and production of ammonia, siderophores, ACC deaminase, catalases, lipases, cellulases, and proteases. Several bacteria were also tolerant to multifarious stresses such as drought, high temperature, extreme pH, and salinity. Among these rhizobacteria, P. cedrina ESR12, P. chlororaphis ESR15, and B. cereus ESD3 impeded the growth of Xanthomonas campestris pv. campestris ATCC 33913, while P. chlororaphis ESR15 and B. cereus ESD21 prevented the progression of Ralstonia solanacearum ATCC® 11696TM. In a pot experiment, tomato plants inoculated with P. azotoformans ESR4, P. poae ESR6, P. gessardii ESR9, P. cedrina ESR12, P. chlororaphis ESR15, S. maltophilia ESR20, P. veronii ESR21, and B. aryabhattai ESB6 exhibited an increased plant growth compared to the non-inoculated plants under water deficit-stressed conditions. Accordingly, the bacterial-treated plants showed a higher antioxidant defense system and a fewer tissue damages than non-inoculated plants under water-limiting conditions. Therefore, biofilm-producing PGPR can be utilized as plant growth promoters, suppressors of plant pathogens, and alleviators of water-deficit stress.


INTRODUCTION
Climate change is one of the key global concerns for sustainable agricultural production. According to the Intergovernmental Panel on Climate Change (Intergovernmental Panel on Climate Change [IPCC], 2014), significant areas of fertile agricultural lands will probably be lost or degraded due to adverse effects of climate change (e.g., sea level raising, salinization, heat, and drought) at the end of the 21st century. Another forecasted is that climate change may limit nutrient bioavailability in soils (Karmakar et al., 2016) and a change in the dynamics of pathogen and pest reproduction . Furthermore, the climate change may cause the emergence of new pathogens and pests. All these effects will ultimately affect the existing crop production system that extensively relies on chemical fertilizers and pesticides. The problem, however, is that an excessive use of agro-chemicals also causes environmental (in the air, water, and soil) pollution (Jarecki et al., 2008). A climate-friendly agriculture, therefore, depends on how well natural biota are used in the agricultural production systems, particularly in waterlimiting conditions (drought).
Bangladesh is one of the most vulnerable countries in the world to climate change, including water-deficit stress. It was reported that in Bangladesh 32.4% of the land during May to October (Kharif season), 27.2% in November to April (Rabi season), and 16.2% between March to May (pre-Kharif season) are under the threat of extreme water shortage, respectively (Alamgir et al., 2019). The scientific community agrees that plant growth and yield is affected by water-deficit stress (Vurukonda et al., 2016;Chandra et al., 2018;Martins et al., 2018;Saikia et al., 2018;Meenakshi et al., 2019). Water-deficit stress aggravates overproduction of reactive oxygen species (ROS) that promote oxidative damage by oxidizing proteins, lipids, nucleic acids, and other cellular macromolecules in plants (Sgherri et al., 2000;Vurukonda et al., 2016). However, plants have an abundant network for ROS detoxification including either enzymatic antioxidants, such as catalases, ascorbate peroxidases, superoxide dismutases, and glutathione reductases or nonenzymatic components through proline, carotenoids, phenolics, and flavonoids (Du et al., 2004;Kaushal and Wani, 2016). In order to combat water-deficit stress, tolerant plant varieties could be used in the future despite the fact that traditional breeding is time consuming. Nevertheless, developing transgenic crops through gene transfer or gene editing using the CRISPR/Cas9 system is associated with ethical and social acceptance issues. In light of these problems, application of beneficial microbes including plant growth-promoting rhizobacteria (PGPR) would be a simple and cheap way for improving plant resiliency against water-deficit stress.
Biofilms are surface-associated microbial cells, encased in a self-produced EPS that predominantly contain proteins, polysaccharide, extracellular DNA, and lipids (Flemming and Wingender, 2010). The literature contains several examples of biofilm PGPR that are much more effective under field conditions than any planktonic PGPR (Seneviratne et al., 2010;Pandin et al., 2017;Backer et al., 2018). Zhang et al. (2015) reported that nitrogenase activity, IAA production, phosphate solubilization, siderophore production, and ammonia production are incredibly higher in biofilm PGPR than the planktonic PGPR. Other advantages of biofilm PGPR are their higher resistance to antibiotics and adverse environmental stresses (e.g., high temperature, extreme pH, salinity, and drought), leading to an improved chance of survival in a competitive soil environment (Mah et al., 2003). Biofilm PGPR also produce remarkably higher amounts of antimicrobial compounds than the planktonic PGPR, leading to suppression of phytopathogens (Pandin et al., 2017). However, quite a few biofilm PGPR (e.g., Rhizobium leguminosarum, Agrobacterium sp., A. vinelandii, Enterobacter cloacae, Xanthomonas sp., Pseudomonas sp., P. polymyxa, Bradyrhizobium sp., Bacillus subtilis, and B. drentensis) were identified till date (Bais et al., 2004;Timmusk et al., 2005;Ude et al., 2006;Mahmood et al., 2016;Gupta et al., 2017). Thus, biofilm-producing bacteria could be advantageous over others to thrive in a new water-stress environment for delivering beneficial effects to plants. However, biofilm-producing bacteria associated with crop species which are naturally adapted to water stress, such as tomato, have not been explored so far.
Tomato (Solanum lycopersicum L.) is one of the most popular vegetables cultivated worldwide. It is rich in micronutrients, antioxidants, phenolics, flavonoids, vitamins, and essential trace elements. Currently, water stress limits the productivity of tomato worldwide, including Bangladesh (Habiba and Shaw, 2013). This study reports the results of a series of experiments we carried out to screen and identify biofilm-producing bacteria from the tomato rhizosphere grown in water stress-prone areas (Rajshahi, Dinajpur, and Bogura districts) of Bangladesh. We also characterized the matrix components, i.e., EPS, generated by different bacterial biofilms by means of Fourier transform infrared spectroscopy, scanning electron microscopy, and different binding assays. The expressions of multiple PGP-and biocontrol-related traits in these bacteria were examined in vitro. Moreover, selected biofilm-producing rhizobacterial strains were evaluated in vivo for their tomato growth promotion under water-deficit stress conditions. This study contributes toward an understanding about how biofilm-producing rhizobacteria may help to promote plant growth, suppress phytopathogens, and reduce water-deficit stress.

Collection of Rhizospheric Soil Samples and Isolation of Bacteria
Tomato plants (Solanum lycopersicum L.) were uprooted on soil during the flowering stage collected from three drought-prone areas of Bangladesh, including Rajshahi, Dinajpur, and Bogura.
Geographical positions (GPS) of the sampling areas are shown in Table 1. Each sample (excluding the aboveground parts) was transferred into a sterile polythene bag and then transported to the laboratory. During transportation, the cold chain was maintained by keeping the samples in an icebox. The collected samples were stored at 4 • C before the isolation of bacteria. In order to isolate rhizobacteria, firstly non-rhizosphere soil was discarded by hand shaking. Then, the sample (roots together with soil) was homogenized using a sterile mortar pestle in sterile distilled water, and a serial dilution was performed. After the serial dilution, 50 µL of each sample was spread on yeast extract peptone [YEP (1% of peptone, 0.5% of yeast extract, pH 6.8)] agar (1.5%) plates and incubated at 28 • C in a stationary condition. After a 36-h incubation, morphologically distinct (e.g., in size, shape, and color) colonies were transferred to fresh YEP agar plates using sterile toothpicks. In order to prepare pure cultures, the repeated-streaking method was followed.

Selection of Biofilm-Producing Rhizobacteria
To screen biofilm-producing rhizobacteria, each strain was initially inoculated in YEP broth and incubated at 28 • C under agitating conditions (160 rpm) for 6 h [until the optical density (OD 660 ) reached 0.6 to 0.8]. Then, 1 mL culture of each strain was collected and centrifuged at 12000 rpm for 10 min. The supernatant was carefully discarded. The pellet was resuspended in sterile distilled water and then diluted (ca. 10 5 colonyforming units (CFU) mL −1 )]. Afterward, 50-µL cultures were inoculated into glass test tubes (Pyrex, flat bottom, Glassco, United Kingdom) containing 5 mL salt-optimized broth plus glycerol (SOBG) medium (per liter: 20 g of tryptone, 5 g of yeast extract, 0.5 g of NaCl, 2.4 g of MgSO 4 · 7H 2 O, 0.186 g of KCl, and 50 mL of 40% glycerol, pH 7.0) and incubated at 28 • C under static conditions. After 72 h of incubation, airliquid (AL) and/or solid-air-liquid (SAL) biofilm-producing rhizobacteria were selected as described in Haque et al. (2012Haque et al. ( , 2015, and photographs were taken. The biomass of biofilms and the enumeration of bacterial cells coupled with biofilm matrices were determined as described in Mosharaf et al. (2018).

16S rRNA Gene Sequencing
The protocol described by Sambrook et al. (1989) was used for the collection of bacterial DNA. The 16S rRNA gene primers of 27F (5 -AGAGTTTGATCMTGGCTCAG-3 ) and 1492R (5 -GGTTACCTTGTTACGACTT-3 ) were used to amplify the 16S rRNA gene by polymeric chain reaction (PCR). To amplify the gene by PCR, the following conditions were set: initial DNA denaturation for 5 min at 94 • C, 35 cycles of denaturation for 30 s at 94 • C, annealing for 45 s at 57 • C, elongation for 1.5 min at 72 • C, and a final extension for 10 min at 72 • C. The QIAquick R Gel Extraction Kit (Qiagen, GmbH, Germany) was used to purify the PCR products and then sequenced by using 3500 Genetic Analyzer (Applied Biosystems). Using the BLASTN (Basic Local Alignment Search for Nucleotide) program, we compared the gene sequences of different bacterial strains against the sequences of bacteria available in NCBI (National Center for Biotechnology

Identification of the Matrix Components of the Biofilms
Fourier transform infrared (FTIR) spectroscopy, scanning electron microscopy (SEM), and Congo red and Calcofluor binding assays were used to identify the components of the biofilm matrices, i.e., EPS. For the FTIR analysis, the pellets were prepared as described in Mosharaf et al. (2018). Using the triglycine sulfate (TGS) detector, 450 to 4000 cm −1 was scanned (16 scans at 4 cm −1 resolution and at 0.2 cm sec −1 scanning speed and Calcofluor binding assays were performed as described in Haque et al. (2009Haque et al. ( , 2017.

Indole-3-Acetic Acid Production
Production of indole-3-acetic acid (IAA) was detected as described by Gordon and Weber (1951) with a few modifications. In brief, one single colony of each rhizobacterium was inoculated in 5 mL Luria-Bertani (LB) broth (1% of tryptone, 0.5% of yeast extract, 0.5% of NaCl, pH 7.0) supplemented with 0.2% of L-tryptophan (Bio Basic Inc., Canada) and incubated at 28 • C under shaking conditions (160 rpm). After a 48-hour incubation, 1 mL culture was collected and centrifuged at 14000 rpm for 10 min. Then, 500 µL supernatant was transferred to a sterile glass test tube and mixed with 1 mL Salkowski reagent (98 mL 35% perchloric acid and 2 mL 0.5 M FeCl 3 ). The test tubes were incubated at room temperature for 45 min in the dark. Development of pink color indicated positive for auxin production. Among the auxins, IAA (µg mL −1 ) production was quantified spectrophotometrically at OD 530 nm and compared to a standard curve prepared from commercial IAA (Duchefa Biochemie, Netherlands) with a concentration range of 0 to 1000 µg mL −1 .
Colorimetric IAA results were also verified and quantified using high-performance liquid chromatography (HPLC). In brief, all the rhizobacterial strains were grown as described in the colorimetric determination section. The cell-free extract was prepared by centrifugation followed by filtration through Einmalfilter (CHROMAFIL R Xtra PTFE-45/25, 0.45 µM, Macherey-Nagel, GmbH and Co. KG, Germany). The cell-free extract was acidified (2.5 to 3) using 1 N HCl and extracted with ethyl acetate (1:1 v/v), then the organic phase was collected. A rotary evaporator was used to evaporate the ethyl acetate fraction. The crude extract was dissolved in methanol (HPLC grade, Sigma-Aldrich, St. Louis, MO, United States) and kept at 4 • C for further use. In order to detect IAA specifically, Shimadzu Prominence HPLC (Japan) using a C18 analytical column (4.6 mm × 250 mm, 5 µm) was used. The column temperature was maintained at 25 • C, and methanol and 1% acetic acid (50:50 v/v) were used as the mobile phase at a flow rate of 1 mL min −1 with an injection volume of 20 µL (Myo et al., 2019). Detection was monitored at 254 and 280 nm, and data were evaluated using Lab Solutions software by comparing with the elution profiles of standard IAA (6.25, 12.5, 25, 50, and 100 µg mL −1 ) injected separately.

Ability of Nitrogen Fixation
Nitrogen fixation assays were performed according to Ker (2011) with minor modifications. In brief, initially, each rhizobacterium strain was grown in YEP broth under shaking conditions (120 rpm) at 28 • C for 16 h, then diluted to 10 7 CFU mL −1 . Two microliters (2 µL) of diluted culture of each rhizobacterium strain was spotted (4 spots plate −1 ) onto N-free solid LG agar (per liter: 10 g sucrose, 0.5 g K 2 HPO 4 , 0.2 g MgSO 4 ·7H 2 O, 0.2 g NaCl, 0.001 g MnSO 4 ·H 2 O, 0.001 g FeSO 4 , 0.001 g Na 2 MoO 4 ·2H 2 O, 5 g CaCO 3 , and 15 g agar) plates and incubated at 28 • C under static conditions for 10 days. For control, non-inoculated N-free solid LG agar plates were incubated at the same condition. Development of colonies indicated a positive nitrogen fixation.
The dinitrogenase reductase gene, nifH, the most widely used biomarker for the study of nitrogen-fixing bacteria, was analyzed by PCR. In this study, the primer pair of Ueda19F-GCIWTYTAYG GIAARGGIGG and Ueda19R-AAICCRCCRCAIACIACRTC was used to amplify a 389-bp nifH fragment (Ueda et al., 1995), and that of KAD3F-ATHGT IGGITGYGAYC CIAARGCIGA and DVVR-ATIGCRAAICCI CCRCAIACIACRTC was used to amplify a 310-bp nifH fragment (Ando et al., 2005). The reaction conditions with 100 ng of template DNA were as follows: one cycle at 95 • C (5 min); 30 cycles at 95 • C (30 s), 51 • C for Ueda19F and Ueda19R or 56 • C for KAD3F and DVVR (30 s) and 72 • C (1 min); plus one cycle at 72 • C (7 min). For negative control, the water was used instead of DNA.

PSI =
Diameter of bacterial colony + Diameter of halo zone Diameter of bacterial colony
For control, each bacterial strain was marked onto potassium aluminum silicate-deficient Aleksandrov agar plates. Inoculated plates were incubated at 28 • C for 7 days. The appearance of a clearing zone around the colonies indicated a positive potassium solubilization.

Zinc Solubilization
Zinc (Zn) solubilization assays were done as described in Saravanan et al. (2007) with a few modifications. In brief, 2 µL (ca. 10 7 CFU mL −1 ) culture of each rhizobacterium strain was spotted (4 spots plate −1 ) onto the basal agar medium (per liter: 10 g glucose, 1 g (NH 4 ) 2 SO 4 , 0.2 g KCl, 0.1 g K 2 HPO 4 , MgSO 4 ·7H 2 O, and 15 g agar, pH 7.0) containing 0.2% insoluble Zn from three sources, such as ZnO, ZnCO 3 , and Zn 3 (PO 4 ) 2 (Wako Pure Chemical Industries Ltd., Japan). For control, the bacterial strain was spotted onto the Zn-free agar plates. The inoculated plates were kept at 28 • C under static conditions. A clearing zone around the colonies indicated positive results. The Zn solubilization index (ZnSI) was also calculated after 7 days using the same equation as for PSI.

Production of Siderophores
Overlay chrome azurol S (O-CAS) medium was used to detect siderophores (Shin et al., 2001). Initially, each bacterial strain was grown in YEP broth under shaking conditions at 28 • C until OD 660 reached 0.6-0.8. Then, 2 µL (ca. 10 7 CFU mL −1 ) diluted culture of each rhizobacterium strain was spotted (1 spot plate −1 ) onto the center of the LB agar plates and incubated at 28 • C for 20 h. Later, 10 mL O-CAS broth was applied over those LB agar plates and incubated at 28 • C under static conditions. Development of colors (e.g., purple, orange, or yellow) indicated positive results. After a 10-hour incubation, photographs were taken.

Production of Volatile Compounds Acetoin
Qualitative acetoin production was examined as described in Dye (1968). In brief, each bacterial culture (16-h old) was inoculated in 5 mL yeast extract salt broth (per liter: 0.5 g of NH 4 H 2 PO 4 , 0.2 g of MgSO 4 ·7H 2 O, 5.0 g of NaCl, and 5.0 g of glucose) and incubated at 28 • C. After a 96-h incubation, 1 mL culture of each bacterium strain was transferred to a sterile glass test tube and 600 µL 5% (w/v) alpha-napthol in absolute alcohol was added and shaken gently. Occurrence of a crimson to ruby color at the top or throughout the mixtures within 4 h indicated a positive acetoin production.

Indole
Qualitative indole production was assessed as described in Lelliott and Dickey (1984). Initially, the broth (per liter: 10 g of tryptone, 1 g of L-tryptophan, and an adequate amount of distilled water) was prepared and autoclaved. Then, 50 µL culture (16-hour old) was inoculated in glass test tubes containing 5 mL broth and incubated at 28 • C under shaking conditions (160 rpm). After a 96-hour incubation, 500 µL Kovac's reagent (HiMedia, India) was added to the culture and shaken gently. The development of a dark-red color on the surface of the medium indicated a positive indole production.

Ammonia
Qualitative ammonia production was examined by the method described in Dinesh et al. (2018) with a few modifications. In brief, each bacterial strain was grown in YEP broth under shaking conditions for 16 h at 28 • C. Then, 50 µL (10 8 CFU mL −1 ) culture of each bacterium strain was inoculated in glass test tubes containing 5 mL peptone water (Sigma-Aldrich, St. Louis, MO, United States) and incubated at 28 • C. After a 72h incubation, 1 mL Nessler's reagent (Sigma-Aldrich, St. Louis, MO, United States) was added. The development of yellow to brown color indicated a positive ammonia production.

Hydrogen Cyanide (HCN)
Each bacterial colony was streaked on LB agar plates containing 0.45% glycine. Two sterilized filter papers were soaked in alkaline picrate solution (0.25% picric acid in 1.25% sodium carbonate) (Lorck, 1948) and then placed on the lids of petri plates. The plates were sealed with parafilm and incubated at 28 • C under static conditions for 24 h. Positive HCN production resulted in a color development on the filter papers from yellow to a reddish-brown.

Production of Hydrolytic Enzymes
Qualitative ACC deaminase activity was analyzed as described in Dworkin and Foster (1958) and in Penrose and Glick (2003). In brief, each bacterial strain was grown in LB broth until OD 660 reached 0.6 to 0.8 and then diluted (ca. 10 7 CFU mL −1 ).  ), and 1.8% bacto agar]. As a negative control, 100 µL suspension was spread onto ACC-deficient minimal DF salt agar plates. The plates were incubated at 28 • C for 4 days. Colonies formed on the plates were considered as ACC-deaminase producers. Qualitative catalase, oxidase, gelatinase, and arginine dihydrolase production was determined as described in Hayward (1992); Shekhawat et al. (1992), Schaad (1988), and Thornley (1960), respectively. Lipase, cellulase, and protease assays were performed as described in Dinesh et al. (2018).

Abiotic-Stress Tolerance
The maximum drought stress was reported to be achieved by adding 25% polyethylene glycol (PEG) 6000 into the broth (Vardharajula et al., 2011). Thus, for drought-tolerance stress, all the bacterial strains were grown in LB broth containing 25% PEG 6000 at 28 • C under shaking conditions. After a 24-h incubation, the optical density was measured by a spectrophotometer. A bacterial OD 660 ≥ 0.1 was considered as drought tolerance. Biofilm-producing rhizobacterial strains were also tested for their ability to grow on LB agar plates at different temperatures (37, 42, and 50 • C), varying pHs (pH 4.0, 7.0, 8.0, 9.0, and 10.0), and different salinities (5, 10, and 20% NaCl). For control, uninoculated plates/test tubes were incubated. Formation of bacterial colonies on the agar plates under the tested conditions after 72 h was recognized as positive for the respective test.

In vitro Antagonistic Activities Against Pathogenic Bacteria
All these rhizobacteria strains were tested for their antagonistic activities against Xanthomonas campestris pv. campestris ATCC 33913 (causal agent of bacterial leaf spot in tomato), Ralstonia solanacearum ATCC R 11696 TM (causal agent of wilt in tomato), and Pectobacterium carotovorum subsp. carotovorum PCC8 (causal agent of soft rot in tomato, accession number KX098362) as described in Furuya et al. (1997). In brief, a single colony of each rhizobacterium was spotted onto an LB agar plate and incubated at 28 • C under static conditions for 36 h. Then, a sheet of sterilized filter paper was soaked in chloroform and then placed in each lid of the Petri dish and incubated at room temperature for 2 h to kill the rhizobacterium. Then, 5 mL melted water agar (1.5% at 50 • C) containing a suspension of each plant pathogenic bacterium [ca. 10 8 CFU mL −1 ] was poured onto the plates and incubated at 28 • C for 48 h. For control, only 5 mL melted water agar (1.5% at 50 • C) was poured. The formation of an inhibition zone around the growing region of rhizobacterium was considered positive for the biocontrol agent.

Experimental design and treatments
The pot experiment was laid out in a complete randomized design (CRD) having nine treatments with four replications. The treatments were T 1 (applied only a standard dose of N:P:K but no bacteria were inoculated, herein termed as non-inoculated control), T 2 (Pseudomonas azotoformans ESR4), T 3 (P. poae ESR6), T 4 (P. gessardii ESR9), T 5 (P. cedrina ESR12), T 6 (P. chlororaphis ESR15), T 7 (Stenotrophomonas maltophilia ESR20), T 8 (P. veronii ESR21), and T 9 (Bacillus aryabhattai ESB6). Information regarding these bacterial strains on promotion of plant growth and alleviation water-deficit stress was not available in the literature. Therefore, these bacterial strains were selected for this study. The standard dose of N:P:K (as a source of urea, triple super phosphate, and murate of potash, respectively) for the high-yield goal of tomato were calculated using the model of Bangladesh Agricultural Research Council (Fertilizer Recommendation Guide, 2012).

Root bacterization, seedling transplantation, and imposing water-deficit stress
For root bacterization, selected rhizobacterial strains were grown in 100 mL YEP broth at 28 • C in agitated conditions (150 rpm) until OD 660 reached 0.6 to 0.8. Then, 1 mL culture of each bacterial strain was harvested and centrifuged at 12000 rpm for 10 min. The supernatant was discarded. The pellets were resuspended in phosphate buffer (pH 7.0) and diluted to ca. 10 8 CFU mL −1 . Healthy-looking, uniform-sized seedlings (20-dayold) were carefully uprooted. Afterward, the roots of the seedlings were immersed in bacterial suspension treatment wise (i.e., T 2 to T 9 ). For the non-inoculated control (i.e., T 1 ), the roots of the seedlings were immersed only in the phosphate buffer. After a 2hour incubation, one seedling was transplanted in each pot (four pots for each treatment) and the pots were kept in an open-field condition. When required, the pots were transferred in the rain shelter. The pot was watered with sterile deionized water to field capacity [when water was leached through bottom holes of the pots, this was considered maximum field capacity (Chandra et al., 2018;Jochum et al., 2019)] every day up to 45 days post planting (DPP). In order to impose water stress, the soil in the root region of each plant was carefully loosened. On 46 DPP, 50 mL bacterial suspension (suspended in phosphate buffer, pH 7.0) containing 10 8 CFU mL −1 bacterial cells was applied to the rhizosphere region of each plant (T 2 to T 9 treatments), while to the control (T 1 ), only 50 mL phosphate buffer was applied. After treatment, the watering in the pot was immediately stopped. Water stress was continued up to the 12th day (58 DPP).

Evaluation of plant growth and biochemical parameters
Plant height, number of primary branches, number of leaves, maximum leaf length, and maximum leaf width were recorded on the 12th day (58 DPP) of water stress. For biochemical parameters, the leaves were collected on this day. Then, each plant was carefully uprooted. The shoots and roots were separated, oven dried at 70 • C for 72 h, and then weighed. To quantify the number of colonized bacteria in the rhizosphere (roots with soil), non-rhizosphere soil was discarded from the roots by hand shaking. Then, 1 g sample (roots with soil) was taken from the composite sample, homogenized, and serial diluted. Then, 100 µL diluted sample was spread on YEP agar plates (4 replications each). After a 48-hour incubation at 28 • C, bacterial populations were counted.

Relative water content
Three fully expanded 3rd leaves from the top of the main stem were collected from each treatment. Five leaf discs (1 cm 2 ) from each treatment were weighed. Then, the leaf discs were placed in distilled water and the turgid weight recorded after 24 h at 4 • C.
The discs were then oven-dried at 72 • C until constant weights were attained, and the dry weight was determined. The relative water content (RWC) was calculated as follows:

Chlorophylls and carotenoids
A UV-Vis spectrophotometer (Ultrospec 3000, Pharmacia Biotech, Cambridge, United Kingdom) was used to quantify chlorophyll (chl) a, chl b, total chl, and carotenoids from fresh leaf samples as described in Khan et al. (2017) with a few modifications. In brief, 50 mg fresh leaf sample was transferred into a glass test tube containing 5 mL 90% acetone, covered tightly, and kept in the dark at room temperature. After 48 h, the absorbance was measured at 663, 645, and 470, respectively. The result was expressed as mg g −1 fresh weight (FW). The formulae for computing chl a, chl b, total chl, and carotenoids were: where A (663, 645, 470) represents the optical density of the chlorophyll extract at the wave length of 663 nm, 645 nm, and 470 nm; V is the final volume (mL) of 90% acetone with chlorophyll extract; and W is the weight of the fresh leaf sample in g.

Electrolyte leakage
Electrolyte leakage (EL) was determined as described in Lutts et al. (1996) with a few modifications. In brief, 1-cm 2 leaf discs (5 discs/treatment) were transferred into glass test tubes containing 10 mL distilled water. Then, electrical conductivity (EC) was recorded at room temperature. The sample-containing test tubes were kept in a water bath at 40 • C for 30 min, and EC was measured. Afterward, the samples were boiled at 100 • C, and EC was recorded after 30 min. Finally, EL was calculated using the following formula:

Measurement of Lipid Peroxidation (MDA Level)
The MDA level was quantified as described in Vemanna et al. (2017) with a few modifications. In brief, 0.15 g leaf sample was homogenized in 2.5 mL of 5% (w/v) trichloroacetic acid (TCA) and 0.25% of thiobarbituric acid (TBA) and then centrifuged at 15000 rpm at 4 • C for 10 min. The supernatant was mixed with an equal volume of TBA (0.5% in 20% TCA) and then boiled for 20 min at 100 • C. The reaction was stopped by incubation on ice. The absorbance of the supernatant was measured at 532 and 600 nm, respectively. Nonspecific turbidity was corrected by subtracting the absorbance at 600 nm. The MDA content was calculated using its molar extinction co-efficient of 155 mM −1 cm −1 and expressed as µmol g −1 FW.

Proline Content
The proline content was estimated as described in Bates et al. (1973) with a few modifications. In brief, 0.1 g leaf was homogenized in 10 mL 3% aqueous sulfosalicylic acid (Sigma-Aldrich, St. Louis, MO, United States) and centrifuged at 15000 rpm for 15 min. Then, 2 mL acid-ninhydrin and 2 mL glacial acetic acid were mixed with 2 mL supernatant and cooked in a water bath at 100 • C. After 30 min, the test tubes were transferred on ice to stop the reaction. Then, 4 mL toluene (Wako Pure Chemicals, Japan) was added. After a 10-min incubation at room temperature, the reaction mixture was vigorously mixed and the absorbance of the upper layer of the mixture containing toluene was measured at 520 nm using an UV-Vis spectrophotometer. The amount of proline was quantified from the standard curve using the following equation: Proline (µg per g fresh weight) = (µg per mL Proline × Volume of Toluene ×Volume of Sulfosalicylic acid) 0.1 × 115.5 where 0.1 is the sample weight (g) and 115.5 is the molecular weight of proline.

Catalase Activity
In order to determine the catalase activity, 0.2 g leaf sample was homogenized in phosphate buffer (100 mM; pH 7.2) and then centrifuged at 14000 rpm for 15 min at 4 • C. The supernatant (0.1 mL) containing the enzyme extract was added to a 1-mL

Statistical Analysis
All the assays were performed in CRD with at least three replications and repeated at least twice unless otherwise stated. However, the effect of water-deficit stress on plant growth and the expression of the biochemical parameters were conducted only once. For the different parameters collected from the bacterial treatments, a one-way ANOVA test was done. The distribution of the data sets was normal and confirmed by the Shapiro-Wilk test. Heteroscedasticity of the data sets was checked by the Bartlett's test and found homogenous/homoscedastic. The graphical distribution of datasets was analyzed by Q-Q plot. ANOVA, distribution of data, homogeneity of variance, and mean comparison of treatment effects were analyzed using the R software version 3.3.6. The Fisher's least significant difference test was applied to compare the means.

Isolation of Biofilm-Producing Rhizobacterial Strains
Based on colony morphology, 78 (seventy-eight) rhizobacterial strains (26 strains from each area) were isolated (data not shown). All these strains were tested for their ability to form AL and/or SAL biofilms on the glass test tubes containing SOBG broth at 28 • C under static conditions. Only 21 (26.92%) strains produced delicate to stout AL biofilms after 72 h of incubation ( Figure 1A). The remaining strains formed neither AL nor SAL biofilms even after 7 days of incubation (data not shown All these biofilm-producing strains produced rough-surfaced biofilms ( Figure 1A). Among these strains, ESR6, ESR12, and ESB9 built very thick and rigid biofilms, and the associated bacterial cells were not dispersed when the aggregates were agitated. Conversely, ESR9, ESR20, ESD16, and ESB18 produced very thin and fragile biofilms, and the cells of these biofilms were easily dispersed when disturbed. Furthermore, those biofilms generated by ESR7, ESR16, ESR23, ESD3, ESD8, and ESB22 were denser and stronger than the biofilms constructed by ESR9, ESR20, ESD16, or ESB18. Interestingly, the biofilms formed by ESR3, ESR13, ESR15, ESR21, ESR25, and ESB6 did not completely cover the surface of the standing culture. Thus, the biofilm characteristics, including thickness, strength, and covering the surface depended on the bacterial strains.

Identification of Biofilm-Producing Rhizobacteria
All the biofilm-producing rhizobacterial strains were identified based on 16S rRNA gene sequencing (  Table 1. A phylogenetic tree was also constructed using the 16S rRNA gene sequence data (Figure 2).

FTIR Spectroscopy
The matrix components of the biofilms produced by the identified rhizobacterial strains used in this study were not yet characterized. In the present study, we characterized the biofilm matrix formed by these bacterial strains by the FTIR spectroscopy (Figure 3). All the matrices were dominated by protein compounds, producing peaks at amide I (1600-1700 cm −1 ), amide II (1500-1600 cm −1 ), and amide III (1240-1350 cm −1 ) regions. Also, these bacterial strains generated a substantial amount of polysaccharides which produced intense peaks near 900-1150 cm −1 . Moreover, the peaks within the 1220-1250cm −1 band region demonstrate the presence of nucleic acids in the biofilm matrices (Naumann, 2001). Nevertheless, the 2800-2970-cm −1 , 3200-cm −1 , and 2955-cm −1 domains also indicate the presence of lipids (Mosharaf et al., 2018), amide A of peptidoglycan (Naumann et al., 1982), and amide B of peptidoglycan (Naumann, 2001), respectively. Thus, proteins, polysaccharides, nucleic acids, lipids, and amide A-and amide B of peptidoglycan are all components of the biofilm matrices for these bacteria.

Congo Red and Calcofluor Binding Assays
All these bacterial strains were tested for their abilities to bind Congo red. Indeed, all these strains were able to bind Congo red and developed the typical red, dry, and rough (rdar) phenotype on Congo red agar plates ( Figure 4A). The results suggested that they produced both curli fimbriae and cellulose (Römling, 2005;Milanov et al., 2015). On Calcofluor agar plates, P. fluorescens ESR7, P. cedrina (ESR12, ESR16, and ESR23), S. maltophilia ESR20, B. cereus (ESD3, ESD21, and ESB22), and B. aryabhattai ESB6 and P. parfulva ESB18 exhibited a strong fluorescence (Figure 4B), while the remaining bacterial strains only showed a weak or moderate fluorescence ( Figure 4B). Hence, all these strains expressed cellulose (Zogaj et al., 2001;Römling, 2005). However, the quantity of cellulose production might vary between these bacterial strains.

Production of IAA
The qualitative assessment of IAA using Salkowski's reagent showed that except for B. cereus ESD3 and S. saprophyticus ESD8, all the analyzed rhizobacterial strains exhibited the color change ranging from light pink to dark red, suggesting the positivity for production of auxin-related compounds (Glickmann and Dessaux, 1995) including IAA (Supplementary Figure S1). When quantified using a spectrophotometer, the IAA production ranged from 4.45 to 59.57 µg mL −1 ( Table 2). The highest amount of IAA was synthesized by B. horikoshii ESD16 (59.57 µg mL −1 ) followed by P. poae ESR6 (31.17 µg mL −1 ).

FIGURE 3 | Detection of organic macromolecules constructed by different biofilm-producing PGPR by FTIR.
The least amount of IAA was produced by B. cereus ESB22 (4.45 µg mL −1 ). We also confirmed IAA production quantitatively by the HPLC system. The results showed that standard IAA (Duchefa, Netherland) elution in HPLC developed a major peak at a retention time of 6.864 min, while IAA extracted from the isolates developed a sharp peak at a retentions time ranging from 6.68 to 6.82 min. We found that the detection sensitivity of IAA by HPLC is 2-to 3-fold higher than spectrometric quantification. In HPLC, we quantified IAA ranging from 19.3 to 71.7 µg mL −1 ( Table 2). The isolate B. aryabhattai ESB6 and P. cedrina ESR16 synthesized the highest and lowest amount of IAA, respectively, as the spectrometric analysis IAA was not detected from the isolates B. cereus ESD3 and S. saprophyticus ESD8. The representative chromatogram of standard IAA (50 µg mL −1 ) and P. cedrina ESR12, P. chlororaphis ESR15, and B. aryabhattai ESB6-extracted IAA is presented in Figure 6.

Fixation and Solubilization of Nutrients
All the bacterial strains formed colonies on N-free solid LG agar plates ( Table 2). Genomic DNAs of these bacterial strains were also amplified for the nifH gene by PCR. The primer pairs of Ueda19F and Ueda19R and KAD3-F and DVV-R were amplified at 389 and 310 bp, respectively ( Table 2 and Supplementary Figure S2). Thus, all these rhizobacterial strains have nitrogen fixation activity.
All the rhizobacterial strains were evaluated for their abilities to solubilize P from two sources including tricalcium phosphate and rock phosphate. Except P. parafulva ESB18 and B. cereus (ESD3 and ESB22), all other strains formed a clear halo zone around the colonies on NBRIP agar containing 0.5% tricalcium phosphate (data not shown). All the rhizobacterial strains also developed a clear halo zone around the colonies on NBRIP containing 0.8% rock phosphate except B. megaterium ESB9, P. parafulva ESB18 and B. cereus ESB22 (data not shown). Thus, most of the rhizobacterial strains are P solubilizers. The PSI value of tricalcium phosphate and rock phosphate was calculated, fluctuating from 2.11 to 6.51 and 2.00 to 4.67, respectively ( Table 2). In the case of tricalcium phosphate, P. veronii ESR13 exhibited the highest PSI value (6.51) which was statistically similar with P. chlororaphis ESR3 (6.43) and P. cedrina ESR23 (6.47). Conversely, the lowest PSI value (2.11) was noted in B. megaterium ESB9. In the case of rock phosphate, the maximum PSI value (4.67) is displayed in P. chlororaphis ESR15, B. cereus ESD3, and B. horikoshii ESD21 which did not significantly differ with that of P. poae ESR6 (4.23), P. fluorescens ESR7 (4.08), P. gessardii ESR7 (4.19), P. veronii ESR13 (4.50), P. cedrina ESR16 (4.21), and S. saprophyticus (4.08). Furthermore, when the P-solubilizing strains were inoculated in NBRIP broth supplemented with 0.5% tricalcium phosphate or 0.8% rock phosphate at pH 7.0 and incubated at 28 • C for 96 h under shaking conditions, the pH value of this broth decreased from 7.0 to a pH of 4.5-5.5 (data not shown). This result indicated that these rhizobacterial strains secreted organic acids to solubilize tricalcium phosphate and rock phosphate (Ahemad and Kibret, 2014).

Expression of Siderophores
Developing varying colors on the O-CAS agar medium is linked with the expression of different types of siderophores: a purple color is associated with catechol-type siderophores, a light orange/orange color is related with hydroxamate-type  siderophores, a light yellow color is connected with carboxylate-type siderophores, and a yellow color is linked with both hydroxamate-and carboxylate-type siderophores (Pérez-Miranda et al., 2007). In this study, P cedrina (ESR12, ESR16, and ESR23), S. maltophilia ESR20, and B. horikoshii ESD16 were incapable of changing colors (data not shown). In contrast P. chlororaphis (ESR3 and ESR15), P. azotoformans ESR4, P. poae ESR6, P. fluorescens (ESR7 and ESR25), P. gessardii ESR9, P. veronii ESR13, and B. cereus ESD21 produced a light orange color (Figure 7), suggesting that these bacterial strains produced hydroxamate-type siderophores. B. aryabhattai ESB6, B. megaterium ESB9, B. cereus ESB22, and P. parafulva ESB18 developed a yellow color, indicating that they generated both hydroxamate-and carboxylate-type siderophores. P. veronii ESR21, B. cereus ESD3, and S. saprophyticus ESD8 induced a light yellow color, specifying that they constructed carboxylate-type siderophores (Figure 7).

Production of Volatile Compounds
Volatile compounds produced by the different rhizobacterial strains are shown in  (Table 3 and Supplementary Figure S4).

Production of Hydrolytic Enzymes
The hydrolytic enzyme production in the different rhizobacteria is shown in

Biocontrol of Phytopathogenic Bacteria in vitro
All these rhizobacterial strains were tested for their antagonistic activities against three in tomato catastrophic phytopathogenic bacteria X. campestris pv. campestris ATCC 33913, R. solanacearum ATCC R 11696 TM , and P. carotovorum subsp. carotovorum PCC8 (Figure 8). None of the rhizobacterial strains were able to restrict the growth of P. carotovorum subsp. carotovorum PCC8 (data not shown). Interestingly, the growth of X. campestris pv. campestris ATCC 33913 was inhibited by P. cedrina ESR12, P. chlororaphis ESR15, and B. cereus ESD3,

Abiotic Stress Tolerance
The abiotic stress tolerance performance of different rhizobacteria is depicted in

Effect of Drought Stress on Biofilm Formation
Some selected biofilm-producing PGPR were further examined for their abilities to form biofilms on SOBG containing 25% PEG 6000 which mimics water-stress conditions (Supplementary Figure S5). Compared to PEG-deficient SOBG, biofilm formation was only slightly reduced in P. azotoformans ESR4 and P. poae ESR6. Interestingly, P. gessardii ESR9, P. cedrina ESR12, P. chlororaphis ESR15, S. maltophilia ESR20, and B. aryabhattai ESB6 produced even thicker biofilms on SOBG containing 25% PEG 6000 as compared to PEG-deficient SOBG. Importantly, most of the bacterial cells were attached to the biofilms in SOBG containing 25% PEG 6000, while the number of planktonic cells was higher in PEG-deficient SOBG. These results suggest that all the tested rhizobacteria might be able to form biofilms in plants under water-deficit conditions but the quantity might vary.

In vivo Plant Growth Promotion Under Water Deficit-Stressed Conditions
We also analyzed the effect of biofilm-producing PGPR on tomato plants grown under water-deficit stress. The noninoculated plants were wilted, and the leaves started curling on the 4th day (50 DPP) of water stress. This type of stress responses occurred on the 6th day of water stress when plants were inoculated with P. azotoformans ESR4 and P. poae ESR6 (data not shown). However, only the leaf-curling symptom but not the wilting one was observed in P. gessardii ESR9-, P. cedrina ESR12-, and S. maltophilia-treated plants on the 11th day (57 DPP) of water stress (Figure 9). Interestingly, P. chlororaphis ESR15and B. aryabhattai ESB6-inoculated plants did not display any wilts and/or leaf-curling symptoms even after 12 days of water stress (Figure 9). This indicates a difference among the PGPR to mitigate water-deficit stress in vivo in field-grown tomato plants.
Data on plant growth-related traits are shown in Table 5. The highest increase in plant height (16.7%) was observed in P. azotoformans ESR4-inoculated plants, though this was statistically akin with S. maltophilia ESR20 and B. aryabhattai ESB6, showing 13.4 and 12.7% of height increase, respectively. Plant height was also improved by inoculation with P. poae ESR6, P. gessardii ESR9, P. cedrina ESR12, P. chlororaphis ESR15, and P. veronii ESR21, displaying a 10.7, 8.7, 8.1, 6.7, and 2.7% increase compared to non-inoculated plants, respectively. Also, the number of primary branches provides an evidence for the health of the tomato plants. The highest number of primary branches (7.0 plant −1 ) was found in B. aryabhattai ESB6-inoculated plants followed by P. azotoformans ESR4 (6.67 plant −1 ), P. poae ESR6 (6.67 plant −1 ), and S. maltophilia ESR20 (6.33 plant −1 ). In total, inoculation with all the biofilmproducing PGPR resulted in an increase in the number of primary branches, though in varying degrees.
Similarly, the number of leaves plant −1 incredibly amplified in P. azotoformans ESR4-inoculated plants (50% increase compared to non-inoculated plants), which was not substantially different to B. aryabhattai ESB6 (46.7%), P. poae ESR6 (36.9%), and S. maltophilia ESR20 (34.8%). Also, the application of other PGPR increased the number of leaves. Comparably, also the leaf width was higher in those plants treated with the bacteria as compared to the non-inoculated plants.
The root and shoot dry matter plant −1 were also significantly and remarkably higher after the application of the different biofilm-producing rhizobacterial strains. The highest root dry matter weight increase plant −1 (2.89 g) was obtained in P. azotoformans ESR4-inoculated plants (49.7% increase compared to non-inoculated plants). Interestingly, inoculation with the same strain also improved the shoot dry matter weight plant −1 by 30.1%, though the increase was even higher when the plants have been treated with B. aryabhattai ESB6 (49.5% increase as compared to the non-inoculated plants). In this study, the largest bacterial populations were detected, as expected, in the rhizosphere (roots along with soil) of bacterized plants and not in the non-inoculated plants.

Production of Malondialdehyde (MDA) and Electrolyte Leakage (EL)
The level of MDA represents the lipid peroxidation of membrane lipids and designates as a marker of oxidative injury. In this study, non-inoculated plant leaves accumulated MDA around 116.77 ± 2.54 µmol g −1 FW. Importantly, MDA accumulation was lowered by 15.5% when inoculated with S. maltophilia to FIGURE 9 | Effect of bacterial inoculation on growth of tomato plants under water deficit stressed conditions. Photographs were taken on 12th day of water stress. T 1 = Without bacterial inoculation, T 2 , T 3 , T 4 , T 5 , T 6 , T 7 , T 8 , and T 9 = Tomato plants inoculated with P. azotoformans ESR4, P. poae ESR6, P. gessardii ESR9, P. cedrina ESR12, P. chlororaphis ESR15, S. maltophilia ESR20, P. veronii ESR21, and B. aryabhattai ESB6, respectively.

Leaf Proline and Catalase Activity
In response to water stress, proline accumulation is common in plant leaves for osmotic adjustment. Also, proline acts as a protective agent of enzymes and antioxidants. In this study, application of P. poae ESR6 resulted in a significant increase in proline content (38.2 µg g −1 FW) followed by non-inoculated plants (30.4 µg g −1 FW) ( Table 6). However, proline content was reduced by 69.4, 63.3, 62.7, 57.2, 47.7, and 23.5% by inoculation of B. aryabhattai ESB6, P. gessardii ESR9, P. chlororaphis ESR15, P. cedrina ESR12, P. veronii ESR21, and S. maltophilia ESR20, respectively, as compared to non-inoculated plants. The catalase (CAT) activity, another plant protectant enzyme, remarkably changed by application of rhizobacteria ( Table 6). The CAT activity was increased by 10.6% for P. gessardii ESR9 to up to 72.1% when inoculated with B. aryabhattai, respectively, as compared to non-inoculated plants.
The characterized rhizobacterial biofilm matrices (i.e., EPS) predominantly contain proteins, polysaccharides, nucleic acids, and lipids (Figure 3). The protein peaks (amide I, II, and III) were much higher than the peaks of polysaccharides, nucleic acids, and lipids (Figure 3). Our results align with the results of Conrad et al. (2003). Indeed, these protein components show the characteristic IR band through C = O stretching at the amide I region, in contrast to the C-N bending and the N-H stretching at the amide II region and the amide III region (Naumann, 2001;Ojeda et al., 2008;Haque et al., 2014;Mosharaf et al., 2018). On the other hand, the band region of polysaccharides principally resulted from a stretching vibration of C-C and C-O bonds and the deformation of C-O-H and C-O-C bonds (Naumann, 2001). Negatively charged functional groups of biofilm matrices were reported to bind with heavy metals, leading to heavy metal remediation from the contaminated environment (Teitzel and Parsek, 2003;Xie et al., 2015). Moreover, various functional groups play a pivotal role in bacterial root colonization (Gupta et al., 2017).
The here analyzed extracellular polymeric substances (EPS) are highly hydrated polymers synthesized by different microorganisms including bacteria (Costa et al., 2018). The contents of the EPS play many important roles such as adhesion, soil aggregation, cohesion, retention of water, protective barrier, absorption of organic compounds, absorption of inorganic ions, nutrient source, exchange of genetic information, electron donor or acceptor, export of cell components, sink of excess energy, and the binding of enzymes (Flemming and Wingender, 2010). Hepper (1975) reported that bacterial EPS creates a microenvironment that holds water and dries up more slowly, thus protecting the bacteria and plant roots against extreme desiccation. Furthermore, Roberson and Firestone (1992) have shown that EPS from Pseudomonas strains increased the water-holding capacity in sandy soil. EPS released from biofilms also improved the permeability by creating soil aggregation and maintaining a higher water potential around the roots, thereby increasing the uptake of nutrients by plants and protecting them from water shortage (Selvakumar et al., 2012). For example, inoculation of sunflower seedlings with EPS-producing P. putida GAP-P45 showed an augmented plant growth and increased the survival rates under drought-stressed condition (Sandhya et al., 2009). EPS production was reported to be positively correlated with drought tolerance of cowpea (Rodríguez-Navarro et al., 2007) and foxtail millet (Niu et al., 2018). In our present study, we observed that rhizobacterial biofilm matrices, i.e., EPS, are sponge-like and fibrous (Figure 5). Thus, these rhizobacterial EPS may hold water and/or increase the water-holding capacity in the soil to assist the bacteria and the plant roots under water-deficit stress. In the present study, inoculation of tomato plants with biofilm-producing PGPR increased plant growth ( Table 5) and resulted in more water-deficit stress tolerance than non-inoculated plants (Figure 9).
Plant growth-promoting rhizobacteria produce the phytohormones which are directly involved in plant growth promotion (Patten and Glick, 1996). Among these phytohormones, IAA increases the number of root hairs and lateral roots, and this increased root surface area promotes nutrient and water uptake by the plants even under drought stress (Gray and Smith, 2005;Mantelin and Touraine, 2009;Ruzzi and Aroca, 2015;Saikia et al., 2018;Ojuederie et al., 2019). Thus, a constitutive synthesis of IAA is required for proper plant growth and development. In contrast, primary root length is stimulated at low concentrations of IAA, while increased at high IAA concentration promotes lateral root formation but decreased primary root lengths (Vacheron et al., 2013). Ethylene, another phytohormone, also controls different physiological responses including senescence, ripening of fruits, initiation of roots, and the abscission and inhibition of storage organ formation. Overproduction of ethylene significantly reduces the root and shoot development (Glick et al., 1998). However, ACC deaminases in PGPR reduced the ethylene production leading to an increased crop productivity under water deficit-stressed conditions (Saikia et al., 2018;Ojuederie et al., 2019). In this study, biofilm-producing PGPR expressed IAA and ACC deaminases in vitro (Tables 2, 3). Importantly, in vivo grown tomato plants bacterized with biofilm-producing P. azotoformans ESR4, P. poae ESR6, P. gessardii ESR9, P. cedrina ESR12, P. chlororaphis ESR15, S. maltophilia ESR20, P. veronii ESR21, and B. aryabhattai ESB6 showed an enhanced plant growth in terms of plant height, number of primary branches, number of leaves, maximum leaf length and width, root and shoot dry matter production ( Table 6), and withstanding water-deficit stress (Figure 9). Therefore, the isolated PGPR are capable of attenuating water-deficit stress by synthesizing IAA and ACC deaminases, thereby reducing the accumulation of ethylene in tomato.
Leaf RWC is the index of physiological water status of plants that improves their tolerance to drought stress (Kadioglu et al., 2011). Reduction of leaf RWC was reported to positively correlate with closure of stomata which leads to decreased CO 2 assimilation in plants (Wezel et al., 2014). Under severe drought stress, a reduced RWC decreased leaf pigments such as chl a and b, resulting in a reduced photosynthesis rate (Kawamitsu et al., 2000;Manivannan et al., 2007;Mafakheri et al., 2010). However, in stressed plants, increased chlorophyll concentration can be used as an index of tissue tolerance (Lutts et al., 1996). Heidari and Golpayegani (2012) reported that leaf chlorophylls were increased in basil (Ocimum basilicum) after inoculation of PGPR even under water deficit-stressed conditions. Saline or water stress-induced osmotic conditions enhanced the chloroplast development (Chang et al., 1997). In this study, a significant reduction of leaf RWC was observed in non-inoculated plants, while it only slightly decreased in plants inoculated with P. azotoformans ESR4 and P. poae ESR6. However, all other PGPR-inoculated plants did not show any reduction in leaf RWC ( Table 6). It was also observed that the photosynthetic leaf pigment contents (such as chl a, chl b, and total chl) in all PGPRinoculated plants were remarkably higher than in non-inoculated plants ( Table 6). Therefore, biofilm-producing PGPR efficiently retained the leaf RWC and the photosynthetic efficiency of plants under water-limiting conditions.
In the presence of environmental stimuli, i.e., drought, high temperature, and salt, plant cells suffer oxidative stress by lipid peroxidation resulting in tissue damages (Catal, 2006). MDA is a product of lipid peroxidation and is used as an indicator of oxidative stress (Esterbauer and Cheeseman, 1990). In this study, non-inoculated plants and the plants inoculated with P. azotoformans ESR4 and P. poae ESR6 produced more MDA as compared to other biofilm-producing PGPRtreated plants ( Table 6). Similar trends were also observed for electrolyte leakage, a measure for cell death. Proline is one of the common compatible osmoprotectants and a stress marker which influences adaptive responses (Maggio et al., 2002;Mafakheri et al., 2010). This component shields cells from dehydration, detoxifies stressed cells from ROS, maintains the integrity of subcellular structures and enzymes, and protects the transcriptional and translational machinery of plants (Maggio et al., 2002). In this study, P. poae ESR6inoculated plants synthesized remarkably higher amounts of proline which was statistically similar with non-inoculated plants, suggesting that non-inoculated-and P. poae ESR6-inoculated plants are more stressed than other plants ( Table 6). The enzymatic and non-enzymatic defense system showed that PGPR-inoculated plants accumulated under water-deficit stress more CAT and carotenoids than non-inoculated control plants ( Table 6). Kohler et al. (2008) reported that inoculation with P. mendocina and Glomus intraradices significantly increased CAT in lettuce plants under severe drought conditions. Therefore, the experimental results showed that our identified biofilm-producing PGPR attenuate water-deficit stress and promoted plant growth which might be associated with multiple mechanisms such as biofilm formation, production of EPS, synthesis of IAA and ACC deaminases, increased enzymatic and non-enzymatic defense systems, and the improved solubilization of nutrients.

CONCLUSION
The tomato plants grown in drought-prone ecosystems of Bangladesh are associated with a variety of rhizobacteria with multifarious plant growth-promoting traits. Thus, all these biofilm PGPR can be used as biofertilizers, plant-growth promoters, suppressors of phytopathogens, and alleviators of abiotic stressors, such as drought, salinity, and heat. Application of biofilm PGPR would reduce environmental pollution as well as global warming leading to climate change. Thus, it would be necessary in the near future to further apply phenotypic screening of bacterial strains and their ability to form biofilms as a rapid selection criterion of PGPR. Our future studies should focus on biofilm formation in PGPR under field conditions.

DATA AVAILABILITY STATEMENT
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/Supplementary Material.

AUTHOR CONTRIBUTIONS
MMH conceived the idea, developed the methodologies, conducted several experiments, wrote the manuscript and collected the research fund. MKM conducted siderophore, K solubilization assays, antagonistic tests, pot experiments and analyzed the data. MK collected the samples and characterized the strains. MAH helped in planning the research and conducted the FTIR spectroscopy. MSB determined MDA, protein and catalase activity. MSI isolated bacterial DNA, identified the bacteria based on 16S rRNA gene sequencing and constructed phylogenetic tree. MMI took the SEM images. HBS and MAS verified and quantified the IAA using HPLC. MMUM and AHM supervised MS student (MK). All the authors read the manuscript.

ACKNOWLEDGMENTS
We would like to thank Dr. Ziban Chandra Das, Department of Gynecology, Obstetrics and Reproductive Health, for supplying sheep blood and Dr. S. M. Rafiquzzaman, Department of Fisheries Biology and Aquatic Environment, Faculty of Fisheries, BSMRAU, Gazipur, Bangladesh, for facilitating hemolytic tests.

SUPPLEMENTARY MATERIAL
The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmicb. 2020.542053/full#supplementary-material Supplementary Figure 1 | Production of auxins by different rhizobacterial strains. Initially each rhizobacterium was grown in LB broth supplemented with 0.2% of L-tryptophan in agitating condition. After 48 h incubation, 1 mL culture was collected and centrifuged. Then 500 µL supernatant was mixed with 1 mL of Salkowski reagent. Development of pink color indicated positive for auxin related compounds including IAA production.
Supplementary Figure 3 | Qualitative ammonia production by different rhizobacterial strains. The 50 µL (10 8 CFU mL −1 ) culture were inoculated in the glass test tubes containing 5 mL peptone water and incubated at 28 • C for 72 h. Then 1 mL Nessler's reagent was added. Development of yellow to brown color indicated positive results.
Supplementary Figure 4 | Qualitative production of HCN. Rhizobacterial strains were streaked on LB agar plates containing 0.45% glycine. Then filter papers dipped in alkaline pirate solution and placed on the lids of petri plates, sealed with parafilm and incubated at 28 • C for 24 h. Positive HCN production resulted in a color development from yellow to reddish-brown.
Supplementary Figure 5 | Production of biofilms by some selected rhizobacterial strains after 72 h incubation at 28 • C in stationary condition. Biofilm formation on SOBG without 25% PEG 6000 (A) and SOBG with 25% PEG 6000 (B).