Isolation, characterization, and identification of bacterial isolate P. rettgeri strain KDM3 (Genbank Accession Number: KC247668.1, GI: 432140652) have been described earlier by Salaskar (2015). For various assays, P. rettgeri strain KDM3 cells were resuspended in fresh lysogeny broth (LB) medium at a starting cell density of OD_530nm = 0.05, and the cells were allowed to grow under standard growth conditions. When the cell density reached OD_530nm = 0.5, various Cd stress assays, such as growth experiments, microscopic investigations, and proteomic and rice ecosystem studies, were performed using CdCl₂.2½ H₂O salt. P. rettgeri cells in the exponential phase (OD₅₃₀ = 1.0–1.2) were resuspended in the LB broth containing 0–250 μg Cd ml⁻¹ (0–250 ppm) and incubated at 30°C in an orbital shaker (140 rpm). The growth of P. rettgeri was monitored at various time intervals till 8 h by measuring the optical density of the culture at 530 nm using UV/Vis spectrophotometer (Jasco-V-530). Plasmid isolation and plasmid curing experiments were performed as per the protocol detailed earlier (Salaskar, 2015).

The concentration of Cd accumulated by P. rettgeri was measured after 20 h of exposure to 0–250 ppm Cd. The bacterial cells were spun down (10,000 g, 10 min) to remove residual Cd and resuspended in 10 ml of distilled water, vortexed, and centrifuged at 10,000 g, 10 min. After centrifugation, the bacterial cell pellet was digested in a microwave digestion system (Milestone Ethos up, Italy) using 6 ml concentrated HNO₃. The cooled digested solution was further concentrated to 2 ml and diluted to 10 ml with water, and the Cd concentration was measured using a Flame Atomic Absorption Spectrometer (GBC 932 plus, Australia). The detection limit of Cd was 0.2–1.8 ppm, the sensitivity was 0.009 ppm, and the wavelength was set at 228.8 nm with D2 correction. The assay was performed in triplicates.

Providencia rettgeri cells were exposed to 100 ppm Cd for 20 h. Following Cd exposure, the cells were harvested, washed, and observed under a light microscope (Carl Zeiss Axiostar plus) with a 100 × oil immersion objective under hydrous conditions. Providencia rettgeri cells were processed similarly, except that exposure to Cd served as control. For scanning electron microscopy (SEM), the cells were washed in normal saline and fixed in 2.5% glutaraldehyde for 2 h. The cells were then dehydrated in a graded ethanol series (30–100%), spotted on aluminum studs, and dried at 37°C for 1 h. The dried samples were gold coated using the thermal evaporation technique and analyzed by SEM using a Tescan VEGA 40 Microscope. For transmission electron microscopy (TEM), cells were washed in normal saline and were fixed with 2.5% glutaraldehyde, a 0.5% paraformaldehyde mixture, dehydrated with increasing concentrations of ethanol, and embedded in epoxy resin. Then sections (50 nm) were cut and visualized under Carl Zeiss Libra 120 keV TEM without staining. Energy-filtered transmission electron microscopy (EF-TEM) was used for elemental mapping on the electron-dense area as described earlier (Saunders and Shaw, 2014; Anaganti et al., 2015), except that the slit width of the Cd-specific energy filter was set at 390 and 503 ev.

Cellular proteins were resolved by 2D gel electrophoresis as described earlier (Basu and Apte, 2012). In brief, the cellular protein samples were prepared from control or Cd-treated cells following 1 h of exposure to 100 ppm (880 μM) Cd. The cup loading method was used to resolve cellular proteins (500 μg) by isoelectric focusing (17 cm IPG strip, pI 4–7, Bio-Rad, India), followed by 12% SDS-PAGE. The gels were stained with Coomassie Brilliant Blue G250 stain and digitized using a Dyversity-6 gel imager (Syngene, UK). Three biological replicate sets were generated by repeating the experiment three times. PDQuest 2D analysis software (version 8.1.0, Bio-Rad) was used for the first level match set generated from three biological 2D gel replicates. The minimum correlation coefficient value between the replicates of group sets was 0.75. Protein spots were detected and matched between the replicates in automatic detection mode, and spuriously detected spots were removed manually. Spot densities were normalized with the local regression method. An independent Student's t-test was applied for statistical analysis and the protein spots with p-values ≤ 0.05 were considered as significantly differentially expressed between control and Cd-treated cells. The protein spots of interest were manually excised from the gel and further identified by mass spectrometry.

The protein gel plugs were destained, reduced with Dithiothreitol (DTT), alkylated with iodoacetamide, in-gel digested with trypsin, and the oligopeptides were eluted as described earlier (Anaganti et al., 2015). Similarly, co-crystallization of the eluted oligopeptides and mass spectrometric (UltraFlex III MALDI-TOF/TOF mass spectrometer, Bruker Daltonics, Germany) analysis were performed as described earlier (Anaganti et al., 2015). Mascot searches were conducted using the NCBI nonredundant database (released in January 2016 or later, with a minimum of 79,354,501 entries actually searched) with the following settings: the number of missed cleavages permitted was 1; fixed modifications such as carbamidomethyl on cysteine, variable modification of oxidation on methionine residue; peptide tolerance as 100 ppm; enzyme used as trypsin; and a peptide charge setting as +1. Providencia rettgeri protein with significant Molecular Weight Search (MOWSE) score considered as successful identification.

The hierarchical cluster was created using the pheatmap toolbox in RStudio (Kolde, 2019). Briefly, raw protein intensity values were imported into the RStudio, row scaled, and plotted using default parameters. Gene Ontology (GO) enrichment analysis was carried out using the Bingo tool in Cytoscape (Maere et al., 2005). For this purpose, a basic ontology file in Open Biomedical Ontologies (OBO) format was downloaded from the GO website (http://geneontology.org/). Providencia rettgeri-specific annotations were retrieved from the Quickgo website (https://www.ebi.ac.uk/QuickGO/) using taxon ID filter (587). Protein identities were mapped to the corresponding UniProt ID for P. rettgeri using the UniProt ID-conversion tool. GOs were FDR corrected, and the selected ontology was plotted in RStudio using “ggplot2” for representation (Wickham, 2016).

A greenhouse glass pot culture experiment was conducted using an agriculture acidic soil (pH, 5.9) in Dapoli, Maharashtra (Table 1). The soil was passed through a 2-mm sieve. Glass tanks (60 cm in diameter and 30 cm in height) were filled with 15 kg of sieved soil, and 60 plants were planted per pot. The experimental sets were (a) control soil with no Cd (C), (b) 100 ppm Cd in soil (T1), and (c) 100 ppm Cd with 1 × 10⁸ cells of P. rettgeri strain g⁻¹ soil (T2). Three replicates were maintained for each set. The pots were maintained at 60% water holding capacity for 1-week for bacterial soil colonization. The soil was treated with CdCl₂.2½ H₂O (solubility in water 135 g/100 ml at 20°C). Bacterial colonization was checked during plant growth and after crop harvest by plating the representative soil samples followed by the serial dilution technique. Providencia rettgeri cells were selected on agar plates containing 50 ppm kanamycin.

Oryza sativa BARC-KKV-13 was used in this study. Seeds were surface sterilized with 0.1% HgCl₂ for 1 min, washed 3–4 times with running tap water, followed by one wash with distilled water, and dried on the blotting paper. The seedlings were grown in plastic trays containing control soil for about 2 weeks, and later seedlings with similar appearance and biomass were carefully transplanted into the, respectively, developed glass ecosystem as stated above. The rice crop was grown till maturity (135 days). Fertilizers were applied at regular intervals as per standard agriculture practice, and flooding condition was maintained throughout the growth period by replenishing water every 2 days. After 135 days, the plants were carefully removed from the pots by separating the roots and the shoots. Shoot samples were rinsed several times with water to remove surface contamination, and the root surfaces were washed several times with tap water, followed by distilled water, and further with 0.5 mM ethylenediaminetetraacetic acid (EDTA) to remove adhering Cd on the root surface.

The various physiological parameters of plants were determined by measuring the lengths and fresh weights of plant roots, shoots, or grains. The plant shoot, root, and grain samples were dried in a hot air oven at 70°C till a constant weight was achieved, and dry matter yield was recorded. To determine the Cd content, the dried samples were finely ground and wet digested in a microwave digestion system (Milestone Ethos UP, Italy) in concentrated HNO₃ (AR grade). The cooled digested solution was further concentrated to 2 ml and diluted to 10 ml with water, and the Cd concentration was measured using a flame atomic absorption spectrometer (GBC 932 plus, Australia), as described in the Section “Cd Concentration Measurements.”

All the experiments were performed in triplicates to generate three biological replicates, and the results have been reported as means ± standard deviation (SD). The software used for statistical analyses was ANOVA in Excel 2007. Statistical analyses for various parameters in the rice ecosystem experiment included comparisons of individual groups using the post-hoc Q-test. The p-values were determined using a post-hoc Q table. The significant difference was determined through Tukey's mean grouping at α = 0.05, i.e., 95% CI, and α = 0.01, i.e., 99% CI.

Providencia species are tolerant to high concentrations of an array of heavy metals (Thacker et al., 2006; Naik et al., 2013; Salaskar, 2015). We found that arsenic-tolerant P. rettgeri strain KDM3 also exhibited remarkable tolerance to high concentrations of Cd. The organism could grow in the increasing concentrations of Cd, while a concentration of 100 ppm Cd (880 μM Cd) resulted in a 50% decrease in growth rate (LD₅₀), as compared to control cells (Figure 1A). An increase in Cd content in the media caused a proportionate increase in the Cd uptake by the growing cells (Figure 1B). At 200-ppm Cd exposure, P. rettgeri cells accumulated ≈ 1 mg/g Cd/dry weight of cells (Figure 1B). Exposure to Cd (100 ppm) induced morphological changes in P. rettgeri. Loosely packed short rods of control cells changed to characteristic chains of coccoidal lenticular shape after being exposed to 100 ppm Cd, as seen under a light microscope as well as SEM (Figure 1C). Transmission electron micrographs showed spread-out electron-dense depositions inside the Cd-exposed cells (Figure 1C), indicating the presence of Cd. The intracellular distribution of Cd was confirmed by the elemental mapping of cells through Energy-Filtered Transmission Electron Microscopy (EFTEM). A large intracellular distribution of Cd was observed in the electron-dense region of Cd-treated cells, which was absent in control cells (Figure 1C). From these data, we concluded that the high Cd tolerance of P. rettgeri KDM3 strain is due to intracellular accumulation of the toxic metal.

Even though Providencia is known to be able to survive high concentrations of heavy metals, its Cd stress response network has not been explored yet. Plasmid-mediated Cd resistance has been reported for P. aeruginosa (Chellaiah, 2018). We earlier found that the arsenic tolerance of P. rettgeri strain KDM3 was plasmid-borne (Salaskar, 2015); however, Cd tolerance of the organism was unaffected even after plasmid curing, indicating that the determinants of Cd tolerance were encoded on the chromosome (data not shown). To systematically investigate the response of P. rettgeri to Cd exposure through a proteomic approach, we first proceeded to generate the 2D proteome map of P. rettgeri since the reference proteome map is not available for any Providencia strains yet. The cellular proteins were extracted from P. rettgeri cells exposed to 100 ppm Cd (test) for 1 h or not (control) and resolved by 2D gel electrophoresis. On average, PDQuest image analysis software detected about 543 high-quality spots on the 2D gel sets. Among the replicates in each group, high correlations were observed for protein spot matching in the individual group (Supplementary Table S1). Since a 2D proteome map is not yet available for P. rettgeri, we picked up a total of 144 protein spots for identification by MALDI mass spectrometry. We identified a total of 147 proteins confidently (Figure 2A, Supplementary Figure S1, Supplementary Table S2, Supplementary MS Data). The identified proteins belonged to 132 unique genes since 15 proteins (isoforms) were identified from more than one spot (Supplementary Table S2).

To decipher the biological processes, molecular functions, and cellular components of the identified proteins, GO enrichment analysis was carried out. Biological processes such as ribose phosphate metabolic process, TCA cycle, glycolytic process, acyl-CoA metabolic process, translation, protein folding, aerobic respiration, and response to oxidative stress were enriched among the identified proteins (Figure 2B, Supplementary Table S3). We also observed enrichment of various molecular functions such as binding of nucleotides or metal ions, translation regulation, ligase activity, pyrophosphatase activity, proton-transporting ATP synthase activity, and peroxiredoxin activity (Figure 2B, Supplementary Table S4). The majority of the identified proteins were cytoplasmic, while a few were part of protein complexes such as ATP synthase complex, TCA cycle enzyme complex, etc. (Figure 2B, Supplementary Table S5).

Furthermore, differential expression analysis on the proteome profiles of the control and test cells revealed a set of 69 proteins differentially expressed (p-value < 0.05) in response to 100 ppm Cd exposure (Figure 3A, Supplementary Table S6, Supplementary Figure S2). The DEPs belonged to 66 unique genes since three proteins were present in at least two spots. In all, 23 proteins were upregulated while 46 proteins were downregulated following exposure to 100 ppm Cd (Figure 3A, Supplementary Table S6). Among the upregulated proteins, de novo expression was observed for the typical bacterial 2-Cys peroxiredoxin AhpC (112.85-fold), while its cognate reductase AhpF showed a 5-fold increase in abundance in the cells exposed to 100 ppm Cd (Figure 3B, Supplementary Table S6). The genes encoding AhpC and AhpF proteins were found to be present in an operonic arrangement on the genome of P. rettgeri (Supplementary Figure 3A). Interestingly, the P. rettgeri cells retained the Cd tolerance property even when cured of the plasmid, suggesting that the ahp operon could be a major determinant of Cd tolerance in this organism. AhpC, belonging to a highly conserved thiol-specific antioxidant (TSA) family, possesses a peroxidase activity against H₂O₂, peroxynitrite, and organic hydroperoxides. In a homodimeric AhpC, the peroxidatic cysteine (C-47) is first oxidized to sulfenic acid intermediate and resolved by the cysteine (C-166) in the other subunit. Both the cysteine residues are completely conserved in AhpC encoded by P. rettgeri (Supplementary Figure 3B). AhpF acts as a reductase that uses electrons from NADH to reduce intermolecular disulfide bonds in an oxidized AhpC dimer (Zhang et al., 2019; Feng et al., 2020). Other noteworthy upregulated antioxidant enzymes include DNA starvation/stationary phase protection protein Dps (5.2-fold induction) and Mn-Superoxide dismutase (3.2-fold induction) (Figure 3B, Supplementary Table S6).

We also observed the de novo induction of 2 membrane transporters. Interestingly, ABC transporter ATP-binding protein (ENTRA1, 109-fold induction) was identified as a protein from Enterococcus raffinosus ATCC 49464, while amino acid ABC transporter substrate-binding protein FliY (49-fold induction) belonged to P. rettgeri (Figure 3B, Supplementary Table S6). Since the P. rettgeri strain used in this study was isolated from a natural habitat, acquisition of the transporter protein from E. raffinosus through horizontal transfer cannot be ruled out. The ATP-binding cassette-type vacuolar membrane transporter HMT1 from Schizosaccharomyces pombe (SpHMT1) was first shown to be involved in Cd ion import and Cd resistance (Ortiz et al., 1992). Later, it was demonstrated that the ATP-binding cassette transporter HMT1 constitutes a Cd detoxification mechanism that is highly conserved from bacteria to plants to humans (Preveral et al., 2009). Interestingly, ABC transporter ATP-binding protein ENTRA1 (260 amino acids), but not ABC transporter substrate-binding protein FliY, showed homology (ClustalW alignment score 211) to the C-terminal domain (254 amino acids) of SpHMT1 protein (Supplementary Figure S4), suggesting that ENTRA1 could be involved in Cd transport and intracellular accumulation in P. rettgeri cells. In Escherichia coli, upregulation of the periplasmic L-cystine-binding protein FliY was observed following exposure to 100 μM Cd or 0.88 mM H₂O₂ (Helbig et al., 2008; Ohtsu et al., 2010). FliY is an integral component of the L-cysteine/L-cystine shuttle system that imports L-cystine, an oxidized product of L-cysteine, from the periplasm to the cytoplasm to limit lipid peroxidation (Ohtsu et al., 2010). Cells overexpressing cystine/glutamate transporter protein acquire oxidative stress resistance independent of glutathione (GSH) (Banjac et al., 2008). ABC transporter substrate-binding protein FliY (279 amino acids) of P. rettgeri showed homology (ClustalW alignment score 398) to the L-cystine-binding protein TcyJ (266 amino acids) of E. coli (Supplementary Figure S5). It is, thus, tempting to speculate that the FliY protein provides an additional layer of oxidative stress resistance to Cd-exposed P. rettgeri cells. Additionally, an increase in abundance was seen for protein disaggregation chaperone ClpB (5.3-fold) and ATP-dependent protease, and ATPase subunit HslU (1.75-fold) (Figure 3B; Supplementary Table S6). While the organism showed a robust response to alleviate oxidative stress and revive or clear the damaged proteins, GO enrichment analysis on the 46 downregulated DEPs showed enrichment of several metabolic pathways such as the TCA cycle, aerobic respiration, and pyruvate metabolism (Figure 3C, Supplementary Tables S7, S8). Reduced metabolism could be either the direct effect of the diversion of NADH for redox reactions related to Cd bioaccumulation or a strategy adopted by the organism to limit metabolically generated ROS. Zhai et al. (2017) have also reported energy conservation and oxidative stress defense as some of the important mechanisms of Cd tolerance in Lactobacillus plantarum. Thus, we conclude that the Cd tolerance of P. rettgeri originates from the upregulation of relevant antioxidant enzymes and protein protection proteins.

A rice ecosystem experiment was performed to evaluate the effect of P. rettgeri on different growth parameters of rice plants and the Cd accumulation in different plant parts. The rice plants grown on control soil exhibited normal growth; however, spiking of the soil with 100-ppm Cd resulted in a significant reduction in plant growth throughout the growth phase till 135 days (Figure 4). In contrast, the application of P. rettgeri cells to the 100 ppm Cd-spiked soil improved the growth of rice plants on par with control plants, demonstrating that the presence of P. rettgeri cells had a beneficial effect on reducing Cd toxicity (Figure 4). Furthermore, we found that the fresh weights of roots, shoots, and grains of rice plants grown in the presence of 100 ppm Cd (T1) were significantly reduced as compared to control (C) plants (Figures 5A,B). The application of P. rettgeri cells to 100 ppm Cd containing soil (T2) caused a significant increase in the fresh weights of roots, shoots, and grains as compared to plants grown in the presence of 100 ppm Cd (T1). The plants exposed to 100 ppm Cd (T1) showed a significant reduction in the elongation of shoots and roots as compared to control plants (C) (Figure 5C). However, the application of P. rettgeri cells to 100 ppm Cd-containing soil (T2) nullified the toxic effects of Cd on the elongation of shoots and roots (Figure 5C). Cd contents in the control plant parts were below detectable limits. In the plants exposed to 100 ppm Cd (T1), roots accumulated the highest amount of Cd (575 ppm), followed by shoots (90 ppm) and grains (36 ppm). When 100 ppm Cd soil was treated with P. rettgeri, Cd content in the roots and the shoots were reduced by 11.2 and 50%, respectively, while the grains showed a 46.1% reduction (Figure 5D). Our results demonstrate that the presence of P. rettgeri cells has synergistic effects on minimizing Cd toxicity in the rice ecosystem.

In-situ bioremediation of soil has gained importance in recent years. To access the feasibility of the application of any metal-tolerant bacteria at the field level, it is necessary to simulate the field conditions and study its potential for in-situ bioremediation. This is the first report of its kind wherein a rice field condition is simulated and developed in the form of a rice ecosystem and the potential of P. rettgeri bacteria for Cd in-situ bioremediation is studied in crops grown till full maturity (135 days). The overall outcome of the study has indicated that the Cd bioaccumulation property of P. rettgeri cells has favorably decreased the Cd contents (~50%) in the edible parts of the rice plant due to reduced phytoavailability of Cd in soils. This is in agreement with in-situ bioremediation reported using other microbes (Chi et al., 2020; Haider et al., 2021). Various strains such as Enterobacter aerogenes MCC 3092, Pseudomonas aeruginosa, Bacillus megaterium H3, and Neorhizobium huautlense T1-17 were found to be potentially useful microbes in reducing phytoavailable Cd from the soil in the range of 40–60% Cd in different rice plant parts (Lin et al., 2016; Pramanik et al., 2018; Zhou et al., 2021).