Differential protein profiling of soil diazotroph Rhodococcus qingshengii S10107 towards low-temperature and nitrogen deficiency

Protein-based biomarkers can be a promising approach for identification and real-time monitoring of the bio-inoculants employed under sustainable agricultural plans. In this perspective, differential proteomics of psychrophilic diazotroph Rhodococcus qingshengii S10107 (JX173283) was performed to unravel its adaptive responses towards low-temperature nitrogen deficiency and identification of a biomarker for respective physiological conditions. LC-MS/MS-based proteome analysis mapped more than 4830 proteins including 77 up-regulated and 47 down-regulated proteins (p ≤ 0.05). Differential expression of the structural genes of nif regulon viz. nifH, nifD, and nifK along with their response regulators i.e. nifA, nifL, and nifB indicated that the nitrogenase complex was activated successfully. Besides up-regulating the biosynthesis of certain amino acids viz. Leucine, Lysine, and Alanine; the expression of the peptidoglycan synthesis proteins were also increased; while, the enzymes involved in Lipid biosynthesis were found to decrease. Furthermore, two important enzymes of the pentose phosphate pathway viz. Transketolase and Transaldolase along with Ribose import ATP-binding protein RbsA were also found to induce significantly under low temperature a nitrogen deficient condition, which suggests the cellular need for ample ribose sugar instantly. Additionally, comparative protein profiling of S10107 strain with our previous studies revealed that CowN protein was significantly up-regulated in all the cases under low-temperature nitrogen deficient conditions and therefore, can be developed as a biomarker. Conclusively, present study for the first time provides an in-depth proteome profiling of R. qingshengii S10107 and proclaims CowN as a potential protein biomarker for monitoring BNF under cold niches.


Results
The proteome of R. qingshengii S10107 was analyzed under two different physiological conditions i.e. low-temperature nitrogen deficient condition (without nitrogen source; BM) and low-temperature nitrogen sufficient condition (with nitrogen source, NB). The analysis was performed with three biological and three technical replicates.
Identification of the differentially expressed proteins. The whole proteomic dataset mapped to a total of 4838 proteins under BM and 4831 proteins under NB with a 1% false detection rate (FDR) (Fig. 1). There were 4813 proteins shared between the BM and NB along with 25 exclusive proteins under BM, and 18 under NB (p ≤ 0.05). Moreover, 77 proteins were found significantly up-regulated, while, 47 were down-regulated with ≥2-fold change under BM in reference to its counterpart.

Protein-protein interaction (PPI) network analysis.
Protein-protein interactions (PPIs) are considered very critical for cell survival. They are essential for the understanding of cellular physiology under different conditions. Therefore, PPI network analysis was performed to emphasize the crucial proteins involved in low-temperature nitrogen fixation (Fig. 2). Among BM proteins, ArgF (Ornithine carbamoyltransferase, Q8YMM6); HisF (Imidazole glycerol phosphate synthase subunit, A8HYT7); ProB (Glutamate 5-kinase, B2JHD6); Eno (Enolase, B2JIX0); PyrG (CTP synthase, B7KF08); PheT (Phenylalanine-tRNA ligase beta subunit, Q8YMH5); PheS (Phenylalanine-tRNA ligase alpha subunit, A0A1D8TAJ1); GlmS (Glutamine-fructose-6-phosphate aminotransferase, P59362) and EcaA (Carbonic anhydrase, P94170) were identified as hub nodes along with higher BC values (Table 1). Besides them, AtpA (ATP synthase subunit alpha, Q98EV6); AtpH (ATP synthase subunit delta, A8HS18); UppP (Undecaprenyl-diphosphatase, Q89WH1); DeaD (ATP-dependent RNA helicase, V9XLR7); TpiA (Triosephosphate isomerase, Q8YP17); TufA (Elongation factor Tu, C0ZVT7); PckG (Phosphoenolpyruvate carboxykinase, B2JJT8); Tkt (Transketolase, Q8YRU9) and SucC (Succinate-CoA ligase, A1KAU3) were identified as hub nodes only while, LeuD1 (3-  Venn diagram depicting the total number of expressed proteins of Rhodococcus qingshengii S10107 strain under Nitrogen deficient (BM) and Nitrogen sufficient (NB) conditions. Cells were harvested at midlogarithmic phase. A total of 4813 proteins are shared by both BM and NB, whereas the BM (25) showed more exclusive proteins than the NB (18). The analysis was performed in triplicates. The diagram shows the average value of proteins obtained from three replicates of the experiment.  GO-based functional characterization of the differentially expressed proteins. All the differentially expressed proteins were characterized on the basis of their biological functions and compared with the earlier studies (Fig. 3). Proteins associated with stress response, nitrogen fixation and energy production were up-regulated while that of biosynthetic processes and energy consuming processes were observed to down-regulate by all the diazotrophs studied under low-temperature nitrogen deficient conditions. Besides them, a good fraction of the up-regulated and down-regulated proteins were found uncharacterized which need a detailed investigation in the future. In case of the present study, 17% of the proteins were related to stress response, followed by nitrogen fixation (16%), protein synthesis/modifications (13%), energy production (10%), gene regulation/transcription (7%) and uncharacterized (5%). Among down-regulated proteins, the majority of the proteins were involved in biosynthetic processes (13%) and energy consuming processes (10%) besides uncharacterized proteins (9%).
Additionally, functional enrichments in the PPI network were explored to provide a more specific description of the differentially expressed proteins in terms of their involvement in amino acid biosynthesis (AAB); metabolic pathways (MP); nitrogen metabolism (NM) and production of secondary metabolites (Fig. 2).  CowN as a protein biomarker. In order to identify the protein biomarker(s) associated with the nitrogen fixation process at lower temperatures, the author group has performed a series of experiments on six different cold-adapted microorganisms viz. P. palleroniana N26-GL 10 ; Dyadobacter psychrophilus B2 and P. jessenii MP1 9 ; P. palleroniana N26-GB 11 and P. migulae S10724 12 including present study (R. qingshengii S10107). Moreover, two distinct proteomic approaches i.e. gel-based 9,11,12 and gel-less approach 10 were also employed to rectify the experimental biases.

Amino acid biosynthesis (AAB
In this perspective, comparative protein profiling of all the diazotrophs revealed that CowN was significantly up-regulated in all the six experimentation followed by Eno which was up-regulated in four cases only (Fig. 4). PyrG was also found to express in four instances, however, in three cases it was a down-regulated and up-regulated only once (SM). Similarly, proteins IlvC, ClpX, MnmE, AtpA, HtpG, MutL, and IleS were expressed in three cases only. Among them, AtpA, HtpG, and MutL were up-regulated in all the three cases while; MnmE and IleS were down-regulated, subsequently.

Discussion
Genus Rhodococcus is well known for its broad metabolic versatility, however, the diazotrophic potential of this actinobacterium is least studied. All the previous proteomic investigations, which were performed on Rhodococcus, only highlighted its catabolic behavior towards different organic compounds viz. Fluoranthene 14 , toluene and phenol 15 , propane 16 , triacylglycerol 17 , 4,4′-Dithiodibutyric Acid 18 , benzoate 19 and polyaromatic compounds 20 . In this perspective, the present study for the first time unravels its proteomic response towards the abiotic stress and identified a protein biomarker for low-temperature nitrogen deficient conditions. The study revealed that R. qingshengii S10107 expressed all the three structural nif genes viz. nifD, nifH and nifK under BM along with their regulators i.e. nifA, nifL, and nifB; thereby, confirming its nitrogen fixation potential 21 . However, complete nif regulon could not be observed, which might be due to the fact that diazotrophs do not express all the nif genes at a time in order to minimize their energy consumption 22 . Additionally, the accumulated Nif proteins may get cleaved by the proteases i.e. ClpX (N1M9I2) as also evident from earlier studies 10 . Therefore, the kinetics of nif regulon must be studied over time to scan the complete network. Besides the Nif proteins, two associated proteins were also observed viz. CO Weal-Nitrogenase (CowN) and CO-responsive regulator (CooA). CowN is known to protect the nitrogenase from Carbon monoxide (CO) stress which is supposed to induce under low-temperature nitrogen deficient conditions 10,23,24 .
R. qingshengii S10107 was found to employ unique strategies for performing important cellular processes under low-temperature nitrogen deficient condition. Under BM, proteins 3-isopropylmalate dehydratase, 2,3,4,5-tetrahydropyridine-2,6-dicarboxylate N-succinyltransferase and Alanine racemase were up-regulated which are responsible for the biosynthesis of the amino acids Leucine, Lysine, and Alanine, respectively. Lysine and Alanine in a peptide help it to get α-helical conformation 25 and thus, contributed towards low-temperature adaptation 26 . Similarly, Tyrosine-tRNA ligase, Isoleucine-tRNA ligase and Aspartate-tRNA(Asp/Asn) ligase were down-regulated which are responsible for the biosynthesis of Tyrosine, Iso-leucine, and Aspartate. These amino acids tend to be decreased when the organism exposed to the cold 27 .
Further, among the up-regulated proteins, Ornithine carbamoyltransferase (ArgF) and N-acetyl-gamma-glutamylphosphate reductase (ArgC) are known for the biosynthesis of Arginine, which mostly serves as a key source of carbon and nitrogen for ornithine, proline, and pyrimidine 28,29 . Up-regulation of ArgC during low-temperature nitrogen fixation needs to be investigated in detail because of the two main reasons -first -it catalyzes high energy consuming reaction which is otherwise very rare event during such conditions 10,30 and second -it causes growth delay as well as inefficient nodule formation in diazotrophs 30 . In the case of Proline biosynthesis, the genes encode for protein Gamma-glutamyl phosphate reductase (ProA) and Glutamate 5-kinase (ProB) are organized in a single operon i.e. proBA. In the present study, S10107 strain increases the expression of ProB while, decreases ProA, thereby, regulated the intracellular concentration of Proline. P. putida, K. aerogenes, and E. coli, are known to enhance the catabolism of Proline under nitrogen-deficient conditions 29,31 . A similar condition was also observed in case of Histidine which showed up-regulation of Imidazole glycerol phosphate synthase subunit (HisF) and down-regulation of Histidine-tRNA ligase (HisS). These expressional patterns can be supported with the fact that nitrogen fixation is an energy consuming process and therefore, down-regulation and/or shutting down the nonessential proteins is pre-requisite to the cell survival.
Cold along with nitrogen deficiency represents dual-stress condition for the microorganisms involving environmental as well as nutritional stress. Therefore, S10107 strain up-regulated stress associated proteins viz. UvrABC system protein A (Q0SI39); DNA repair protein RecO (Q985A3), Chaperone protein HtpG (Q0S467), and Error-prone DNA polymerase (Q98E34). UvrABC and RecO are the multienzyme systems which aid in DNA repair; while Chaperone HtpG protects the cellular proteins from the stress conditions 9,10 . Also, up-regulation of Error-prone DNA polymerase, UvrABC, and RecO revealed the cellular need for DNA maintenance under given physiological conditions 32,33 . Conclusively, these proteins are responsible for tight regulation of the housekeeping genes along with their protection to sustain the life under multi-stress conditions. Under BM conditions, S10107 strain showed a distinct metabolic behavior by up-regulating the proteins of peptidoglycan synthesis (Undecaprenyl-diphosphatase, Q89WH1 and D-alanine-D-alanine ligase, B2JHF8); Glutamine synthesis (CTP synthase, B7KF08; Glutamine-fructose-6-phosphate aminotransferase, P59362); Biotin metabolism (ATP-dependent dethiobiotin synthetase, B7K5E6 and Biotin carboxylase, Q06862), Nitrogen metabolism (Urease enzyme complex) and Nucleotide metabolism (dITP/XTP pyrophosphatase, Q98DN4 and Thymidylate synthase, Q0SEI1) while down-regulating the enzymes for Lipid biosynthesis (Acyl-[acyl-carrier-protein]-UDP-N-acetylglucosamine O-acyltransferase, A8I491 and 3-hydroxybutyryl-CoA dehydrogenase, Q45223) and Pantothenate biosynthesis (3-methyl-2-oxobutanoate hydroxymethyltransferase, Q0SHJ0). Up-regulation of the peptidoglycan synthesis proteins while down-regulation of Lipid biosynthesis-related proteins revealed that R. qingshengii S10107 tend to thicken their cell wall to respond towards external stress. Further, enhancement of Glutamine synthesis proteins can be justified by the general tendency of the microbial cell to prefer ammonium and glutamine as a nitrogen source 29 . Furthermore, CTP synthase is responsible to regulate cytidine triphosphate (CTP) concentration within a cell. Killer et al., (2018) has developed a CTP synthase based tool to identify the family Bifidobacteriaceae. Glutamine-fructose-6-phosphate aminotransferase has glutaminase activity that links hexosamine biosynthetic pathway to the hexose pathway and thereby, plays a regulatory role for the nutrient-sensing system 34 .
The up-regulation of the two enzymes of the Pentose phosphate pathway viz. Transketolase (Q8YRU9) and Transaldolase (P58561) along with Ribose import ATP-binding protein RbsA (Q0S9A4) under BM suggests the cellular need of excessive ribose sugar instantly. It can again be supported with the down-regulation of Ribose-5-phosphate isomerase (B7K6D3) which is responsible for the conversion of ribose-5-phosphate to ribulose-5-phosphate. In addition to these, three enzymes of Gluconeogenesis (2,3-bisphosphoglycerate-dependent phosphoglycerate mutase, B2JC95; Phosphoenolpyruvate carboxykinase, B2JJT8, and Triosephosphate isomerase, Q8YP17) and two of glucose catabolism (Enolase, B2JIX0 and Succinate-CoA ligase, A1KAU3) were also found to up-regulate under BM. Contrary to these, proteins related to the glycogen metabolism (glycogen debranching enzyme, Q0SHV3 and Glycogen phosphorylase, Q0SGS1) and glycerol metabolism (Glycerol kinase, A1K962) were found down-regulated along with Isocitrate dehydrogenase (Q0S371), an important enzyme of TCA cycle. It might be the part of the adaptation strategies of the microorganism so that it could utilize carbohydrates as well as non-carbohydrates precursors for fulfilling its energy demands. The similar results were also observed by Ma et al. 35 . This group have used SILAC method for analysis the proteome of Edwardsiella tarda ATCC 15947 under prolonged cold stress and reported that the enzymes of Gluconeogeneis were significantly enhanced under cold. Furthermore, Tullio et al. 36 have analysed the proteome of the diazotroph Rhizobium freirei and observed the up-regulation of Gluconeogeneis associated enzymes under metabolic stress conditions.
The expression of CowN was reported in the proteome of all the six previously studied cold adapted diazotrophs viz. P. palleroniana N26-GL 10 ; Dyadobacter psychrophilus B2 and P. jessenii MP1 9 ; P. palleroniana N26-GB 11 and P. migulae S10724 12 including this study (R. qingshengii S10107) under cold and nitrogen deficient conditions. Therefore, this protein is being identified as a protein biomarker for monitoring BNF under cold niches. Several protein-based biomarkers are already being used for the clinical purpose 6 , veterinary science 37 , bioremediation 7 and heavy metal detection 8 . Saito et al. 38 detected the multiple nutrient stresses at Pacific Ocean biomes by using the protein biomarkers. Recently, Andrade-Herrera et al. 39 have developed an earthworm based biomarker for pesticides and toxicity assessment in agricultural soils. However, information is not available about such biomarkers which can monitor any biogeochemical processes under abiotic stress that is going to be imperative for precision farming in the future.
In conclusion, the present study provides a detailed investigation on the adaptive responses of R. qingshengii S10107 towards cold diazotrophy which can be explored for further advance research. Further, CowN can be proclaimed as a protein biomarker for diazotrophic identification and monitoring of BNF under cold and nitrogen deficient conditions. Moreover, the role of carbon monoxide and ribose sugar in low-temperature diazotrophy should be explored for detailed investigation.

Methods
Bacterial strain and growth conditions. R. qingshengii S10107 (JX173283) was originally isolated from Proteome extraction and LC-MS/MS analysis. The proteome of S10107 strain was extracted at the mid-log phase in triplicates as per the earlier reports 9,11 (Supplementary material). ACQUITY UPLC system (Waters, UK) having ACQUITY UPLC BEH C18 column (Waters, UK)(150mm X 2.1mm X 1.7µm) was used to perform liquid chromatography. A gradient of two mobile phases A (0.1% Formic Acid in WATER), and B (0.1% Formic Acid in Acetonitrile) were used for the chromatographic separation. Further, mass spectrometric detection was performed by using SYNAPT G2 QTOF (Waters, UK) having an electrospray ionization (ESI) source.
Database searching and analysis. The observed proteins were matched to Pseudomonas protein database which was downloaded from Swiss-Prot through PLGS software 3.0.2. The data analysis was performed in triplicates using the following parameters: peptide tolerance = 50 ppm, fragment tolerance = 100 ppm, minimum no. of fragment matches (for Proteins) = 5, minimum no. of fragment matches (for peptides) = 2,, minimum no. of peptide matches (for proteins) = 2, No. of missed cleavages = 2 and modifications include oxidation-m and carbamidomethyl-c. The FDR (false discovery rate) was set as <1% on both protein and peptide levels. The data were normalized across the conditions including the replicates using spectral counts method 40,41 . For identifying differentially expressed proteins, the normalized peptide matches of the treatment were divided by that of control conditions. Statistical analysis was performed through pair-wise Student's t-test (p ≤ 0.05).
Construction of the PPI network. Among the total expressed proteins, only two types of proteins were selected for the construction of the PPI (protein-protein interactions) Networks: (1) unique and (2) which showed 2-fold or higher change in their expression. PPI information of proteins was obtained from the String database (version 9.1; http://string-db.org/) having the confidence score > 0.7. It was further imported in Cytoscape (version 3.6.0; http://www.cytoscape.org/) and the union calculation was performed, followed by removal of the duplicated edges by using Advanced Network Merge 40,42 . Gene ontology (GO) enrichment analysis. The ontologies and their genes from the PPI network of selected proteins were identified using BiNGO plugin (version 3.0.3; http://apps.cytoscape.org/apps/bingo) for the Cytoscape which are: molecular functions, biological processes, and cellular components. For a detailed description of the biological process, the ClueGo plugin (version 2.5.0; http://apps.cytoscape.org/apps/cluego) of Cytoscape was used to integrate several ontology sources by extracting the non-redundant biological information from several databases viz. KEGG, REACTOME, GO, Wiki Pathways and BioCarta 40,43 . Statistical analysis. Significance was determined using unpaired two-tailed t-test or linear regression analysis (GraphPad Prism software, version 6.01). Differences were noted as significant *p ≤ 0.05 for t-test or linear regression analysis. The scatter plot analysis was performed by using the software PERMANOVA 44 .

Data availability
The datasets used and/or analyzed during the current study are available from the corresponding authors on reasonable request.