Unraveling adaptation of Pontibacter korlensis to radiation and infertility in desert through complete genome and comparative transcriptomic analysis

The desert is a harsh habitat for flora and microbial life due to its aridness and strong radiation. In this study, we constructed the first complete and deeply annotated genome of the genus Pontibacter (Pontibacter korlensis X14-1T = CCTCC AB 206081T, X14-1). Reconstruction of the sugar metabolism process indicated that strain X14-1 can utilize diverse sugars, including cellulose, starch and sucrose; this result is consistent with previous experiments. Strain X14-1 is also able to resist desiccation and radiation in the desert through well-armed systems related to DNA repair, radical oxygen species (ROS) detoxification and the OstAB and TreYZ pathways for trehalose synthesis. A comparative transcriptomic analysis under gamma radiation revealed that strain X14-1 presents high-efficacy operating responses to radiation, including the robust expression of catalase and the manganese transport protein. Evaluation of 73 novel genes that are differentially expressed showed that some of these genes may contribute to the strain’s adaptation to radiation and desiccation through ferric transport and preservation.

irradiation, respectively, than Escherichia coli 8 , and the complete genome of D. radiodurans was first published in 1999 9 . To elucidate the extreme resistance phenotype of D. radiodurans R1, various research strategies have been combined 10 , and three hypotheses regarding DNA repair have been proposed 11 . The lack of novelty in DNA repair-related genes/proteins and the greater efficiency of specific bacteria to use conventional repair pathways are partially supported by the findings from previous studies [12][13][14][15][16][17] . An in-depth analysis of the D. radiodurans R1 genome and its gene expression profile revealed that many undefined genes, including ddrA, ddrB, ddrC, ddrD and pprA, are involved in DNA repair [18][19][20][21] , suggesting that repair functions are encoded by these hypothetical genes. The last hypothesis is that ring-like nucleoids (RNs) contribute to DNA repair 22 .
There are also three assumptions regarding the maintenance of a low ROS concentration in bacteria 10 , most of which are detoxifying and scavenging ROS, including small catalase, superoxide dismutase, and antioxidant molecules, and exhibit an increased Mn(II)/Fe ratio intermediated by manganese complexes 11,13,23 . Daly and Krisko found that molecules smaller than 3 kDa in the extract of Deinococcus radiodurans R1 can impose antioxidant protection on E. coli proteins 23,24 . The promotion of metabolic activities with decreased ROS production (e.g., glyoxylate bypass of the TCA cycle 21,25 ) is an alternative to the response to oxidative damage and single antioxidant pathways through high ROS production, which could be inactivated due to redundant ROS-tolerance mechanisms. To lower the ROS, it is also helpful to reduce proteins with Fe-S clusters and the number of respiratory chain enzymes 25 . In addition to maintaining a low ROS concentration, many other metabolic activities, such as proteolysis and glucose metabolism, contribute to the robustness of D. radiodurans R1 15,23,26,27 . In addition to D. radiodurans R1, additional genome sequences of the genus Deinococcus have been published 26,[28][29][30][31][32] , and comparative analyses have been performed to elucidate the diverse molecular mechanisms and physiological determinants underlying the extreme resistance phenotype 33,34 .
We isolated the strain Pontibacter korlensis X14-1 T (X14-1) from the surface layer of a desert in Xinjiang, China, and identified it as a new species of the genus Pontibacter 2 . This study provides the first complete genome of the genus Pontibacter and attempted to delineate genomic components related to radiation and desiccation resistance in comparison with other species from the genus Pontibacter. A comparative analysis of the gene expression profile under radiation was also conducted to unravel the complicated mechanisms of strain X14-1 involved in its adaptation to the arid environment of the desert. This work will provide referable information for the comprehensive understanding of the evolution and adaptation of the genus Pontibacter as well as various radiation and desiccation resistances.

Results
Genomic characteristics and phylogeny of strain X14-1. The complete genome sequence of strain X14-1 was produced based on high-quality reads and corrected by read mapping and PCR verification. Strain X14-1 has a larger genome size (5.46 MB) and a lower GC content (47.3%) than three other Pontibacter strains distributed in different species (summarized in Table 1). We found that most of the transposase-related genes are located near genomic islands and next to DNA repair-and ROS detoxification-related genes ( Fig. 1), which implied that mobile genetic elements (MGEs) play an important role in the adaptation to radiation and desiccation in the desert. Differences in genome size and MGEs between strain X14-1 and other Pontibacter strains may be attributed to the genomic evolution or gapped assembly of P. actiniarum DSM 19842, Pontibacter sp. BAB1700 and P. roseus DSM 17521.
To confirm the phylotype of strain X14-1, we downloaded 40 genomes of the family Cytophagaceae (higher taxonomic classification of the genus Pontibacter) available in the NCBI database. Phylogeny analysis indicated the same results as those previously reported based on the 16S rDNA sequence 2 , and P. actiniarum DSM 19842 was found to be the most homologous to strain X14-1 (Fig. 2), a finding that is also supported by the following functional analysis. Sugar metabolism in strain X14-1. In comparison with three other genomes from the genus Pontibacter, we found that only strain X14-1 harbors genes encoding D-fructokinase, which is essential for sucrose and fructose utilization, and this is consistent with previous experimental results 2 . Beta-galactosidase, which is essential for strain X14-1 to use lactose as an alternative carbon source by catalyzing lactose to galactose and glucose, is specific to strain X14-1 compared with other Pontibacter strains. Although comparative analysis revealed a common dispersion of cellobiose glucohydrolase in Pontibacter, enzymes responsible for degrading cellulose to cellobiose are only distributed in strain X14-1. Starch could be degraded to amylose and alpha-D-galactose-1-phosphate, which is an intermediate in the production of UDP-glucose that could link pentose and glucuronate interconversion. This is important for the utilization of D-galactose as a carbon resource. Mannose can enter glycolysis through beta-D-fructose-6-phosphate with the help of hexokinase and mannose-6-phosphate isomerase, which could be encoded by genes in strain X14-1. The ability to utilize versatile sugars as described above (Fig. 3) could partly explain how strain X14-1 survives in an infertile desert.
Determinants in the genome for the adaptation of strain X14-1 to the radiation and aridness of the desert. Genes related to DNA repair and the stress response, including ROS detoxification and the osmotic response, were analyzed to understand whether and how strain X14-1 protect itself against to desiccation and radiation. In strain X14-1, recombination repair-related genes are the most abundant, followed by base excision repair (BER) and nucleotide excision repair (NER) (Fig. 4, detailed in Supplementary File 1). There are no specific corresponding genetic determinants for the radiation resistance of strain X14-1, implying that the robustness of the resistance of strain X14-1 to radiation and desiccation is not attributed to new genes but to the high-efficacy operation of systems relevant to recovery from radiation. Trehalose is a natural product that can form a protective film outside the cell under low and high temperatures, osmotic pressure and aridness, preventing proteins from becoming inactivated. The analyses conducted in this study elucidated that intermediate products from the metabolism of several sugars, such as starch and lactose, in strain X14-1 could be transferred to trehalose through the OstAB and TreYZ pathways, as summarized in a previous publication 35 (Fig. 5). Trehalase-encoding genes, which are homologous to the trehalase-like protein-encoding genes of Gramella forsetii KT0803, were also identified, suggesting the trehalose consumption of strain X 14-1.
Responses of strain X14-1 to gamma radiation revealed by RNA-Seq. After radiation, the average survival rate of strain X14-1 ranged from 43.4% to 55.1%, as determined from triplicate analyses. To better understand the specific responses of strain X14-1 to gamma radiation, we determined the expression level of all possible determinants responsible for radiation adaptation and ROS detoxification. BER, NER and homologous recombination-related genes, which are mainly required for recovery from gamma radiation, are apparently upregulated (Supplementary File 2).
Catalase-encoding genes exhibit intensive upregulation after radiation, but a similar finding was not obtained for superoxide dismutase, which is also important for ROS detoxification (Supplementary File  2). The expression of the manganese transport protein is 10-fold higher after radiation, and a similar trend was identified for subunit A-F of the multicomponent Na + :H + antiporter as well as other Initial evaluation of unannotated genes with differential expression. Apparent differential expression was found for 394 of 893 unclassified genes as a result of gamma radiation. To explore whether these unknown genes contribute to the adaptation of strain X14-1 to radiation, an alignment of these genes to the conserved domain database (CDD) was conducted. Of the 394 new genes, 73 could be  assigned to specific domains in CDD, and the corresponding hypothesized functions are summarized in Supplementary File 3, indicating that these 73 genes are commonly found in the FecR, DUF and FlgD_ig (associated with ferric transport and preservation) superfamily.

Discussion
The desert is characterized by a low diversity of inhabitants due to fluctuations in temperature, aridness, strong radiation and poor nutrition 1,36 . Elucidating the microbial life in the desert could be important because it could help us understand the dry limit of life and improve our research on extremophiles 37 . Contributors to radiation resistance and adaptation to poor nutrition in microbes could be applied to improve plant culturing in the desert. This will decrease sand storms mainly induced by the desert and increase arable lands to aid the avoidance of hunger worldwide. Determinants to radiation resistance in microbes residing in the desert could also be industrially applied in cosmetic and soil remediation under radiation.
Strain X14-1, which was isolated from the sand surface in the desert, belongs to the poorly understood genus Pontibacter, and there are no publications on an in-depth analysis of the Pontibacter genome. In this study, we performed a high-coverage sequencing of strain X14-1 and produced the first complete map, which may be used as a valuable reference to promote research on the genus Pontibacter. A detailed annotation of the strain X14-1 genome in combination with a comparative transcriptomic analysis was conducted to understand how strain X14-1 utilizes diverse alternative nutrition sources and recovers from desiccation and strong radiation.
Previous studies have revealed that strain X14-1 can utilize versatile carbon sources 2 , and this finding is supported by the reconstruction of the metabolic pathways of different sugars. D-fructokinase, beta-galactosidase, endoglucanase and beta-glucosidase are key for the utilization of sucrose, fructose, lactose, cellulose and glycoside, respectively. Enzymes that are important for metabolizing starch, galactose and mannose were also found in strain X14-1. In addition to their robust competence for utilizing diverse available sugars, another interesting finding is that only strain X14-1 and P. actiniarum DSM19842 possess phosphoenolpyruvate (PEP) carboxylase, which is key for CO 2 fixation, mainly in plants, through the C4 cycle and may be used to resemble the autotrophic lifestyle in environments with extremely poor nutrition. The fact that most of the hypothetical proteins in strain X14-1 are homologous to those in P. actiniarum DSM19842 also demonstrates the homology of these two Pontibacter strains revealed by the phylogenetic construction (Fig. 2).
In addition to the various metabolic pathways required for the utilization of limited resources, strain X14-1 also harbors an intensive arsenal of DNA repair-and stress response-related determinants that allow it to survive in the desert. An in-depth analysis of the genomic components of strain X14-1 revealed possible broad-spectrum contributors to resistance to radiation (Supplementary File 1). This is consistent with the genetic factors found in the genus Deinococcus, and we also found new genes that  (Fig. 3) could be catalyzed to trehalose through trehalose 6-phosphate synthase and trehalose 6-phosphate phosphatase. Glycogen operon protein could degrade starch synthesized from alpha-D-Glucose-6P available from many metabolic pathways to trehalose. encode resistance agents, indicating supplementary mechanisms to those found in Deinococcus. Based on changes in the annotated genes after gamma radiation, we found that catalases but not dismutase exhibited a marked increase, indicating the robustness of catalase in oxidative detoxification in strain X14-1. An increased Mn(II)/Fe ratio has been demonstrated to play an important role in detoxification from oxidative damage 11 , and this is also supported by our findings: upregulation of the manganese transport protein, ferric uptake regulator and LysR family transcriptional regulator. The expression of gene gshB encoding glutathione synthase was high during radiation, but glutathione peroxidase and reductase, which are responsible for the circulation of GSS (reduced glutathione) and GSSG (oxidized glutathione) exhibited no change, implying that glutathione is not the intermediate in strain X14-1during radiation resistance. An interesting finding is the robustly elevated expression of transposase during gamma radiation (Supplementary File 2), which implies the potently active role of widely distributed transposases in strain X14-1 for desert adaptation. Some of the conserved domains of upregulated unknown genes could be assigned to the FecR and ferritin-like superfamily. This finding may imply that some unknown genes could contribute to oxidative detoxification because these proteins are related with ferric transport or preservation, but further analysis or experiments are required to confirm this hypothesis.

Materials and Methods
Strains and culture conditions. All of the strains were obtained from the China Center for Type Culture Collection (CCTCC). P. korlensis X14-1 was grown at 30 °C on marine broth 2216 (Difco). To determine the tolerance of the culture to gamma radiation, the strains were grown in the appropriate liquid medium to the exponential phase, and 2.5 ml of the cell culture was then subjected to 60Co radiation with a continuous dose rate of 1000 Gray/h at room temperature. The control culture cells were not treated with radiation. Samples were collected 3 h after culture with (radiated samples) or without (controls) radiation treatment. Biological triplicates were harvested from each treatment. These samples were immediately frozen in RNAlater (Qiagen, German) and stored at -80 °C for RNA analysis.
DNA extraction, sequencing and assembly. Total DNA was extracted through the traditional CTAB method, and its quality was tested by Qubit. Sequencing was performed for a 470-bp insert size library (1.65-Gb clean reads with a length of 90 bp) and mate-paired libraries with insert sizes of 2222 bp and 6200 bp (477-Mb clean reads, each with a length of 49 bp), respectively. Clean reads were used to assemble the complete genome sequence as previously described 38 . Gene prediction, annotation and comparative analysis. Genes were predicted based on the assembled and confirmed sequence using GLIMMER [39][40][41] , which was developed for microorganisms including bacteria, archaea, and viruses. This was also used for the prediction of genes from other three Pontibacter strains downloaded from the NCBI ftp site. The annotations were conducted by assigning the predicted genes to the Kyoto Encyclopedia of Genes and Genomes (KEGG) database (http://www. genome.jp/kegg/) 42 , Cluster of Orthologous Groups of proteins (COG) 43,44 , and the protein database in NCBI (http://www.ncbi.nlm.nih.gov/protein/) through a BLASTP search (e-value ≤ 1e-5, query coverage ≥ 40%). The strategy described by Tian et al. 45 was used to identify the orthologous proteins in four Pontibacter genomes.
In the phylogenetic analysis, homologous genes from 41 strains were used to construct the gene family and phylogenetic tree: (i) the gene set was first generated using BLAST (V2.2.23) with the genes from 41 strains; (ii) the gene family was constructed using OrthoMCL (V1.4) with the gene set data; (iii) to construct the phylogenetic tree, single-copy genes in the gene family were screened, and the genes were aligned according to the strains using Muscle (V3.8.31); and (IV) the phylogenetic tree was drawn using TreeBeST (V1.9.2) with 1000 bootstrap replicates.

Identification of MGEs.
The prophages in the genome sequences were identified using the Phage Search Tool (PHAST) 46 . Genomic islands were predicted with multiple methods (IslandPath-DIOMB, SIGI-HMM, and IslandPicker using Islandviewer [47][48][49] . We searched repeat sequences with the RepeatMasker program 50 , and transposases were selected from the annotation results. RNA-Seq analysis. Three replicates of each group were used for total RNA extraction and mRNA enrichment using the Ribo-Zero TM kit. The qualified mRNA was sequenced on the HiSeq2000 platform, and the expression level of each gene was calculated using the reads per kilo bases per million reads (RPKM) 51 . The genes that were differentially expressed between the two groups were identified as described by Audicet et al. 52 . The analysis results with a p-value < 0.05 were corrected by the FDR (false discovery rate), which was set to ≤ 0.001.