Rhizodegradation of Petroleum Oily Sludge-contaminated Soil Using Cajanus cajan Increases the Diversity of Soil Microbial Community

Most components of petroleum oily sludge (POS) are toxic, mutagenic and cancer-causing. Often bioremediation using microorganisms is hindered by the toxicity of POS. Under this circumstance, phytoremediation is the main option as it can overcome the toxicity of POS. Cajanus cajan a legume plant, was evaluated as a phyto-remediating agent for petroleum oily sludge-spiked soil. Culture dependent and independent methods were used to determine the rhizosphere microorganisms’ composition. Degradation rates were estimated gravimetrically. The population of total heterotrophic bacteria (THRB) was significantly higher in the uncontaminated soil compared to the contaminated rhizosphere soil with C. cajan, but the population of hydrocarbon-utilizing bacteria (HUB) was higher in the contaminated rhizosphere soil. The results show that for 1 to 3% oily sludge concentrations, an increase in microbial counts for all treatments from day 0 to 90 d was observed with the contaminated rhizosphere CR showing the highest significant increase (p  < 0.05) in microbial counts compared to other treatments. The metagenomic study focused on the POS of 3% (w/w) and based on the calculated bacterial community abundance indices showed an increase in the values for Ace, Cho, Shannon (Shannon-Weaver) and the Simpson’s (measured as InvSimpson) indices in CR3 compared to CN3. Both the Simpson’s and the Shannon values for CR3 were higher than CN3 indicating an increase in diversity upon the introduction of C. cajan into the contaminated soil. The PCoA plot revealed community-level differences between the contaminated non-rhizosphere control and contaminated rhizosphere microbiota. The PCoA differentiated the two treatments based on the presence or absence of plant. The composition and taxonomic analysis of microbiota-amplified sequences were categorized into eight phyla for the contaminated non-rhizosphere and ten phyla for the contaminated rhizosphere. The overall bacterial composition of the two treatments varied, as the distribution shows a similar variation between the two treatments in the phylum distribution. The percentage removal of total petroleum hydrocarbon (TPH) after 90 days of treatments with 1, 2, 3, 4, and 5% (w/w) of POS were 92, 90, 89, 68.3 and 47.3%, respectively, indicating removal inhibition at higher POS concentrations. As the search for more eco-friendly and sustainable remediating green plant continues, C. cajan shows great potential in reclaiming POS contaminated soil. Our findings will provide solutions to POS polluted soils and subsequent re-vegetation.


planting of Seeds
Seed viability test and breaking of dormancy. The viability of the seed of C. cajan was tested using the floatation method. The seeds were tested for viability by soaking it in distilled water for 5 min. After that, all floating seeds are sieved out as non-viable and the water was drained immediately from the seeds that sank which are now considered viable seeds. The seeds were then surface sterilized with 10% hydrogen peroxide solution before planting. The seeds was sown to a depth recommended for C. cajan, which is from 1.5 to 3 cm 22 . For an excellent establishment of the plants, five seeds were planted in a hole. Likewise, the respective pots were irrigated every day and emergence was observed subsequently. The plants were moderately watered every two days with tap water. The appearance of the plants in response to the presence of POS in soil was monitored to determine if there is any phytotoxicity of POS to the plants.
Microbiological analysis. The media used were Nutrient Agar (NA); for the enumeration and isolation of bacteria, mineral salts media and a modified Bushnell and Haas media the for isolation of hydrocarbon-utilizing bacteria referred to here as oil agar (OA) where diesel oil (0.1% w/v) was used as the carbon source in the medium 23 .
Culture dependent microbial analysis. The bacterial population of the soil samples for 0, 30, 60, and 90 days was enumerated by serially diluting 1 g of soil sample collected from the rhizosphere of C. cajan. Suitably diluted samples (10 −5 , 10 −6 and 10 −7 ) were transferred into the already prepared media using a dropper pipette, 500 µL of each dilution was inoculated on NA and OA which contained 0.1% v/v diesel agar surface 23 . The plates were incubated at 30 °C for 24 h, for NA, and five days for OA, respectively 23 . The number of viable total heterotrophic rhizobacteria and hydrocarbon-utilizing bacteria in the soil samples was estimated from the number of colonies formed using a colony counter.
Culture-independent microbial analysis. Culture-independent bacterial community in C. cajan rhizosphere after 90 days was determined using metagenomics analysis of the rhizospheric soil collected from the contaminated rhizosphere CR of 3% oily sludge and contaminated non-rhizosphere CN of 3% oily sludge as control.
Soil extraction and PCR amplification. Microbial DNA was extracted from samples in treatments CN3 and CR3 after 90 days using Power soil ® DNA Isolation Kits (MOBIO, USA) following the manufacturer's instruction. The marker region of the bacteria was amplified by PCR using a PCR (BioRad Thermal Cycler, United Kingdom) with the following conditions; 95 °C for 2 min, followed by 25 cycles at 95 °C for 30 s, 55 °C for 30 s, and 72 °C for 30 s, and a final extension for 5 min using the primers sequences 5'-barcode 2 -(reverse primer sequence) −3'and 5'-barcode 1-(forward primer sequence) -3', where the barcode consists of an eight-base sequence that is sample-specific and distinctive to each sample. The PCR reaction was carried out in triplicate 20 µL-mixtures containing 2 µL of 2.5 mM dNTPs, 4 µL of 5 × FastPfu buffer, 0.4 µL of FastPfu Polymerase, 0.8 µL of each primer (5 µM) and 10 ng of template DNA. Extraction of amplicons was carried out from a 2% agarose gel. Purification of the amplicons was carried out utilizing AxyPrep DNA Gel Extraction Kit (Axygen Bioscience, www.nature.com/scientificreports www.nature.com/scientificreports/ Union City, CA, USA) in accordance to the instruction furnished by the manufacturer and quantified using QuantiFluor TM -ST (Promega, USA). Construction of the library was carried out utilizing Linked 'Y' adapter; with adapter dimer removed by utilizing beads and the library was concentrated via PCR and hydroxides utilized in the generation of single-stranded DNA fragments. Sample libraries were then pooled in equimolar concentrations, and paired-end sequenced using the Illumina MiSeq platform (2 × 250/300 bp) according to standard protocols.

Data analysis
Several criteria were utilized for demultiplexing raw Fasta files and QIIME-based quality-filtration (version 1.9.1). The criteria include overlapped relationship and merging of paired-reads into a single read. The merged reads were then utilized for the clustering of operational taxonomic units (OUT), classification of taxonomy and assessment of community diversity. The software Trimmomatic was utilized for processing sequence reads 24 , and then the reads were assembled utilizing the software Flash 25 followed up by more analysis using the software MOTHUR v 1.33.0 26 . Alignment of unique sequences was based to the SILVA database with the settings put to default. In addition, removal of chimeric sequences was carried out. Obtained sequences that passed the screening process were then classified using the Ribosomal Database Project naïve Bayesian rRNA classifier at a confidence of 80%. The proportion of sequence identities was calculated at each taxonomic level as the percentage of all sequences classified in that particular sample. Classification of OTUs was based on the similarities of 97%. The software MOTHUR was utilized in the classification of the alpha-diversity indices, which include observed OTUs (Sobs), Ace, Chao, Shannon and InvSimpson indices. PCoA or principal coordinates analysis conducted in R (Version 3.1.2) was utilized to find the bacterial community structure, which is based on the OTU composition.

Measurement of biodegradation in the soil
Gravimetric method. Five grams of soil from all treatments were collected and transferred into 100 mL conical flasks. Then 50 mL of hexane was added and shaken on a Protech orbital shaker (USA) at 150 rpm for 24 h. The layer separating the solvent with oil and the soil was transferred to a pre-weighted clean conical flask, the sample was left overnight in a fume hood for the evaporation of the solvent, and the amount of residual TPH was gravimetrically determined using the formula.
Weight of oil (control) Weight of oil (degraded) Weight of oil (control) 100
In general, the results show a significant (p < 0.05) increase in the numbers of THRB counts in the uncontaminated rhizosphere (UR) compared to the uncontaminated non-rhizospheric (UN) bacterial counts. Although the microbial community in the rhizosphere is about 10-100 times higher than that of the non-rhizosphere 10 , the increase in the THRB counts in UR compared to UN as found in this study is marginal. As anticipated, the Scientific RepoRtS | (2020) 10:4094 | https://doi.org/10.1038/s41598-020-60668-1 www.nature.com/scientificreports www.nature.com/scientificreports/ presence of the contaminant POS significantly reduces the number of THRB counts in both of the contaminated rhizosphere (CR) and non-rhizosphere (CN) treatments over the uncontaminated treatments (UN and UR). The microbial counts were much more pronounced in the uncontaminated soil than the contaminated soil because of the inhibitory effects of POS to the microorganism in general in the contaminated soil. However, there was a significant increase in the number of bacterial populations over time in the rhizosphere treatments (CR), as there were higher bacterial counts after 90 days of treatment compared to that of 30 days while a decrease in THRB counts was generally observed in CN. This shows that the presence of C. cajan favours the growth of THRB bacteria in the soil as found in other studies 22,27 . The result is also supported by a previous finding 28 , who reported a total decrease in the number of viable heterotrophic bacteria in soil contaminated with oily sludge.
Hydrocarbon-utilizing rhizospheric bacterial counts revealed higher numbers in the contaminated rhizosphere (CR) for all of the POS concentrations tested, although the density tends to decrease as the POS concentration was increased. This result is similar to previous findings [29][30][31][32] , which include studies on the phytoremediation www.nature.com/scientificreports www.nature.com/scientificreports/ of soil amended with waste lubricating oil with Jatropha curcaa 31 and Hibiscus cannabinus 29 where significant increase in hydrocarbon-utilizing bacteria was observed after 30 days of incubation. However, both of these studies require the addition of organic wastes such as brewery spent grain (BSG) and spent mushroom compost (SMC) as additional carbon and nitrogen sources. The results obtained in this work demonstrate the stimulatory effect of rhizosphere to the hydrocarbon-utilizing bacterial population without the need for additional carbon or nitrogen sources.
Considering the complexity of the rhizosphere, a considerable number of microbial communities tend to withstand the toxic effect of the contaminant and are capable of using hydrocarbon as the source of carbon and energy compared to the community in the uncontaminated control. This is similarly reported in a study of epiphytic hydrocarbon-utilizing bacteria where the average number of bacterial density in a given contaminated soil is significantly greater than in the corresponding control, directly indicating that the contaminant is being utilized by the soil bacteria 33 . The results suggest that microbial enumeration is a direct indicative method to prove the response of microorganisms to hydrocarbons 34,35 . For a successful and effective phytoremediation process, the bacterial community in the hydrocarbon-contaminated soil must be well connected to the plant's ability to enhance microbial association in the rhizosphere, resulting in a higher number of hydrocarbon-utilizing bacteria and enhancing their degradative capacity 36 .
Apart from the presence of petroleum oil which serves as the carbon and energy source, the higher population of hydrocarbon-degrading bacteria in the contaminated soil may also be attributed to the additive effect of the C. cajan roots which release organic compounds to further stimulate the degradation and bacterial growth. The increase in HURB counts in the presence of C. cajans observed in this study are also observed in several similar studies where higher counts of heterotrophic and oil-degrading bacteria were observed in contaminated rhizospheric soil than in the unplanted contaminated soil 8,37 . Culture-independent metagenomics analysis. A sum of 59,873 and 59,756 sequences for CN3 and CR3 was found after a sequence optimization method, with an average number of 442.56 and 435.39 sequences for CN3 and CR3, respectively. A sum of 50821 and 47999 reads on CN3 and CR3, respectively, were subsampled from each replicate for further analysis. The calculated bacterial community abundance indices showed an increase in the values for Ace, Cho, Shannon (Shannon-Weaver) and the Simpson's index in CR3 compared to CN3 (Table 1). In phylogeny, OTU is the most commonly applied microbial diversity unit where OTU is clustered with a cutoff of 97% similarity for the investigation of the abundance of group or species in the microbial community. The difference between CN3 and CR3 was seen in the number of OTU, which were 512 and 650 for CN3 and CR3, respectively, indicating an increase in richness upon the addition of C. cajan. A similar increase in OTU from 48 (control) to 62 (addition of rhizobacteria) is reported during the rhizoremediation of hexachlorobenzene in constructed wetlands 38 . The coverage indices showed a significant difference in the two treatments. The coverage did not change by much in this study. In general, a reduction in coverage rate indicates higher diversity. In a similar study of rhizoremediation of hexachlorobenzene using T. angustifolia rhizosphere and P. australis rhizosphere in constructed wetlands 38 , little reduction in the coverage index was observed for T. angustifolia rhizosphere treatment (from 50 to 47.5) while greater reduction in coverage was observed in P. australis rhizosphere soil treatment (from 50 to 29) despite both treatments showing nearly equal efficacy in remediating hexachlorobenzene. This may imply that coverage change alone may not be adequate in describing potential remediating ability of rhizodegraders. Of all the indices used in population diversity studies, the robust Shannon and Simpson indices have been recommended in measuring microbial diversity 39 , and it was observed that both of the Simpson's (measured as InvSimpson) and the Shannon values for CR3 were higher than CN3. A change in both indicating an increase in diversity upon the introduction of C. cajan into the contaminated soil, which is a common theme seen in several studies involving phytoremediation using legumes 11,38,40 . Both measurements of population were lower in CN3, which may be due to the toxic effect of the contaminant on the bacterial community. The shift of soil bacterial community organization is also seen in the metagenomics sequences, and the results showed (Fig. 4a) that the contaminated rhizosphere (CR3) shows a diverse community of bacterial phyla, in comparison to the change of the microbial community structure seen in the contaminated non-rhizosphere (CN3). A lower diversity in CN3 may be the cause of a lower removal rate of petroleum hydrocarbon which has similarly been reported in previous studies 40,41 where with an increase in the concentration of the contaminant, this results in an increase in the toxicity which reduces the efficiency of microbial degradation. This shows the important role of plant-like C. cajan, which stabilizes the C:N:P ratio for the effective degradation of hydrocarbon by the microbial community 22 . On the other hand, some study reported the inability of plant growth-promoting rhizobacteria to acquire nutrient for growth in severely polluted environments 41 .
The composition and taxonomic analysis of microbiota amplified sequences were categorized into eight phyla (CN3) and ten phyla in (CR3). The overall bacterial composition of the two treatments varied, as the distribution shows a similar variation between the two treatments in the phylum, class, family and genus level distributions, represented by Fig. 4a to d, respectively. This variation in microbial community is also shown by the PCoA plot, www.nature.com/scientificreports www.nature.com/scientificreports/ which revealed community-level differences between the contaminated non-rhizosphere control (CN3) and contaminated rhizosphere (CR3) microbiota (Fig. 5).
The contaminated non-rhizosphere (CN3) shows a trend of bacterial phylum such as Proteobacteria, Actinobacteria, Acidobacteria, Bacteroidetes, Firmicutes, Chloroflexi, Saccharibacteria and some uncategorized group. The contaminated rhizosphere (CR3) shows Actinobacteria, Proteobacteria, Bacteroidetes, Acidobacteria, Firmicutes, Gemmatimonadetes, Saccharibacteria, Chloroflexi and Verrucomicrobia to be the dominant phyla, and  www.nature.com/scientificreports www.nature.com/scientificreports/ an unclassified group. In both of the treatments, the phylum Proteobacteria constitute the phylum with higher relative abundance accounting for > 60% in CN3 and almost 42% in CR3. In general, the phylum Proteobacteria dominated the communities of CN3 and CR3 at 70% and 42%, respectively. In the CN3, the Proteobacteria was dominated by the genus Rhizobium, Sphingomonas, and Herbaspirillum. The two phyla were not observed in CN3, but the phyla observed were Verrucomicrobia and Gemmatimonadetes. The presence of Verrucomicrobia is an indicator to the rhizospheric effect created by the presence of C. cajan as bacteria from this phylum are mostly found inhabiting grasslands and in subsurface soil horizons, where they were habitually the prevailing bacterial phylum 42 . Similarly, Gemmatimonadetes tend to dominant in soil with high rhizosphere activities, although their ecology remains poorly understood, and appear to be the dominant phyla in many soil bacterial communities; with bacteria from the phylum Gemmatimonadetes featuring nearly 2% of soil bacterial communities. Nevertheless, very little is understood of their ecology as a result of an insufficient study on the occurrence and ecology of this bacterial group 43 . The degradation of petroleum oily sludge hydrocarbons, in general, is accredited to indigenous microorganisms which are found in soil, but the presence of C. cajan will stimulate the habitat for the formation of favourable conditions of metabolisms to the microbial communities as demonstrated in this study where the culture-independent metagenomics technique to access a much more in-depth knowledge of the biological processes of petroleum oily sludge degradation during rhizodegradation shows promising results that agrees in principal to what was observed in the experiments. An assessment at the phylum level identifies the bacteria belonging to the Proteobacteria as the richest community, and the richness was significantly increased. Proteobacteria covers a group of Gram-negative bacteria that have been widely reported to be able to degrade POS 41 .
Biodegradation of petroleum oily sludge in soil. Gravimetric analysis. The result of oily sludge degradation in the soil shows the effectiveness of C. cajan in plant-microbe bioremediation process. The result of the gravimetric analysis shows that almost 50% biodegradation of the oily sludge was observed at lower concentrations of oily sludge (CR1%, CR2% and CR3%) after 30 days of planting C. cajan (Fig. 6).
A low percentage of biodegradation (19%) was observed at the highest concentration of oily sludge tested (CR5%) after 30 days of planting with C. cajan. This may be as a result of inhibition at high concentrations of POS. The biodegradation shows significant (p < 0.05) increase after 60 days of planting of C. cajan for all oily sludge concentrations with CR1%, CR2%, CR3% and CR4% showing 74, 65, 63 and 52% biodegradation of POS, respectively, while CR5% was at 37% which also shows a significant increase albeit at a much lower percentage of degradation. A significant (p < 0.05) increase in biodegradation of POS was shown at the 90 days of the planting of C. cajan where CR1%, CR2%, CR3% and CR4% show 92, 90, 89 and 68% degradation, respectively. Likewise, CR5% had the lowest biodegradation at 47%. This result corresponds to the pattern of many studies observed in different plants, and the biodegradation also varies. In a previous study 31 , they reported that the phytoremediation of soil contaminated with 2.5 and 1% of spent engine oil using J. curcas at day 180 results in the 56.6% and 67.3% biodegradation of waste lubricating oil, respectively. Once the addition of organic waste to J. curcas was carried out, the remediation rapidly increases the removal of 2.5 and 1% spent engine oil by 89.6 and 96.6%, respectively. The variation is that the plant was stimulated with organic waste to achieve 96% biodegradation at 1% spent engine oil concentration whereas in this study, C. cajan being a legume plant, the addition of organic source is not necessary making C. cajan a better phytoremediating plant. Another study in the plant Hibiscus cannabinus for soil contaminated with 2.5 and 1% used lubricating oil for 90 days show the same pattern of biodegradation where the stimulation of the plant with organic waste resulted in the biodegradation of 86.4 and 91.8%, respectively, while in the unstimulated plant much lower biodegradations were observed at 52.5 and 58.9%, respectively, indicating the need for the addition of organic waste in non-nitrogen fixing plants 29 . Legumes have been shown to independently stimulate biodegradation of various forms of hydrocarbons including poly-aromatic hydrocarbons (PAHs) and their constituents 44 . In all of the phytoremediation studies, legumes are more effective at remediation