Illumina MiSeq sequencing reveals the community composition of NirS-Type and NirK-Type denitrifiers in Zhoucun reservoir – a large shallow eutrophic reservoir in northern China

Abstract

Denitrification is a major biological process that reduces nitrate to nitrogen gas (N₂ or N₂O). In shallow eutrophic reservoirs, this process can remove the largest portion of fixed nitrogen and play an important role in the self-purification of this ecosystem. To understand the structure of denitrifying communities in a shallow eutrophic reservoir, NirS-Type and NirK-Type denitrifier communities were explored using a MiSeq high-throughput sequencing technique. NirK and NirS are two functional marker genes for denitrification that encode copper- and cytochrome cd1-containing nitrite reductases, respectively, which catalyze the reduction of nitrite to nitric oxide. MiSeq sequencing revealed a total of 13 102 and 3490 OTUs with 97% similarity for NirS-Type and NirK-Type denitrifiers, which contained 4 and 5 phyla, and 22 and 21 genera for NirS-Type and NirK-Type denitrifiers, respectively. The ACE, Chao diversity, and Shannon richness estimates for NirS-Type denitrifiers were higher than those for NirK-Type denitrifiers. However, the Simpson richness exhibited the opposite trend. The denitrification community structure exhibited significantly seasonal variations during the whole experiment period based on the PCA, Hcluster analysis, OTUs distributions, community composition, and heat map analysis. Meanwhile, the RDA indicated that the total sediment nitrogen, aerobic denitrifiers, functional genes, denitrification rates, and temperature were critical environmental factors influencing the seasonal variation in the bacterial denitrification community.

---

1. Introduction

Denitrification is a major cause of nitrogen loss from fixed nitrogen in ecosystems, and contributes to the production of gaseous end products (N₂O, and N₂), especially production of N₂O which is involved in global warming.¹ Denitrification could remove more than 50% of the nitrogen input into water systems, and plays an important role in water quality control and in self-purification of polluted water. The process of denitrification occurs through a sequence of intermediates (nitrate, nitrite, nitric oxide, and nitrous oxide), finally ending with gaseous end products. Denitrification is widespread among phylogenetically unrelated groups, which suggests that denitrifying bacteria are phylogenetically diverse, and hence cannot to be investigated by 16S rDNA analysis methods.² Fortunately, the nitrite reductases that reduce nitrite into nitric oxide are the main biocatalysts of the denitrification process. Recently, the functional nitrite reductase genes nirS and nirK have been proven useful molecular markers to resolve the community structure of denitrifying microbes. The choice for nirS and nirK as molecular markers for detecting denitrifying communities is strengthened by the fact that the common difference between true denitrifiers and other microorganisms with nitrate-reducing ability is that true denitrifiers have two different types of enzymes, a cytochrome cd1-containing nitrite reductase encoded by nirS (cdNIR) or a Cu-dependent nitrite reductase (CuNIR) encoded by nirK. Furthermore, the information gained from the sequences of nirS and nirK have provided a comprehensive measure of community diversity, and have been effectively adopted to elucidate the community composition of denitrifiers in various environmental samples, such as estuary,^3,4 lake,⁵ soil,⁶ wetland.⁷ In addition, the quantity of the denitrification functional genes (NirS and NirK) can be evaluated by PCR.⁸ Therefore, we selected NirS and NirK as gene markers to study the denitrifier community in the surface sediment of the Zhoucun reservoir.

Over the past few decades, denitrification and denitrifier communities have received widespread attention for their important contributions to the nitrogen cycle.⁹ Previous investigation of denitrifier diversity was based on clone libraries,^10,11 terminal-restriction fragment length polymorphism (T-RFLP),^12–14 and polymerase chain reaction-denaturing gradient gel electrophoresis (PCR-DGGE).¹⁵ In the past few years, denitrifier communities were studied by high-throughput sequencing technology. For example, Yu et al.¹⁶ investigated the denitrifier community in the oxygen minimum zone of a subtropical deep reservoir by 454 pyrosequencing. While Fan et al.¹⁷ studied the diversity of denitrifiers in Lake Taihu through the Illumina MiSeq high-throughput sequencing technology, and Wang et al.¹⁸ used the Illumina HiSeq 2000 high-throughput sequencing to analyze the diversity of denitrifiers in tannery wastewater treatment plants.

Zhoucun reservoir was a large shallow eutrophic reservoir, located in the northern region of China, with a surface area of 8.54 km², a maximum depth of about 20 m, with a corresponding storage capacity of 84.29 × 10⁶ m³. Fish breeding in the reservoir started in the 1980s, and the scale expanded rapidly. The fish breeding area reached 20% of the reservoir and the net cases have exceeded 10 000. Net case fish breeding has caused serious deterioration of the water quality in the reservoir. After clearing up the network cases, the water quality of the reservoir improved, but the sediments of the reservoir remained polluted. However, there is little research to date on the denitrifier community by high-throughput sequencing (NirS and NirK genes), especially on a water source reservoir with seriously endogenous pollution.

Therefore, samples of surface sediment of the main reservoir and environmental parameters were simultaneously measured over a whole year. Real-time quantitative PCR (qPCR) was applied to assess the abundance of denitrifiers based on the presence of the NirS gene. The objectives of this study were: (i) to investigate the diversity and composition of denitrification using the NirS and NirK genes as molecular markers; (ii) to determine the abundance of NirS- and NirK-Type denitrifiers and explore their seasonal distribution; and (iii) to investigate the relationship between the denitrification community structure and environmental driving factors across the whole experimental period by redundancy analysis (RDA). This study could enrich our understanding of the denitrifying bacterial community structure and provide references to the research, protection, and pollution control of freshwater environments.

2. Methods

2.1 Sample collection and preparation

In this study, surface sediment sampling was performed monthly over twelve months from September of 2014 to August 2015, and water samples (2.5 L) were collected in association with the sediment site at depths of 0.5 m, 2.5 m, 5.0 m, 7.5 m, 10 m, 12.5 m, and bottom water layer (the maximum depth of the reservoir was approximately 13–15 m) for water chemistry analysis. Sampling locations are shown in Fig. S1.† All samples were kept in iceboxes and shipped to the water research laboratory within 6 h.

Meanwhile, environmental parameters of the sampling site were measured in situ in 0.5 m increments using a multi-parameter water quality analyzer (Hydrolab DS5, HACH Company, USA). In detail, T (temperature), DO (dissolved oxygen), pH, ORP (oxidation–reduction potential), EC (electrical conductivity), and CHl-a (chlorophyll-a) were analyzed. The water parameters were measured using a spectrophotometer (DR6000; HACH Company, USA). Specifically, the TN and nitrate concentrations were measured using hydrochloric acid photometry.¹⁹ The TP concentration was measured by ammonium molybdate spectrophotometric method.¹⁹ The COD_Mn was analyzed using the potassium permanganate method.¹⁹ The TN and TP (surface sediment) were determined by persulfate. The MC (moisture content) was determined by weighing the sediment samples in an oven at 105 °C for 12 h. Aerobic denitrifier abundance was calculated by plate count on a solid screening medium (in g L⁻¹: 0.1 CH₃CO₂Na, 0.02 NaNO₃, 0.02 K₂HPO₄·3H₂O, 0.01 CaCl₂, 0.01 MgCl₂·6H₂O, 20 agar; pH 7.2).²⁰ Finally, surface sediments were collected at a deep layer of 0–2 cm using a sterilized Petersen stainless steel grab sampler.

2.2 Sediment DNA extraction

In order to obtain total DNA, ∼50 mL of surface sediment (0–2 cm) was collected. Microbial DNA was extracted from sediment samples using the E.Z.N.A.® Soil DNA Kit (Omega Bio-tek, Norcross, GA, U.S.) according to the manufacturer's protocols. DNA was purified using the AxyPrep DNA Gel Extraction Kit (Axygen Biosciences, Union City, CA, U.S.) according to the manufacturer's instructions. The extracted DNA was stored at −80 °C for PCR amplification analysis.

2.3 Quantification of nirS and nirK gene abundance

The NirS gene was used as a molecular marker to determine the abundance of denitrifying bacteria in the surface sediment system during the entire experimental period. Quantitative PCR was performed in an ABI7500 Real-Time PCR Machine (Life Technologies, USA) using the SYBR Green method with two primers pair for nirS (cd3aF, 5′-GTSAACGTSAAGGARACSGG-3′/R3cd: 5′-GASTTCGGRTGSGTCTTGA-3′) and nirK (F1aCu: 5′-ATYGGCGGVCAYGGCGA-3′/R3Cu: 5′-GCCTCGATCAGRTTRTGGTT-3′).²⁰ Each 25 μL qPCR mixture contained 12.5 μL of SYBR Green qPCR Master Mix, 0.5 μL of each primer (10 μM), 9.5 μL of dd H₂O, and 2.0 μL of template (cDNA). The reaction was initially denatured at 95 °C for 10 min, followed by 40 cycles of denaturation at 95 °C for 15 s, annealing at 60 °C for 60 s and extension at 60 °C for 60 s. To determine gene abundance in one ng of extracted DNA, all targeted genes were cloned to plasmids following the method recommended by Zhang et al.²¹ to generate qPCR standard curves. In order to correct for potential variations in DNA extraction efficiencies, eubacterial 16S rRNA genes were also quantified using the method recommended by López-Gutiérrez et al.²² All samples and standards were analyzed in triplicate, and the specificity of products was checked by observation of the melting curve and agarose electrophoresis.

2.4 Microbial community analysis using Illumina MiSeq sequencing

To explore the microbial community composition of NirS-Type and NirK-Type denitrifiers in the surface sediment of the Zhoucun reservoir system, Illumina MiSeq high-throughput sequencing was performed by Shanghai Majorbio Bio-pharm Technology Co., Ltd (Shanghai, China). DNA extracted from surface sediment samples (as described above) was amplified by PCR using primers nirS (cd3aF: 5′-GTSAACGTSAAGGARACSGG-3′/R3cd: 5′-GASTTCGGRTGSGTCTTGA-3′)²³ and nirK (nirK1aCu: 5′-ATCATGGTSCTGCCGCG-3′/nirKR3Cu: 5′-GCCTCGATCAG(A/G)TTGTGGTT-3′).²⁴ PCR amplification was carried out in triplicate using ABI GeneAmp® 9700 PCR System in a total volume of 20 μL PCR reaction mix containing 5× FastPfu buffer (4 μL), 2.5 mM dNTPs (2 μL), 5 μM forward primer (0.8 μL), 5 μM reverse primer (0.8 μL), FastPfu polymerase 0.4 μL, BSA 0.2 μL, template DNA 2 μL, dd H₂O 9.8 μL for NirS denitrifier; 10× buffer (2 μL), 2.5 mM dNTPs (2 μL), 5 μM forward primer (0.8 μL), 5 μM reverse primer (0.8 μL), rTaq polymerase (0.2 μL), BSA 0.2 μL, template DNA 10 ng, dd H₂O 14 μL for NirK denitrifier. Thermal cycling (NirS and NirK) was performed at 95 °C for 3 min, with 40 cycles at 95 °C for 0.5 min, 55 °C for 0.5 min, 72 °C for 1 min, and a final extension at 72 °C for 10 min. All PCR products were identified by gel electrophoresis and purification. The purified amplicons were sequenced on an Illumina MiSeq platform. The amplicons with sequences shorter than 200 bps and of low quality (quality score < 25) were removed.²⁵ The taxonomic classification of effective sequences was determined using the RDP (Ribosomal Database Project) database (http://rdp.cme.msu.edu/). For fair comparison, the size of each surface sample was normalized to the same sequencing depth by randomly removing the redundant reads.

2.5 Statistical analysis

Rarefaction curves (RC), abundance-based coverage estimators (ACE), and community diversity indices (Chao richness estimator, the Shannon index and Simpson) were calculated by MOTHUR.²⁶ The advanced analysis, the difference in composition of species of samples was compared with Venn. PCA, PCoA, and Nmds were performed using R project for statistical computation and were used to describe the relationship between different species or samples. The gradient length of the longest axis was calculated using Canoco 5.0 with detrended correspondence analyses (DCA); if the maximum length was shorter than 3 SD (standard deviation) units, PCA and RDA (linear model) were more suitable.²⁷ In our study, the length of the first gradient was 0.58 for nirS and 0.83 for nirK, so redundancy analysis (RDA) was chosen to construct the relationship between species and environment factors.

Analysis of variance (one-way ANOVA) was used to determine the significance of difference (P < 0.05) in the abundance of functional genes among samples during the whole experimental period using the software SPSS 20. Pearson correlation coefficients were also performed to describe the relationship between denitrifying bacteria and environmental factors.

3. Results and discussion

3.1 Environmental factors and nutrient conditions

The Zhoucun reservoir in northern China is a typical large shallow eutrophic reservoir. The water column was well mixed in winter and spring (Dec.–Apr.), whereas stratification events appeared in spring (May) and lasted until autumn (Fig. 1). In detail, the metalimnion was observed from 4–9 m in spring (May) and then dropped to 6–9 m in summer, and then reached between 9 and 12 m depths in autumn. Meanwhile, during the stratification period, the DO reached 0 mg L⁻¹ at the 8 m water layer in spring (May) and then appeared between 4 and 6 m water layers in summer (Jun.–Aug.), and then dropped to 9 to 10 m in autumn (Sep. and Oct.). The DO concentration of bottom water remained at 0 mg L⁻¹ from September–October 2014 and May–August 2015, and the thickness of the anaerobic layer ranged from 1.0 to 9.5 m with the highest value at 9.5 m (Jun. 2015). The temperature of bottom water ranged from 3.35 °C to 12.82 °C during the whole experimental period, the temperature of stratification and mixed periods maintained 11.73 ± 1.13 °C and 4.81 ± 1.22 °C, respectively. Moreover, nutrients presented significant temporal and spatial variation in the Zhoucun reservoir, and an obvious increase of TP and COD was observed in the stratification period, though the trend of TN was unclear. The STN (sediment total nitrogen), STP (sediment total phosphorus), and MC (moisture content) exhibited significant seasonal variations during the whole experimental period (P < 0.05).

Fig. 1 Distribution of environmental factors and nutrient concentration in Zhoucun reservoir. (A) Changes of temperature; (B) changes of DO concentration; (C) changes of ORP concentration; (D) changes of pH concentration; (E) changes of EC concentration; (F) changes of CHl-a concentration; (G) changes of TN concentration; (H) changes of TP concentration; (I) changes of COD concentration.

3.2 Abundance of aerobic denitrification bacteria and denitrification function gene

The average abundance of aerobic denitrifying bacteria in the fresh surface sediment of the Zhoucun reservoir ranged from (3.55 ± 0.21) × 10³ cfu g⁻¹ (Feb. 2015) to (1.50 ± 0.11) × 10⁹ cfu g⁻¹ (Aug. 2015). The abundance of the denitrification functional gene (nirS and nirK) was also explored during the whole experimental period. In detail, the abundance of the nirS gene ranged from (6.13 ± 0.61) × 10⁴ copies per μL (Dec. 2014) to (9.15 ± 1.13) × 10⁶ copies per μL (May 2015); the abundance of the nirK gene ranged from (3.47 ± 0.31) × 10⁴ copies per μL (Dec. 2014) to (3.63 ± 2.05) × 10⁶ copies per μL (May 2015). Meanwhile, nirS had higher abundance than nirK in each of the surface sediment samples during the whole experiment period and the abundance of nirS and nirK genes were higher for stratification period than that for mixing period in Zhoucun reservoir (Table 1).

Table 1 Changes in sediment moisture content (MC), sediment total nitrogen (STN), sediment total phosphorus (STP), aerobic denitrifying bacteria and the denitrification functional genes in the surface sediment systems of the Zhoucun reservoir^a

System	NirS	NirK	AD	STN	STP	MC
	Mean ± S.D.	Mean ± S.D.	Mean ± S.D.	Mean ± S.D.	Mean ± S.D.	Mean ± S.D.
a Note, NirS mean NirS functional genes (copies per μL); NirK mean NirK functional genes (copies per μL); AD mean aerobic denitrification bacteria (lg(cfu g^−1)), we collected 1 g fresh sediment, and counted the number of aerobic denitrification bacteria through gradient dilution.
Sep.	5.72 ± 0.01	5.52 ± 0.01	8.32 ± 0.01	3287.20 ± 54.77	2258.44 ± 141.64	0.71 ± 0.00
Oct.	6.21 ± 0.07	6.11 ± 0.02	5.01 ± 0.03	4857.29 ± 188.72	2048.52 ± 155.95	0.75 ± 0.00
Nov.	5.22 ± 0.03	5.02 ± 0.03	4.38 ± 0.03	1515.12 ± 60.16	652.46 ± 36.38	0.37 ± 0.00
Dec.	4.79 ± 0.04	4.54 ± 0.02	4.01 ± 0.02	3761.95 ± 66.36	1153.25 ± 91.81	0.74 ± 0.01
Jan.	5.47 ± 0.02	5.13 ± 0.04	3.65 ± 0.02	3715.65 ± 156.73	1413.53 ± 33.62	0.68 ± 0.01
Feb.	5.54 ± 0.02	5.03 ± 0.02	3.56 ± 0.04	4250.35 ± 64.76	2251.96 ± 181.12	0.76 ± 0.00
Mar.	6.70 ± 0.05	5.80 ± 0.05	6.02 ± 0.03	1199.20 ± 36.67	655.48 ± 30.59	0.85 ± 0.00
Apr.	5.71 ± 0.08	5.75 ± 0.06	6.86 ± 0.03	4477.59 ± 215.85	708.86 ± 17.97	0.83 ± 0.00
May.	6.96 ± 0.05	6.56 ± 0.05	7.49 ± 0.05	4166.10 ± 124.40	644.20 ± 17.63	0.83 ± 0.00
Jun.	6.54 ± 0.05	6.44 ± 0.03	8.14 ± 0.04	4136.18 ± 161.11	722.52 ± 60.47	0.84 ± 0.00
Jul.	6.63 ± 0.04	6.23 ± 0.02	8.45 ± 0.06	4025.07 ± 98.31	679.07 ± 12.67	0.81 ± 0.00
Aug.	6.54 ± 0.06	6.14 ± 0.03	9.17 ± 0.03	3509.77 ± 194.49	536.97 ± 68.78	0.84 ± 0.00
Whole	6.00 ± 0.69	5.69 ± 0.64	6.25 ± 2.07	3575.12 ± 1120.5	1143.77 ± 677.03	0.75 ± 0.13
P	0.35	0.45	0.44	0.00	0.00	0.00
F	0.90	0.59	0.63	103.57	31.99	38.98

---

The Pearson correlation showed that nirS positively correlated with AD (aerobic denitrification bacteria) (R = 0.641, P < 0.05) and T (R = 0.325, P > 0.05) during the whole experimental period. However, it could not be concluded that the abundance of nirS played a greater role in the nitrogen removal. Thus, further studies based on mRNA and protein expression have to be conducted.

3.3 Pyrosequencing overview of NirS and NirK denitrifiers

The microbial diversity and community structure of NirS-Type and NirK-Type denitrifiers were investigated using MiSeq high-throughput sequencing. After removing low-quality sequences, a total of 250 694 sequences (no. Dec. 2014 sample site) with an average length of 393 bp for NirS-Type denitrifiers and 257 058 sequences with an average length of 450 bp for NirK-Type denitrifiers were obtained in each sample. After quality trimming, there were 23 samples in total (11 samples for NirS-Type denitrifiers and 12 samples for NirK-Type denitrifiers) analyzed in triplicate. Each sample generated 756–1537 OTUs (operational taxonomic units) for NirS-Type and 245–341 OTUs for NirK-Type denitrifiers, with coverage of 0.97–0.99 and 1.00 for NirS-Type and NirK-Type denitrifiers, respectively, at 97% similarity (Table 2). These results seem to reflect the real structure of the microbial communities.²⁸ The highest number of OTUs in the surface sediment system was found in Jun. 2015 for both NirS-Type and NirK-Type denitrifiers. The ACE and Chao diversity estimators for NirS-Type and NirK-Type denitrifiers, respectively presented significant seasonal variation over the experimental period, and the diversity for NirS-Type denitrifiers was obviously higher than that for NirK-Type denitrifiers. Meanwhile, the Shannon richness of NirS-Type denitrifiers was also higher than that for NirK-Type denitrifiers. However, the Simpson richness exhibited an opposite trend.

Table 2 Spatial and temporal distribution of NirS-Type and NirK-Type denitrifier's community diversity and richness estimators in the surface sediment systems of the Zhoucun reservoir^a

Period	Water depth	NirS-Type denitrifier							NirK-Type denitrifier
		Reads number	0.97 level						Reads number	0.97 level
			OTUs	Diversity		Coverage	Richness			OTUs	Diversity		Coverage	Richness
				ACE	Chao1		Shannon	Simpson			ACE	Chao1		Shannon	Simpson
a Note, ACE, abundance based-coverage estimator; OTUs, operational taxonomic units; diversity, diversity estimator; richness, richness estimator.
Autumn	Sep.	34 064	1286	1610	1628	0.9895	5	0.0291	20 786	290	326	320	0.9973	2.01	0.3126
	Oct.	17 704	1028	1351	1365	0.9815	4.8	0.0444	17 560	311	376	382	0.9953	1.96	0.4168
	Nov.	11 317	756	1009	984	0.9784	4.46	0.07	18 616	328	351	347	0.9976	2.64	0.2383
Winter	Dec.	—	—	—	—	—	—	—	16 295	293	309	306	0.9979	2.54	0.2098
	Jan.	9896	819	1553	1304	0.9664	4.85	0.036	21 725	267	305	308	0.9975	1.76	0.4549
	Feb.	13 659	990	1325	1287	0.9765	5.17	0.0213	17 020	245	318	301	0.9955	1.42	0.5662
Spring	Mar.	35 710	1435	1815	1818	0.9890	5.5	0.012	26 988	290	324	315	0.9981	2.69	0.1871
	Apr.	18 537	1344	1731	1751	0.9787	5.94	0.006	22 193	255	302	314	0.9974	2.84	0.1825
	May.	21 093	1244	1699	1641	0.9808	5.45	0.0128	20 962	302	328	317	0.9979	2.65	0.1846
Summer	Jun.	35 216	1537	1930	2008	0.9878	5.59	0.0122	29 467	341	378	379	0.9979	2.8	0.1675
	Jul.	23 419	1247	1635	1661	0.9842	5.63	0.01	21 068	253	277	272	0.9981	2.63	0.2007
	Aug.	30 079	1416	1740	1693	0.9880	5.55	0.0127	24 378	315	339	333	0.9981	2.77	0.1725

---

Rarefaction curves, Shannon–Wiener curves, and rank-abundance curves for NirS-Type and NirK-Type denitrifiers at 97% similarity (Fig. 2) showed that the MiSeq high-throughput sequencing of NirS-Type and NirK-Type denitrifiers could supply enough information to explore variations in the bacterial denitrification community. All OTUs with 97% similarity were assigned to 4 phyla for NirS-Type and 5 phyla for NirK-Type denitrifiers, and 22 genera and 21 genera for NirS-Type and NirK-Type denitrifiers, respectively.

Fig. 2 Rarefaction curves of OTUs (operational taxonomic units) number at 97% similarity boxplot for spatial and temporal distributions of microbial community of water and surface sediment systems in enclosure experiment. (A) Rarefaction curves for NirS-Type denitrifier (a) rarefaction curves of samples for NirK-Type denitrifier (B) Shannon index of species for NirS-Type denitrifier (b) Shannon index of species for NirK-Type denitrifier (C) rank-abundance for NirS-Type denitrifier (c) rank-abundance for NirK-Type denitrifier.

3.4 Seasonal changes of the NirS-Type denitrifier community

The microbial compositions of NirS-Type and NirK-Type denitrifiers were significantly different at different periods during the experimental period. Based on the NirS-Type denitrifier community, the main phyla included Proteobacteria (dominant 1, 66.28 ± 3.52%, P < 0.05), Bacteria_unclassified (dominant 2, 26.33 ± 2.76%, P < 0.05), and environmental_samples (dominant 3, 7.39 ± 1.72%, P = 0.23) (Fig. S1-A†); the β-proteobacteria (dominant 1, 51.08 ± 4.34%, P = 0.00) was the largest class of the Proteobacteria phyla (Fig. S1-B†). Dechloromonas was the main genus (31.43 ± 9.40%, P = 0.00) (Fig. S1-C and S4-A†), and as shown in Fig. S1-D,† the largest species was Dechloromonas aromatica (11.16 ± 3.55%, P = 0.00) during the whole experiment period, which was consistent with the strains isolated from Pearl River sediment.⁵ The variations of order for NirS-Type denitrifiers are shown in Fig. 3-A; the main order exhibited significant seasonal changes. In detail, Rhodocyclales ranged from 48.02 ± 1.80% for autumn (2014), to 30.63 ± 11.80% for winter (2014), to 32.95 ± 7.91% for spring (2015), and to 35.43 ± 3.89% for summer (2015); the Burkholderiales ranged from 0.24% to 2.46%; the unclassified bacteria (Bacteria_unclassified, Alphaproteobacteria_unclassified, Betaproteobacteria_unclassified, Proteobacteria_unclassified, and environmental_samples_norank) accounted for 49.30–76.10%. Megumi et al. (2009) showed that some clones from Rhodocyclales and Burkholderiales bacteria in a rice paddy field soil were related to the NirS gene.²⁹ Saito et al. (2008) reported that the majority of NirS clones in a succinate-assimilating denitrifying population were Burkholderiales and Rhodocyclales bacteria.³⁰ Zeng et al. (2016) also showed that the NirS gene was related to Rhodocyclales and Burkholderiales from municipal wastewater.³¹ Dechloromonas aromatica^32,33 (soil system and sequencing batch reactor) and Leptothrix (soil system)³² were also isolated. Zeng's study showed that the Dechloromonas sp. (AM230913) belong to the order Rhodocyclales at 71–94% identity in a constructed wetland.³¹ The Azospira exhibited significant seasonal difference with 0.84 ± 2.06% (P = 0.00) over the whole experimental period, which was similar to Azospira oryzae.³⁴ Some of the nirS clones (5.02 ± 1.28%, P = 0.37) were similar to Rhodocyclaceae bacteria from agricultural soil.^11,35,36 Meanwhile, the Betaproteobacteria_unclassified accounted for 12.51 ± 5.31% (P = 0.00) during the whole experimental period. Previous studies showed that a large amount of Betaproteobacteria (Acidovorax,³⁷ Thauera,³⁸ Alcaligenes faecalis,³⁹ Zoogloea sp.,^40,41 and Azoarcus⁴²) were involved in nitrogen cycling,⁴³ however this was not consistent with our study. Therefore, further research is required to explain this discrepancy.

Fig. 3 A color-scale heat map of surface sediment based on NirS-Type and NirK-Type denitrifiers showing the top 100 representative predominant 16S rRNA gene-based microbial sequences (at order level) in Zhoucun reservoir (red colors indicate higher abundance; blue and green colors indicate lower abundance).

3.5 Seasonal changes of the NirK-Type denitrifier community

Nitrite reductase occurs in two structurally different, but functionally equivalent, forms: nirS and nirK. Therefore, it is also important to analyze the community changes for NirK-Type denitrifiers during the experimental period. As shown in Fig. S1-E, S1-F, and S4-B,† the Bacteria_unclassified accounted for 92.49 ± 2.89% (P = 0.00) at phyla and genus levels, which was different from the NirS-Type denitrifier. The variations of NirK-Type denitrifiers at the order level are exhibited in Fig. 3-B. In detail, the Rhizobiales was the main order (0.26–1.21%) aside from unclassified_bacteria. Guo et al. (2011) showed that the majority of clones were related to Rhizobiales (Rhizobiaceae, Bradyrhizobiaceae and Phyllobacteriaceae) based on the nirK clone library analysis.³⁶

It is well known that denitrification plays a critical role in the nitrogen cycle. N-Functional bacteria are rarely cultured and studied in their natural ecosystem, especially in the sediment of drinking water reservoirs. Therefore, future research using high-throughput GeoChip functional gene microarray analysis, investigating denitrification activity, denitrifying enzymes, and nitrogen metabolic genes is necessary to further study the community structure and function of denitrifying bacteria (Table 3).

Table 3 The information of denitrification bacteria (high-through sequence)^a

Denitrifiers (genus)	P.A. (class, order, family)	Descriptions	References
a Note, P.A. means phylogenetic affiliation.
Hyphomicrobium	α-Proteobacteria, Rhizobiales, Hyphomicrobiaceae	Wastewater (industrial, municipal, seawater)	44 and 45
Paracoccus	α-Proteobacteria, Rhodobacterales, Rhodobacteraceae	Wastewater (municipal, seawater)	46
Azoarcus	β-Proteobacteria, Rhodocyclaceae, Rhodocyclaceae	Wastewater (industrial, municipal)	42
Thauera	β-Proteobacteria, Rhodocyclaceae, Rhodocyclaceae	Wastewater (industrial, municipal)	38
Methylophaga	β-Proteobacteria, Rhodocyclaceae, Methylophaga	Wastewater (seawater)	47 and 48
Accumulibacter	β-Proteobacteria, Rhodocyclaceae, Rhodocyclaceae	Wastewater (municipal)	49 and 50
Pseudomonas	γ-Proteobacteria, Pseudomonadales, Pseudomonadaceae	Wastewater (industrial, municipal)	51
Acidovorax	β-Proteobacteria, Burkholderiales, Comamonadaceae	Wastewater (industrial, municipal)	37
Dechloromonas	β-Proteobacteria, Rhodocyclales	Sediment (reservoir)	32 and 33, this study
Azospira	β-Proteobacteria, Rhodocyclales	Aquifer and sediment (reservoir)	34, this study
Leptothrix	β-Proteobacteria	Soil system and sediment (reservoir)	32, this study
Rhizobiales	α-Proteobacteria, Rhizobiales	Sediment (reservoir)	This study

---

3.6 Comparison of microbial community structure

Several statistical methodologies were used to identify the relationship between NirS-Type and NirK-Type denitrifying bacterial communities. Surface sediment samples were collected from the main reservoir area during the whole experimental period. Community structure comparisons were done by principal component analysis (PCA) using Canoco 5.0.

The PCA results are shown in Fig. 4-A and a. Results revealed that the first two principle components (PC1 and PC2) explained 58.63% and 48.43% of the variability for NirS-Type and NirK-Type denitrifiers in the surface sediment, respectively. The accumulated contribution ratios of PC1 and PC2 for NirS-Type denitrifiers achieved 40.40% and 18.23%, respectively (Fig. 4-A), whereas, for NirK-Type denitrifiers, the accumulated contribution ratios of PC1 and PC2 achieved 29.48% and 18.95%, respectively (Fig. 4-a). The microflora of the sediment system were well separated at different time periods for NirS-Type and NirK-Type denitrifiers over the experimental period. Samples from the same season showed tighter clustering, especially spring and summer samples, while the others were relatively distributed. In detail, as shown in Fig. 4-A, PCA1 of samples from the stratification period (Mar.–Aug. 2015) were all negative for NirS-Type denitrifiers (quadrants 2 and 3), while the samples of autumn and winter were positive. Meanwhile, PCA1 of samples from the same period (Mar.–Aug. 2015) were positive for NirK-Type denitrifiers (quadrants 4 and 1), while the samples from autumn and winter were negative (Fig. 4-a). The distributions of species for NirS-Type and NirK-Type denitrifiers are shown in Fig. 4-B and b, respectively. The denitrifiers were distributed in different sample sites and from different seasons during the experimental period. Moreover, the PC1 and PC2 values for NirS-Type and NirK-Type denitrifiers were significantly different during the whole experiment period.

Fig. 4 Principal component analysis (PCA) of NirS-Type and NirK-Type denitrifiers of surface sediments in Zhoucun reservoir. (A) Distribution of samples for NirS-Type denitrifier (a) distribution of samples for NirK-Type denitrifier (B) distribution of species for NirS-Type denitrifier (b) distribution of species for NirK-Type denitrifier (C) PC1 and PC2 values for NirS-Type denitrifier (c) PC1 and PC2 values for NirK-Type denitrifier.

The observed difference in the surface sediment at different periods, and their low similarity, suggested different and rapidly changing bacterial communities throughout the experiment period (Fig. S2†). The results showed that the numbers of shared genera for NirS-Type and NirK-Type denitrifiers did not vary much in the same season (395, 540, 554, and 558 OTUs for NirS-Type denitrifiers; 104, 84, 112, and 110 OTUs for NirK-Type denitrifiers). The hierarchical clustering based on OTU information was also generated (Fig. S3†). Sediment samples for NirS-Type and NirK-Type denitrifiers from the same season were tightly grouped, and could be well separated between different seasons.

3.7 Environmental effect on the community classification of the NirS-Type and NirK-Type bacteria

In order to explore the effects of environmental factors on NirS-Type and NirK-Type bacterial communities, multiple statistical analyses (RDA) were used to identify the relationship between the microbial functional community and environmental variables based on the genus level. This helped identify the missing link between diversity and activity using denitrifying bacteria as a model organism. The different bacterial communities for NirS-Type and NirK-Type denitrifiers were well discriminated at the genus level (Fig. 5-A and a).

Fig. 5 Redundancy analyses of NirS-Type and NirK-Type denitrifiers in the surface sediment of Zhoucun reservoir. (A) RDA of NirS-Type denitrifiers community based on samples and environment factors; (a) RDA of NirK-Type denitrifiers community based on samples and environment factors; (B) RDA of NirS-Type denitrifiers community based on species and environment factors; (b) RDA of NirK-Type denitrifiers community based on species and environment factors; (C) RDA of NirS-Type denitrifiers community based on species and samples factors; (c) RDA of NirK-Type denitrifiers community based on species and samples factors. Other details please refer to the above descriptions.

Based on samples and environmental variations in the NirS-Type denitrifier community, the first two RDA dimensions using the 7 parameters (VIF < 20, except T = 22) explained 55.75% of the microbial community variation (Fig. 5-A, F = 1.68, P = 0.066). Physical and chemical parameters including STN, DR, AD, nirS, and T significantly influenced the water bacterial community composition (Fig. 5A and Table 4). Temperature (T, environment factor) was the critical factor for the variation of the denitrifier community, and exhibited positive correlations with the abundance of AD (R = 0.665, P < 0.05) (Table S2†) and nirS (R = 0.325, P > 0.05) (Table S2†). The temperature effect on denitrifiers was Dechloromonas > environmental_samples_norank > Rubrivivax > Burkholderiales_unclassified > Leptothrix³² (Fig. 5-B). Moreover, the main denitrifiers (Dechloromonas, Bacteria_unclassified, Proteobacteria_unclassified, and Betaproteobacteria_unclassified) were distributed during the stratification period of the reservoir (Fig. 5-C).

Table 4 Critical environmental factors influencing the spatial and temporal variation for NirS-Type and NirK-Type denitrifiers during the whole experiment period in Zhoucun reservoir^a

Samples	Variations	RDA1	RDA2
a Note, STN, TN (sediment); STP, TP (sediment); MC, moisture content; DR, denitrification rate (sediment) was showed in Table S1; AD, aerobic denitrification bacteria (sediment); NirS, mean nirS function gene; NirK mean nirK functional genes; T, temperature; * means r > 0.50.
NirS-Type	STN	−0.18	−0.52*
	STP	0.49	−0.17
	MC	−0.34	−0.39
	DR	0.38	0.58*
	AD	−0.90*	0.14
	NirS	−0.80*	0.10
	T	−0.52*	0.66*
NirK-Type	STN	0.30	−0.12
	STP	0.87*	0.07
	MC	−0.17	−0.24
	DR	−0.33	0.62*
	AD	−0.40	−0.31
	NirK	−0.33	−0.82*
	T	−0.59*	−0.04

---

The effects of environmental factors on the NirK-Type denitrifier community are shown in Fig. 5-a (samples – environment factors), Fig. 5-b (species – environment factors), and Fig. 5-c (samples – species). As shown in Fig. 5-a, the first two RDA dimensions using the 7 parameters (VIF < 10) explained 40.98% (F = 1.4, P = 0.112) of the variability for NirK-Type denitrifiers in the surface sediment. The accumulated contribution ratios of RDA1 and RDA2 for NirK-Type denitrifiers achieved 26.50% and 14.48%, respectively. The STP and T were critical environmental factors for the NirK-Type denitrifiers; the factors (T and STP) exhibited a significant effect on the variation of the main denitrifiers (Fig. 5-b). In detail, the T factor affected environmental_samples_norank > Afipia > Proteobacteria_unclassified > Bradyrhizobium > Bradyrhizobiaceae_unclassified > _unclassified and the STP factor affected Bacteria_unclassified > Rhizobiales_unclassified > Alphaproteobacteria_unclassified. Furthermore, the main denitrifiers were also found during the stratification period (Fig. 5-c).

From this study it is clear that the physical factor (temperature) and nutrients (nitrogen and phosphorus) were the most important factors affecting the function and composition of the denitrifier community. Therefore, changes in N-functional bacteria have important implications in the geochemical cycle. Future research is needed to further explore the mechanism.

4. Conclusions

The densities of aerobic denitrifying bacteria, STN concentration, and denitrification rate all exhibited significantly seasonal differences during the whole experiment period. Based on the qPCR of denitrifying functional genes (nirS and nirK), the results indicated that the abundance of the nirS gene in the surface sediment was higher than that for the nirK in Zhoucun reservoir during the experiment period. All OTUs with 97% similarity were assigned to 4 phyla and 5 phyla for NirS-Type and NirK-Type denitrifiers, and 22 genera and 21 genera for NirS-Type and NirK-Type denitrifiers, respectively, through the high MiSeq throughput sequencing technology. The Dechloromonas (31.43 ± 9.40%, P = 0.00) and Rhizobiales (0.26–1.21%) were the main genus and order for NirS-Type and NirK-Type denitrifiers. Meanwhile, the NirS-Type and NirK-Type denitrifying bacterial community presented significant seasonal difference by the PCA analysis. Moreover, STN, denitrification functional genes, AD, T, and DR were the most important factors affecting the bacterial community function and composition.

Acknowledgements

This work was supported by the National Science and Technology Pillar Program (Grant No. 2012BAC04B02) and National Natural Science Foundation of China (No. 51478378).

References

1. K. Maeda, R. Morioka, D. Hanajima and T. Osada, Microb. Ecol., 2010, 59, 25–36.
2. L. Philippot, S. Hallin and M. Schloter, Adv. Agron., 2007, 96, 249–305.
3. G. C. Abell, A. T. Revill, C. Smith, A. P. Bissett, J. K. Volkman and S. S. Robert, ISME J., 2009, 4, 286–300.
4. Y. Zhang, X. Xie, N. Jiao, S.-Y. Hsiao and S.-J. Kao, Biogeosciences, 2014, 11, 2131–2145.
5. J.-K. Yang, Z.-B. Cheng, J. Li and L.-H. Miao, Microb. Ecol., 2013, 66, 796–805.
6. S. Wertz, C. E. Dandie, C. Goyer, J. T. Trevors and C. L. Patten, Appl. Environ. Microbiol., 2009, 75, 7365–7377.
7. G. Ji, R. Wang, W. Zhi, X. Liu, Y. Kong and Y. Tan, Ecol. Eng., 2012, 44, 179–188.
8. E. Kandeler, K. Deiglmayr, D. Tscherko, D. Bru and L. Philippot, Appl. Environ. Microbiol., 2006, 72, 5957–5962.
9. B. B. Oakley, C. A. Francis, K. J. Roberts, C. A. Fuchsman, S. Srinivasan and J. T. Staley, Environ. Microbiol., 2007, 9, 118–130.
10. Y. Zheng, L. Hou, M. Liu, J. Gao, G. Yin, X. Li, F. Deng, X. Lin, X. Jiang and F. Chen, Microb. Ecol., 2015, 70, 30–40.
11. Y. Chen, W. Zhou, Y. Li, J. Zhang, G. Zeng, A. Huang and J. Huang, Appl. Microbiol. Biotechnol., 2014, 98, 4233–4243.
12. E. Tatti, C. Goyer, D. L. Burton, S. Wertz, B. J. Zebarth, M. Chantigny and M. Filion, Microb. Ecol., 2015, 70, 795–808.
13. J. R. Deslippe, H. Jamali, N. Jha and S. Saggar, Soil Biol. Biochem., 2014, 71, 48–60.
14. L. Guo, Z. Hu, F. Fang, T. Liu, X. Chuai and L. Yang, Ann. Microbiol., 2014, 64, 999–1006.
15. K. Hamonts, T. J. Clough, A. Stewart, P. W. Clinton, A. E. Richardson, S. A. Wakelin, M. O'Callaghan and L. M. Condron, FEMS Microbiol. Ecol., 2013, 83, 568–584.
16. Z. Yu, J. Yang and L. Liu, PLoS One, 2014, 9, e92055.
17. Z. Fan, R.-M. Han, J. Ma and G.-X. Wang, Environ. Sci. Pollut. Res., 2016, 1–13.
18. Z. Wang, X.-X. Zhang, X. Lu, B. Liu, Y. Li, C. Long and A. Li, PLoS One, 2014, 9, e113603.
19. N. Chinese, Water and Wastewater Monitoring Methods, Chinese Environmental Science Publishing House, Beijing, China, in chinese, 2002.
20. S. Zhou, T. Huang, H. Zhang, M. Zeng, F. Liu, S. Bai, J. Shi, X. Qiu and X. Yang, Bioresour. Technol., 2016, 201, 195–207.
21. X.-X. Zhang and T. Zhang, Environ. Sci. Technol., 2011, 45, 2598–2604.
22. J. C. López-Gutiérrez, S. Henry, S. Hallet, F. Martin-Laurent, G. Catroux and L. Philippot, J. Microbiol. Methods, 2004, 57, 399–407.
23. L. Liu, G. Shen, M. Sun, X. Cao, G. Shang and P. Chen, J. Air Waste Manage. Assoc., 2014, 64, 894–902.
24. M. A. C. Valero, L. F. Read, D. D. Mara, R. J. Newton, T. P. Curtis and R. J. Davenport, Nitrification-denitrification in WSP: a mechanism for permanent nitrogen removal in maturation ponds, in 8th IWA Specialist Group Conference on Waste Stabilization Ponds, Belo Horizonte, MG, Brazil, 2009.
25. C. Quince, A. Lanzen, R. J. Davenport and P. J. Turnbaugh, BMC Bioinf., 2011, 12, 1.
26. N. J. Loman, R. V. Misra, T. J. Dallman, C. Constantinidou, S. E. Gharbia, J. Wain and M. J. Pallen, Nat. Biotechnol., 2012, 30, 434–439.
27. P. Šmilauer and J. Lepš, Multivariate analysis of ecological data using CANOCO 5, Cambridge university press, 2014.
28. J. Luo, H. Liang, L. Yan, J. Ma, Y. Yang and G. Li, Bioresour. Technol., 2013, 148, 189–195.
29. M. Yoshida, S. Ishii, S. Otsuka and K. Senoo, Soil Biol. Biochem., 2009, 41, 2044–2051.
30. T. Saito, S. Ishii, S. Otsuka, M. Nishiyama and K. Senoo, Microb. Environ., 2008, 23, 192–200.
31. W. Zeng, J. Zhang, A. Wang and Y. Peng, Bioresour. Technol., 2016, 207, 322–331.
32. S. Demanèche, L. Philippot, M. M. David, E. Navarro, T. M. Vogel and P. Simonet, Appl. Environ. Microbiol., 2009, 75, 534–537.
33. M. Langone, J. Yan, S. C. Haaijer, H. J. O. den Camp, M. S. Jetten and G. Andreottola, Front. Microbiol., 2014, 5, 28.
34. M. I. Bellini, L. Gutiérrez, S. Tarlera and A. F. Scavino, Syst. Appl. Microbiol., 2013, 36, 505–516.
35. T. Huang, B. Gao, X.-K. Hu, X. Lu, R. Well, P. Christie, L. R. Bakken and X.-T. Ju, Sci. Rep., 2014, 4, 3950.
36. G.-X. Guo, H. Deng, M. Qiao, Y.-J. Mu and Y.-G. Zhu, Environ. Pollut., 2011, 159, 1886–1895.
37. K. Heylen, L. Lebbe and P. De Vos, Int. J. Syst. Evol. Microbiol., 2008, 58, 73–77.
38. K. Jiang, J. Sanseverino, A. Chauhan, S. Lucas, A. Copeland, A. Lapidus, T. G. Del Rio, E. Dalin, H. Tice and D. Bruce, Stand. Genomic Sci., 2012, 6, 325–335.
39. L. A. Robertson and J. G. Kuenen, Microbiology, 1983, 129, 2847–2855.
40. T. Lukow and H. Diekmann, Biotechnol. Lett., 1997, 19, 1157–1159.
41. T.-L. Huang, S.-L. Zhou, H.-H. Zhang, S.-Y. Bai, X.-X. He and X. Yang, Int. J. Mol. Sci., 2015, 16, 10038–10060.
42. T. Mechichi, E. Stackebrandt, N. Gad'on and G. Fuchs, Arch. Microbiol., 2002, 178, 26–35.
43. Q. Yan, Y. Bi, Y. Deng, Z. He, L. Wu, J. D. Van Nostrand, Z. Shi, J. Li, X. Wang and Z. Hu, Sci. Rep., 2015, 5, 8605.
44. G. T. Sperl and D. S. Hoare, J. Bacteriol., 1971, 108, 733–736.
45. A. Layton, P. Karanth, C. Lajoie, A. Meyers, I. Gregory, R. Stapleton, D. Taylor and G. Sayler, Appl. Environ. Microbiol., 2000, 66, 1167–1174.
46. B. Baumann, M. Snozzi, A. Zehnder and J. R. Van Der Meer, J. Bacteriol., 1996, 178, 4367–4374.
47. J. D. Neufeld, H. Schäfer, M. J. Cox, R. Boden, I. R. McDonald and J. C. Murrell, ISME J., 2007, 1, 480–491.
48. H. G. Kim, G. H. Han, D. Kim, J. S. Choi and S. W. Kim, J. Basic Microbiol., 2012, 52, 141–149.
49. G. Carvalho, P. C. Lemos, A. Oehmen and M. A. Reis, Water Res., 2007, 41, 4383–4396.
50. S. He, V. Kunin, M. Haynes, H. G. Martin, N. Ivanova, F. Rohwer, P. Hugenholtz and K. D. McMahon, Environ. Microbiol., 2010, 12, 1205–1217.
51. M. Kornaros, C. Zafiri and G. Lyberatos, Water Environ. Res., 1996, 68, 934–945.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra18017e

---