Received Date: 08-Dec-2013
2
Accepted Date: 16-Mar-2014
Accepted Article 1
3 4
Revised EMI-2013-1527
5 6
Identification of key N2O production pathways in aerobic partial nitrifying
7 8
granules
1
9
10
Running title: N2O production pathways in aerobic PN granules
11 12
Satoshi Ishii1*, Yangjun Song1, Lashitha Rathnayake1, Azzaya Tumendelger2, Hisashi Satoh1,
13
Sakae Toyoda3, Naohiro Yoshida2,3,4, and Satoshi Okabe1*
14 15 16
1
17
West 8, Kita-ku, Sapporo, Hokkaido 060-8628, Japan
18
2
19
Nagatsuta, Midori-ku, Yokohama 226-8502, Japan
20
3
21
Nagatsuta, Midori-ku, Yokohama 226-8502, Japan
22
4
23
Japan
Division of Environmental Engineering, Faculty of Engineering, Hokkaido University, North 13,
Department of Environmental Chemistry and Engineering, Tokyo Institute of Technology, 4259
Department of Environmental Science and Technology, Tokyo Institute of Technology, 4259
Earth-Life Science Institute, Tokyo Institute of Technology, 2-12-1 Meguro-ku, Tokyo 152-8551,
This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as doi: 10.1111/1462-2920.12458 This article is protected by copyright. All rights reserved.
Accepted Article
24 25 26
Correspondence: Satoshi Ishii or Satoshi Okabe
27
Division of Environmental Engineering, Faculty of Engineering, Hokkaido University, North 13,
28
West 8, Kita-ku, Sapporo, Hokkaido 060-8628, Japan
29
Tel & Fax: +81-(0)11-706-7162 or +81-(0)11-706-6266
30
Email: [email protected] or [email protected]
31
Abstract
32
The identification of the key nitrous oxide (N2O) production pathways is important to establish a
33
strategy to mitigate N2O emission. In this study, we combined real-time gas monitoring analysis, 15N
34
stable isotope analysis, denitrification functional gene transcriptome analysis, and microscale N2O
35
concentration measurements to identify the main N2O producers in a partial nitrification aerobic
36
granule reactor, which was fed with ammonium and acetate. Our results suggest that heterotrophic
37
denitrification was the main contributor to N2O production in our partial nitrification (PN) aerobic
38
granule reactor. The heterotrophic denitrifiers were probably related to Rhodocyclales bacteria,
39
although different types of bacteria were active in the initial and latter stages of the PN reaction
40
cycles, most likely in response to the presence of acetate. NH2OH oxidation and nitrifier
41
denitrification occurred, but their contribution to N2O emission was relatively small (20–30%)
42
compared with heterotrophic denitrification. Our approach can be useful to quantitatively examine
43
the relative contributions of the three pathways (hydroxylamine oxidation, nitrifier denitrification,
44
and heterotrophic denitrification) to N2O emission in mixed microbial populations.
45 46
1
This article is protected by copyright. All rights reserved.
Introduction
Accepted Article
47 48
Nitrous oxide (N2O) emission is of great concern because N2O has a >300-fold stronger effect
49
on global warming than carbon dioxide (CO2), and it is also responsible for ozone layer destruction
50
(Ravishankara et al 2009). Agriculture is considered to be the major source of global N2O emission,
51
but N2O is also emitted by wastewater treatment facilities, especially during biological nitrogen
52
removal processes (Kampschreur et al., 2009). The conventional biological nitrogen removal process
53
employs nitrification (NH4+ NO2− NO3−) followed by denitrification (NO3− NO2− NO
54
N2O N2), whereas the alternative nitrogen removal process involves partial nitrification (PN;
55
NH4+ NO2−) and the anaerobic ammonium oxidation (anammox; NH4+ + NO2− N2). The
56
amount of N2O evolved is believed to be smaller with the PN-anammox process than the
57
conventional nitrification-denitrification process (Kampschreur et al., 2009); however, the N2O
58
emission from the PN process is not well understood at present.
59
Three main biological pathways produce N2O during biological nitrogen removal processes:
60
hydroxylamine (NH2OH) oxidation, nitrifier denitrification, and heterotrophic denitrification
61
(Kampschreur et al., 2008; Kampschreur et al., 2009; Okabe et al., 2011; Law et al., 2012). N2O can
62
be produced either as a by-product of NH2OH oxidation during nitrification and PN processes (NH4+
63
NH2OH NOH N2O; Law et al., 2012), or as an end-product or intermediate product of
64
nitrite reduction during the denitrification process. Some nitrifiers are capable of the latter pathway
65
under oxygen-limiting conditions and this pathway is called nitrifier denitrification. Therefore, the
66
identification of the key pathways involved with N2O emission during biological nitrogen removal
67
processes is essential for establishing a strategy to mitigate N2O emission.
68 69
We previously established a stable PN process using aerobic granules in an internal-circulating
sequence batch airlift reactor (SBAR) (Song et al., 2013). In this reactor, the NH4+ conversion rate 2
This article is protected by copyright. All rights reserved.
was 1.22 kg N m−3 day−1, whereas the nitrite production rate was 0.64 kg N m−3 day−1. The
71
difference in these two values indicates the loss of nitrogen by denitrification. Denitrification may
72
have been promoted by the presence of acetate in the influent, most (95%) of which was removed
73
during the process. A significant amount of N2O was detected in this reactor, but its source has not
74
been identified. Because nitrification and denitrification co-occurred in this reactor, it was difficult to
75
identify the major sources of N2O emission based on the information regarding the reactor
76
operational conditions [e.g., pH, dissolved oxygen (DO) concentration, and NO2− concentration].
Accepted Article
70
77
N2O isotopomer analysis is a powerful tool to distinguish whether N2O originates from NH2OH
78
oxidation or NO2- reduction (nitrifier denitrification and heterotrophic denitrification). This technique
79
is based on the analysis of the intramolecular distribution of 15N in the central position (14N15N16O)
80
and the end position (15N14N16O) of the asymmetric N2O molecules (Toyoda and Yoshida, 1999;
81
Yoshida and Toyoda, 2000). The 15N-site preference (SP) is defined as the difference in the bulk
82
nitrogen isotope ratios of N2O between δ15Nα and δ15Nβ, where 15Nα and 15Nβ represent the 15N/14N
83
ratios at the center (α) and end (β) sites of the nitrogen atoms, respectively (Toyoda and Yoshida,
84
1999; Yoshida and Toyoda, 2000). Because N2O produced through NH2OH oxidation and NO2-
85
reduction have different SP values, the analysis of the SP allows us to identify the sources of N2O
86
produced during these two processes (Sutka et al., 2006; Maeda et al., 2010; Maeda et al., 2011).
87
This technique has been used to distinguish the N2O produced through NH2OH oxidation and NO2-
88
reduction from various environments, including a municipal wastewater treatment plant (Toyoda et
89
al., 2011a) and an autotrophic PN reactor (Rathnayake et al., 2013). However, it is important to note
90
that N2O reduction also increases the SP values slightly (Ostrom et al., 2007; Wunderlin et al., 2012).
91
In addition, it is difficult to distinguish N2O derived from nitrifier denitrification and from 3
This article is protected by copyright. All rights reserved.
heterotrophic denitrification because both of these pathways produce N2O through NO2- reduction.
93
Several isotopic analyses have been conducted (e.g., Wunderlin et al., 2012), but isotopic analysis
94
alone is not sufficient to clearly identify the sources of N2O, particularly under low DO conditions,
95
where NH2OH oxidation, nitrifier denitrification, and heterotrophic denitrification may occur
96
simultaneously.
Accepted Article
92
97
Analyses of denitrification functional genes and their transcripts is a useful approach to identify
98
key denitrifiers (Philippot and Hallin, 2005). The following genes have been frequently used as
99
functional gene markers for denitrification: nirK encoding copper-containing nitrite reductase, nirS
100
encoding cytochrome cd1-containing nitrite reductase, cnorB encoding cytochrome bc-containing
101
nitric oxide reductase (cNOR), and nosZ encoding nitrous oxide reductase (Zumft, 1997; Philippot
102
and Hallin 2005). In addition, reverse transcription quantitative PCR analysis (RT-qPCR) targeting
103
these functional genes allow us to examine the relative transcription levels of denitrification
104
functional genes in response to environmental stimuli (Yoshida et al., 2012). However, it is difficult
105
to identify denitrifiers based solely on this functional gene sequence information because
106
phylogenetically distantly related bacteria may carry highly similar functional gene sequences (Jones
107
et al., 2008; Ishii et al 2011a). Exceptions are the cnorB sequences from Nitrosomonas and
108
Nitrosococcus AOB, which are clustered together and are distinct from the cnorB sequences of
109
heterotrophic denitrifiers (Casciotti and Ward, 2005). Therefore, the sequence analysis of cnorB
110
transcripts is a potentially useful approach to distinguish the relative contributions of AOB and
111
heterotrophic denitrifiers to N2O production. In addition to cNOR, AOB is recently recognized to
112
carry another NO reductase called sNOR (Klotz and Stein, 2008; Stein et al., 2007). In the previous
113
pure culture studies, sNOR or the genes encoding sNOR (norYS) were actively expressed or 4
This article is protected by copyright. All rights reserved.
transcribed in Nitrosomonas europaea (Cho et al., 2006), Nitrosomonas eutropha (Kartal et al.,
115
2012), and Nitrosococcus oceani (Stein et al., 2013). Therefore, norYS can be additional gene of
116
target when contribution of AOB to N2O production is examined.
Accepted Article
114
117
Consequently, the objective of this study was to identify the major N2O production pathways in
118
the PN aerobic granule reactor (Song et al., 2013). To achieve this, we used time-course water and
119
gas quality analyses, N2O isotopomer analysis, and functional gene transcriptome analyses. In
120
addition, N2O microsensor measurement was performed to understand the mechanism of N2O
121
emission from PN aerobic granules.
122 123
Results
124
N2O emission from the PN aerobic granule reactor
125
Figure 1 shows a typical time course of the concentrations of NH4+, NO2−, N2O, and N2 during
126
one SBAR operation cycle (4 h) in the PN aerobic granule reactor. The concentration of NH4+
127
decreased along with the increase in NO2− concentration, indicating the occurrence of PN. However,
128
a decrease in the NO2− concentration was observed during the initial 20 min, which was probably
129
attributable to denitrification. Total organic carbon (TOC) was utilized almost completely within
130
about 20 min (data not shown).
131
The N2O concentration increased rapidly after the beginning of aeration (0 min), but declined
132
within 10–20 min, and then it increased again and reached a stable concentration level
133
(approximately 400 ppm [v/v]) at 1 h after the beginning of aeration. Almost the same N2O
134
concentrations were obtained by photoacoustic real-time monitoring and GC-ECD measurements.
135
The time-course concentration profiles of the aforementioned chemical species were reproducible. 5
This article is protected by copyright. All rights reserved.
According to the N mass balance calculation, N2 production probably occurred during the initial
Accepted Article
136 137
period of the reaction cycle, especially 10–20 min after substrate feeding. However, most of the
138
gaseous N was in the form of N2O during the latter period of the reaction cycle (1 h after substrate
139
feeding). Overall, the NH4+ conversion rate in our PN aerobic granule reactor was 1.28 kg N m−3
140
day−1, where the production rates for NO2−, N2O gas, and N2 gas were 0.67, 0.10, and 0.51 kg N m−3
141
day−1, respectively. This suggests that 5.6% of the NH4+ load was converted into N2O as an off-gas.
142 143
15
N isotope analysis
144
The N2O concentration greatly changed over time, but the SP was relatively constant within the
145
N2O molecule (4.9–14.1‰) during the entire period (Fig. 2A). The SP vs. δ15Nbulk diagram shown in
146
Fig. 2B and Supplementary Fig. S1 suggests that N2O was produced through NH2OH oxidation and
147
NO2- reduction, and N2O reduction occurred particularly during the initial stage of the SBAR
148
operation cycle. The Monte Carlo calculation suggested that the majority (70–80%) of the N2O
149
produced in this study originated from NO2- reduction rather than from NH2OH oxidation
150
(Supplementary Table S2).
151 152 153
Transcriptome analysis of key denitrification functional genes The different transcription levels of nitrification and denitrification functional genes were
154
determined for PN aerobic granules collected at different time during a single SBAR operation cycle
155
(Fig. 3). The transcription levels of nirS and cnorB increased rapidly after substrate feeding, peaked
156
at 10 min after substrate feeding, and decreased gradually thereafter. The quantities of nirS and
157
cnorB transcripts at 10 and 20 min after substrate feeding were significantly (p < 0.01) larger than 6
This article is protected by copyright. All rights reserved.
those at 0, 120, 180, or 233 min after substrate feeding. In contrast, the transcription levels of amoA,
159
Nitrosomonas-specific nirK, and nosZ clade I remained relatively constant throughout the SBAR
160
operation. The results were also similar when the transcription levels were normalized against the
161
quantity of 16S rRNA. The quantities of nirS transcripts were always greater than those of
162
Nitrosomonas-specific nirK transcripts. The transcription activities of narG, general nirK, qnorB,
163
norY, and nosZ clade II were not detected. Quantity of general nirK was performed using nirK876
164
and nirK1040 primers (Henry et al., 2004), which cannot amplify nirK from Nitrosomonas (data not
165
shown). In contrast, only Nitrosomonas-related nirK sequences were amplified using nirK_166F and
166
nirK_665R primers (Cantera and Stein, 2007), as confirmed by cloning and sequencing analysis
167
(n=26).
Accepted Article
158
168
To identify the microbes responsible for N2O emission, we performed cnorB transcript
169
pyrosequencing analysis. A total of 79,730 cnorB sequences were obtained from 27 cDNA samples
170
(Table S3). Based on the 95% nucleotide sequence similarity, 868 operational taxonomic units
171
(OTUs) were identified; however, majority (>90%) of the cnorB transcripts were classified into one
172
of the four OTUs. One OTU (2A_63087) was closely related to the cnorB from Nitrosomonas spp.
173
and Nitrosococcus spp. (Fig. 4). The other three OTUs were related to cnorB from Rhodocyclales
174
bacteria (e.g., Azoarcus spp., Dechloromonas spp., and Thauera spp.).
175
A temporal shift was observed in the proportion of cnorB sequences affiliated to the four main
176
OTUs (Fig. S2). The proportion of cnorB sequences affiliated to OTU 3B_67163 was large (>60%)
177
during 0–1 h after substrate feeding. The proportion of OTU 3B_67163 sequences peaked (80%) at
178
10 min after feeding but thereafter decreased gradually to 15% at 233 min after feeding. In contrast,
179
the proportions of the other three OTUs (2A_63087, 7A_13064, and 7A_78700) increased gradually 7
This article is protected by copyright. All rights reserved.
from 10 to 233 min after feeding. The proportion of AOB-related cnorB (OTU 2A_63087) ranged
181
from 1.24% to 16.5%. We estimated the quantity of each cnorB OTU based on the cnorB RT-qPCR
182
results and the proportion of each OTU in the cnorB transcript pyrosequencing results, which
183
revealed that the quantities of OTUs 2A_63087, 7A_13064, and 7A_78700 were relatively constant
184
throughout the reaction cycle (Fig. 5).
Accepted Article
180
185 186 187
Microscale N2O analysis Microsensor measurements were conducted under two representative conditions imitating the
188
beginning and end of one SBAR operation cycle (Fig. 6). The net N2O production in the presence of
189
acetate (i.e., imitating the beginning of the cycle) was low and found >500 μm from the surface of
190
the granules. In this condition, N2O consumption was observed near the surface (300–400 µm) of the
191
granules. In the absence of acetate (i.e., imitating the end of the cycle), N2O production also occurred
192
throughout the granules, with a peak at 600 μm depth, although no significant N2O consumption was
193
detected.
194
Based on the FISH analysis, Nitrosomonas-related AOB were present near the surface (0–300
195
µm) of the granules (Fig. 6). Bacteria other than AOB, most of which were probably heterotrophs,
196
were present near the surface as well as in the deeper part of the granules.
197 198 199
DISCUSSION
The amount of N2O emitted by our PN aerobic reactor changed greatly over time. The initial
200
peak of N2O was attributed partly to the stripping of accumulated N2O after the resumption of
201
aeration, which agreed with the observations in another N2O monitoring study (Rathnayake et al., 8
This article is protected by copyright. All rights reserved.
2013). However, denitrification may have also occurred because the NO2− and TOC concentrations
203
declined with the increase in the pH. In contrast, no apparent correlation was found between N2O
204
emissions and major water chemistry parameters.
Accepted Article
202
205
In our PN aerobic granule reactor, 5.6% of the NH4+ load was converted to N2O, which was
206
larger than the previous PN studies (e.g., 1.7% in Kampschreur et al., 2008; 4.0% in Okabe et al.,
207
2011; 0.8% in Rathnayake et al., 2013) but smaller than the conventional nitrification-denitrification
208
systems (e.g., >20% in Itokawa et al., 2001). The TOC concentration may have affected the amount
209
of N2O produced and consumed (Itokawa et al., 2001).
210
Stable isotope techniques have been used to identify the key pathways of N2O emissions from
211
wastewater treatment plants, agricultural fields, forests, oceans, and other environments (Maeda et
212
al., 2010; Toyoda et al., 2011a; Toyoda et al., 2011b; Wunderlin et al., 2012). N2O isotopomer
213
analysis is particularly useful for distinguishing N2O derived from NH2OH oxidation and NO2-
214
reduction. However, there are several limitations with this approach. First, N2O reduction increases
215
the SP value; therefore, isotopomer analysis alone may underestimate the effect of NO2- reduction in
216
environments where N2O reduction is intense (Ostrom et al., 2007). Second, isotopomer analysis
217
cannot distinguish NH2OH oxidation and fungal denitrification (NO2- reduction) because fungal
218
denitrification, although based on the tests performed using limited number of strains, produces a
219
similar SP value to that of NH2OH oxidation (approximately 37‰) (Sutka et al., 2008). Third, it is
220
not possible to distinguish the relative contributions of nitrifier denitrification and heterotrophic
221
denitrification using isotopomer analysis because both pathways occur through NO2- reduction.
222 223
To overcome the first problem, we used both SP and δ15Nbulk values to calculate the occurrence
of N2O reduction using the Monte Carlo method (Toyoda et al., 2011b). With this calculation, we can 9
This article is protected by copyright. All rights reserved.
account for the 15N isotope effect by N2O reduction, and therefore, can estimate the relative
225
contributions of NH2OH oxidation and NO2- reduction more precisely. Our results suggest that the
226
majority (approximately 70–80%) of the N2O produced in our PN reactor originated from NO2-
227
reduction rather than from NH2OH oxidation. Fungi were not detected in our PN reactor according to
228
the fungi-specific PCR (data not shown). Therefore, occurrence of fungal denitrification, the second
229
problem mentioned above, was not relevant in this study. Thus, we did not consider that the
230
contribution of NO2- reduction to N2O production was underestimated in our PN reactor. NO2-
231
reduction was the major pathway for N2O production in our heterotrophic PN reactor, which
232
contrasts with autotrophic PN reactors where 65% of the N2O is produced through NH2OH oxidation
233
(Rathnayake et al., 2013). This suggests that the N2O production mechanism differs in autotrophic
234
and heterotrophic PN reactors.
Accepted Article
224
235
Based on the N2O isotopomer analysis alone, it was difficult to estimate the relative contribution
236
of nitrifier denitrification and heterotrophic denitrification to N2O production (the third problem
237
mentioned above). To overcome this, we conducted transcriptome analyses. mRNA-based
238
transcriptome analysis is useful to assess microbial activity because mRNA is degraded within a
239
short time (Philippot and Hallin, 2005). The half-lives of the nirS, cnorB, and nosZ transcripts in
240
active denitrifiers are approximately 13 min (Härtig and Zumft, 1999). In our study, the increased
241
transcription levels of nirS and cnorB during the initial stage of the SBAR operation cycle (10–20
242
min after substrate feeding) agreed with the sharp rise in N2O and N2 emission during the same
243
period. Both nirK and nirS encode nitrite reductase, but the transcription levels of
244
Nitrosomonas-specific nirK (Cantera and Stein, 2007) were relatively constant during the reaction
245
cycle. Heterotrophic denitrifiers, but not AOB, possess nirS; therefore, these results indicate that 10
This article is protected by copyright. All rights reserved.
heterotrophic denitrifiers were responsive to the addition of substrates and might be active during the
247
initial 30 min of the PN reaction.
Accepted Article
246
248
We did not detect transcripts of norY, a gene encoding sNOR of AOB and some other bacteria
249
(Klotz and Stein, 2008; Stein et al., 2007), although the PCR amplification was successful from the
250
PN granule DNA samples. This was in contrast to the report by Kartal et al. (2012), in which sNOR,
251
but not cNOR, was expressed in NO2--exposed cells of Nitrosomonas eutropha C91 under both oxic
252
and anoxic conditions. In Nitrosomonas europaea nirK-deficient mutant and Nitrosococcus oceani
253
ATCC 19707, transcription levels of norY also increased in the nitrifying cells under oxic conditions
254
(Cho et al., 2006; Stein et al., 2013). However, Beyer et al. (2009) reported that the transcription of
255
cnorB increased in Nitrosomonas europaea ATCC 19718 under nitrifier denitrification conditions.
256
Therefore, expression of sNOR and cNOR may depend on strains or incubation conditions. In this
257
study, Nitrosomonas eutropha-related AOB actively performed ammonia oxidation in the PN aerobic
258
granules, but some AOB cells might face anoxic conditions in the granules. In addition, heterotrophic
259
denitrifiers present in the PN granules might have influenced the microenvironment around AOB
260
(e.g., removal of NO), which could also have influenced the gene expression of AOB.
261
We also did not detect qnorB, a gene encoding quinol-dependent nitric oxide reductase (qNOR),
262
and its transcripts. The qNOR is considered as NO detoxification enzyme and is found in the
263
genomes of various bacteria and archaea, mostly in non-denitrifiers (Jones et al., 2008). Because our
264
PN aerobic granules contained mostly AOB and denitrifiers (Song et al., 2013), these results
265
indicated that the contribution of qNOR-containing microbes was very small, if any, in our reactor.
266 267
In contrast to the increased transcription levels of nirS and cnorB, the transcription level of nosZ
did not change over time. The gene transcription levels do not necessarily reflect the activity of the 11
This article is protected by copyright. All rights reserved.
coding enzymes, but our results indicated that the N2O reducing activity might have been constant
269
throughout the reaction cycle. The failure to detect narG suggests that nitrite, rather than nitrate, was
270
the substrate for denitrification, which agreed with the low concentration of NO3− in our reactor. A
271
previous batch experiment also showed that denitrifiers in the PN granule reactor preferred NO2− to
272
NO3−, which was probably due to the long-term acclimation to a high concentration of NO2− (Song et
273
al., 2013).
Accepted Article
268
274
The contribution of denitrification to N2O emission was analyzed by pyrosequencing of the
275
cnorB transcript. The cnorB sequences from Nitrosomonas and Nitrosococcus AOB clustered
276
together and were distantly related to other cnorB sequences, similar to the previous studies
277
(Casciotti and Ward, 2005; Jones et al., 2008). This suggests that the cnorB sequences affiliated to
278
OTU 2A_63087 originated from Nitrosomonas and Nitrosococcus AOB. The representative sequence
279
of OTU 2A_63087 was most closely related to the cnorB from Nitrosomonas eutropha, in agreement
280
with the presence of AOB closely related to N. eutropha in our PN aerobic granules (Song et al.,
281
2013). The quantity of Nitrosomonas-related cnorB, which was estimated based on the quantity of
282
cnorB transcripts and the proportion of Nitrosomonas-related cnorB, remained relatively constant
283
during the SBAR operation cycle, similar to the results obtained by RT-qPCR targeting
284
Nitrosomonas-specific nirK. Nitrosomonas-related cnorB comprised a relatively small proportion of
285
the cnorB libraries, which indicates that the contribution of nitrifier denitrification to N2O emission
286
may be small. If we assume that the abundance of the Nitrosomonas-related cnorB sequence
287
reflected the proportion of N2O produced by AOB, the contribution of nitrifier denitrification to N2O
288
production was
2
Accepted Date: 16-Mar-2014
Accepted Article 1
3 4
Revised EMI-2013-1527
5 6
Identification of key N2O production pathways in aerobic partial nitrifying
7 8
granules
1
9
10
Running title: N2O production pathways in aerobic PN granules
11 12
Satoshi Ishii1*, Yangjun Song1, Lashitha Rathnayake1, Azzaya Tumendelger2, Hisashi Satoh1,
13
Sakae Toyoda3, Naohiro Yoshida2,3,4, and Satoshi Okabe1*
14 15 16
1
17
West 8, Kita-ku, Sapporo, Hokkaido 060-8628, Japan
18
2
19
Nagatsuta, Midori-ku, Yokohama 226-8502, Japan
20
3
21
Nagatsuta, Midori-ku, Yokohama 226-8502, Japan
22
4
23
Japan
Division of Environmental Engineering, Faculty of Engineering, Hokkaido University, North 13,
Department of Environmental Chemistry and Engineering, Tokyo Institute of Technology, 4259
Department of Environmental Science and Technology, Tokyo Institute of Technology, 4259
Earth-Life Science Institute, Tokyo Institute of Technology, 2-12-1 Meguro-ku, Tokyo 152-8551,
This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as doi: 10.1111/1462-2920.12458 This article is protected by copyright. All rights reserved.
Accepted Article
24 25 26
Correspondence: Satoshi Ishii or Satoshi Okabe
27
Division of Environmental Engineering, Faculty of Engineering, Hokkaido University, North 13,
28
West 8, Kita-ku, Sapporo, Hokkaido 060-8628, Japan
29
Tel & Fax: +81-(0)11-706-7162 or +81-(0)11-706-6266
30
Email: [email protected] or [email protected]
31
Abstract
32
The identification of the key nitrous oxide (N2O) production pathways is important to establish a
33
strategy to mitigate N2O emission. In this study, we combined real-time gas monitoring analysis, 15N
34
stable isotope analysis, denitrification functional gene transcriptome analysis, and microscale N2O
35
concentration measurements to identify the main N2O producers in a partial nitrification aerobic
36
granule reactor, which was fed with ammonium and acetate. Our results suggest that heterotrophic
37
denitrification was the main contributor to N2O production in our partial nitrification (PN) aerobic
38
granule reactor. The heterotrophic denitrifiers were probably related to Rhodocyclales bacteria,
39
although different types of bacteria were active in the initial and latter stages of the PN reaction
40
cycles, most likely in response to the presence of acetate. NH2OH oxidation and nitrifier
41
denitrification occurred, but their contribution to N2O emission was relatively small (20–30%)
42
compared with heterotrophic denitrification. Our approach can be useful to quantitatively examine
43
the relative contributions of the three pathways (hydroxylamine oxidation, nitrifier denitrification,
44
and heterotrophic denitrification) to N2O emission in mixed microbial populations.
45 46
1
This article is protected by copyright. All rights reserved.
Introduction
Accepted Article
47 48
Nitrous oxide (N2O) emission is of great concern because N2O has a >300-fold stronger effect
49
on global warming than carbon dioxide (CO2), and it is also responsible for ozone layer destruction
50
(Ravishankara et al 2009). Agriculture is considered to be the major source of global N2O emission,
51
but N2O is also emitted by wastewater treatment facilities, especially during biological nitrogen
52
removal processes (Kampschreur et al., 2009). The conventional biological nitrogen removal process
53
employs nitrification (NH4+ NO2− NO3−) followed by denitrification (NO3− NO2− NO
54
N2O N2), whereas the alternative nitrogen removal process involves partial nitrification (PN;
55
NH4+ NO2−) and the anaerobic ammonium oxidation (anammox; NH4+ + NO2− N2). The
56
amount of N2O evolved is believed to be smaller with the PN-anammox process than the
57
conventional nitrification-denitrification process (Kampschreur et al., 2009); however, the N2O
58
emission from the PN process is not well understood at present.
59
Three main biological pathways produce N2O during biological nitrogen removal processes:
60
hydroxylamine (NH2OH) oxidation, nitrifier denitrification, and heterotrophic denitrification
61
(Kampschreur et al., 2008; Kampschreur et al., 2009; Okabe et al., 2011; Law et al., 2012). N2O can
62
be produced either as a by-product of NH2OH oxidation during nitrification and PN processes (NH4+
63
NH2OH NOH N2O; Law et al., 2012), or as an end-product or intermediate product of
64
nitrite reduction during the denitrification process. Some nitrifiers are capable of the latter pathway
65
under oxygen-limiting conditions and this pathway is called nitrifier denitrification. Therefore, the
66
identification of the key pathways involved with N2O emission during biological nitrogen removal
67
processes is essential for establishing a strategy to mitigate N2O emission.
68 69
We previously established a stable PN process using aerobic granules in an internal-circulating
sequence batch airlift reactor (SBAR) (Song et al., 2013). In this reactor, the NH4+ conversion rate 2
This article is protected by copyright. All rights reserved.
was 1.22 kg N m−3 day−1, whereas the nitrite production rate was 0.64 kg N m−3 day−1. The
71
difference in these two values indicates the loss of nitrogen by denitrification. Denitrification may
72
have been promoted by the presence of acetate in the influent, most (95%) of which was removed
73
during the process. A significant amount of N2O was detected in this reactor, but its source has not
74
been identified. Because nitrification and denitrification co-occurred in this reactor, it was difficult to
75
identify the major sources of N2O emission based on the information regarding the reactor
76
operational conditions [e.g., pH, dissolved oxygen (DO) concentration, and NO2− concentration].
Accepted Article
70
77
N2O isotopomer analysis is a powerful tool to distinguish whether N2O originates from NH2OH
78
oxidation or NO2- reduction (nitrifier denitrification and heterotrophic denitrification). This technique
79
is based on the analysis of the intramolecular distribution of 15N in the central position (14N15N16O)
80
and the end position (15N14N16O) of the asymmetric N2O molecules (Toyoda and Yoshida, 1999;
81
Yoshida and Toyoda, 2000). The 15N-site preference (SP) is defined as the difference in the bulk
82
nitrogen isotope ratios of N2O between δ15Nα and δ15Nβ, where 15Nα and 15Nβ represent the 15N/14N
83
ratios at the center (α) and end (β) sites of the nitrogen atoms, respectively (Toyoda and Yoshida,
84
1999; Yoshida and Toyoda, 2000). Because N2O produced through NH2OH oxidation and NO2-
85
reduction have different SP values, the analysis of the SP allows us to identify the sources of N2O
86
produced during these two processes (Sutka et al., 2006; Maeda et al., 2010; Maeda et al., 2011).
87
This technique has been used to distinguish the N2O produced through NH2OH oxidation and NO2-
88
reduction from various environments, including a municipal wastewater treatment plant (Toyoda et
89
al., 2011a) and an autotrophic PN reactor (Rathnayake et al., 2013). However, it is important to note
90
that N2O reduction also increases the SP values slightly (Ostrom et al., 2007; Wunderlin et al., 2012).
91
In addition, it is difficult to distinguish N2O derived from nitrifier denitrification and from 3
This article is protected by copyright. All rights reserved.
heterotrophic denitrification because both of these pathways produce N2O through NO2- reduction.
93
Several isotopic analyses have been conducted (e.g., Wunderlin et al., 2012), but isotopic analysis
94
alone is not sufficient to clearly identify the sources of N2O, particularly under low DO conditions,
95
where NH2OH oxidation, nitrifier denitrification, and heterotrophic denitrification may occur
96
simultaneously.
Accepted Article
92
97
Analyses of denitrification functional genes and their transcripts is a useful approach to identify
98
key denitrifiers (Philippot and Hallin, 2005). The following genes have been frequently used as
99
functional gene markers for denitrification: nirK encoding copper-containing nitrite reductase, nirS
100
encoding cytochrome cd1-containing nitrite reductase, cnorB encoding cytochrome bc-containing
101
nitric oxide reductase (cNOR), and nosZ encoding nitrous oxide reductase (Zumft, 1997; Philippot
102
and Hallin 2005). In addition, reverse transcription quantitative PCR analysis (RT-qPCR) targeting
103
these functional genes allow us to examine the relative transcription levels of denitrification
104
functional genes in response to environmental stimuli (Yoshida et al., 2012). However, it is difficult
105
to identify denitrifiers based solely on this functional gene sequence information because
106
phylogenetically distantly related bacteria may carry highly similar functional gene sequences (Jones
107
et al., 2008; Ishii et al 2011a). Exceptions are the cnorB sequences from Nitrosomonas and
108
Nitrosococcus AOB, which are clustered together and are distinct from the cnorB sequences of
109
heterotrophic denitrifiers (Casciotti and Ward, 2005). Therefore, the sequence analysis of cnorB
110
transcripts is a potentially useful approach to distinguish the relative contributions of AOB and
111
heterotrophic denitrifiers to N2O production. In addition to cNOR, AOB is recently recognized to
112
carry another NO reductase called sNOR (Klotz and Stein, 2008; Stein et al., 2007). In the previous
113
pure culture studies, sNOR or the genes encoding sNOR (norYS) were actively expressed or 4
This article is protected by copyright. All rights reserved.
transcribed in Nitrosomonas europaea (Cho et al., 2006), Nitrosomonas eutropha (Kartal et al.,
115
2012), and Nitrosococcus oceani (Stein et al., 2013). Therefore, norYS can be additional gene of
116
target when contribution of AOB to N2O production is examined.
Accepted Article
114
117
Consequently, the objective of this study was to identify the major N2O production pathways in
118
the PN aerobic granule reactor (Song et al., 2013). To achieve this, we used time-course water and
119
gas quality analyses, N2O isotopomer analysis, and functional gene transcriptome analyses. In
120
addition, N2O microsensor measurement was performed to understand the mechanism of N2O
121
emission from PN aerobic granules.
122 123
Results
124
N2O emission from the PN aerobic granule reactor
125
Figure 1 shows a typical time course of the concentrations of NH4+, NO2−, N2O, and N2 during
126
one SBAR operation cycle (4 h) in the PN aerobic granule reactor. The concentration of NH4+
127
decreased along with the increase in NO2− concentration, indicating the occurrence of PN. However,
128
a decrease in the NO2− concentration was observed during the initial 20 min, which was probably
129
attributable to denitrification. Total organic carbon (TOC) was utilized almost completely within
130
about 20 min (data not shown).
131
The N2O concentration increased rapidly after the beginning of aeration (0 min), but declined
132
within 10–20 min, and then it increased again and reached a stable concentration level
133
(approximately 400 ppm [v/v]) at 1 h after the beginning of aeration. Almost the same N2O
134
concentrations were obtained by photoacoustic real-time monitoring and GC-ECD measurements.
135
The time-course concentration profiles of the aforementioned chemical species were reproducible. 5
This article is protected by copyright. All rights reserved.
According to the N mass balance calculation, N2 production probably occurred during the initial
Accepted Article
136 137
period of the reaction cycle, especially 10–20 min after substrate feeding. However, most of the
138
gaseous N was in the form of N2O during the latter period of the reaction cycle (1 h after substrate
139
feeding). Overall, the NH4+ conversion rate in our PN aerobic granule reactor was 1.28 kg N m−3
140
day−1, where the production rates for NO2−, N2O gas, and N2 gas were 0.67, 0.10, and 0.51 kg N m−3
141
day−1, respectively. This suggests that 5.6% of the NH4+ load was converted into N2O as an off-gas.
142 143
15
N isotope analysis
144
The N2O concentration greatly changed over time, but the SP was relatively constant within the
145
N2O molecule (4.9–14.1‰) during the entire period (Fig. 2A). The SP vs. δ15Nbulk diagram shown in
146
Fig. 2B and Supplementary Fig. S1 suggests that N2O was produced through NH2OH oxidation and
147
NO2- reduction, and N2O reduction occurred particularly during the initial stage of the SBAR
148
operation cycle. The Monte Carlo calculation suggested that the majority (70–80%) of the N2O
149
produced in this study originated from NO2- reduction rather than from NH2OH oxidation
150
(Supplementary Table S2).
151 152 153
Transcriptome analysis of key denitrification functional genes The different transcription levels of nitrification and denitrification functional genes were
154
determined for PN aerobic granules collected at different time during a single SBAR operation cycle
155
(Fig. 3). The transcription levels of nirS and cnorB increased rapidly after substrate feeding, peaked
156
at 10 min after substrate feeding, and decreased gradually thereafter. The quantities of nirS and
157
cnorB transcripts at 10 and 20 min after substrate feeding were significantly (p < 0.01) larger than 6
This article is protected by copyright. All rights reserved.
those at 0, 120, 180, or 233 min after substrate feeding. In contrast, the transcription levels of amoA,
159
Nitrosomonas-specific nirK, and nosZ clade I remained relatively constant throughout the SBAR
160
operation. The results were also similar when the transcription levels were normalized against the
161
quantity of 16S rRNA. The quantities of nirS transcripts were always greater than those of
162
Nitrosomonas-specific nirK transcripts. The transcription activities of narG, general nirK, qnorB,
163
norY, and nosZ clade II were not detected. Quantity of general nirK was performed using nirK876
164
and nirK1040 primers (Henry et al., 2004), which cannot amplify nirK from Nitrosomonas (data not
165
shown). In contrast, only Nitrosomonas-related nirK sequences were amplified using nirK_166F and
166
nirK_665R primers (Cantera and Stein, 2007), as confirmed by cloning and sequencing analysis
167
(n=26).
Accepted Article
158
168
To identify the microbes responsible for N2O emission, we performed cnorB transcript
169
pyrosequencing analysis. A total of 79,730 cnorB sequences were obtained from 27 cDNA samples
170
(Table S3). Based on the 95% nucleotide sequence similarity, 868 operational taxonomic units
171
(OTUs) were identified; however, majority (>90%) of the cnorB transcripts were classified into one
172
of the four OTUs. One OTU (2A_63087) was closely related to the cnorB from Nitrosomonas spp.
173
and Nitrosococcus spp. (Fig. 4). The other three OTUs were related to cnorB from Rhodocyclales
174
bacteria (e.g., Azoarcus spp., Dechloromonas spp., and Thauera spp.).
175
A temporal shift was observed in the proportion of cnorB sequences affiliated to the four main
176
OTUs (Fig. S2). The proportion of cnorB sequences affiliated to OTU 3B_67163 was large (>60%)
177
during 0–1 h after substrate feeding. The proportion of OTU 3B_67163 sequences peaked (80%) at
178
10 min after feeding but thereafter decreased gradually to 15% at 233 min after feeding. In contrast,
179
the proportions of the other three OTUs (2A_63087, 7A_13064, and 7A_78700) increased gradually 7
This article is protected by copyright. All rights reserved.
from 10 to 233 min after feeding. The proportion of AOB-related cnorB (OTU 2A_63087) ranged
181
from 1.24% to 16.5%. We estimated the quantity of each cnorB OTU based on the cnorB RT-qPCR
182
results and the proportion of each OTU in the cnorB transcript pyrosequencing results, which
183
revealed that the quantities of OTUs 2A_63087, 7A_13064, and 7A_78700 were relatively constant
184
throughout the reaction cycle (Fig. 5).
Accepted Article
180
185 186 187
Microscale N2O analysis Microsensor measurements were conducted under two representative conditions imitating the
188
beginning and end of one SBAR operation cycle (Fig. 6). The net N2O production in the presence of
189
acetate (i.e., imitating the beginning of the cycle) was low and found >500 μm from the surface of
190
the granules. In this condition, N2O consumption was observed near the surface (300–400 µm) of the
191
granules. In the absence of acetate (i.e., imitating the end of the cycle), N2O production also occurred
192
throughout the granules, with a peak at 600 μm depth, although no significant N2O consumption was
193
detected.
194
Based on the FISH analysis, Nitrosomonas-related AOB were present near the surface (0–300
195
µm) of the granules (Fig. 6). Bacteria other than AOB, most of which were probably heterotrophs,
196
were present near the surface as well as in the deeper part of the granules.
197 198 199
DISCUSSION
The amount of N2O emitted by our PN aerobic reactor changed greatly over time. The initial
200
peak of N2O was attributed partly to the stripping of accumulated N2O after the resumption of
201
aeration, which agreed with the observations in another N2O monitoring study (Rathnayake et al., 8
This article is protected by copyright. All rights reserved.
2013). However, denitrification may have also occurred because the NO2− and TOC concentrations
203
declined with the increase in the pH. In contrast, no apparent correlation was found between N2O
204
emissions and major water chemistry parameters.
Accepted Article
202
205
In our PN aerobic granule reactor, 5.6% of the NH4+ load was converted to N2O, which was
206
larger than the previous PN studies (e.g., 1.7% in Kampschreur et al., 2008; 4.0% in Okabe et al.,
207
2011; 0.8% in Rathnayake et al., 2013) but smaller than the conventional nitrification-denitrification
208
systems (e.g., >20% in Itokawa et al., 2001). The TOC concentration may have affected the amount
209
of N2O produced and consumed (Itokawa et al., 2001).
210
Stable isotope techniques have been used to identify the key pathways of N2O emissions from
211
wastewater treatment plants, agricultural fields, forests, oceans, and other environments (Maeda et
212
al., 2010; Toyoda et al., 2011a; Toyoda et al., 2011b; Wunderlin et al., 2012). N2O isotopomer
213
analysis is particularly useful for distinguishing N2O derived from NH2OH oxidation and NO2-
214
reduction. However, there are several limitations with this approach. First, N2O reduction increases
215
the SP value; therefore, isotopomer analysis alone may underestimate the effect of NO2- reduction in
216
environments where N2O reduction is intense (Ostrom et al., 2007). Second, isotopomer analysis
217
cannot distinguish NH2OH oxidation and fungal denitrification (NO2- reduction) because fungal
218
denitrification, although based on the tests performed using limited number of strains, produces a
219
similar SP value to that of NH2OH oxidation (approximately 37‰) (Sutka et al., 2008). Third, it is
220
not possible to distinguish the relative contributions of nitrifier denitrification and heterotrophic
221
denitrification using isotopomer analysis because both pathways occur through NO2- reduction.
222 223
To overcome the first problem, we used both SP and δ15Nbulk values to calculate the occurrence
of N2O reduction using the Monte Carlo method (Toyoda et al., 2011b). With this calculation, we can 9
This article is protected by copyright. All rights reserved.
account for the 15N isotope effect by N2O reduction, and therefore, can estimate the relative
225
contributions of NH2OH oxidation and NO2- reduction more precisely. Our results suggest that the
226
majority (approximately 70–80%) of the N2O produced in our PN reactor originated from NO2-
227
reduction rather than from NH2OH oxidation. Fungi were not detected in our PN reactor according to
228
the fungi-specific PCR (data not shown). Therefore, occurrence of fungal denitrification, the second
229
problem mentioned above, was not relevant in this study. Thus, we did not consider that the
230
contribution of NO2- reduction to N2O production was underestimated in our PN reactor. NO2-
231
reduction was the major pathway for N2O production in our heterotrophic PN reactor, which
232
contrasts with autotrophic PN reactors where 65% of the N2O is produced through NH2OH oxidation
233
(Rathnayake et al., 2013). This suggests that the N2O production mechanism differs in autotrophic
234
and heterotrophic PN reactors.
Accepted Article
224
235
Based on the N2O isotopomer analysis alone, it was difficult to estimate the relative contribution
236
of nitrifier denitrification and heterotrophic denitrification to N2O production (the third problem
237
mentioned above). To overcome this, we conducted transcriptome analyses. mRNA-based
238
transcriptome analysis is useful to assess microbial activity because mRNA is degraded within a
239
short time (Philippot and Hallin, 2005). The half-lives of the nirS, cnorB, and nosZ transcripts in
240
active denitrifiers are approximately 13 min (Härtig and Zumft, 1999). In our study, the increased
241
transcription levels of nirS and cnorB during the initial stage of the SBAR operation cycle (10–20
242
min after substrate feeding) agreed with the sharp rise in N2O and N2 emission during the same
243
period. Both nirK and nirS encode nitrite reductase, but the transcription levels of
244
Nitrosomonas-specific nirK (Cantera and Stein, 2007) were relatively constant during the reaction
245
cycle. Heterotrophic denitrifiers, but not AOB, possess nirS; therefore, these results indicate that 10
This article is protected by copyright. All rights reserved.
heterotrophic denitrifiers were responsive to the addition of substrates and might be active during the
247
initial 30 min of the PN reaction.
Accepted Article
246
248
We did not detect transcripts of norY, a gene encoding sNOR of AOB and some other bacteria
249
(Klotz and Stein, 2008; Stein et al., 2007), although the PCR amplification was successful from the
250
PN granule DNA samples. This was in contrast to the report by Kartal et al. (2012), in which sNOR,
251
but not cNOR, was expressed in NO2--exposed cells of Nitrosomonas eutropha C91 under both oxic
252
and anoxic conditions. In Nitrosomonas europaea nirK-deficient mutant and Nitrosococcus oceani
253
ATCC 19707, transcription levels of norY also increased in the nitrifying cells under oxic conditions
254
(Cho et al., 2006; Stein et al., 2013). However, Beyer et al. (2009) reported that the transcription of
255
cnorB increased in Nitrosomonas europaea ATCC 19718 under nitrifier denitrification conditions.
256
Therefore, expression of sNOR and cNOR may depend on strains or incubation conditions. In this
257
study, Nitrosomonas eutropha-related AOB actively performed ammonia oxidation in the PN aerobic
258
granules, but some AOB cells might face anoxic conditions in the granules. In addition, heterotrophic
259
denitrifiers present in the PN granules might have influenced the microenvironment around AOB
260
(e.g., removal of NO), which could also have influenced the gene expression of AOB.
261
We also did not detect qnorB, a gene encoding quinol-dependent nitric oxide reductase (qNOR),
262
and its transcripts. The qNOR is considered as NO detoxification enzyme and is found in the
263
genomes of various bacteria and archaea, mostly in non-denitrifiers (Jones et al., 2008). Because our
264
PN aerobic granules contained mostly AOB and denitrifiers (Song et al., 2013), these results
265
indicated that the contribution of qNOR-containing microbes was very small, if any, in our reactor.
266 267
In contrast to the increased transcription levels of nirS and cnorB, the transcription level of nosZ
did not change over time. The gene transcription levels do not necessarily reflect the activity of the 11
This article is protected by copyright. All rights reserved.
coding enzymes, but our results indicated that the N2O reducing activity might have been constant
269
throughout the reaction cycle. The failure to detect narG suggests that nitrite, rather than nitrate, was
270
the substrate for denitrification, which agreed with the low concentration of NO3− in our reactor. A
271
previous batch experiment also showed that denitrifiers in the PN granule reactor preferred NO2− to
272
NO3−, which was probably due to the long-term acclimation to a high concentration of NO2− (Song et
273
al., 2013).
Accepted Article
268
274
The contribution of denitrification to N2O emission was analyzed by pyrosequencing of the
275
cnorB transcript. The cnorB sequences from Nitrosomonas and Nitrosococcus AOB clustered
276
together and were distantly related to other cnorB sequences, similar to the previous studies
277
(Casciotti and Ward, 2005; Jones et al., 2008). This suggests that the cnorB sequences affiliated to
278
OTU 2A_63087 originated from Nitrosomonas and Nitrosococcus AOB. The representative sequence
279
of OTU 2A_63087 was most closely related to the cnorB from Nitrosomonas eutropha, in agreement
280
with the presence of AOB closely related to N. eutropha in our PN aerobic granules (Song et al.,
281
2013). The quantity of Nitrosomonas-related cnorB, which was estimated based on the quantity of
282
cnorB transcripts and the proportion of Nitrosomonas-related cnorB, remained relatively constant
283
during the SBAR operation cycle, similar to the results obtained by RT-qPCR targeting
284
Nitrosomonas-specific nirK. Nitrosomonas-related cnorB comprised a relatively small proportion of
285
the cnorB libraries, which indicates that the contribution of nitrifier denitrification to N2O emission
286
may be small. If we assume that the abundance of the Nitrosomonas-related cnorB sequence
287
reflected the proportion of N2O produced by AOB, the contribution of nitrifier denitrification to N2O
288
production was
