Subscriber access provided by UNIVERSITY OF ADELAIDE LIBRARIES
Ecotoxicology and Human Environmental Health
Carbonaceous Nanomaterials Have Higher Effects on Soybean Rhizosphere Prokaryotic Communities During the Reproductive Growth Phase than During Vegetative Growth Yuan Ge, Congcong Shen, Ying Wang, Yao-Qin Sun, Joshua P Schimel, Jorge L. Gardea-Torresdey, and Patricia A. Holden Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.8b00937 • Publication Date (Web): 02 May 2018 Downloaded from http://pubs.acs.org on May 5, 2018
Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.
Page 1 of 36
Environmental Science & Technology
1
Carbonaceous Nanomaterials Have Higher Effects
2
on Soybean Rhizosphere Prokaryotic Communities
3
During the Reproductive Growth Phase than During
4
Vegetative Growth
5
Yuan Ge1,2,3,4, Congcong Shen1, Ying Wang2,3,4, Yao-Qin Sun1, Joshua P. Schimel3,4,5, Jorge L.
6
Gardea-Torresdey4,6, Patricia A. Holden2,3,4,*
7
1
8
Sciences, Chinese Academy of Sciences, Beijing 100085, China
9
State Key Laboratory of Urban and Regional Ecology, Research Center for Eco-Environmental
2
Bren School of Environmental Science and Management, University of California, Santa
10
Barbara, California 93106, United States
11
3
12
States
13 14 15 16
Earth Research Institute, University of California, Santa Barbara, California 93106, United
4
University of California Center for the Environmental Implications of Nanotechnology (UC
CEIN), University of California, Santa Barbara, California 93106, United States 5
Department of Ecology, Evolution and Marine Biology, University of California, Santa
Barbara, California 93106, United States
ACS Paragon Plus Environment
1
Environmental Science & Technology
17 18
6
Page 2 of 36
Department of Chemistry, University of Texas at El Paso, El Paso, Texas 79968, United States.
*
Corresponding Author:
[email protected]; Tel: 805-893-3195; Fax: 805-893-7612
ACS Paragon Plus Environment
2
Page 3 of 36
Environmental Science & Technology
19
Abstract: Carbonaceous nanomaterials (CNMs) can affect agricultural soil prokaryotic
20
communities, but how the effects vary with crop growth stage is unknown. To investigate this,
21
soybean plants were cultivated in soils amended with 0, 0.1, 100, or 1000 mg kg-1 of carbon
22
black, multi-walled carbon nanotubes (MWCNTs), or graphene. Soil prokaryotic communities
23
were analyzed by Illumina sequencing at day 0 and at the soybean vegetative and reproductive
24
stages. The sequencing data were functionally annotated using the Functional Annotation of
25
Prokaryotic Taxa (FAPROTAX) database. The prokaryotic communities were unaffected at day
26
0, and were altered at the plant vegetative stage only by 0.1 mg kg-1 MWCNTs. However, at the
27
reproductive stage—when pods were filling—most treatments (except 1000 mg kg-1 MWCNTs)
28
altered prokaryotic community composition, including functional groups associated with C, N,
29
and S cycling. The lower doses of CNMs—which were previously shown to be less
30
agglomerated and thus more bioavailable in soil relative to the higher doses—were more
31
effective towards both overall communities and individual functional groups. Taken together,
32
prokaryotic communities in the soybean rhizosphere can be significantly phylogenetically and
33
functionally altered in response to bioavailable CNMs, especially when soybean plants are
34
actively directing resources to seed production.
35
ACS Paragon Plus Environment
3
Environmental Science & Technology
36
Page 4 of 36
TOC/Abstract Art
37 38
ACS Paragon Plus Environment
4
Page 5 of 36
Environmental Science & Technology
39
Introduction
40
Carbonaceous nanomaterials (CNMs), e.g., carbon black, carbon nanotubes, and graphene, are
41
used in diverse commercial applications such as pigments, automotive tires, composite bicycle
42
frames, antifouling coatings, solar cells, capacitors, and water filters.1-4 CNMs may be released
43
into soils with field application of CNM-containing biosolids,5 or through the intentional use of
44
CNM-containing agrochemicals (e.g., fertilizer products).6-8 The accumulation of CNMs in soils
45
raises concerns about CNM effects on soil microorganisms—the main catalysts of soil nutrient
46
cycling in terrestrial ecosystems.
47
Most studies addressing CNM effects on soil microorganisms have been conducted in
48
unplanted soil microcosms. Such studies have examined the effects of various CNMs including
49
fullerenes,9-12 single-walled carbon nanotubes (SWCNTs),11, 13-17 multi-walled carbon nanotubes
50
(MWCNTs),11, 17-21 graphene,21, 22 graphene oxide,23, 24 and carbon black.21 Results from these
51
studies have been mixed, indicating that CNM effects on soil bacterial communities might be
52
tempered by CNM properties, soil conditions, exposure times, exposure doses, and CNM types.
53
For example, Tong et al. (2007) reported limited effects of fullerenes on soil microbial biomass,
54
community structure, respiration, and enzymatic activities.10 In contrast, Johansen et al. (2008)
55
found 20–30% of the community changed in response to fullerenes, although soil respiration and
56
microbial biomass were unaffected.9 The magnitude of effects seems to vary with soil conditions;
57
for example, fresh, unmodified SWCNTs altered microbial communities and metabolic activity
58
in low, but not high, organic matter soils.13 CNM effects also change with exposure time as
59
evidenced by short-term (3-4 days)
60
CNM type may also relate to the magnitude of effects. Shan et al. (2015) found that biochar had
61
no effect on catechol mineralization; however, activated carbon at all amendment doses (0.2, 20,
22, 25
and long-term (1 year)
21
studies. Exposure dose and
ACS Paragon Plus Environment
5
Environmental Science & Technology
Page 6 of 36
62
and 2000 mg kg-1) and SWCNTs at 2000 mg kg-1 significantly reduced mineralization, whereas
63
MWCNTs at 0.2 mg kg-1 significantly stimulated mineralization.17 Oyelami et al. (2015) found
64
higher glucose mineralization in fullerene amended soils, compared to MWCNT-, SWCNT-, or
65
fullerene soot-amended soils.11 In a 1-year exposure experiment where the effects of multiple
66
CNMs were compared, some treatments (e.g., biochar, carbon black, narrow MWCNTs, and
67
graphene) altered bacterial communities when compared to the no amendment control, but there
68
were no significant differences across the amendment treatments.21 CNMs differ in size,
69
morphology, surface chemistry and other physiochemical properties; thus, they may undergo
70
agglomeration, sorption, migration, and surface modification differently when they are released
71
into soils.26, 27 Such physicochemical changes to CNMs could alter their exposures to organisms
72
and thus observed effects including to soil microbial communities.
73
While multiple studies have examined CNM effects on microbial communities in unplanted
74
soils, little is known about CNM effects on microbial communities in planted soils.28 Yet
75
exposing soil microbial communities to CNMs in mesocosms with plant cultivation represents a
76
societally-relevant situation, where the interactions between crop plant roots and soils may
77
modify nanomaterial effects on soil microbial communities and their associated functions.29, 30
78
For example, plants exude into the soil 5−10% of their fixed carbon in the forms of sugars,
79
amino acids, organic acids, mucilage and organic chelators,31, 32 which may modify nanomaterial
80
bioavailability and toxicity.33, 34 Also, the quantity and composition of root exudates change with
81
plant growth,
82
CNMs on soil microbial communities during plant growth, including as a consequence of plant
83
growth stage, are mostly unknown.
35, 36
which may further temper nanomaterial effects. However, the effects of
ACS Paragon Plus Environment
6
Page 7 of 36
Environmental Science & Technology
84
Here, we studied the effects of carbon black, MWCNTs, and graphene on soil prokaryotic
85
communities during the course of soybean plant growth. This work builds on a prior
86
publication37 from the same mesocosm study in which soybean plants were cultivated to maturity
87
in soils amended with 0, 0.1, 100, or 1000 mg kg-1 of either carbon black, MWCNTs, or
88
graphene. In the previous publication, Wang et al. (2017) reported that all three CNMs affected
89
soybean growth, nodulation, and dinitrogen fixation potential, with stronger effects more
90
frequently observed at lower CNM doses. Through separate studies on CNM concentration-
91
dependent agglomeration in soil water extracts, the authors demonstrated that the greater CNM
92
agglomeration at higher CNM concentrations likely decreased CNM dispersal and bioavailability
93
in the soil, and thereby decreased CNM effects on soybean plants and dinitrogen fixing
94
symbioses (Supporting Information).37 This preceding publication provides the foundation for
95
the present study. Here we asked: How did the same three CNMs affect rhizosphere prokaryotic
96
communities? How did the effects on bacterial communities change with time during the course
97
of soybean plant growth which itself was affected by CNM exposure?37 How did the effects on
98
bacterial communities vary with CNM concentration, including if—as observed for soybean
99
plants by Wang et al. (2017)37—there were stronger effects at lower CNM concentrations? To
100
answer these questions, soil prokaryotic communities were analyzed at day 0, and at the
101
vegetative (day 20), and reproductive (day 39) plant growth stages to compare the temporal
102
variation of treatment effects. Prokaryotic taxa were mapped to metabolic or other ecologically
103
relevant functional groups using the Functional Annotation of Prokaryotic Taxa (FAPROTAX)
104
database, which allows for comparing CNM effects on individual functional groups at different
105
soybean growth stages. The findings of this study newly show that CNMs have higher effects on
106
soybean rhizosphere prokaryotic communities with associated functional groups during the plant
ACS Paragon Plus Environment
7
Environmental Science & Technology
Page 8 of 36
107
reproductive growth phase relative to vegetative growth. Among the effects are changes to taxa
108
associated with C, N, and S biogeochemical cycling. That such effects are more pronounced
109
when soybean plants are actively directing resources to seed production is suggestive of the
110
potential for ecosystem consequences of CNM exposure at late plant growth stages.
111 112
Materials and Methods
113
Soil. Surface soil (0-20 cm depth) was collected from the University of California Sedgwick
114
Reserve (34°40’32”N, 120°2’27”W), sieved (4 mm) and stored (4 oC) for less than two weeks
115
before the exposure experiment. Soil properties, including texture, pH, saturation, cation
116
exchange capacity, soluble salts, organic matter, total nutrients (C, Cu, Fe, Mn, N, Zn),
117
extractable nutrients (B, Ca, Cl, Cu, Fe, Mg, Mn, Na, P, Zn, HCO3-, CO32-, NH4+, NO3-), and
118
exchangeable nutrients (Ca, K, Mg, Na), were characterized by the UC Davis Analytical
119
Laboratory (Davis, CA; http://anlab.ucdavis.edu/) and reported previously.37 The soil is a Pachic
120
Argiustoll in the Botella series, with a sandy clay loam texture (50% sand, 25% silt, and 20%
121
clay) containing 3.03% organic matter, 1.53% total C, and 0.15% total N; the pH was 7.38.37
122
Carbonaceous Nanomaterials (CNMs). The three CNMs used in this study had distinct
123
morphologies and sizes.37, 38 Carbon black (Printex 30, Orion Engineered Carbons, Kingwood,
124
TX) was spherical, 36.6 ± 8.3 nm diameter; MWCNTs (Cheap Tubes, Grafton, VT) had outer
125
diameters of 18.8 ± 4.1 nm; graphene (Cheap Tubes, Grafton, VT) was a two-dimensional sheet
126
with an average diameter of 350 ± 320 nm and a thickness of 8-12 nm.37, 38 The three CNMs had
127
high purities, as reflected by low non-carbon impurities (< 2.2% for all) and low metal contents
128
(< 0.1% for all metals measured, except 0.9% of nickel in the MWCNTs).37 The primary
ACS Paragon Plus Environment
8
Page 9 of 36
Environmental Science & Technology
129
oxidation temperatures (an indicator of thermal stability) for CB, MWCNTs, and GNPs differed
130
and were approximately 620, 585, and 623 °C, respectively.37
131
Soybean Seedlings. As described before,37 to cultivate soybean seedlings, soybean seeds
132
(Glycine max, Midori Giant variety, Lot No. WA15060001, Park Seed Co., Hodges, SC) were
133
soaked in a Bradyrhizobium japonicum USDA 110 inoculum (1.0 optical density at 600 nm) for
134
10 min. Then, the inoculated seeds were sowed into rehydrated peat-filled seed starter pellets
135
(Park Seed Co., Hodges, SC), with an additional 100 µL of the B. japonicum inoculum dispensed
136
onto the sowed seeds. The pellets were watered daily and incubated on a heating mat (23 °C) for
137
ten days before seedling transplantation.37
138
Exposure Experiment. To homogenously distribute CNMs into soils, a previously reported 39
139
10-fold dilution method
140
mg kg-1 for either carbon black, MWCNTs, or graphene). The exposure doses were chosen,
141
based on the evaluation of exposure concentrations in previous publications, to represent a range
142
of possible environmental exposures: the predicted environmental concentration, the possible
143
presence of CNM hotspots in soil, and a relatively high exposure concentration to explore
144
potential toxicity.3, 40, 41 In brief, powder nanomaterials were initially added to soil in doses of
145
0.01, 10, and 100 g kg-1, and mechanically mixed for 10 min with handheld kitchen mixers. Then
146
each mixture was diluted 10-fold twice using unspiked soils, and mechanically mixed each time
147
as described above. Soil without CNMs was used as the control. After mixing, triplicate soil
148
samples of each treatment were immediately subsampled, denoted as 0-day samples, and frozen
149
at -80°C for later DNA-based analyses.
150 151
was used to achieve the targeted exposure doses (0.1, 100, and 1000
For each of the ten treatments (control, and three doses of each CNM), eight experimental pots were prepared following the procedures reported previously.37, 39 In each experimental pot (2.84
ACS Paragon Plus Environment
9
Environmental Science & Technology
Page 10 of 36
152
L container with bottom perforations, high density polyethylene) there was a layer of
153
polyethylene WeedBlock fabric (Easy Gardener Products, Waco, TX) at the bottom, and then 0.4
154
kg of washed gravel on top to maintain drainage. Then, 2.3 kg of soil in a perforated
155
polyethylene bag (40 holes) was overlain on the gravel. One soybean seedling was transplanted
156
into each experimental pot. Each seedling was inoculated with B. japonicum (10 mL, prepared as
157
above) during transplantation to ensure effective inoculation.37
158
The planted pots were placed in a greenhouse for up to 39 days wherein the soybean plants had
159
grown to full maturity (before senescence). The greenhouse was under full sunlight, with daily
160
temperatures ranging from 15oC to 34oC, and photosynthetically active radiation (PAR)
161
fluctuating between 21 and 930 µmol m-2 s-1. The soil water content averaged 0.25 m3 m-3 by
162
watering the experimental pots with tap water.37
163
Soil Sampling. During the exposure experiment, soil and plant samples were destructively
164
harvested twice based on soybean plant growth stages: vegetative (20 days post transplantation,
165
before flowering) or reproductive (39 days post transplantation, full seed production, before
166
senescence).37 For each of the ten treatments, three and five replicates were sacrificed at the
167
vegetative and reproductive stages, respectively. This sampling strategy minimized the
168
possibility of insufficient experimental replicates (i.e., less than 3) at the second sampling time,
169
in case of plant death. As reported previously, plant samples were examined for CNM effects.37
170
For this study, soil samples were preprocessed by manually removing most main roots and
171
sieving (2 mm) out any remaining fine roots. The sieved soils were stored at -80o C for later
172
DNA extraction to characterize soil prokaryotic communities. Because the plant roots extended
173
throughout the pots and were of similar masses at the intermediate and reproductive growth
174
stages,37 all soils were rhizospheric.
ACS Paragon Plus Environment
10
Page 11 of 36
Environmental Science & Technology
175
Soil DNA Extraction and Illumina Sequencing. Total DNA was extracted from 0.3 g of
176
subsampled soil from each of the replicates, for each treatment and sampling time, using the
177
Powersoil DNA Isolation Kit (Mo Bio, Carlsbad, CA, USA). The sequencing library was
178
prepared by the UC Santa Barbara Biological NanoStructures Lab
179
(https://www.cnsi.ucsb.edu/resources/facilities/bnl/ngs-core) following a standard protocol. In
180
brief, prokaryotic primers 341F (CCT ACG GGN GGC WGC AG) and 805R (GAC TAC HVG
181
GGT ATC TAA TCC) were used to amplify the V3-V4 region of genes encoding 16S rRNA for
182
each DNA sample, with each primer set containing a unique barcode. The unique barcodes were
183
used to assign sequences to samples post-sequencing. After size and quality verification by
184
TapeStation (Agilent, Santa Clara, CA), PCR products from each sample were quantified using a
185
Qubit Fluorometer (Invitrogen, Eugene, OR) and equally pooled by mass. The pooled PCR
186
products were shipped to the UC Riverside IIGB Genomics Core facility
187
(http://illumina.ucr.edu/ht/) on dry ice for paired-end sequencing with 300 cycles (PE300) on a
188
MiSeq platform (Illumina, San Diego, CA). There were four sequencing outliers defined as
189
having a low sequencing depth (< 8500 sequence counts): two were from the medium and high
190
doses of the MWCNT treatment at day 0, and the other two were from the medium and high
191
doses of the carbon black treatment at the reproductive stage (day 39). These outliers were not
192
included in the statistical analyses (Table S1).
193
Sequence
Preprocessing.
Sequences
were
preprocessed
using
QIIME
software
194
(http://qiime.org/).42 Briefly, all sequences that passed the quality controls of the MiSeq platform
195
were assigned to samples by examining the unique barcodes, and the primers were trimmed.
196
Then, the forward and reverse sequences from the same sample were merged based on the
197
overlap between them (10 bp of minimum overlap). The merged sequences were further screened
ACS Paragon Plus Environment
11
Environmental Science & Technology
Page 12 of 36
198
to remove low quality sequences that had an average quality score of < 20 and contained any
199
ambiguous characters. After initial trimming and screening, the PCR chimeras were removed,
200
and similar sequences were clustered into operational taxonomic units (OTUs, a cutoff
201
dissimilarity of 0.03) using the “usearch” method. Because the number of final qualified
202
sequences varied according to the sample, the sequence counts of all samples (i.e., 15000
203
sequence counts) were similarly rarefied through a random subsampling process to increase the
204
reliability of community comparisons across samples. The rarefied sample-OTU matrix was used
205
for the OTU-based community analysis. The OTU abundance was counted as the number of
206
sequences that clustered into a specific OTU. The community richness was estimated by the
207
number of observed OTUs per sample.
208
The OTUs were assigned to a set of hierarchical taxa (phylum, class, order, family, and genus)
209
using the Ribosomal Database Project (RDP) Classifier. Then the taxa were classified to
210
different functional groups using the Functional Annotation of Prokaryotic Taxa (FAPROTAX)
211
database (http://www.zoology.ubc.ca/louca/FAPROTAX/). The FAPROTAX is a software tool
212
designed to map prokaryotic taxa (e.g., genera or species) to established metabolic or other
213
ecologically relevant functions based on the functional descriptions of cultured strains in the
214
peer-reviewed literature.43 In FAPROTAX, functional groups may be nested since a taxon may
215
be affiliated with multiple functions; for example, all taxa associated with nitrate denitrification
216
are also associated with nitrate respiration and nitrate reduction. Currently, the FAPROTAX
217
includes 90 functional groups covering the main processes of C, N, and S cycling (e.g.,
218
cellulolysis, methanotrophy, methanogenesis, dinitrogen fixation, nitrification, denitrification,
219
sulfur oxidation and respiration; Table S2).
ACS Paragon Plus Environment
12
Page 13 of 36
Environmental Science & Technology
220
Statistical Analysis. After testing the normality and variance homogeneity, one-way analysis
221
of variance (ANOVA) was performed to test the global effect of exposure dose on measured
222
variables at each sampling time. Where the global ANOVA was significant (P < 0.05), a post-
223
hoc least significant difference (LSD) test was further conducted to test the significance (P
0.18 for all pairs, Figure 1a), by the
240
convergence of 0-day treatments in the PCoA graph (Figure 2a), and by the PERMANOVA test
241
of community shifts (P > 0.09, Table S3). These results corroborate other studies,21, 45 indicating
ACS Paragon Plus Environment
13
Environmental Science & Technology
Page 14 of 36
242
the negligible effects of CNM exposure on DNA extraction and bacterial community analysis
243
methods.
244
The CNM effects became more substantial with soybean growth (exposure time). At the
245
vegetative stage (day 20), only the low dose of MWCNTs significantly changed soil prokaryotic
246
communities (P < 0.05, Figure 1b). At the reproductive stage (day 39), almost all CNM-amended
247
treatments significantly affected soil prokaryotic communities (P < 0.05 for all except the high
248
dose of MWCNTs, Figure 1c). This was also supported by the PCoA graphs (Figure 2) and the
249
associated PERMANOVA test (Table S3), showing the greater separation of CNM-amended
250
treatments from the control at the reproductive stage when compared to day 0 and the vegetative
251
stage.
252
A similar time-dependent trend was observed for community richness. The community
253
richness was not significantly affected by any of the CNM treatments at either day 0 (P > 0.39)
254
or the vegetative stage (20 days, P > 0.12), but the CNM effects tended to be more distinct after
255
39 days at the soybean reproductive stage (Figure 3). At that later sampling, community richness
256
was significantly reduced by the medium and high doses of carbon black (10.7% and 9.6%), as
257
well as the medium dose of MWCNTs (12.2%, P < 0.05, Figure 3c). These three treatments were
258
also the treatments that caused the most distinct community shifts (Figure 1c), indicating that the
259
CNM-induced richness reduction might explain the community shifts.
260
Exposure time has been found to be an important factor in modulating nanomaterial effects on
261
soil microbial communities in unplanted soil microcosms. For example, functionalized SWCNTs
262
and graphene have been found to cause more distinct effects on soil bacterial communities at the
263
early stages of exposure (3-4 days),14, 22 while another study reported the effects of carbon black,
264
narrow MWCNTs, and graphene on soil bacterial communities even after 1-year exposure.21
ACS Paragon Plus Environment
14
Page 15 of 36
Environmental Science & Technology
265
Previous studies focusing on metal oxide nanoparticles also found that the effects on soil
266
bacterial communities increased with exposure time.45,
267
studies may be partially attributed to differences in nanomaterial types, soil properties, associated
268
communities, and experimental conditions. In this study, because soybean plants were cultivated
269
in the exposure experiment, it was also possible that the observed time-dependent CNM effects
270
on soil bacterial communities were due to soybean plant growth effects. Plants exude into the
271
soil 5−10% of their fixed carbon which may modify nanomaterial bioavailability and toxicity.33,
272
34
273
growth,35, 36 further altering nanomaterial effects. Besides the natural variations of root exudates
274
with plant growth, CNMs may have affected the quantity and composition of root exudates by
275
affecting plant growth,37 which in turn caused the time-dependent effects on soil prokaryotic
276
communities. In addition, the CNMs, e.g., carbon black, may be oxidized after deposition,47
277
which may also contribute to the effects. Although it is difficult to differentiate whether the
278
cause is purely time—independent of live plants—versus plant mediation, our results showed
279
that CNMs had higher effects on soybean rhizosphere prokaryotic communities during the
280
reproductive growth phase (day 39)—the most crucial stage of plant development in terms of
281
seed yield, when compared to the vegetative growth period (day 20).
46
The different time effects between
Also, the quantity and composition of root exudates may have been changed with soybean
282
CNM Effects Vary with CNM type. Besides the time-dependent CNM effects, we also found
283
that the three CNMs had different effects on soil prokaryotic communities. For example, only the
284
low dose of MWCNTs induced significant shifts of soil prokaryotic communities at the soybean
285
vegetative stage (day 20, P < 0.05, Figures 1b). At the soybean reproductive stage (day 39),
286
although almost all CNM treatments clearly affected soil prokaryotic communities (P < 0.05
287
except the high dose of MWCNTs), the effects differed in magnitude, as reflected by varied
ACS Paragon Plus Environment
15
Environmental Science & Technology
Page 16 of 36
288
community dissimilarities induced by different CNM treatments (Figure 1c). For example, the
289
community dissimilarities between the control and carbon black or MWCNT treatments (except
290
the high dose of MWCNTs) were around 1.9 times of the within-control community dissimilarity,
291
while the community dissimilarities between the control and graphene treatments were around
292
1.2 times of the within-control community dissimilarity (Figure 1c). This was also illustrated by
293
the PCoA graphs: the samples exposed to carbon black or MWCNTs (except the high dose of
294
MWCNTs) separated from the control more clearly than those from the graphene treatments
295
(Figure 2c, d, e).
296
These results indicate that the effects of CNMs on soil microbial communities are driven by
297
the type of material. In monocultures, metal oxide nanomaterials have been demonstrated to be
298
different in their toxicity to cell lines and microorganisms, yet nano-ZnO and nano-CuO appear
299
to be more toxic than other metal oxide nanoparticles.48-50 Herein, the differences in CNM effects
300
among different types of CNMs could derive from differing toxicity mechanisms. The toxic
301
mechanisms include membrane disorganization, surface coating-related photocatalytic oxidation
302
and associated cell damage, toxic ion release, and cell-damaging reactive oxygen species (ROS)
303
accumulation.51, 52 The specific toxicity mechanisms could be related to CNM physicochemical
304
properties (e.g., size and shape), which could vary with CNM type. For example, MWCNTs
305
were more toxic to Bacillus subtilis than graphene,53 which is consistent with our observation
306
that MWCNTs were more effective on soil prokaryotic communities than graphene. Even when
307
comparing within each type of CNM, the differences in particle size have been shown to alter the
308
toxicity to microorganisms. For example, the antimicrobial activities of graphene oxide sheets
309
were reported to vary with sheet size.54 Meanwhile, the variation in CNM effects could also be
310
attributed to different CNM exposures due to differing CNM agglomeration behaviors in soil,
ACS Paragon Plus Environment
16
Page 17 of 36
Environmental Science & Technology
37
311
and thus, differing bioavailabilities. In a prior publication
from the same mesocosm study,
312
Wang et al. (2017) examined the relationships between CNM concentrations and agglomeration
313
in soil extracts (Supporting Information). Graphene was found to be significantly less stable in
314
soil water extracts than carbon black at the same concentration (Table S4).37 Therefore, it is
315
possible that more extensive agglomeration of graphene significantly decreased its
316
bioavailability in soil and mitigated toxicity-related effects on soil prokaryotic communities.
317
Consequently, both the colloidal stability of CNMs in soil water and their differing toxicity
318
mechanisms (likely related to their physicochemical properties) are critical to the bioavailability
319
and effects of CNMs in soils.
320
CNM Effects Vary with Exposure Dose. We found that the low or medium doses of CNMs
321
tended to cause the most distinct effects on soil prokaryotic communities. For example, at the
322
vegetative stage, only the low dose of MWCNTs significantly altered soil prokaryotic
323
communities (P < 0.05, Figure 1b). At the reproductive stage, although all doses of carbon black
324
and graphene significantly affected soil prokaryotic communities (P < 0.05), the medium dose of
325
carbon black and low dose of graphene showed the highest effects (Figure 1c). In addition, with
326
MWCNTs, the lower doses significantly changed soil prokaryotic communities (P < 0.05), while
327
the high dose did not significantly affect soil prokaryotic communities (P = 0.79, Figure 1c).
328
These results indicate that the CNM effects on soil prokaryotic communities did not follow the
329
typical dose-response relationship in which the response increases with toxicant dose.
330
Previously, Wang et al. (2017) showed that the low dose of MWCNTs inhibited soybean
331
growth, as reflected by shorter plants, slower leaf cover expansion, and less final leaf area.37
332
Further, CNMs negatively affected nodulation and dinitrogen fixation potential, with stronger
333
effects at lower CNM doses.37 To explain the observed inverse dose-response relationships,
ACS Paragon Plus Environment
17
Environmental Science & Technology
Page 18 of 36
334
Wang et al. (2017) conducted separate experiments to examine the effect of CNM concentration
335
on CNM dispersal in soil water, and thus on CNM bioavailability in soil.37 Specifically, they
336
studied CNM stability at two CNM concentrations (10 and 300 mg L-1) in water extracts of the
337
control soil (Supporting Information).37 The two CNM concentrations studied were relevant to
338
the CNM concentrations in the soil solution of the mesocosm experiment (Supporting
339
Information). By measuring CNM agglomeration and sedimentation dynamics, and imaging
340
CNMs using environmental scanning electron microscopy, Wang et al. (2017) found that CNM
341
agglomeration increased with increasing CNM concentration (Table S4).37 The strength of CNM
342
effects depends partly on CNM bioavailability (defined as “the accessibility of a chemical for
343
assimilation and possible toxicity” 55), which in turn is affected by CNM colloidal stability in soil
344
pore water.56 CNM stability in soil water could determine the amount (i.e., available dose) and
345
physicochemical characteristics (e.g., size and shape) of CNMs that would disperse in soil water
346
and directly interact with plant roots and soil microorganisms. At higher CNM concentrations,
347
greater CNM agglomeration resulted in increased average sizes of CNM agglomerates (shown as
348
the larger hydrodynamic diameters) and decreased CNM concentrations (reflected by the lower
349
derived count rates) remaining in soil water (Table S4).37 Therefore, CNM bioavailability and
350
observed effects were reduced at higher CNM concentrations.37
351
In the present study, focused on soil prokaryotic communities, we found that lower doses of
352
CNMs caused more distinct effects on soil prokaryotic communities. This result may be
353
explained by higher dispersion-mediated bioavailability of inherently toxic CNMs, similarly to
354
the effects on soybean plants as was concluded in the previous study.37 The trend was better
355
evidenced at the reproductive plant growth stage, i.e. after 39 days of plant growth (Figures 1c
356
and 3c). If this was an effect of growth stage, rather than merely time, then the effect could
ACS Paragon Plus Environment
18
Page 19 of 36
Environmental Science & Technology
357
derive from plants exerting a differential influence on rhizosphere prokaryotic communities if
358
plants changed their root physiology while adjusting resource allocation during pod filling.
359
Functional Implications of Prokaryotic Community Shifts Responding to CNM Exposure.
360
We used Functional Annotation of Prokaryotic Taxa (FAPROTAX) to map prokaryotic clades
361
(e.g. genera or species) to metabolic or other ecologically-relevant functions. Among the 90
362
functional groups in the FAPROTAX database, 69 groups (76.7%) were present in at least one of
363
the samples (Table S2). These functional groups included taxa (Table S5) associated with
364
processes in the biogeochemical cycling of C (e.g., cellulolysis, ligninolysis, methanotrophy,
365
methanogenesis, hydrocarbon degradation), N (e.g., dinitrogen fixation, nitrification,
366
denitrification), S (e.g. the oxidation and reduction of sulfur, sulfite, and sulfate), and other
367
elements (e.g., the oxidation and reduction of iron, manganese, and arsenic, Table S2).
368
Pearson correlations between prokaryotic community dissimilarities, and the relative
369
abundance of each functional group for each CNM type at each sampling time, were examined.
370
This allowed determining the functional groups that significantly (adjusted P < 0.05) contributed
371
to the community shifts. We further examined how these functional groups were altered by CNM
372
exposure using ANOVA in conjunction with an LSD test. None of the 69 functional groups
373
represented in the samples was significantly correlated with bacterial community shifts for all
374
three CNM types at either day 0 or the vegetative stage (adjusted P > 0.05 for all, Table 1).
375
However, at the reproductive stage, 18, 25, and 5 functional groups, whose relative abundances
376
were significantly correlated to the community shifts (adjusted P < 0.05, Table S6), were also
377
significantly altered by carbon black, MWCNTs, and graphene, respectively (P < 0.05, Figures
378
4-6). Therefore, these functional groups were defined as sensitive functional groups (Table 1).
379
These results (i.e., more sensitive functional groups at the reproductive stage) mirrored the
ACS Paragon Plus Environment
19
Environmental Science & Technology
Page 20 of 36
380
community analysis results, showing that key functional groups affected by CNMs were changed
381
more at the soybean reproductive stage (Table 1). Further, the significant correlations (Table S6)
382
suggested a linkage between community shifts induced by CNM exposure and functional shifts.
383
This result may appear intuitive, given that affected functional groups are determined from
384
analyzing the phylogenetic data using the FAPROTAX database. However, given that less than 1%
385
of all microbial community taxa have been studied in culture,57 the functional implications in the
386
prokaryotic community shifts herein could be larger than what can be currently inferred using the
387
FAPROTAX database.
388
In this study, sensitive functional groups were not related linearly to NM dose, differently from 29
389
previous studies of metal oxide ENM effects on bacterial communities in soils with
390
without 46 plants. Herein, the medium doses of carbon black and MWCNTs, and the low dose of
391
graphene, tended to cause the highest effects on soil prokaryotic communities at the soybean
392
reproductive stage (Figures 1 and 2). Consistently, almost all of the sensitive functional groups
393
were significantly (P < 0.05) affected by the medium dose of carbon black (100% of the
394
sensitive functional groups, Figure 4) and MWCNTs (96%, Figure 5), and the low dose of
395
graphene (100%, Figure 6), while fewer functional groups were significantly altered by other
396
treatments (Table 1).
and
397
Interestingly, some of the functional groups associated with C, N, and S cycling were
398
significantly changed at the reproductive stage with some groups increasing while others
399
decreasing. For example, with carbon black exposure, 9 of the 18 sensitive functional groups,
400
such as those associated with methanogenesis, chitinolysis, aerobic nitrite oxidation, and nitrate
401
reduction, were significantly decreased (by 27-74%, P < 0.05), while functional groups
402
associated with cellulolysis, xylanolysis and aerobic ammonia oxidation were significantly
ACS Paragon Plus Environment
20
Page 21 of 36
Environmental Science & Technology
403
increased (P < 0.05, Figure 4). Of the 25 sensitive functional groups in the MWCNT treatments,
404
15 functional groups—including those associated with nitrification (aerobic nitrite oxidation),
405
denitrification (nitrate, nitrite, and nitrous oxide denitrification), nitrogen respiration (nitrate and
406
nitrite respiration), and anoxygenic photoautotrophic S oxidation—were significantly decreased
407
due to exposure to a certain dose (by 22-70%, P < 0.05, Figure S1), while the functional groups
408
associated with chitinolysis, xylanolysis and sulfur respiration were significantly increased (P