AEM Accepted Manuscript Posted Online 18 August 2017 Appl. Environ. Microbiol. doi:10.1128/AEM.01363-17 Copyright © 2017 American Society for Microbiology. All Rights Reserved.
1
Differences in temperature and water chemistry shape distinct diversity patterns in
2
thermophilic microbial communities
3 Cecilia M. Chiriaca,b, Edina Szekeresa,b, Knut Rudic, Andreea Baricza,b, Adriana Hegedusa,b, Nicolae
5
Dragoșa,b, Cristian Comana#
6 7
a
8
Biotechnology Department, Faculty of Biology and Geology, Babeş-Bolyai University, Cluj-
9
Napoca, Romania;
10
NIRDBS, Institute of Biological Research, Cluj-Napoca, Romania; bMolecular Biology and
c
Chemistry, Biotechnology and Food Science Department, Norwegian
University of Life Sciences, Aas, Norway.
11 12
Running title: Diversity patterns in hot spring microbial mats.
13 14
# Address correspondence to Cristian Coman, [email protected]
15 16
1
Downloaded from http://aem.asm.org/ on August 19, 2017 by FUDAN UNIVERSITY
4
17
Abstract This study presents the biodiversity and ecology of microbial mats developed in thermal
19
gradients (20-65 oC) in the surroundings of three drillings (CH, CI, MB) tapping a hyperthermal
20
aquifer from Romania. Using a metabarcoding approach, 16S rRNA genes were sequenced from
21
both DNA and RNA transcripts (cDNA) and compared. The relationships between the microbial
22
diversity and the physico-chemical factors were explored. Additionally, the cDNA data was used
23
for in silico functionality predictions, bringing new insights into the functional potential and
24
dynamics of these communities. The results showed that each hot spring determined the formation
25
of distinct microbial communities. In the CH mats (40-53 oC), the abundance of Cyanobacteria
26
decreased
27
Ectothiorhodospira, Oscillatoria and methanogenic archaea dominated the CI communities (20-
28
65 oC), while
29
Chloroflexi, Hydrogenophilus, Thermi and Aquificae. Alpha diversity was negatively correlated
30
with the increase in water temperature, while beta diversity was shaped in each hot spring by the
31
unique combination of physico-chemical parameters, regardless of the type of nucleic acid analyzed
32
(DNA vs. cDNA). The rank correlation analysis revealed a unique model that associated
33
environmental data with community composition, consisting in the combined effect of Na+, K+,
34
HCO3-, PO43- concentrations, together with temperature and electrical conductivity. These factors
35
seem to determine the grouping of samples according to location, rather than with the similarities in
36
thermal regimes, showing that other parameters beside temperature alone are significant drivers of
37
biodiversity.
38
Importance
with
the
temperature,
MB
opposite
microbial
to
that
of
(53-65 oC)
mats
Chloroflexi
were
and
mainly
Proteobacteria.
composed
of
39
Hot spring microbial mats are remarkable manifestation of life on Earth, and were
40
intensively studied for decades. Moreover, as hot spring areas are isolated and have a limited
41
exchange of organisms, nutrients and energy with the surrounding environments, hot spring
42
microbial communities can be used as model studies to elucidate the colonizing potential within
2
Downloaded from http://aem.asm.org/ on August 19, 2017 by FUDAN UNIVERSITY
18
extreme settings. Thus, they are of great importance in evolutionary biology, microbial ecology and
44
exobiology. With all the efforts that have been made, the current understanding of the influence of
45
temperature and water chemistry on the microbial community composition, diversity and abundance
46
in microbial mats is limited. In this study, the composition and diversity of microbial communities
47
developed in thermal gradients in the vicinity of three hot springs from Romania were investigated,
48
each geothermal water having particular physico-chemical characteristics. Our results expose new
49
factors that could determine the formation of these ecosystems, expanding the current knowledge in
50
this regard.
51 52
Introduction
53
Microbial mats are stratified communities of functional groups that are determined and
54
maintained by steep physico-chemical gradients (1). Hot spring microbial mats are of particular
55
interest as they are considered model ecosystems for the study of early life on Earth. By now, the
56
oldest evidence of life on land came from the ca. 3.48 billion year old Archaean rocks from Pilbara
57
Craton, Western Australia. These structures include stromatolites, or mineralized microbial mats,
58
that have developed in an ancient, hot spring environment (2). Modern hot spring habitats,
59
including various microbial mat ecosystems have been extensively studied, especially in the
60
Yellowstone National Park, USA (3 - 6), in the Tibetan Plateau, China (7 - 9), in Iceland (10, 11),
61
and Russia (12, 13). These studies have considerably improved our understanding of microbial
62
composition and diversity in terrestrial hot spring ecosystems (7). Thus, it is known that these
63
microbial mats lack seasonality and undergo changes in response to differences in physico-chemical
64
parameters, such as temperature, pH, sulfide concentrations etc. (1). These environments are
65
dominated by prokaryotes, as extreme temperatures usually exclude eukaryotic organisms (14). In
66
hyperthermal conditions that are not suitable for photosynthesis (>75oC) (15), thermal and
67
hyperthermal members of Bacteria are common inhabitants of microbial mats, being predominantly
68
affiliated to the Aquificae, Thermatogae, Deinococcus-Thermus ([Thermi]) phyla, or to certain
3
Downloaded from http://aem.asm.org/ on August 19, 2017 by FUDAN UNIVERSITY
43
thermophilic members of Proteobacteria (7). At lower temperatures and under light conditions,
70
oxygenic and anoxygenic photosynthesis usually support the primary production in the mat,
71
allowing a diversification of the microbial communities (16). Photosynthetic hot spring microbial
72
mats may be sustained by Cyanobacteria, Chloroflexi or Chlorobi, or by a combination of these
73
groups (17). Beside the bacterial domain, Archaea is also present in hot spring mats, being
74
represented especially by the Crenarchaeota and Euryarchaeota phyla (18, 19).
75
To better understand the ecology and microbial functions leading to the establishment of
76
microbial mats, a number of studies have focused on the links between the physico-chemical
77
parameters of the water and the microbial composition and diversity (7, 11, 16, 20 - 22). Within
78
those factors, temperature received much of the attention. While some studies explore the
79
temperature effect on small scale gradients (16, 20), others have selected springs within a wide
80
temperature range (11, 23). The results showed that the microbial mat composition and the
81
complexity of the ecosystem seem to be mainly determined by the rise in temperature (1). In the
82
study performed by Wang et al. (7), mat samples were collected from thermal gradients in different
83
areas and no location-specific variations were observed. Instead, the temperature was sufficient to
84
explain the variability in community composition. In contrast, other studies found that the sulphide
85
concentration (21) or the pH values (24) were important factors that control the similarity among
86
samples. Nevertheless, more effort is required to better understand the effects of physico-chemical
87
conditions on the development of hot spring microbial ecosystems.
88
As a result of geothermal resource exploitation in the second half of the 20th century, eight
89
geothermal areas were identified in Romania, with over 200 drillings ranging in depth from 800 to
90
3500 m, having the outflow water temperatures from 40oC to 120oC (25). Among these, a number
91
of drillings (80) are tapping the Pannonian hyperthermal aquifer (50 – 85oC) from the Western Plain
92
of Romania, the water that comes from 800 to 2100 m deep being used mostly for domestic heating,
93
health and recreational bathing, industrial processes or fish farming (25). Some wells have been
94
abandoned over the years and the continuous hot water flow has led to the development of specific
4
Downloaded from http://aem.asm.org/ on August 19, 2017 by FUDAN UNIVERSITY
69
mineralized communities and other photosynthetic microbial mats (26, 27). Even though the
96
biodiversity in some of these microbial mats has been previously described using 16S rDNA clone
97
libraries (26), the current study applied high-throughput sequencing techniques for a more thorough
98
investigation. It focuses on microbial mats formed along thermal gradients in hot springs from three
99
different locations, each with particular physico-chemical characteristics. An integrative approach
100
was performed that included both 16S rRNA transcripts and 16S rDNA gene sequencing. Although
101
the amount of ribosomal RNA that is detected in a microorganism is no longer viewed as a clear
102
indicator for microbial growth, a large number of copies does indicate the potential for protein
103
synthesis, for a rapid shift in metabolic function, or for an overall higher fitness of that species in
104
the ecosystem (28). The inferences of metabolic potential based on 16S rRNA transcripts may fall
105
under some degree of uncertainty (29). But, taking into consideration that extracellular RNA is less
106
stable in the environment compared with free DNA (30), functional roles can be attributed in this
107
way to intact organisms. By contrast, the DNA extracted from environmental samples is a
108
continuum of DNA derived from whole living organisms and extraorganismal, free DNA (31). A
109
more precise view on the microbial community composition, its diversity and its function may be
110
proposed by comparing the 16S rDNA gene libraries with those of 16S rRNA transcripts.
111
Thus, the aim of this research was to characterize the prokaryotic composition and
112
biodiversity in the newly investigated hot springs microbial mats associated with three hot springs
113
situated in the Western Plain of Romania and to explore their relationship with the physico-
114
chemical properties of the thermal water. As these microbial communities may bring new
115
information on the distribution of microbial groups in thermal environments, an additional effort
116
was made to compare their microbial composition with other microbial mats that have been
117
previously investigated through a metabarcoding approach.
118
Materials and Methods
119
Sampling procedure
120
Samples were taken in 2013 from microbial mats developed nearby three abandoned
5
Downloaded from http://aem.asm.org/ on August 19, 2017 by FUDAN UNIVERSITY
95
drillings from the Chiraleu (CH – drilling number 4095), Ciocaia (CI - drilling number 4045) and
122
Mihai Bravu (MB - drilling number 605) villages, at temperatures ranging from 20 to 65 oC (Table
123
1). Colorful microbial mats were observed and sampled at each location along with the temperature
124
gradient (Figure 1). In the surroundings of the Chiraleu drilling (Figure 1A), green and slimy
125
microbial mat developed at 53°C, without a clear vertical stratification (Figure 1B), as well as a
126
yellow microbial mat at 46°C which presented a typical laminated structure (Figure 1C). A thin
127
microbial mat was sampled from 40°C, with a smooth orange surface and green layers underneath
128
(Figure 1D). Three distinct microbial mats were observed in the proximity of the Ciocaia drilling
129
(Figure 1E), including green, mucilaginous and stratified microbial mats forming at 20°C (Figure
130
1F), red and slimy microbial mats developing at 35°C that do not show a clear stratification (Figure
131
1G) and white microbial mats at 65°C, more consolidated and firm near to the substrate and with a
132
mucilaginous, translucent surface layer (Figure 1H). Microbial mats and streamers were identified
133
near the Mihai Bravu drilling opening (Figure 1I), including a green, compact and smooth microbial
134
mat forming at 53°C, with thin layers (Figure 1J), a stratified microbial mat at 59°C, with an orange
135
top layer and green layers underneath (Figure 1K) and long, flexible structures forming cream -
136
colored streamers at 65°C along with the flowing hot water (Figure 1L).
137
The stable characteristics of the water sources were previously reported (32, 33) (Table 2).
138
These three drilling (CH, CI, MB) are tapping the Pannonian hyperthermal aquifer, a deposit of
139
neutral to slightly alkaline waters, with pH values ranging from 7.75 to 7.91. The high
140
concentrations of Na + (676 - 4120 mg/L), Cl- (537 - 1160 mg/L) and HCO3- (895 – 1980 mg/L) ions
141
are probably the result of the static nature of this deposit, which resulted in prolonged water - rock
142
interactions. The collecting rocks of the aquifer were formed in lagoon conditions that allowed the
143
accumulation of dissolvable chloride rocks, which ultimately impact the chemical composition of
144
the thermal deposit (32, 33). The temperature at the surface and at the bottom of each microbial mat
145
was determined using a field multiparameter (WTW Multi 340i Multiparameter, Weilheim,
146
Germany). As temperature was similar (±0.1 oC) between top and bottom (data not shown), only the
6
Downloaded from http://aem.asm.org/ on August 19, 2017 by FUDAN UNIVERSITY
121
surface values were reported. For DNA and RNA extraction, approx. 3 g of each microbial mat was
148
collected in 15 mL Falcon tubes in triplicate and TRIzolTM Reagent (Invitrogen) was immediately
149
added to fill up the tubes in order to maintain the integrity of the total RNA. All the samples were
150
transported on ice to the laboratory, and processed within the same day. After DNA and RNA
151
extraction, followed by cDNA synthesis, a total of 18 samples were used for prokaryotic cell
152
abundance estimation and metabarcoding (Table 1).
153
Nucleic acids extraction
154
Total DNA and RNA were extracted in triplicate for each microbial sample according to the
155
manufacturer’s instructions (TRIzolTM Reagent, Invitrogen) with slight modifications. 750 mg of
156
each sample (wet weight) were incubated at 60oC for 10 min, being vortexed prior to and at the end
157
of the incubation time. Then, 150 μL of chloroform were added and the tubes were vortexed for 10
158
s. The tubes were centrifuged at 10,000 × g for 10 min. The mixture was separated in three phases,
159
an upper aqueous phase, an inter-phase and a lower phenol-chloroform phase. The upper phase was
160
transferred in a new tube for RNA extraction, while the other two phases were used for DNA
161
purification. For DNA extraction, 300 μL of 100% ethanol were added to the lower phases, and the
162
components were gently mixed by inverting the tubes. Following a 3 min incubation period, the
163
tubes were centrifuged for 15 min at 12000 × g and 4°C to pellet the DNA. The pellet was
164
resuspended in1 mL of 0.1 M sodium citrate in 10% ethanol, pH 8.5, and was incubated at room
165
temperature for 30 min. The tubes were centrifuged for 15 min at 12000 × g at 4°C. This washing
166
step was repeated one more time, and the DNA pellets were resuspended in 1.5 mL of 75% ethanol.
167
The tubes were incubated for another 20 min at room temperature, and then were centrifuged for 15
168
min at 12000 × g and 4°C. The supernatant was discarded and the DNA was air dried for 15 min.
169
The pellet was resuspended in 500 μL of 8 mM NaOH by pipetting up and down. Then, the tubes
170
were centrifuged for 10 min at 12,000 × g and 4°C, the supernatant was removed, and the DNA
171
pellet was eluted in 35 µl PCR Grade water. The DNA replicates of the same sample were pooled,
172
and the DNA concentration was determined based on the Qubit® dsDNA BR Assay Kit, using a
7
Downloaded from http://aem.asm.org/ on August 19, 2017 by FUDAN UNIVERSITY
147
173
Qubit® Fluorometer (data not shown).
174
RNA extraction and cDNA synthesis For RNA purification, an equal amount of phenol – chloroform – isoamyl alcohol (25:24:1)
176
was added to the aqueous supernatant (see previous section) and the tubes were vortexed and
177
centrifuged at 10,000 x g for 2 min. The upper phase was carefully transferred to a new tube and
178
mixed with an equal amount of isopropyl alcohol. The samples were incubated at room temperature
179
for 10 min and then centrifuged at 4oC and 10,000 x g for 10 min. The supernatant was removed,
180
and 1 mL of 70% ethanol was added to the tubes placed on ice. After 5 min of incubation the tubes
181
were centrifuged at 4oC and 8,000 x g for 5 min. The ethanol was discarded and the RNA was
182
resuspended in 30 μL of PCR Grade water. Next, the DNA was removed using the TURBO DNA-
183
freeTM kit (ThermoFisher Scientific) and the RNA replicates of the same sample were pooled. The
184
RNA concentration was measured using the Qubit® Fluorometer and the Qubit® RNA BR Assay
185
Kit (data not shown). Total RNA reverse transcription was performed starting with 200 ng of
186
purified RNA for each sample using the First Strand cDNA Synthesis kit (Thermo Scientific)
187
according to the manufacturer’s instructions.
188
Quantification of prokaryotic cell abundance by qPCR
189
The PRK341F/PRK806R primers were used to target the 16S rRNA gene in the total
190
extracted DNA and cDNA by quantitative real time PCR (qPCR) (34). The reactions were
191
performed in triplicate using the SsoFast Eva Green Supermix (Bio-Rad, Hercules, CA, USA) on
192
the CFX96 Touch™ Real-Time PCR Detection System (Bio-Rad). The following components were
193
added in the reaction mix: 7 μL 1X Sso Fast EvaGreen SuperMix (Bio-Rad), 0.4 μM of the forward
194
and reverse primers, 20 ng of DNA/RNA and RNase/DNase-free water to a final volume of 14 μL.
195
The reaction program began with an initial denaturation at 98°C for 120 s, followed by 45 cycles of
196
10 s at 98°C and an annealing/elongation step of 30 s at 55°C. The gene copy number was
197
calculated by comparing the amplification results to a standard ten-fold serial dilution of known
198
quantities of recombinant plasmids (109 – 103) carrying the targeted gene. The plasmids were
8
Downloaded from http://aem.asm.org/ on August 19, 2017 by FUDAN UNIVERSITY
175
199
prepared as previously described (35). Additionally, three negative controls were analyzed in the
200
same run. A Ct cutoff value corresponding to the lowest Ct reported for the negative controls was
201
used, and the average number of gene copies detected in these controls were subtracted from the
202
sample values.
203
Preparation of sequencing libraries The V3-V4 regions of the 16S rRNA gene were amplified from the extracted DNA and
205
cDNA using the PRK341F/PRK806R primers (34) modified by addition of Illumina-specific
206
adaptors. The coverage of the primers against the SILVA reference database was 84.1% for
207
Bacteria and 58.5% for Archaea. Each PCR reaction (25 μL) contained 1X HOT FIREPol® PCR
208
mix (Solis BioDyne, Estonia), 200 nM uniquely tagged forward and reverse primers, 1 μL of
209
sample DNA and 18 μL water. The reaction conditions were: 95°C for 15 min, followed by 25
210
cycles of 95°C for 30s, 50°C for 30s, 72°C for 45s, ending with 72°C for 7 min. The amplicon
211
concentration was measured with the Qubit®dsDNA HS Assay Kit using the Qubit® Fluorometer
212
(Thermo Fisher Scientific, USA) and equal amounts of amplicons were pooled for each sample into
213
a normalized library. The pooled library concentration was measured using the PerfeCta® NGS
214
Quantification Kit for Illumina (Quanta BioSciences, USA) and the library was diluted in Tris pH
215
8.5 to a final 4 nM concentration. Sequencing was performed on a MiSeq platform (Illumina, USA)
216
using V3 sequencing chemistry with 300 bp paired-end reads.
217
Amplicon analysis
218
Raw sequence data (SRA accession number PRJNA389363) was processed and quality
219
filtered through a combination of Usearch v8 and QIIME pipelines (36, 37). Briefly, QIIME was
220
used to extract the barcodes from the sequence data, to join the forward and reverse Illumina reads
221
and to demultiplex the sequence data. Singletons removal and quality control filtering were
222
performed using the Usearch v8 pipeline, by discarding sequences with less than 350 nucleotides
223
and those with more than 0.2 total expected errors (-fastq_maxee E option of the fastq_filter
224
command). Remaining quality control filtration included de novo and reference chimera removal
9
Downloaded from http://aem.asm.org/ on August 19, 2017 by FUDAN UNIVERSITY
204
via the Usearch v8 pipeline, using the latest version of the Greengenes database ('13_8') as a
226
reference (38). The resulting Operational Taxonomic Unit (OTU) table was converted into the
227
Biological Observational Matrix (BIOM) format (39). Taxonomy was assigned using the default
228
classifier in QIIME (Greengenes) against the updated '13_8' version of the Greengenes database at a
229
97% similarity threshold. Next, the mitochondrial and plastidial sequences were filtered out of the
230
OTU table. A multiple sequence alignment was generated with PyNAST (36) in QIIME, and the
231
alignment was used to build a phylogeny with the FastTree algorithm (40).
232
Additionally, a comparison of the alpha and beta-diversity encountered in the Romanian
233
samples to other sequences from hot spring microbial mats deposited in the Sequence Read Archive
234
(SRA) was performed, including the following entries: SRA061437 (7), SRS932137, SRS932073
235
(41), PRJEB7059 (42). As in these studies different variable regions of the 16S rRNA gene were
236
amplified and sequenced, a closed-reference OTU picking was performed in QIIME (36) using a
237
97% similarity threshold. Prior to this step, sequences were first quality filtered as described in their
238
original studies. The final data set retained only those sequences that match a full-length
239
representative sequence within the last version ('13_8') of the Greengenes database.
240
Statistical analysis
241
Rarefaction was performed at a depth of 3499 sequences per sample, followed by alpha- and
242
beta-diversity estimation in QIIME (36). Euclidean distances for physico-chemical parameters were
243
computed with the vegdist function of the 'vegan' package in R, and the Mantel test (999
244
permutations) was run to evaluate the environmental impact on the microbial diversity using the
245
weighted Unifrac distances (43). Environmental physico-chemical data was log transformed prior to
246
the analysis. PCoA plots were generated for the weighted and unweighted Unifrac matrices, using
247
the cmdscale and envfit functions of the 'stats' and 'vegan' packages in R. The environmental factors
248
were fitted onto the ordination with the envfit function of the 'vegan' package in R, being scaled by
249
their correlation to the distance matrix. The significance of the sample grouping according to
250
distinct sampling sites or distinct type of investigated nucleic acids were tested by the analysis of
10
Downloaded from http://aem.asm.org/ on August 19, 2017 by FUDAN UNIVERSITY
225
251
similarity (ANOSIM) using 999 permutations (44) in QIIME. Phyla and classes with abundances
252
higher than 1% in the sequencing libraries were plotted using the inkspot function of the 'rioja'
253
package in R. The multilevel indicator OTUs (45) for each thermal spring were established using
254
the multipatt function of the 'perm' package in R.
255
Functional prediction For functionality prediction with PICRUSt (46), the OTUs were picked using an open
257
reference approach, and de-novo OTUs were removed, keeping in the final OTU-table only the
258
OTUs that had matching Greengenes IDs ('13_8'). Data in the BIOM OTU-table was normalized
259
using the 16S rRNA gene copy number for each OTU, and the normalized table was used for
260
KEGG functional orthologs predictions. In order to evaluate the PICRUSt predictions accuracy for
261
each sample, the weighted Nearest Sequenced Taxon Index (NSTI) was computed, an index that
262
describes the extent to which microorganisms from each sample are related to fully annotated
263
genomes, a value of 0.03 meaning that on average, microorganisms in the sample were predicted
264
using a relative from the same species.
265
Results and discussion
266
Abundance of prokaryotic 16S rRNA genes and transcripts
267
The abundance of prokaryotic microorganisms in the DNA and cDNA samples was
268
estimated through qPCR by targeting the 16S rRNA gene. Generally, the copy numbers in the
269
cDNA transcripts were found to be two to three orders of magnitude lower than the numbers
270
observed in the DNA sample (Figure 2). This difference may suggest that evaluation of cell
271
abundances should not be based solely on the 16S rRNA genes. Some studies proposed that up to
272
90% of the free DNA is stable in sediments and may not be affected by nucleases (47, 48). Thus,
273
the total extracted DNA most probably reflects not only the living microbes present in the
274
environment, but also the extracellular DNAs (49). Considering the instability of the free RNA, it
275
seems that it could provide a better estimation of the actual abundance of microbial cells that can
276
synthesize proteins (1, 50). In the CH samples, the number of 16S rDNA copies decreased with
11
Downloaded from http://aem.asm.org/ on August 19, 2017 by FUDAN UNIVERSITY
256
temperature from 1.1 ± 0.01 × 108 at 40oC to 3.4 ± 0.6 × 106 copies/mg at 53oC, whereas in the other
278
sampling sites no clear pattern was observed. In the CI DNA samples, the abundance of 16S rDNA
279
genes (1.2 ± 0.05 × 107 - 2.3 ± 0.27 × 108 copies/mg) were comparable with those reported in a
280
previous study focused on the microbialites described at temperatures between 32 and 65oC in the
281
proximity of the currently investigated microbial mats (27). The copy number of 16S rRNA in the
282
MB samples (2.89 ± 0.1 × 107 – 9.9 ± 0.2 × 108 copies/mg) were similar with the other two
283
locations, and no direct relationship with temperature was observed. As in the case of rDNA, the
284
16S rRNA copy number in the cDNA samples did not show a monotonic relationship with
285
temperature fluctuations in none of the investigated hot springs (Pearson rcDNA = 0.0921; rDNA = -
286
0.211; p > 0.05) (Figure 2). Whereas a linear relationship was reported in literature between the
287
abundance of specific microbial groups and the increase in temperature, this was not the case for the
288
samples within this study. Hence, more effort is needed to examine the effect of temperature on the
289
overall microbial abundance (7).
290
Microbial communities revealed by 16S rRNA gene sequencing
291
Following raw sequences quality filtration, a number of 498,539 sequences were kept in the
292
final dataset that included both the 16S rDNA and 16S cDNA libraries. Each sample was
293
represented by 3,499 to 78,280 high quality sequences that were clustered at a 97% identity
294
threshold in 639 OTUs. Figure 3 summarizes the phyla and class-level composition of samples,
295
showing that overall, up to 13.5% of the sequencing libraries were unassigned, underlining once
296
again the importance of such extreme environments in the discovery of novel microbial groups with
297
unknown features. Archaea were present in all samples except MB65r, being represented mainly by
298
aceticlastic methanogens from the Methanosaeta (31.6%) genus in CI65, together with
299
Methanoculleus (5.3%) and Methanothermobacter (3.2%) in CI35r. Regarding the bacterial
300
taxonomic composition, the groups with the highest abundances were Proteobacteria (13.2 –
301
66.2%), Cyanobacteria (up to 40.3%), Chloroflexi (up to 24%), Bacteroidetes (up to12.8%), and
302
Firmicutes (up to 11.3%). The remaining sequences were widespread affiliated across the Bacteria
12
Downloaded from http://aem.asm.org/ on August 19, 2017 by FUDAN UNIVERSITY
277
303
domain, in the Deinococcus-Thermus (or [Thermi]) (5.7 – 11.2%), Thermatogae (up to 9.03%),
304
Nitrospirae (2.6 – 15.75%), Spirochaetes (up to 5.94%), Verrucomicrobia (up to 4.32%) and
305
Aquificae (up to 10%) phyla. Chiraleu (CH) hot spring microbial mats
307
The CH samples were mainly dominated by Proteobacteria and Cyanobacteria (Figure 3).
308
The abundance of Proteobactaria increased with temperature, an inverse relationship being observed
309
for the oxygenic phototrophic group (Figure 3). While photosynthesis can take place at
310
temperatures up to 75oC (15, 51), distinct taxonomic groups are adapted to perform this function at
311
different concentrations of oxygen, at variable temperatures or light intensities by using various
312
electron acceptors and donors (52). Moderately thermophilic cyanobacteria from the genera
313
Leptolyngbya and Oscillatoria and from the order Nostocales dominated the oxygenic mats at 40
314
and 46oC and are most probably the primary producers (1). In contrast, the Cloroflexus genus
315
became dominant at 53oC (Figure 3, Supplementary Figure 1), members of this group being able to
316
assemble a fully functional photosynthetic apparatus in anoxic conditions. Such anoxic conditions
317
may be caused by the increased water temperature which lowers the solubility of various gases,
318
including oxygen (53). In the absence of O2 or in fluctuating oxic/anoxic circumstances, and in the
319
presence of H2 and H2S, Chloroflexus may grow photoautotrophically by assimilation of CO2
320
through the 3-hydroxypropionate pathway (54). In some cases, Cyanobacteria and Chloroflexi may
321
compete for limited resources (55), but their association was previously documented in thermal
322
conditions, where the oxygenic photosynthesis was dependent on sulfide consumption by
323
Chloroflexus sp. (Portillo et al., 2009; Kim, 1999). As Cyanobacteria possess chlorophyll a (680
324
nm) and Chloroflexi synthesize bacteriochlorophyll c and a (740 nm, 790 nm), they are able to
325
coexist by harvesting different wavelengths of light (52, 56). Thus, it can be assumed that the two
326
groups form distinct assemblages in the mat layers, and both can contribute to the primary
327
production of the ecosystem (52).
328
Even though the members of the Methylophilaceae family (Betaproteobacteria) are usually
13
Downloaded from http://aem.asm.org/ on August 19, 2017 by FUDAN UNIVERSITY
306
restricted to use methanol or methylamine as carbon sources and are common inhabitants of mud,
330
lake and pond water and sludge (57), this group makes almost 8% of the CH46 library. Although
331
not typical for a hot spring environment, members of this family were previously reported in wet
332
sediments collected from the Magadi Lake (Kenya), a hypersaline lake that is supplied by a series
333
of thermal springs with temperatures up to 86oC (58), and in a pond from the Kirishima geothermal
334
area, Japan (Satoh, 2013). Another abundant phylum in the CH microbial mats was Nitrospirae
335
(CH53 – 7%, CH46 – 15.75%), being mainly represented by the Nitrospira genus. It seems that the
336
moderately thermophilic conditions from CH create an environment that is suitable for nitrifying
337
organisms such as Nitrospira, a widespread group that thrives in habitats with temperatures below
338
55oC in Europe, Asia and Australia (13).
339
Ciocaia (CI) hot spring microbial mats
340
Microbial mats developed at lower temperatures (20 and 35oC) near the CI hot spring were
341
represented by almost equal proportions of Cyanobacteria (34.2 – 40.4%) and Proteobacteria (32.2
342
– 40.7%) (Figure 3). Bacteroidetes was the next most abundant group (5.6 – 6.5%) in both samples.
343
While in the sample CI20 Actinobacteria (2.8%) and Chrysiogenetes (2.9) were more abundant, in
344
the CI35 sample higher abundances of Chloroflexi (5.5%) and Firmicutes (4.2%) were detected.
345
Almost all cyanobacterial sequences in these two samples were affiliated to the Oscillatoria genus,
346
and show a decrease in abundance with the increase in water temperature (Supplementary Figure 1).
347
Usually, Oscillatoria dominated microbial mats tend to have a limited distribution, communities
348
from higher temperatures being usually formed by the Pseudoanabaena, Synechococcus and
349
Chloroflexus genera (59). Microbial mats dominated by Oscillatoria spp. were found in the waters
350
from the Chocolate Pots Hot Springs (36 to 45oC) (Yellowstone National Park, USA), rich in iron,
351
bicarbonate and silica (60). In mesothermal and saline environments, nitrogen fixing members of
352
Oscillatoria were found to be the first in colonizing new surfaces, as in the case of the microbial
353
mats from the Mellum island in the North Sea and in the Laguna Figueroa (Pacific Ocean) (61, 62).
354
Also, Oscillatoria boryana formed a predominant population in the mat developed around a sulfide-
14
Downloaded from http://aem.asm.org/ on August 19, 2017 by FUDAN UNIVERSITY
329
rich geothermal spring in Rotorua (New Zealand) (63). Within Proteobacteria, the Alpha - and
356
Gamma - classes prevailed in samples CI20 and CI35. Salinarimonas rosea and Ectothiorhodospira
357
spp. were the most dominant taxa in sample CI20, both of them being formerly encountered in hot
358
spring microbial mats. The first one is a halotolerant, facultatively anaerobic bacterium isolated
359
from a salt mine in Yunnan (China) (64), and was also encountered in a phototrophic microbial mat
360
from Hot Lake, Washington (USA) (65). The second taxon includes purple sulfur bacteria that may
361
form a red layer in microbial mats under the oxic, green layer dominated by Cyanobacteria (61, 62).
362
Like all purple sulfur bacteria, Ectothiorhodospira members are capable of photosynthesis, using
363
hydrogen sulfide or monothioarsenate instead of water as electron donors, oxidizing them to
364
elemental sulfur or arsenate, respectively (66, 67). Because the synthesis of bacteriochlorophyll in
365
these microorganisms is inhibited in the presence of oxygen and under high light intensities, they
366
are found in mats developed in the photic zone of aquatic environments at the oxic/anoxic interface
367
(68). Similar hot spring biofilms composed primarily of Oscillatoria-like and Ectothiorhodospira-
368
like organisms were previously described in Mono Lake, California (USA) (69). Other
369
microorganisms with a high abundance in the sample collected from 35oC were included in the
370
Chloronema genus (5% of the sequencing library) of the Chloroflexi phylum. This is a surprising
371
observation since these anoxygenic filamentous phototrophs usually require low temperatures (4 –
372
15oC) for growth (70). The presence of Chloronema members in CI35, that can tolerate higher
373
temperatures than those previously reported in literature, may indicate the existence of a new taxon
374
within this genus. The Flexibacter genus was also dominant (4.5%), some species (i.e. F. flexis, F.
375
roseolus, F. ruber) being common for hot spring environments (71, 72).
376
The CI65 sample comprised 101 OTUs with highly uneven proportions. The Methanosaeta
377
genus comprised 31.6% of the library. This group, even though is usually encountered in anaerobic
378
sediments and sewage sludge digesters (73), was also detected in thermal settings such as the
379
geothermal waters of the Sungai Klan spring (Malaysia) (42), and in an alkaline hot spring from
380
Iceland (74). While in the other CI samples Gamaproteobacteria were dominant, it seems that they
15
Downloaded from http://aem.asm.org/ on August 19, 2017 by FUDAN UNIVERSITY
355
381
are replaced by Betaproteobacteria at 65oC. The second abundant OTU (18.5%) was affiliated to
382
Hydrogenophilus genus of Betaproteobacteria, a group of thermophilic and aerobic organisms, that
383
can have a facultatively chemolithoautotrophic metabolism, using H2 as electron donor and CO2 as
384
a carbon source (75). Mihai Bravu (MB) hot spring microbial mats
386
The microbial mats from MB were encountered at the highest temperature range, from 53 to
387
65oC. In these microbial mats, Cyanobacteria is present as a major group only at 53oC (5.4%)
388
(Figure 3). It seems that the environmental conditions favor the Chloroflexi phylum that rises from
389
15% in MB53 and MB65, to 24% of the library at 59oC. In contrast, the proportion of
390
Betaproteobacteria (Rhodocyclales) increased with temperature (Supplementary Figure 1), being
391
mostly represented by Hydrogenophilus members. These results imply that the MB mats may be
392
sustained by either anoxigenic photosynthesis, or by chemolithoautotrophy through the oxidation of
393
hydrogen, or by both of these strategies. Compared with the CI and CH communities, some
394
particular and dominant phyla in MB samples are [Thermi] (Deinoccocus - Thermus) (5.7 – 11.2%)
395
and Aquificae (10%). Both of these phyla are a collection of bacteria adapted to harsh and extreme
396
environmental settings (76, 77). Aquificae representatives became abundant only in the MB65
397
sample, as they are strictly thermophiles with an optimum growth temperature above 65oC (78).
398
They were encountered in volcanic and geothermal heated environments, such as the hot ponds in
399
Iceland, the Yellowstone National Park, USA (79), in Tuscany, Italy (80), or in neutral and acidic
400
hot springs from Philippines (81). The vast majority of these organisms are obligate autotrophs,
401
using hydrogen and carbon dioxide as electron and carbon sources, and oxygen as final electron
402
acceptor (82). Moreover, the wide distribution of Aquificae may indicate that these bacteria have an
403
important role in high temperature habitats as primary producers (79, 82).
404
The Meiothermus and Thermus genera within the [Thermi] phylum are dominant members
405
of the MB mats. The different optimal growth temperatures for the two taxa may explain why the
406
Meiothermus proportion is higher in the MB53 and MB57 samples, whereas the Thermus prevails at
16
Downloaded from http://aem.asm.org/ on August 19, 2017 by FUDAN UNIVERSITY
385
65oC. A similar trend, a decrease of Meiothermus abundances with temperature and an increase of
408
Thermus proportion, was observed in microbial mats colonizing an aquifer bore runoff channel
409
from Australia (83). They are commonly encountered genera in hot springs with moderate to high
410
temperatures (50 - 99oC) and slightly alkaline or acidic waters, as those from Kamchatka (84),
411
Yellowstone National Park (85) or Thailand (21). Although these microorganisms do not usually
412
become major components in surface environments, situations similar to MB mats have been
413
encountered in several other springs in Iceland and Tibetan Plateau, China (7, 11).
414
Indicator species in the hot spring microbial mats
415
As previously mentioned, a number of 639 OTUs were observed in the microbial mats
416
collected from the three distinct locations. A multilevel-indicator species analysis (Supplementary
417
Table 2) revealed that only few of them (
1
Differences in temperature and water chemistry shape distinct diversity patterns in
2
thermophilic microbial communities
3 Cecilia M. Chiriaca,b, Edina Szekeresa,b, Knut Rudic, Andreea Baricza,b, Adriana Hegedusa,b, Nicolae
5
Dragoșa,b, Cristian Comana#
6 7
a
8
Biotechnology Department, Faculty of Biology and Geology, Babeş-Bolyai University, Cluj-
9
Napoca, Romania;
10
NIRDBS, Institute of Biological Research, Cluj-Napoca, Romania; bMolecular Biology and
c
Chemistry, Biotechnology and Food Science Department, Norwegian
University of Life Sciences, Aas, Norway.
11 12
Running title: Diversity patterns in hot spring microbial mats.
13 14
# Address correspondence to Cristian Coman, [email protected]
15 16
1
Downloaded from http://aem.asm.org/ on August 19, 2017 by FUDAN UNIVERSITY
4
17
Abstract This study presents the biodiversity and ecology of microbial mats developed in thermal
19
gradients (20-65 oC) in the surroundings of three drillings (CH, CI, MB) tapping a hyperthermal
20
aquifer from Romania. Using a metabarcoding approach, 16S rRNA genes were sequenced from
21
both DNA and RNA transcripts (cDNA) and compared. The relationships between the microbial
22
diversity and the physico-chemical factors were explored. Additionally, the cDNA data was used
23
for in silico functionality predictions, bringing new insights into the functional potential and
24
dynamics of these communities. The results showed that each hot spring determined the formation
25
of distinct microbial communities. In the CH mats (40-53 oC), the abundance of Cyanobacteria
26
decreased
27
Ectothiorhodospira, Oscillatoria and methanogenic archaea dominated the CI communities (20-
28
65 oC), while
29
Chloroflexi, Hydrogenophilus, Thermi and Aquificae. Alpha diversity was negatively correlated
30
with the increase in water temperature, while beta diversity was shaped in each hot spring by the
31
unique combination of physico-chemical parameters, regardless of the type of nucleic acid analyzed
32
(DNA vs. cDNA). The rank correlation analysis revealed a unique model that associated
33
environmental data with community composition, consisting in the combined effect of Na+, K+,
34
HCO3-, PO43- concentrations, together with temperature and electrical conductivity. These factors
35
seem to determine the grouping of samples according to location, rather than with the similarities in
36
thermal regimes, showing that other parameters beside temperature alone are significant drivers of
37
biodiversity.
38
Importance
with
the
temperature,
MB
opposite
microbial
to
that
of
(53-65 oC)
mats
Chloroflexi
were
and
mainly
Proteobacteria.
composed
of
39
Hot spring microbial mats are remarkable manifestation of life on Earth, and were
40
intensively studied for decades. Moreover, as hot spring areas are isolated and have a limited
41
exchange of organisms, nutrients and energy with the surrounding environments, hot spring
42
microbial communities can be used as model studies to elucidate the colonizing potential within
2
Downloaded from http://aem.asm.org/ on August 19, 2017 by FUDAN UNIVERSITY
18
extreme settings. Thus, they are of great importance in evolutionary biology, microbial ecology and
44
exobiology. With all the efforts that have been made, the current understanding of the influence of
45
temperature and water chemistry on the microbial community composition, diversity and abundance
46
in microbial mats is limited. In this study, the composition and diversity of microbial communities
47
developed in thermal gradients in the vicinity of three hot springs from Romania were investigated,
48
each geothermal water having particular physico-chemical characteristics. Our results expose new
49
factors that could determine the formation of these ecosystems, expanding the current knowledge in
50
this regard.
51 52
Introduction
53
Microbial mats are stratified communities of functional groups that are determined and
54
maintained by steep physico-chemical gradients (1). Hot spring microbial mats are of particular
55
interest as they are considered model ecosystems for the study of early life on Earth. By now, the
56
oldest evidence of life on land came from the ca. 3.48 billion year old Archaean rocks from Pilbara
57
Craton, Western Australia. These structures include stromatolites, or mineralized microbial mats,
58
that have developed in an ancient, hot spring environment (2). Modern hot spring habitats,
59
including various microbial mat ecosystems have been extensively studied, especially in the
60
Yellowstone National Park, USA (3 - 6), in the Tibetan Plateau, China (7 - 9), in Iceland (10, 11),
61
and Russia (12, 13). These studies have considerably improved our understanding of microbial
62
composition and diversity in terrestrial hot spring ecosystems (7). Thus, it is known that these
63
microbial mats lack seasonality and undergo changes in response to differences in physico-chemical
64
parameters, such as temperature, pH, sulfide concentrations etc. (1). These environments are
65
dominated by prokaryotes, as extreme temperatures usually exclude eukaryotic organisms (14). In
66
hyperthermal conditions that are not suitable for photosynthesis (>75oC) (15), thermal and
67
hyperthermal members of Bacteria are common inhabitants of microbial mats, being predominantly
68
affiliated to the Aquificae, Thermatogae, Deinococcus-Thermus ([Thermi]) phyla, or to certain
3
Downloaded from http://aem.asm.org/ on August 19, 2017 by FUDAN UNIVERSITY
43
thermophilic members of Proteobacteria (7). At lower temperatures and under light conditions,
70
oxygenic and anoxygenic photosynthesis usually support the primary production in the mat,
71
allowing a diversification of the microbial communities (16). Photosynthetic hot spring microbial
72
mats may be sustained by Cyanobacteria, Chloroflexi or Chlorobi, or by a combination of these
73
groups (17). Beside the bacterial domain, Archaea is also present in hot spring mats, being
74
represented especially by the Crenarchaeota and Euryarchaeota phyla (18, 19).
75
To better understand the ecology and microbial functions leading to the establishment of
76
microbial mats, a number of studies have focused on the links between the physico-chemical
77
parameters of the water and the microbial composition and diversity (7, 11, 16, 20 - 22). Within
78
those factors, temperature received much of the attention. While some studies explore the
79
temperature effect on small scale gradients (16, 20), others have selected springs within a wide
80
temperature range (11, 23). The results showed that the microbial mat composition and the
81
complexity of the ecosystem seem to be mainly determined by the rise in temperature (1). In the
82
study performed by Wang et al. (7), mat samples were collected from thermal gradients in different
83
areas and no location-specific variations were observed. Instead, the temperature was sufficient to
84
explain the variability in community composition. In contrast, other studies found that the sulphide
85
concentration (21) or the pH values (24) were important factors that control the similarity among
86
samples. Nevertheless, more effort is required to better understand the effects of physico-chemical
87
conditions on the development of hot spring microbial ecosystems.
88
As a result of geothermal resource exploitation in the second half of the 20th century, eight
89
geothermal areas were identified in Romania, with over 200 drillings ranging in depth from 800 to
90
3500 m, having the outflow water temperatures from 40oC to 120oC (25). Among these, a number
91
of drillings (80) are tapping the Pannonian hyperthermal aquifer (50 – 85oC) from the Western Plain
92
of Romania, the water that comes from 800 to 2100 m deep being used mostly for domestic heating,
93
health and recreational bathing, industrial processes or fish farming (25). Some wells have been
94
abandoned over the years and the continuous hot water flow has led to the development of specific
4
Downloaded from http://aem.asm.org/ on August 19, 2017 by FUDAN UNIVERSITY
69
mineralized communities and other photosynthetic microbial mats (26, 27). Even though the
96
biodiversity in some of these microbial mats has been previously described using 16S rDNA clone
97
libraries (26), the current study applied high-throughput sequencing techniques for a more thorough
98
investigation. It focuses on microbial mats formed along thermal gradients in hot springs from three
99
different locations, each with particular physico-chemical characteristics. An integrative approach
100
was performed that included both 16S rRNA transcripts and 16S rDNA gene sequencing. Although
101
the amount of ribosomal RNA that is detected in a microorganism is no longer viewed as a clear
102
indicator for microbial growth, a large number of copies does indicate the potential for protein
103
synthesis, for a rapid shift in metabolic function, or for an overall higher fitness of that species in
104
the ecosystem (28). The inferences of metabolic potential based on 16S rRNA transcripts may fall
105
under some degree of uncertainty (29). But, taking into consideration that extracellular RNA is less
106
stable in the environment compared with free DNA (30), functional roles can be attributed in this
107
way to intact organisms. By contrast, the DNA extracted from environmental samples is a
108
continuum of DNA derived from whole living organisms and extraorganismal, free DNA (31). A
109
more precise view on the microbial community composition, its diversity and its function may be
110
proposed by comparing the 16S rDNA gene libraries with those of 16S rRNA transcripts.
111
Thus, the aim of this research was to characterize the prokaryotic composition and
112
biodiversity in the newly investigated hot springs microbial mats associated with three hot springs
113
situated in the Western Plain of Romania and to explore their relationship with the physico-
114
chemical properties of the thermal water. As these microbial communities may bring new
115
information on the distribution of microbial groups in thermal environments, an additional effort
116
was made to compare their microbial composition with other microbial mats that have been
117
previously investigated through a metabarcoding approach.
118
Materials and Methods
119
Sampling procedure
120
Samples were taken in 2013 from microbial mats developed nearby three abandoned
5
Downloaded from http://aem.asm.org/ on August 19, 2017 by FUDAN UNIVERSITY
95
drillings from the Chiraleu (CH – drilling number 4095), Ciocaia (CI - drilling number 4045) and
122
Mihai Bravu (MB - drilling number 605) villages, at temperatures ranging from 20 to 65 oC (Table
123
1). Colorful microbial mats were observed and sampled at each location along with the temperature
124
gradient (Figure 1). In the surroundings of the Chiraleu drilling (Figure 1A), green and slimy
125
microbial mat developed at 53°C, without a clear vertical stratification (Figure 1B), as well as a
126
yellow microbial mat at 46°C which presented a typical laminated structure (Figure 1C). A thin
127
microbial mat was sampled from 40°C, with a smooth orange surface and green layers underneath
128
(Figure 1D). Three distinct microbial mats were observed in the proximity of the Ciocaia drilling
129
(Figure 1E), including green, mucilaginous and stratified microbial mats forming at 20°C (Figure
130
1F), red and slimy microbial mats developing at 35°C that do not show a clear stratification (Figure
131
1G) and white microbial mats at 65°C, more consolidated and firm near to the substrate and with a
132
mucilaginous, translucent surface layer (Figure 1H). Microbial mats and streamers were identified
133
near the Mihai Bravu drilling opening (Figure 1I), including a green, compact and smooth microbial
134
mat forming at 53°C, with thin layers (Figure 1J), a stratified microbial mat at 59°C, with an orange
135
top layer and green layers underneath (Figure 1K) and long, flexible structures forming cream -
136
colored streamers at 65°C along with the flowing hot water (Figure 1L).
137
The stable characteristics of the water sources were previously reported (32, 33) (Table 2).
138
These three drilling (CH, CI, MB) are tapping the Pannonian hyperthermal aquifer, a deposit of
139
neutral to slightly alkaline waters, with pH values ranging from 7.75 to 7.91. The high
140
concentrations of Na + (676 - 4120 mg/L), Cl- (537 - 1160 mg/L) and HCO3- (895 – 1980 mg/L) ions
141
are probably the result of the static nature of this deposit, which resulted in prolonged water - rock
142
interactions. The collecting rocks of the aquifer were formed in lagoon conditions that allowed the
143
accumulation of dissolvable chloride rocks, which ultimately impact the chemical composition of
144
the thermal deposit (32, 33). The temperature at the surface and at the bottom of each microbial mat
145
was determined using a field multiparameter (WTW Multi 340i Multiparameter, Weilheim,
146
Germany). As temperature was similar (±0.1 oC) between top and bottom (data not shown), only the
6
Downloaded from http://aem.asm.org/ on August 19, 2017 by FUDAN UNIVERSITY
121
surface values were reported. For DNA and RNA extraction, approx. 3 g of each microbial mat was
148
collected in 15 mL Falcon tubes in triplicate and TRIzolTM Reagent (Invitrogen) was immediately
149
added to fill up the tubes in order to maintain the integrity of the total RNA. All the samples were
150
transported on ice to the laboratory, and processed within the same day. After DNA and RNA
151
extraction, followed by cDNA synthesis, a total of 18 samples were used for prokaryotic cell
152
abundance estimation and metabarcoding (Table 1).
153
Nucleic acids extraction
154
Total DNA and RNA were extracted in triplicate for each microbial sample according to the
155
manufacturer’s instructions (TRIzolTM Reagent, Invitrogen) with slight modifications. 750 mg of
156
each sample (wet weight) were incubated at 60oC for 10 min, being vortexed prior to and at the end
157
of the incubation time. Then, 150 μL of chloroform were added and the tubes were vortexed for 10
158
s. The tubes were centrifuged at 10,000 × g for 10 min. The mixture was separated in three phases,
159
an upper aqueous phase, an inter-phase and a lower phenol-chloroform phase. The upper phase was
160
transferred in a new tube for RNA extraction, while the other two phases were used for DNA
161
purification. For DNA extraction, 300 μL of 100% ethanol were added to the lower phases, and the
162
components were gently mixed by inverting the tubes. Following a 3 min incubation period, the
163
tubes were centrifuged for 15 min at 12000 × g and 4°C to pellet the DNA. The pellet was
164
resuspended in1 mL of 0.1 M sodium citrate in 10% ethanol, pH 8.5, and was incubated at room
165
temperature for 30 min. The tubes were centrifuged for 15 min at 12000 × g at 4°C. This washing
166
step was repeated one more time, and the DNA pellets were resuspended in 1.5 mL of 75% ethanol.
167
The tubes were incubated for another 20 min at room temperature, and then were centrifuged for 15
168
min at 12000 × g and 4°C. The supernatant was discarded and the DNA was air dried for 15 min.
169
The pellet was resuspended in 500 μL of 8 mM NaOH by pipetting up and down. Then, the tubes
170
were centrifuged for 10 min at 12,000 × g and 4°C, the supernatant was removed, and the DNA
171
pellet was eluted in 35 µl PCR Grade water. The DNA replicates of the same sample were pooled,
172
and the DNA concentration was determined based on the Qubit® dsDNA BR Assay Kit, using a
7
Downloaded from http://aem.asm.org/ on August 19, 2017 by FUDAN UNIVERSITY
147
173
Qubit® Fluorometer (data not shown).
174
RNA extraction and cDNA synthesis For RNA purification, an equal amount of phenol – chloroform – isoamyl alcohol (25:24:1)
176
was added to the aqueous supernatant (see previous section) and the tubes were vortexed and
177
centrifuged at 10,000 x g for 2 min. The upper phase was carefully transferred to a new tube and
178
mixed with an equal amount of isopropyl alcohol. The samples were incubated at room temperature
179
for 10 min and then centrifuged at 4oC and 10,000 x g for 10 min. The supernatant was removed,
180
and 1 mL of 70% ethanol was added to the tubes placed on ice. After 5 min of incubation the tubes
181
were centrifuged at 4oC and 8,000 x g for 5 min. The ethanol was discarded and the RNA was
182
resuspended in 30 μL of PCR Grade water. Next, the DNA was removed using the TURBO DNA-
183
freeTM kit (ThermoFisher Scientific) and the RNA replicates of the same sample were pooled. The
184
RNA concentration was measured using the Qubit® Fluorometer and the Qubit® RNA BR Assay
185
Kit (data not shown). Total RNA reverse transcription was performed starting with 200 ng of
186
purified RNA for each sample using the First Strand cDNA Synthesis kit (Thermo Scientific)
187
according to the manufacturer’s instructions.
188
Quantification of prokaryotic cell abundance by qPCR
189
The PRK341F/PRK806R primers were used to target the 16S rRNA gene in the total
190
extracted DNA and cDNA by quantitative real time PCR (qPCR) (34). The reactions were
191
performed in triplicate using the SsoFast Eva Green Supermix (Bio-Rad, Hercules, CA, USA) on
192
the CFX96 Touch™ Real-Time PCR Detection System (Bio-Rad). The following components were
193
added in the reaction mix: 7 μL 1X Sso Fast EvaGreen SuperMix (Bio-Rad), 0.4 μM of the forward
194
and reverse primers, 20 ng of DNA/RNA and RNase/DNase-free water to a final volume of 14 μL.
195
The reaction program began with an initial denaturation at 98°C for 120 s, followed by 45 cycles of
196
10 s at 98°C and an annealing/elongation step of 30 s at 55°C. The gene copy number was
197
calculated by comparing the amplification results to a standard ten-fold serial dilution of known
198
quantities of recombinant plasmids (109 – 103) carrying the targeted gene. The plasmids were
8
Downloaded from http://aem.asm.org/ on August 19, 2017 by FUDAN UNIVERSITY
175
199
prepared as previously described (35). Additionally, three negative controls were analyzed in the
200
same run. A Ct cutoff value corresponding to the lowest Ct reported for the negative controls was
201
used, and the average number of gene copies detected in these controls were subtracted from the
202
sample values.
203
Preparation of sequencing libraries The V3-V4 regions of the 16S rRNA gene were amplified from the extracted DNA and
205
cDNA using the PRK341F/PRK806R primers (34) modified by addition of Illumina-specific
206
adaptors. The coverage of the primers against the SILVA reference database was 84.1% for
207
Bacteria and 58.5% for Archaea. Each PCR reaction (25 μL) contained 1X HOT FIREPol® PCR
208
mix (Solis BioDyne, Estonia), 200 nM uniquely tagged forward and reverse primers, 1 μL of
209
sample DNA and 18 μL water. The reaction conditions were: 95°C for 15 min, followed by 25
210
cycles of 95°C for 30s, 50°C for 30s, 72°C for 45s, ending with 72°C for 7 min. The amplicon
211
concentration was measured with the Qubit®dsDNA HS Assay Kit using the Qubit® Fluorometer
212
(Thermo Fisher Scientific, USA) and equal amounts of amplicons were pooled for each sample into
213
a normalized library. The pooled library concentration was measured using the PerfeCta® NGS
214
Quantification Kit for Illumina (Quanta BioSciences, USA) and the library was diluted in Tris pH
215
8.5 to a final 4 nM concentration. Sequencing was performed on a MiSeq platform (Illumina, USA)
216
using V3 sequencing chemistry with 300 bp paired-end reads.
217
Amplicon analysis
218
Raw sequence data (SRA accession number PRJNA389363) was processed and quality
219
filtered through a combination of Usearch v8 and QIIME pipelines (36, 37). Briefly, QIIME was
220
used to extract the barcodes from the sequence data, to join the forward and reverse Illumina reads
221
and to demultiplex the sequence data. Singletons removal and quality control filtering were
222
performed using the Usearch v8 pipeline, by discarding sequences with less than 350 nucleotides
223
and those with more than 0.2 total expected errors (-fastq_maxee E option of the fastq_filter
224
command). Remaining quality control filtration included de novo and reference chimera removal
9
Downloaded from http://aem.asm.org/ on August 19, 2017 by FUDAN UNIVERSITY
204
via the Usearch v8 pipeline, using the latest version of the Greengenes database ('13_8') as a
226
reference (38). The resulting Operational Taxonomic Unit (OTU) table was converted into the
227
Biological Observational Matrix (BIOM) format (39). Taxonomy was assigned using the default
228
classifier in QIIME (Greengenes) against the updated '13_8' version of the Greengenes database at a
229
97% similarity threshold. Next, the mitochondrial and plastidial sequences were filtered out of the
230
OTU table. A multiple sequence alignment was generated with PyNAST (36) in QIIME, and the
231
alignment was used to build a phylogeny with the FastTree algorithm (40).
232
Additionally, a comparison of the alpha and beta-diversity encountered in the Romanian
233
samples to other sequences from hot spring microbial mats deposited in the Sequence Read Archive
234
(SRA) was performed, including the following entries: SRA061437 (7), SRS932137, SRS932073
235
(41), PRJEB7059 (42). As in these studies different variable regions of the 16S rRNA gene were
236
amplified and sequenced, a closed-reference OTU picking was performed in QIIME (36) using a
237
97% similarity threshold. Prior to this step, sequences were first quality filtered as described in their
238
original studies. The final data set retained only those sequences that match a full-length
239
representative sequence within the last version ('13_8') of the Greengenes database.
240
Statistical analysis
241
Rarefaction was performed at a depth of 3499 sequences per sample, followed by alpha- and
242
beta-diversity estimation in QIIME (36). Euclidean distances for physico-chemical parameters were
243
computed with the vegdist function of the 'vegan' package in R, and the Mantel test (999
244
permutations) was run to evaluate the environmental impact on the microbial diversity using the
245
weighted Unifrac distances (43). Environmental physico-chemical data was log transformed prior to
246
the analysis. PCoA plots were generated for the weighted and unweighted Unifrac matrices, using
247
the cmdscale and envfit functions of the 'stats' and 'vegan' packages in R. The environmental factors
248
were fitted onto the ordination with the envfit function of the 'vegan' package in R, being scaled by
249
their correlation to the distance matrix. The significance of the sample grouping according to
250
distinct sampling sites or distinct type of investigated nucleic acids were tested by the analysis of
10
Downloaded from http://aem.asm.org/ on August 19, 2017 by FUDAN UNIVERSITY
225
251
similarity (ANOSIM) using 999 permutations (44) in QIIME. Phyla and classes with abundances
252
higher than 1% in the sequencing libraries were plotted using the inkspot function of the 'rioja'
253
package in R. The multilevel indicator OTUs (45) for each thermal spring were established using
254
the multipatt function of the 'perm' package in R.
255
Functional prediction For functionality prediction with PICRUSt (46), the OTUs were picked using an open
257
reference approach, and de-novo OTUs were removed, keeping in the final OTU-table only the
258
OTUs that had matching Greengenes IDs ('13_8'). Data in the BIOM OTU-table was normalized
259
using the 16S rRNA gene copy number for each OTU, and the normalized table was used for
260
KEGG functional orthologs predictions. In order to evaluate the PICRUSt predictions accuracy for
261
each sample, the weighted Nearest Sequenced Taxon Index (NSTI) was computed, an index that
262
describes the extent to which microorganisms from each sample are related to fully annotated
263
genomes, a value of 0.03 meaning that on average, microorganisms in the sample were predicted
264
using a relative from the same species.
265
Results and discussion
266
Abundance of prokaryotic 16S rRNA genes and transcripts
267
The abundance of prokaryotic microorganisms in the DNA and cDNA samples was
268
estimated through qPCR by targeting the 16S rRNA gene. Generally, the copy numbers in the
269
cDNA transcripts were found to be two to three orders of magnitude lower than the numbers
270
observed in the DNA sample (Figure 2). This difference may suggest that evaluation of cell
271
abundances should not be based solely on the 16S rRNA genes. Some studies proposed that up to
272
90% of the free DNA is stable in sediments and may not be affected by nucleases (47, 48). Thus,
273
the total extracted DNA most probably reflects not only the living microbes present in the
274
environment, but also the extracellular DNAs (49). Considering the instability of the free RNA, it
275
seems that it could provide a better estimation of the actual abundance of microbial cells that can
276
synthesize proteins (1, 50). In the CH samples, the number of 16S rDNA copies decreased with
11
Downloaded from http://aem.asm.org/ on August 19, 2017 by FUDAN UNIVERSITY
256
temperature from 1.1 ± 0.01 × 108 at 40oC to 3.4 ± 0.6 × 106 copies/mg at 53oC, whereas in the other
278
sampling sites no clear pattern was observed. In the CI DNA samples, the abundance of 16S rDNA
279
genes (1.2 ± 0.05 × 107 - 2.3 ± 0.27 × 108 copies/mg) were comparable with those reported in a
280
previous study focused on the microbialites described at temperatures between 32 and 65oC in the
281
proximity of the currently investigated microbial mats (27). The copy number of 16S rRNA in the
282
MB samples (2.89 ± 0.1 × 107 – 9.9 ± 0.2 × 108 copies/mg) were similar with the other two
283
locations, and no direct relationship with temperature was observed. As in the case of rDNA, the
284
16S rRNA copy number in the cDNA samples did not show a monotonic relationship with
285
temperature fluctuations in none of the investigated hot springs (Pearson rcDNA = 0.0921; rDNA = -
286
0.211; p > 0.05) (Figure 2). Whereas a linear relationship was reported in literature between the
287
abundance of specific microbial groups and the increase in temperature, this was not the case for the
288
samples within this study. Hence, more effort is needed to examine the effect of temperature on the
289
overall microbial abundance (7).
290
Microbial communities revealed by 16S rRNA gene sequencing
291
Following raw sequences quality filtration, a number of 498,539 sequences were kept in the
292
final dataset that included both the 16S rDNA and 16S cDNA libraries. Each sample was
293
represented by 3,499 to 78,280 high quality sequences that were clustered at a 97% identity
294
threshold in 639 OTUs. Figure 3 summarizes the phyla and class-level composition of samples,
295
showing that overall, up to 13.5% of the sequencing libraries were unassigned, underlining once
296
again the importance of such extreme environments in the discovery of novel microbial groups with
297
unknown features. Archaea were present in all samples except MB65r, being represented mainly by
298
aceticlastic methanogens from the Methanosaeta (31.6%) genus in CI65, together with
299
Methanoculleus (5.3%) and Methanothermobacter (3.2%) in CI35r. Regarding the bacterial
300
taxonomic composition, the groups with the highest abundances were Proteobacteria (13.2 –
301
66.2%), Cyanobacteria (up to 40.3%), Chloroflexi (up to 24%), Bacteroidetes (up to12.8%), and
302
Firmicutes (up to 11.3%). The remaining sequences were widespread affiliated across the Bacteria
12
Downloaded from http://aem.asm.org/ on August 19, 2017 by FUDAN UNIVERSITY
277
303
domain, in the Deinococcus-Thermus (or [Thermi]) (5.7 – 11.2%), Thermatogae (up to 9.03%),
304
Nitrospirae (2.6 – 15.75%), Spirochaetes (up to 5.94%), Verrucomicrobia (up to 4.32%) and
305
Aquificae (up to 10%) phyla. Chiraleu (CH) hot spring microbial mats
307
The CH samples were mainly dominated by Proteobacteria and Cyanobacteria (Figure 3).
308
The abundance of Proteobactaria increased with temperature, an inverse relationship being observed
309
for the oxygenic phototrophic group (Figure 3). While photosynthesis can take place at
310
temperatures up to 75oC (15, 51), distinct taxonomic groups are adapted to perform this function at
311
different concentrations of oxygen, at variable temperatures or light intensities by using various
312
electron acceptors and donors (52). Moderately thermophilic cyanobacteria from the genera
313
Leptolyngbya and Oscillatoria and from the order Nostocales dominated the oxygenic mats at 40
314
and 46oC and are most probably the primary producers (1). In contrast, the Cloroflexus genus
315
became dominant at 53oC (Figure 3, Supplementary Figure 1), members of this group being able to
316
assemble a fully functional photosynthetic apparatus in anoxic conditions. Such anoxic conditions
317
may be caused by the increased water temperature which lowers the solubility of various gases,
318
including oxygen (53). In the absence of O2 or in fluctuating oxic/anoxic circumstances, and in the
319
presence of H2 and H2S, Chloroflexus may grow photoautotrophically by assimilation of CO2
320
through the 3-hydroxypropionate pathway (54). In some cases, Cyanobacteria and Chloroflexi may
321
compete for limited resources (55), but their association was previously documented in thermal
322
conditions, where the oxygenic photosynthesis was dependent on sulfide consumption by
323
Chloroflexus sp. (Portillo et al., 2009; Kim, 1999). As Cyanobacteria possess chlorophyll a (680
324
nm) and Chloroflexi synthesize bacteriochlorophyll c and a (740 nm, 790 nm), they are able to
325
coexist by harvesting different wavelengths of light (52, 56). Thus, it can be assumed that the two
326
groups form distinct assemblages in the mat layers, and both can contribute to the primary
327
production of the ecosystem (52).
328
Even though the members of the Methylophilaceae family (Betaproteobacteria) are usually
13
Downloaded from http://aem.asm.org/ on August 19, 2017 by FUDAN UNIVERSITY
306
restricted to use methanol or methylamine as carbon sources and are common inhabitants of mud,
330
lake and pond water and sludge (57), this group makes almost 8% of the CH46 library. Although
331
not typical for a hot spring environment, members of this family were previously reported in wet
332
sediments collected from the Magadi Lake (Kenya), a hypersaline lake that is supplied by a series
333
of thermal springs with temperatures up to 86oC (58), and in a pond from the Kirishima geothermal
334
area, Japan (Satoh, 2013). Another abundant phylum in the CH microbial mats was Nitrospirae
335
(CH53 – 7%, CH46 – 15.75%), being mainly represented by the Nitrospira genus. It seems that the
336
moderately thermophilic conditions from CH create an environment that is suitable for nitrifying
337
organisms such as Nitrospira, a widespread group that thrives in habitats with temperatures below
338
55oC in Europe, Asia and Australia (13).
339
Ciocaia (CI) hot spring microbial mats
340
Microbial mats developed at lower temperatures (20 and 35oC) near the CI hot spring were
341
represented by almost equal proportions of Cyanobacteria (34.2 – 40.4%) and Proteobacteria (32.2
342
– 40.7%) (Figure 3). Bacteroidetes was the next most abundant group (5.6 – 6.5%) in both samples.
343
While in the sample CI20 Actinobacteria (2.8%) and Chrysiogenetes (2.9) were more abundant, in
344
the CI35 sample higher abundances of Chloroflexi (5.5%) and Firmicutes (4.2%) were detected.
345
Almost all cyanobacterial sequences in these two samples were affiliated to the Oscillatoria genus,
346
and show a decrease in abundance with the increase in water temperature (Supplementary Figure 1).
347
Usually, Oscillatoria dominated microbial mats tend to have a limited distribution, communities
348
from higher temperatures being usually formed by the Pseudoanabaena, Synechococcus and
349
Chloroflexus genera (59). Microbial mats dominated by Oscillatoria spp. were found in the waters
350
from the Chocolate Pots Hot Springs (36 to 45oC) (Yellowstone National Park, USA), rich in iron,
351
bicarbonate and silica (60). In mesothermal and saline environments, nitrogen fixing members of
352
Oscillatoria were found to be the first in colonizing new surfaces, as in the case of the microbial
353
mats from the Mellum island in the North Sea and in the Laguna Figueroa (Pacific Ocean) (61, 62).
354
Also, Oscillatoria boryana formed a predominant population in the mat developed around a sulfide-
14
Downloaded from http://aem.asm.org/ on August 19, 2017 by FUDAN UNIVERSITY
329
rich geothermal spring in Rotorua (New Zealand) (63). Within Proteobacteria, the Alpha - and
356
Gamma - classes prevailed in samples CI20 and CI35. Salinarimonas rosea and Ectothiorhodospira
357
spp. were the most dominant taxa in sample CI20, both of them being formerly encountered in hot
358
spring microbial mats. The first one is a halotolerant, facultatively anaerobic bacterium isolated
359
from a salt mine in Yunnan (China) (64), and was also encountered in a phototrophic microbial mat
360
from Hot Lake, Washington (USA) (65). The second taxon includes purple sulfur bacteria that may
361
form a red layer in microbial mats under the oxic, green layer dominated by Cyanobacteria (61, 62).
362
Like all purple sulfur bacteria, Ectothiorhodospira members are capable of photosynthesis, using
363
hydrogen sulfide or monothioarsenate instead of water as electron donors, oxidizing them to
364
elemental sulfur or arsenate, respectively (66, 67). Because the synthesis of bacteriochlorophyll in
365
these microorganisms is inhibited in the presence of oxygen and under high light intensities, they
366
are found in mats developed in the photic zone of aquatic environments at the oxic/anoxic interface
367
(68). Similar hot spring biofilms composed primarily of Oscillatoria-like and Ectothiorhodospira-
368
like organisms were previously described in Mono Lake, California (USA) (69). Other
369
microorganisms with a high abundance in the sample collected from 35oC were included in the
370
Chloronema genus (5% of the sequencing library) of the Chloroflexi phylum. This is a surprising
371
observation since these anoxygenic filamentous phototrophs usually require low temperatures (4 –
372
15oC) for growth (70). The presence of Chloronema members in CI35, that can tolerate higher
373
temperatures than those previously reported in literature, may indicate the existence of a new taxon
374
within this genus. The Flexibacter genus was also dominant (4.5%), some species (i.e. F. flexis, F.
375
roseolus, F. ruber) being common for hot spring environments (71, 72).
376
The CI65 sample comprised 101 OTUs with highly uneven proportions. The Methanosaeta
377
genus comprised 31.6% of the library. This group, even though is usually encountered in anaerobic
378
sediments and sewage sludge digesters (73), was also detected in thermal settings such as the
379
geothermal waters of the Sungai Klan spring (Malaysia) (42), and in an alkaline hot spring from
380
Iceland (74). While in the other CI samples Gamaproteobacteria were dominant, it seems that they
15
Downloaded from http://aem.asm.org/ on August 19, 2017 by FUDAN UNIVERSITY
355
381
are replaced by Betaproteobacteria at 65oC. The second abundant OTU (18.5%) was affiliated to
382
Hydrogenophilus genus of Betaproteobacteria, a group of thermophilic and aerobic organisms, that
383
can have a facultatively chemolithoautotrophic metabolism, using H2 as electron donor and CO2 as
384
a carbon source (75). Mihai Bravu (MB) hot spring microbial mats
386
The microbial mats from MB were encountered at the highest temperature range, from 53 to
387
65oC. In these microbial mats, Cyanobacteria is present as a major group only at 53oC (5.4%)
388
(Figure 3). It seems that the environmental conditions favor the Chloroflexi phylum that rises from
389
15% in MB53 and MB65, to 24% of the library at 59oC. In contrast, the proportion of
390
Betaproteobacteria (Rhodocyclales) increased with temperature (Supplementary Figure 1), being
391
mostly represented by Hydrogenophilus members. These results imply that the MB mats may be
392
sustained by either anoxigenic photosynthesis, or by chemolithoautotrophy through the oxidation of
393
hydrogen, or by both of these strategies. Compared with the CI and CH communities, some
394
particular and dominant phyla in MB samples are [Thermi] (Deinoccocus - Thermus) (5.7 – 11.2%)
395
and Aquificae (10%). Both of these phyla are a collection of bacteria adapted to harsh and extreme
396
environmental settings (76, 77). Aquificae representatives became abundant only in the MB65
397
sample, as they are strictly thermophiles with an optimum growth temperature above 65oC (78).
398
They were encountered in volcanic and geothermal heated environments, such as the hot ponds in
399
Iceland, the Yellowstone National Park, USA (79), in Tuscany, Italy (80), or in neutral and acidic
400
hot springs from Philippines (81). The vast majority of these organisms are obligate autotrophs,
401
using hydrogen and carbon dioxide as electron and carbon sources, and oxygen as final electron
402
acceptor (82). Moreover, the wide distribution of Aquificae may indicate that these bacteria have an
403
important role in high temperature habitats as primary producers (79, 82).
404
The Meiothermus and Thermus genera within the [Thermi] phylum are dominant members
405
of the MB mats. The different optimal growth temperatures for the two taxa may explain why the
406
Meiothermus proportion is higher in the MB53 and MB57 samples, whereas the Thermus prevails at
16
Downloaded from http://aem.asm.org/ on August 19, 2017 by FUDAN UNIVERSITY
385
65oC. A similar trend, a decrease of Meiothermus abundances with temperature and an increase of
408
Thermus proportion, was observed in microbial mats colonizing an aquifer bore runoff channel
409
from Australia (83). They are commonly encountered genera in hot springs with moderate to high
410
temperatures (50 - 99oC) and slightly alkaline or acidic waters, as those from Kamchatka (84),
411
Yellowstone National Park (85) or Thailand (21). Although these microorganisms do not usually
412
become major components in surface environments, situations similar to MB mats have been
413
encountered in several other springs in Iceland and Tibetan Plateau, China (7, 11).
414
Indicator species in the hot spring microbial mats
415
As previously mentioned, a number of 639 OTUs were observed in the microbial mats
416
collected from the three distinct locations. A multilevel-indicator species analysis (Supplementary
417
Table 2) revealed that only few of them (
