Subscriber access provided by University of Pennsylvania Libraries
Environmental Processes
Altered Microbiome Leads to Significant Phenotypic and Transcriptomic Differences in a Lipid Accumulating Chlorophyte Lubna Richter, Cresten B. Mansfeldt, Michael M. Kuan, Alexandra Cesare, Stephen T. Menefee, Ruth E. Richardson, and Beth A Ahner Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.7b06581 • Publication Date (Web): 11 May 2018 Downloaded from http://pubs.acs.org on May 13, 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 30
Environmental Science & Technology
70x39mm (300 x 300 DPI)
ACS Paragon Plus Environment
Environmental Science & Technology
1 2 3 4 5 6 7
Altered Microbiome Leads to Significant Phenotypic and Transcriptomic Differences in a Lipid Accumulating Chlorophyte
8
T. Menefee†, Ruth E. Richardson‡, and Beth A. Ahner†*
Lubna V. Richter†§, Cresten B. Mansfeldt‡§, Michael M. Kuan†, Alexandra E. Cesare†, Stephen
9 10 11 12 13 14 15
†
Department of Biological and Environmental Engineering, Cornell University, Ithaca, NY, USA
‡
School of Civil and Environmental Engineering, Cornell University, Ithaca, NY, USA
16 17
§
Authors contributed equally to this work
18 19 20 21 22 23 24
*Correspondence to: B. A. Ahner Tel: 607- 255- 4677; Fax: 607- 255- 3679 Email:
[email protected] 25 26 27 28
Keywords: Microalgae, Bacteria, Symbiosis, 16S rRNA, Lipid, Chlorella, Transcriptome
ACS Paragon Plus Environment
Page 2 of 30
Page 3 of 30
29
Environmental Science & Technology
ABSTRACT
30
Given the challenges facing the economically favorable production of products from
31
microalgae, understanding factors that might impact productivity rates including growth rates
32
and accumulation of desired products, e.g., triacylglycerols (TAG) for biodiesel feedstock,
33
remains critical. Although operational parameters such as media composition and reactor design
34
can clearly effect growth rates, the role of microbe-microbe interactions is just beginning to be
35
elucidated. In this study an oleaginous marine algae Chlorella spp. C596 culture is shown to be
36
better described as a microbial community. Perturbations to this microbial community showed a
37
significant impact on phenotypes including sustained differences in growth rate and TAG
38
accumulation of 2.4 and 2.5 fold, respectively. Characterization of the associated community
39
using Illumina 16S ribosomal RNA amplicon and random shotgun transcriptomic analyses
40
showed that the fast growth rate correlated with two specific bacterial species (Ruegeria and
41
Rhodobacter spp). The transcriptomic response of the Chlorella species revealed that the slower
42
growing algal consortium C596-S1 upregulated genes associated with photosynthesis and
43
resource scavenging and decreased the expression of genes associated with transcription and
44
translation relative to the initial C596-R1. Our studies advance the appreciation of the effects
45
microbiomes can have on algal growth in bioreactors and suggest that symbiotic interactions are
46
involved in a range of critical processes including nitrogen, carbon cycling and oxidative stress.
47 48 49 50 51 2 ACS Paragon Plus Environment
Environmental Science & Technology
52 53
Page 4 of 30
INTRODUCTION Select
species
of microalgae biosynthesize and
accumulate high
levels
of
54
triacylglycerides (TAG) as a means of energy storage in response to particular environmental
55
stimuli such as nutrient limitation.1 In particular, members of the green algae Chlorella genus
56
have been reported to bioaccumulate high levels of TAG and have therefore been studied for
57
biodiesel production.2-4 Use of such algae species for biofuel production is being explored in
58
part because they can be cultivated on non-arable land using non-potable or saline water
59
resources with high areal productivity.
60
production of biodiesel from algae,5-7 ultimate success will require that algae are grown at near
61
optimal growth rates and within a microbial community that is stable and conducive to lipid
62
production.
63
Although challenges remain to achieve sustainable
Bacteria have been shown to be beneficial for algal growth,2,
8
in part via nutrient
64
recycling and exchange. Bacteria can also provide algae with essential vitamins, co-factors 2, 9, 10
65
or growth hormones,11, 12 participate in signal transduction,13 and may help protect algae from
66
pathogens14 in exchange for algae-excreted dissolved organic carbon. Interactions appear to
67
involve complex communication mechanisms15, 16 and may involve bacterial colonization of the
68
algal cell surface.17,
69
communities are structured in nature19 and in photobioreators.17 These studies have focused on
70
the identification of prevalent microbial gene families or metabolic pathways as a means to
71
identify particular symbiotic relationships. Bacteria belonging to alpha-Proteobacteria, beta-
72
Proteobacteria, and Bacteroidia classes (including members of the Flavobacteriales and
73
Sphingomonadales orders) have been found to be associated with green algae.20
18
Genomic tools are enabling us to learn more about how these
3 ACS Paragon Plus Environment
Page 5 of 30
Environmental Science & Technology
74
Although positive interactions between phototrophic microorganisms and some bacteria
75
are well established, and we are only beginning to understand how specific metabolic pathways
76
change in algae
77
attributes of the algae cell, in particular those of interest for biofuel production. Recent “omics”
78
studies have revealed significant changes in key metabolic pathways when bacteria are
79
introduced or re-introduced to pure cultures of photosynthetic cyanobacteria.21-23 For example,
80
in the presence of the bacterium Alteromonas macleodii, algal (Prochlorococcus) cells
81
upregulated components of their photosynthetic apparatus to increase the export of reduced
82
carbon compounds for use by its heterotrophic partner.23
83
(Synechococcus) expressed membrane transport proteins increased, and vitamin metabolism
84
related proteins decreased in the presence of the bacterium Roseobacter, suggestive of metabolite
85
provision by the bacteria.21
86
in response to variable microbial communities and how this may affect key
In a separate study, algal
Strain C596 is a subtropical marine Chlorella spp. that accumulates lipids (up to 45%
87
TAG by weight) under nitrogen and phosphorous limiting conditions.3,
24
88
versatile growth conditions including across a range of salinities and has the ability to utilize a
89
wide range of carbon substrates for growth in the dark.3 Stock cultures of C596 were not axenic,
90
and during attempts to create axenic stocks, algal cells exhibited lower growth rates that we
91
hypothesized resulted from an altered microbiome in the new consortium. In this study, we
92
created and studied a set of C596 consortia to understand how changes in community structure
93
affect Chlorella growth and lipid accumulation.
94
metatranscriptomic analysis on two consortia with distinct microbiomes to better understand
95
their synergistic roles during co-culture with Chlorella C596.
C596 tolerates
We also conducted a comparative
96
4 ACS Paragon Plus Environment
Environmental Science & Technology
97
METHODS
98
Chlorella consortia growth and plating. Chlorella spp. “SAG 211-18” strain C596 with its
99
initial microbial community3 (designated C596-R1 hereafter) was grown in the synthetic
100
seawater medium Aquil25 under constant illumination (80 µmol photon m-2 s -1, GE Ecolux Cool
101
White bulbs, East Cleveland, OH, USA) at 25˚C.
102
fluorescence (10-AU Fluorometer, Turner Designs, San Jose, CA, USA) and cell counts using a
103
Neubauer hemacytometer (Spencer Bright Line) at 40× magnification with an optical microscope
104
(IX83, Olympus Corp., Center Valley, PA, USA). Various C596 consortia were plated/streaked
105
on solid Aquil medium, and isolated algal colonies were returned to liquid medium as above
106
yielding new consortia (e.g., C596-S1 and C596-S2). To isolate bacterial strains associated with
107
Chlorella, C596-R1 was plated/streaked on solid Aquil medium supplemented with 2, 5, 10 or
108
20 mM of either acetate, sucrose or glycerol. Colony PCR using universal 16S primers (8F: 5’-
109
AGAGTTTGATCCTGGCTCAG- ‘3, 1496R: 5’ -GGCTACCTTGTTACGACTT- ‘3) followed
110
by sequencing were conducted to screen bacterial colonies.
Growth was monitored via chloroplast
111 112
Resuspension experiments. In one set of experiments, C596-R1 was grown till late-exponential
113
phase and harvested by filtration using a 3-µm membrane filter to separate the algae (~5 µm)
114
from the bulk of the bacterial community. Filtrates were then passed through 0.2-µm membrane
115
filters to capture bacteria. These 0.2-µm filters were added to separate mid-exponential phase
116
C596-R1 and -S1 cultures as well as to fresh Aquil media tubes (filter control C596-F1).
117
Membrane filters were removed with sterile forceps after 24 hours, and cultures were monitored
118
for growth thereafter.
119
exponential phase and spent media was filtered through a 0.2-µm membrane. Filtrates were
In a parallel experimental set, C596-R1 culture was grown to late-
5 ACS Paragon Plus Environment
Page 6 of 30
Page 7 of 30
Environmental Science & Technology
120
divided in half, one half was autoclaved for 35 minutes. Both halves were then amended with
121
300 µM NaNO3, 10 µM NaH2PO4, trace metals and vitamins stock solutions to final
122
concentrations as recommended for Aquil medium, and pH was verified. Autoclaved filtrate,
123
non-autoclaved filtrate and fresh Aquil media were used to re-suspend pelleted C596-R1 and -S1
124
consortia harvested at mid-exponential phase.
125
florescence measurements.
Growth was monitored via chlorophyll
126 127
Nile red fluorescence. Nile red stain was dissolved in dimethyl sulfoxide (HPLC grade, Sigma
128
Aldrich, St. Louis, MO, USA) to a final concentration of 1 mg/mL, and stored at -20°C in the
129
dark.
130
fluorescence was measured utilizing filters with excitation/ emission wavelengths of 530 ± 25/
131
590 ± 20 nm, respectively. Resulting measurements were used to quantify the relative cellular
132
lipid content26 which was subsequently normalized by the cell concentration (determined
133
microscopically) to determine cellular lipid content. Triplicate measurements were taken at late
134
exponential, early stationary and senescent growth stages.
Algae were stained in 96-well plates as previously reported,26 and the Nile red
135 136
16S rRNA sequencing and microbial community analysis. Biological triplicates of algal
137
consortia were harvested at mid-exponential growth by passing through 0.2-µm membrane filters
138
(Whatman Nuclepore Track-Etch membrane, Sigma-Aldrich). Total RNA was extracted and
139
purified using the RNeasy Plant Mini Kit (Qiagen, Germantown, MD, USA),3 and its integrity
140
and concentration were evaluated using a RNA 6000 Nano Kit on an Agilent Bioanalyzer 2100
141
(Agilent Technologies, Santa Clara, CA, USA). Complimentary DNA (cDNA) was generated
142
using an iScript reverse transcriptase and a blend of olig(dt) and random hexamer primers
6 ACS Paragon Plus Environment
Environmental Science & Technology
143
(iScript cDNA Synthesis Kit; Bio-Rad, Hercules, CA, USA). The 16S rRNA gene was amplified
144
using the universal bacterial primers 341F (5’- CCTACGGGNGGCWGCAG -3’) and 805R (5’-
145
GACTACHVGGGTATCTAATCC -3’), with overhang sequences compatible for index
146
attachment as described elsewhere.27 PRIME HotMasterMix (5 PRIME Inc., Gaithersburg, MD,
147
USA) was used in PCR reactions, and cycle conditions were set as previously described.28 The
148
16S rRNA gene amplicon pool was sequenced on an Illumina MiSeq (Cornell Genomics
149
Facility, Ithaca, NY) using a 600-cycle MiSeq Reagent Kit v.3 (Illumina, San Diego, CA, USA).
150
Sequences were processed using the Brazilian Microbiome Project pipeline with modifications
151
as described by Howard et al.28 Briefly, paired-end sequences were merged, primers trimmed,
152
and singleton sequences removed using Mothur v.1.36.1.
153
Taxonomic Units (OTUs) and chimera removal were performed using USEARCH v.7 (Edgar
154
2010). Representative OTU sequences were classified (classify.seqs, cutoff = 80) using the
155
GreenGenes v.13.8 database for 16S rRNA gene sequences, and OTUs that were suspected to
156
not be of bacterial origin were removed. All fastq sequences are available at the NCBI’s
157
Sequence Read Archive (SRA) under the BioProgect PRJNA397076. Statistical analysis were
158
performed using R (v.3.2.1). OTUs were first randomly subsampled to yield an equal number of
159
sequences for all samples (rarefaction curves are shown in Figure S1). Heatmaps were generated
160
via the gplots package in R29 to display the OTU relative abundances. OTUs were clustered
161
hierarchically (average linkage) based on the Bray-Curtis dissimilarity index.
Clustering of 97% Operational
162 163
Random shotgun metatranscriptome sequencing and analyses. For the C596-R1 and -S1
164
consortia, an additional rRNA depletion step was performed on the resulting RNA from the
165
extraction and purification methods described.
This step employed the GeneRead rRNA
7 ACS Paragon Plus Environment
Page 8 of 30
Page 9 of 30
Environmental Science & Technology
166
Depletion Kit (Qiagen) specific for eukaryotic rRNA following the manufacturer’s instructions
167
to enhance the sequencing coverage of mRNA.
168
bacterial rRNA.
There is a potential cross reactivity with
169
Random-hexamer based RNA sequencing, transcript assembly, and analyses that resulted
170
in a total of 7,557 C596-annotated transcripts were performed as described elsewhere.3 All raw
171
and processed data files are freely available (NCBI’s Sequence Read Archive accessions
172
SRX1282877 and SRX1281663 under the BioProject ID PRJNA294811). The SortMeRNA
173
program30 was used to bin the reads as either of potential rRNA or mRNA origin by comparing
174
the reads to the SILVA Eukaryotic (18S and 28S), SILVA Bacteria (16S and 23S), SILVA
175
Archaea (16S and 23S), and RFAM (5S and 5.8S) databases. The rRNA sequences were then
176
filtered from the processed files using the SILVA RNA databases and assigned to an Operational
177
Taxonomic Unit (OTU).
178
Reads assigned to mRNA transcripts annotated as of Chlorella were analyzed using
179
edgeR31 in the R software suite (v 3.1.2) to determine transcripts that were differentially
180
regulated using the exactTest function. A fold-ratio change of greater than 4.0 or less than 0.25
181
with a multiple-testing corrected p-value of 6.6 x 10-6 (Bonferroni-Holm established corrected
182
threshold) was set to select for the statistically-significant differentially-expressed genes. A
183
looser ratio cutoff (fold ratio greater than 2.0 or less than 0.5 with an uncorrected p-value < 0.05)
184
was established to select a subset of genes for a bulk profile analysis. These genes were assigned
185
to KEGG categories32 using a tblast search against the protein database for Chlorella variabilis
186
(BLAST e-value of 10-10), and then the binned transcripts were enumerated to compare relative
187
expression in -S1 and -R1 cells.
188
8 ACS Paragon Plus Environment
Environmental Science & Technology
189
Page 10 of 30
RESULTS AND DISCUSSION
190
Algal culture variability appears modulated by associated microbiome.
As part of a
191
previous study, we noted that algal cultures generated by inoculation of algal colonies from a
192
plate appeared to be phenotypically different and we hypothesized an altered microbial
193
consortium was responsible.
194
consortium, generated from the initial consortium C596-R1, exhibited a significant decrease in
195
growth rate, 0.7 ± 0.08 d-1 versus 1.3 ± 0.12 d-1 respectively (Figure 1). This difference was
196
maintained in subsequent culture transfers for over one year.
One of the algal cultures, hereafter designated the C596-S1
197 198
Further modification of the microbial background in Chlorella culture. Our attempts to
199
isolate individual bacteria from the C596-R1 consortium using various carbon sources (see
200
Methods) led to the repeated isolation of only one Ruegeria species as confirmed with 16S rRNA
201
gene sequencing. Addition of this Ruegeria strain to cultures of C596-S1 did not stimulate
202
growth (Figure S2). Because only a small fraction of bacteria from an environmental sample can
203
be isolated and cultivated in a laboratory under a single set of growth conditions,33 we next filter-
204
separated the planktonic microbial community (i.e., 0.2 µm < cell size < 3 µm) from the C596-
205
R1 consortium and added the microbial-laden filters to C596-S1 and -R1 consortia (Figure 2A).
206
The addition of the collected community marginally increased the growth rates of both cultures
207
(Figure 2 B & 2C). Triplicate biological controls comprised solely of the filters (collected from
208
500 mL of C596-R1 medium) placed in Aquil media exhibited no measurable algal growth over
209
the period during which we measured increased growth rates in the aforementioned C596-S1 and
210
-R1 cultures. After four days however, the few algae that were present on the filter grew to a
211
sufficient density to be observed, and their growth rate was quantified (Figure 2B). These newly
9 ACS Paragon Plus Environment
Page 11 of 30
Environmental Science & Technology
212
generated triplicate Chlorella consortia, designated C596-F1, exhibited a 39% increase in algal
213
growth relative to -R1 (1.67 ± 0.07 d-1 vs. 1.35 ± 0.04 d-1 for C596-R1) and maintained this rate
214
for over six months of subsequent culturing (Figure 2C).
215
To investigate this apparent Chlorella-bacteria symbioses further, triplicate C596-S1 and
216
-R1 consortia pellets were resuspended in nutrient-amended C596-R1 filtrate (Figure 2D) which
217
increased algal growth rates relative to the original C596-S1 and -R1 by 29% and 10%,
218
respectively (Figure 2E & 2F). Resuspension in autoclaved nutrient-amended filtrate or fresh
219
Aquil medium had no effect on Chlorella growth (Figure 2 E & 2F). Both C596-S1 and -R1
220
algal cells benefitted from some growth-promoting factor or factors present in the -R1 filtrate
221
that were destroyed by autoclaving. Importantly, two subsequent transfers of the resulting C596-
222
S1 consortium to Aquil media retained the elevated growth rate (data not shown). It is possible
223
that yet another change in microbial community was generated via resuspension and
224
recombination with small biological agents in subsequent transfers (viruses and/or small
225
bacteria) that passed through the 0.2-µm filters or there was the persistence of a dilute chemical
226
factor(s).34
227 228
Microbial community influences Chlorella cellular lipid content. To investigate whether the
229
composition of the microbial community has an influence on C596 lipid accumulation, Nile red
230
assays were conducted on C596-R1, -S1 and -F1. In all consortia, the cellular lipid content
231
profile was coupled to decreasing nutrient availability over time in the medium, with no
232
measurable lipid accumulation at mid exponential phase (data not shown), detectable levels
233
emerging at late exponential phase and maximum levels in senescence (Figure 3). Algal cells in
234
the C596-S1 consortium, however, displayed the lowest cellular lipid content in all three stages
10 ACS Paragon Plus Environment
Environmental Science & Technology
Page 12 of 30
235
relative to -R1 and -F1 cells. Conversely, C596-F1 cells outperformed -R1 and -S1 at each time
236
point; levels were 36% and 58% higher in -F1 cells respectively (p < 0.05) in senescence (Figure
237
3).
238 239
Microbiota analysis reveals pattern of microbial changes that alter phenotype.
To
240
investigate the particular microbial changes in our C596 cultures, extracted RNA was reverse
241
transcribed and 16S rRNA gene sequencing using bacteria-specific primers was employed. Five
242
consortia of C596 were tested: the parent C596-R1, C596-S1 (Figure 1), C596-S2 (a colony
243
culture that maintained the parent growth rate, 1.34 ± 0.03 d-1), C596-F1 (Figure 2A-C) and
244
C596-FS1 (a colony culture of C596-F1 that exhibited a reduced growth rate of 0.84 ± 0.15 d-1).
245
We collected 2,502,720 paired reads ranging from 88,088 to 179,118 reads per sample. The
246
sequencing were then processed for merging and removal of low quality sequencing and
247
singletons. The resulting number of reads after the following elimination of chimeras and non-
248
bacterial OTUs ranged from 30,840 to 1,447 reads per sample with a median OUTs of 7,665. In
249
total, we observed 103 operational taxonomic units (OTUs) from high quality sequence reads,
250
among them 20 OTUs were determined to be of non-bacterial origin. Of the 83 bacterial OTUs,
251
17 OTUs were found to exceed a relative abundance of 1% in one or more of the three biological
252
replicates, of which 14 OTUs belong to the Proteobacteria phylum. Those 17 OTUs were
253
hierarchically grouped based on the average relative abundance of sequencing reads across the
254
tested samples (Figure 4).
255
Higher algal growth rate appears to be linked to the presence of two particular microbes,
256
with the presence of each being important. Both C596-R1 and -S2, which have comparable
257
growth rates (Figure 4), contain an abundant Ruegeria spp. Type 2 (OTU 670577169) and a
11 ACS Paragon Plus Environment
Page 13 of 30
Environmental Science & Technology
258
Rhodobacter spp. (Figures 4 and S3). The consortium with the highest algal growth rate, C596-
259
F1, also displays both of these microbes in high abundance, but the relative abundance of the
260
Rhodobacter spp. is greater compared to C596-R1 and -S2 (Figures 4 and S3). In contrast, the
261
microbiome of C596-FS1, which had a lower algal growth rate than all three of these consortia,
262
was predominantly Rhodobacter spp. (99.8% abundance). C596-S1, which was slowest of all,
263
had neither of these OTUs. Notably, both Ruegeria spp. (Type 1: OTU 169121548 and Type 2)
264
were detected among the abundant OTUs, but in the fast growing C596 cultures, only Type 2
265
was present in high abundance. The previously discussed laboratory Ruegeria isolate which did
266
not improve Chlorella growth (Figure S2) was closely related to the less abundant Ruegeria spp.
267
Type 1.
268 269
Metatranscriptome supports the 16S gene-amplicon sequencing results. Random shotgun
270
transcriptomic analyses of the C596-R1 and -S1 algae were conducted via Illumina sequencing
271
of RNA samples harvested at mid-logarithmic phase.
272
performed to enhance the coverage of the mRNA and samples were reverse transcribed using
273
random-hexamer primers. Nevertheless, approximately 85% of the reads for each sample were
274
of rRNA origin which enabled us to screen for 16S rRNA sequences as an additional means to
275
compare the microbial communities of these two isolates.
276
sequences from potential bacterial community members (Figure S4) were separated from those
277
of algal origin (Figure S4). Reads from Rhodobacterales, Rhizobiales, Caulobacterales, and
278
Cytophagales orders were less abundant in the C596-S1 consortium, whereas those from the
279
order Bacillales were greater in -S1 (Figure S4B). Other orders were relatively unchanged as a
280
percent of reads between the two cultures.
Eukaryotic rRNA depletion was
Reads that mapped to rRNA
Consistent with the 16S rRNA gene-amplicon
12 ACS Paragon Plus Environment
Environmental Science & Technology
Page 14 of 30
281
sequencing results, the active presence of members of the Rhodobacterales order (which includes
282
Ruegeria spp. and Rhodobacter spp.) is linked to higher growth rates.
283 284
Differential regulation of Chlorella transcripts in response to changes in microbial
285
community. An edgeR analysis with an uncorrected p-value cutoff of < 0.05 and a fold change
286
cutoff of > 2 or < 0.5 was performed to broadly compare transcript expression differences
287
between algal cells from C596-S1 and C596-R1 consortia.
288
differentially expressed, 449 of which fell within standard KEGG categories (Figure 5). Genes
289
encoding for proteins in pathways within the Transcription and Translation categories are less
290
transcribed in C596-S1 relative to -R1 (21 higher in -S1 vs. 48 higher in -R1 and 36 vs. 86 genes
291
respectively), suggesting less active protein biosynthesis in -S1 relative to -R1, which is
292
consistent with a slower growth rate. In particular, genes encoding for proteins in pathways
293
involving RNA degradation, spliceosome activity, ribosome biogenesis, and RNA transport were
294
less abundant in the C596-S1 cells. Consistent with this observation, there was overall lower
295
transcription of genes encoding for proteins involved in amino acid biosynthesis in -S1 compared
296
to -R1, and although DNA replication was generally increased in expression in -S1, gene
297
expression specifically involved in purine metabolism was greater in -R1. Additionally, C596-
298
S1 cells exhibited substantially more expression of genes encoding for proteins within
299
Replication and Repair (24 vs. 15), perhaps reflecting more stress in -S1.
A total of 1878 genes were
300
The potential roles of the bacteria consortia were also explored with this analysis. Within
301
broad metabolism categories, Energy Metabolism (42 higher in -S1 vs. 15 higher in R1) and
302
Lipid Metabolism (32 vs. 6) related genes were more likely to be expressed at a higher level in
303
the slower-growing -S1, suggesting an investment required by -S1 to compensate for metabolites
13 ACS Paragon Plus Environment
Page 15 of 30
Environmental Science & Technology
304
no longer provided by the bacterial symbionts. Within the category of Metabolism of Co-factors
305
and Vitamins, the genes involved in the biosynthesis of thiamine (vitamin B1) and biotin
306
(vitamin B7) metabolism were more expressed in -S1, although we include both of these co-
307
factors in our growth medium. We also include cobalamin (vitamin B12), a co-factor known to
308
be provided to algae by bacteria, in our medium. We see no evidence of cobalamin deficiency in
309
either consortium as there is low expression in both -S1 and -R1 of the metE gene (c12716_g1,
310
Table S1), the B12- independent isoform of methionine synthase in C. reinhardtii that is
311
upregulated when B12 is absent.35
312
The application of a more stringent cutoff in the edgeR analysis of significant
313
differentially regulated genes (a fold ratio > 4 or < 0.25; a multiple-testing Bonferroni-Holm
314
corrected p-value < 0.05 which is equivalent to a raw p-value < 6.6 x 10-6) resulted in a subset
315
143 transcripts, 73 of which were “highly” upregulated in C596-S1 (Table S1). Nineteen of
316
these genes displayed significant BLASTx hits (e-value < 10-10 for all hits) to proteins encoded
317
on the C. variabilis NC64A chloroplast or mitochondria,36 whereas only 2 of this type (out of a
318
total of 70 “highly” downregulated gene transcripts) were downregulated in -S1. Consistent with
319
the previous analysis, of those gene transcripts “highly” upregulated in C596-S1, genes encoding
320
photosystem-related proteins are overrepresented (n = 10; Table S2).
321
Also among the restricted list of genes “highly” upregulated in -S1 are several
322
homologues of genes that are known components of a CO2 concentrating mechanism (CCM) in
323
Chlamydomonas reinhardtii. These include the low-CO2 inducible protein LCIC (c94329_g1)
324
and the ABC transporter HLA3 (c10889_g1) (Table S2). In C. reinhardtii, the LCIC gene is
325
expressed under limiting CO2 conditions but their exact function is unknown.37 HLA3, an ATP-
326
binding cassette ABC transporter associated with the HCO3- uptake system, is also upregulated
14 ACS Paragon Plus Environment
Environmental Science & Technology
327
under low CO2 concentrations.38,
39
328
CAH3 (c5724_g1), the enzyme in C. reinhardtii known to catalyze the conversion of HCO3- to
329
CO2 in the thylakoid lumen, did not vary between the transcriptomes of C596-R1 and -S1 (data
330
not shown). However this enzyme is known to be regulated at the post-translational level via
331
phosphorylation.40
Transcript levels of the carbonic anhydrase homologue
332
With respect to nitrogen metabolism, genes encoding for proteins involved in the nitrate
333
assimilation pathway such as nitrate (NO3-) reductase (c11833_g1_i1) and nitrite (NO2-)
334
reductase (c6958_g1_i2) are similarly highly expressed in both C596-R1 and -S1 cells (data not
335
shown). This result is not surprising since our medium contains nitrate that is plentiful during
336
exponential growth. However, we note that a glutamine synthetase gene (c12245_g1) and
337
several genes encoding for proteins involved in secondary nitrogen metabolism, such as an urea
338
active transporter-like protein41 (c13324_g1) and a carbon-nitrogen family hydrolase42
339
(c11710_g1), are upregulated by 4.9, 5.5 and 4.2 fold respectively in the C596-S1 cells (Table
340
S2). This suggests less efficient N recycling by the microbial community in -S1.
341
In addition to facilitating carbonic and nitrogen exchange via the degradation and
342
recycling of carbohydrates, fatty acids and other biomolecules on or near the surface of algal
343
cells, bacteria also consume dissolved oxygen which decreases oxidative stress in the algae.22, 23
344
Transcriptomic evidence of greater oxidative stress in the -S1 cells includes higher expression
345
(5.8 fold) of transcripts encoding for the glutathione reductase43 (c13123_g1, Table S2) and
346
proteins involved in thiamine (vitamin B1) biosynthesis (Figure 5). 44 Increased gene expression
347
in DNA replication and repair in -S1, as well as increased processing of lipids (in the absence of
348
greater TAG accumulation, Figure 5) are also consistent with more oxidative stress in -S1.
15 ACS Paragon Plus Environment
Page 16 of 30
Page 17 of 30
Environmental Science & Technology
349
Similar patterns were observed in cyanobacteria with and without added bacteria, suggesting that
350
the bacterial community can assist the growth of algae by reducing oxidative stress.22
351
Algal cells are known to use chemical signals to recruit bacteria to colonize their
352
surfaces.45, 46 The osmolyte dimethylsulphoniopropionate (DMSP), which is released from algal
353
cells, has recently been shown to serve as a chemoattractant for members of the Roseobacter
354
clade of the Rhodobacterales order, and high concentrations of DMSP can induce a non-motile
355
state in bacteria facilitating biofilm formation.47 Although the final catalytic steps in DMSP
356
production are not characterized,48 it is known that cysteine and methionine are required for the
357
synthesis of DMSP.49
358
upregulation of both the cysteine and methionine metabolism pathways (Figure 5). We speculate
359
that DMSP levels, although typically low in Chlorophytes,50 may be elevated in the -S1 cells in
360
response to the loss of important microbial species from the consortium. Previous studies have
361
noted increased levels of cellular DMSP under nutrient-limiting conditions such as nitrogen, and
362
to a lesser extent CO2 limitation.51
We note in our transcriptome that the C596-S1 cells exhibit an
363
Algal cells may also attempt to recruit bacteria by increasing production of signaling
364
molecules that are anchored in the cell envelope. We see an upregulation of transcripts encoding
365
for general lipid metabolism proteins as well as glycan and carbohydrate metabolism proteins in
366
C596-S1 (Figure 5). Within these broader categories, we note a substantial number of transcripts
367
predicted to be involved in the metabolism of polar lipids are upregulated.
368
phosphoglycerolipids are best known to influence cell membrane structure, they are also known
369
to play a role in plant hormone signal transduction, as are sphingolipids, which can also serve as
370
signal molecules directly.52 Notably, sphingolipids are involved as specific recognition sites for
16 ACS Paragon Plus Environment
Although
Environmental Science & Technology
Page 18 of 30
371
bacterial attachment in animal cells.53 Further work will be needed to determine whether these
372
lipids are also important in symbiotic interactions involving algae.
373 374
IMPLICATIONS
375
Microalgae, in particular lipid-accumulating Chlorella species potentially viable for
376
biofuel feedstock, are well known to have both positive and negative interactions with microbes.
377
Algae are unlikely to remain in pure cultures at the growth scale that will be required for biofuel
378
feedstock production (in open pond or even photobioreactor systems). Particular algae-microbial
379
consortia therefore will exhibit measurable and relevant phenotypic differences. In this study we
380
utilized next-generation sequencing and bioinformatics tools to characterize the algal
381
microbiomes of consortia exhibiting significant differences in algal growth rate and lipid
382
accumulation. We also performed a comparative analysis on algal transcriptomes derived from
383
two of these consortia, the original C596-R13 and the slower growing C596-S1.
384
hypothesized that the expression of genes in various metabolic pathways would help us to
385
identify more precisely which synergistic bacterial interactions have been disrupted in -S1.
We
386
Our work reveals a complex symbiotic relationship between Chlorella C596 cells and its
387
microbiota. Algal cells appear to be carbon limited and stressed when they lose beneficial
388
bacterial partners. Transcriptomic data suggest that cells likely use multiple strategies to recruit
389
bacteria back into a consortium. Our data demonstrate that particular shifts in the microbial
390
community can lead to lower growth rates and less lipid accumulation, which could equate to
391
highly variable productivity and yields for large-scale cultivation facilities.
392
underscores that the performance of axenic algae will not be relevant to production facilities.
393
Clearly, any characterization of lipid-producing algae and/or elucidation of strain-specific
17 ACS Paragon Plus Environment
Our work
Page 19 of 30
Environmental Science & Technology
394
differences in phenotype must be performed with the recognition that microbial consortium are
395
critical to the outcome.
396
phenotype and metabolic activity will also lead to a better understanding of natural ecosystems
397
and may enable future practitioners to manipulate algal phenotypes necessary for the robust large
398
scale algal cultivation required for biofuel feedstocks, including managing the overall growth
399
rate and lipid accumulation.
Understanding how particular microbiological interactions impact
400 401
ACKNOWLEDGMENTS
402
We thank Jenny Kao-Kniffin at Cornell University for providing material for 16S rRNA sample
403
preparation. We also thank Catherine Spirito and Terrence Bell for help in 16S rRNA sample
404
preparation and sequencing analysis. This work was supported by the Department of Energy
405
(DE-EE0003371).
406 407
ASSOCIATED CONTENT
18 ACS Paragon Plus Environment
Environmental Science & Technology
408
Supporting Information
409
Figure S1. Rarefaction curves of the Operational Taxonomic Units (OTUs) detected in MiSeq
410
sequencing of the 16S rRNA of C596-R1, -S1, -S2, -F1, and -FS1 consortia.
411
Figure S2. Effect of one Ruegeria strain isolate on growth of C596-S1. Ruegeria spp. Type 1
412
(OTU169121548) was isolated on solid Aquil medium supplemented with 20 mM acetate. Two
413
colony isolates (A and B) were confirmed by sequencing and independently tested for growth
414
recovery of the C596-S1. Error bars represent the standard error of the mean from triplicate
415
growth experiments.
416
Figure S3. Relative abundance (%) of Rhodobacter spp. and Ruegeria spp. Type 2 (OTU
417
670577169) in C596-F1, -R1, -S2, -FS1, and -S1 cultures. Error bars represent the standard error
418
of the means from triplicate samples.
419
Figure S4. (A) Bar chart displaying the percent of 16S rRNA sequences in the transcriptomes
420
mapping to the chloroplast and mitochondria of Chlorella for the C596-R1 and -S1 cultures;
421
error bars represent the 95% confidence interval based on the biological duplicates. (B) Bar
422
chart displaying 16S rRNA sequences mapping to individual bacterial orders (that represent
423
greater than 0.1% of the total 16S rRNA in either sample); error bars represent the 95%
424
confidence interval based on replicate cultures.
425
Table S1. The identifiers, assembled nucleic acid sequences, annotation descriptions, raw counts
426
and normalized fold changes ratios of all detected algal transcripts with a Bonferonni-Holm
427
adjusted p-value