doi:10.1128/mBio.00869-13.
4(6): .
mBio.
Treponema primitia
Uncoupling on Transcription in the Termite Gut Symbiont
Genome-Wide Effects of Selenium and Translational
2013.
Eric G. Matson, Adam Z. Rosenthal, Xinning Zhang, et al.
Treponema primitia
Symbiont
Transcription in the Termite Gut
Translational Uncoupling on
Genome-Wide Effects of Selenium and
http://mbio.asm.org/content/4/6/e00869-13.full.html
Updated information and services can be found at:
MATERIAL
SUPPLEMENTAL
http://mbio.asm.org/content/4/6/e00869-13.full.html#SUPPLEMENTAL
REFERENCES
http://mbio.asm.org/content/4/6/e00869-13.full.html#ref-list-1
This article cites 44 articles, 24 of which can be accessed free at:
CONTENT ALERTS
more>>
article),
Receive: RSS Feeds, eTOCs, free email alerts (when new articles cite this
http://journals.asm.org/subscriptions/
To subscribe to another ASM Journal go to:
http://mbio.asm.org/misc/contentdelivery.xhtml
Information about Print on Demand and other content delivery options:
http://mbio.asm.org/misc/reprints.xhtml
Information about commercial reprint orders:
mbio.asm.org
on January 23, 2014 - Published by
mbio.asm.org
Downloaded from
mbio.asm.org
on January 23, 2014 - Published by
mbio.asm.org
Downloaded from
Genome-Wide Effects of Selenium and Translational Uncoupling on
Transcription in the Termite Gut Symbiont
Treponema primitia
Eric G. Matson
,
Adam Z. Rosenthal
,
Xinning Zhang
,
Jared R. Leadbetter
Ronald and Maxine Linde Center for Global Environmental Science, California Institute of Technology, Pasadena, California, USA
E.G.M. and A.Z.R. contributed equally to this work..
ABSTRACT
When prokaryotic cells acquire mutations, encounter translation-inhibiting substances, or experience adverse environ-
mental conditions that limit their ability to synthesize proteins, transcription can become uncoupled from translation. Such uncou-
pling is known to suppress transcription of protein-encoding genes in bacteria. Here we show that the trace element selenium controls
transcription of the gene for the selenocysteine-utilizing enzyme formate dehydrogenase (
fdhF
Sec
) through a translation-coupled
mechanism in the termite gut symbiont
Treponema primitia
, a member of the bacterial phylum
Spirochaetes
. We also evaluated
changes in genome-wide transcriptional patterns caused by selenium limitation and by generally uncoupling translation from tran-
scription via antibiotic-mediated inhibition of protein synthesis. We observed that inhibiting protein synthesis in
T. primitia
influ-
ences transcriptional patterns in unexpected ways. In addition to suppressing transcription of certain genes, the expected consequence
of inhibiting protein synthesis, we found numerous examples in which transcription of genes and operons is truncated far downstream
from putative promoters, is unchanged, or is even stimulated overall. These results indicate that gene regulation in bacteria allows for
specific post-initiation transcriptional responses during periods of limited protein synthesis, which may depend both on translational
coupling and on unclassified intrinsic elements of protein-encoding genes.
IMPORTANCE
A large body of literature demonstrates that the coupling of transcription and translation is a general and essential
method by which bacteria regulate gene expression levels. However, the potential role of noncanonical amino acids in regulating
transcriptional output via translational control remains, for the most part, undefined. Furthermore, the genome-wide transcrip-
tional state in response to translational decoupling is not well quantified. The results presented here suggest that the noncanoni-
cal amino acid selenocysteine is able to tune transcription of an important metabolic gene via translational coupling. Further-
more, a genome-wide analysis reveals that transcriptional decoupling produces a wide-ranging effect and that this effect is not
uniform. These results exemplify how growth conditions that impact translational processivity can rapidly feed back on tran-
scriptional productivity of prespecified groups of genes, providing bacteria with an efficient response to environmental changes.
Received
10 October 2013
Accepted
15 October 2013
Published
12 November 2013
Citation
Matson EG, Rosenthal AZ, Zhang X, Leadbetter JR. 2013. Genome-wide effects of selenium and translational uncoupling on transcription in the termite gut symbiont
Treponema primitia
. mBio 4(6):e00869-13. doi:10.1128/mBio.00869-13.
Editor
Margaret McFall-Ngai, University of Wisconsin
Copyright
© 2013 Matson et al. This is an open-access article distributed under the terms of the
Creative Commons Attribution-Noncommercial-ShareAlike 3.0 Unported
license
, which permits unrestricted noncommercial use, distribution, and reproduction in any medium, provided the original author and source are credited.
Address
correspondence to Eric G. Matson, matsone@uwosh.edu, or Adam Z. Rosenthal, arosenthal@caltech.edu.
T
he direct coupling of transcription and translation is a charac-
teristic unique to prokaryotes because both processes occur
simultaneously in the same compartment. The coupling and un-
coupling of these processes are responsible for the well-
characterized regulatory mechanism known as attenuation, ex-
emplified by the tryptophan biosynthesis (
trp
) operon of
Escherichia coli
(1). In this mechanism of gene regulation, com-
mon in several amino acid biosynthesis pathways and in many
bacteria, alternative stem-loop mRNA secondary structures can
be formed or resolved as a result of stalled or actively translating
ribosomes. These alternative RNA structures regulate the tran-
scription of downstream genes by inhibiting or allowing proces-
sive transcription by RNA polymerase (RNAP) (1). The relation-
ship between transcription and translation also gives rise to polar
effects in bacteria, first observed by Zipser in 1969 (2). In this
phenomenon, point mutations and reading frame shifts that in-
troduce translational stop codons in protein-coding genes signal
the end of translation and thus prematurely uncouple translation
from transcription, inhibiting transcription of genes downstream
of the site of an introduced lesion within an operon.
Interactions between translational machinery and transcriptional
machinery can also govern the rate of transcription of bacterial genes.
In bacterial cells, RNAP functions via a Brownian ratchet mechanism
wherein forward progression and backtracking both occur in an
ATP-independent manner (3). In highly transcribed but nontrans-
lated genes, such as rRNAs, multiple RNAPs operating in succession
mechanically limit backtracking and contribute to high transcrip-
tional throughput (4). The capacity of ribosomes to influence tran-
scriptional rates has also been documented. The mechanics of ribo-
some and RNAP interactions can involve direct contact or indirect
contact through other protein factors, including NusG and NusE (5)
and the transcriptional termination factor Rho. Proshkin and col-
leagues (6) have demonstrated the finely tuned coupling of transla-
tion and transcription by comparing genes containing frequently
RESEARCH ARTICLE
November/December 2013 Volume 4 Issue 6 e00869-13
®
mbio.asm.org
1
mbio.asm.org
on January 23, 2014 - Published by
mbio.asm.org
Downloaded from
used codons with genes containing infrequently used codons, which
are translated at a lower rate and are consequently transcribed at a
lower rate, matching that of translation. The authors thus concluded
that macromolecular interactions between translational and tran-
scriptional machinery are the fundamental mechanism of post-
initiation gene regulation and adaptation to environmental changes
in bacteria (6).
The synthesis of some proteins inevitably depends on the abil-
ity of the translational machinery to decode extremely rare
codons. This is the case for proteins that utilize the noncanonical,
twenty-first amino acid, selenocysteine, found in the catalytic ac-
tive site of a variety of enzymes. Selenium substitutes for the sulfur
moiety in the thiol R group of cysteine in the formation of seleno-
cysteine (7). The incorporation of selenocysteine into proteins
requires dedicated cellular machinery because it is encoded by
UGA, a codon normally used to signal a translational stop. A spe-
cialized translation elongation factor (SelB) recognizes a stem-
loop mRNA structure known as a selenocysteine incorporation
sequence (SECIS) element located downstream from selenocys-
teine codons and directs ribosomes to incorporate selenocysteine
into the nascent polypeptide chain (8, 9). A lack of selenium (and
thus selenocysteine) should prevent ribosomes from correctly
translating UGA codons as selenocysteine and instead pause or
terminate translation.
The homoacetogenic spirochete
Treponema primitia
has two
genes that encode formate dehydrogenase isoenzymes, allowing
this organism to grow lithotrophically on H
2
plus CO
2
(10, 11).
One version of the gene (
fdhF
Cys
) encodes an enzyme that utilizes
cysteine in the active site, while the other (
fdhF
Sec
) encodes an
enzyme that uses the noncanonical amino acid selenocysteine.
Studies have shown that selenocysteine-containing enzymes can
be orders of magnitude more active than cysteine-containing ho-
mologs (12–16). Consistent with the hypothesis that the
selenocysteine-containing enzyme is preferred in
T. primitia
be-
cause it may be more active, we previously showed that transcrip-
tion of the selenocysteine version of the gene is favored over the
cysteine version when a source of selenium is available. Although
selenium levels influenced the transcription of
fdhF
Cys
and
fdhF
Sec
,
it appeared to specifically prevent full-length transcription of
fdhF
Sec
. The transcriptional pattern of
fdhF
Sec
suggested that sele
-
nium regulation did not affect transcriptional initiation but in-
stead influenced transcriptional elongation downstream from the
promoter such that production of the full-length transcript is de-
pendent on the presence of the trace element in the medium. A
report by Liu et al. (17) in which the authors synthetically engi-
neered an attenuation mechanism based on the ability of bacterial
cells to incorporate a rare (unnatural) amino acid provides an
elegant example of how a transcription/translation coupling
mechanism might function within the open reading frame of a
gene, but such a system has not been reported to occur in nature.
In this study, we demonstrate that the naturally occurring sele-
nium regulation of
fdhF
Sec
transcription proceeds via a similar
mechanism of uncoupling of transcription from translation,
which explains our previous observation of the transcriptional
patterns of this gene.
To examine the influence of translation on transcription of
protein-encoding genes in
T. primitia
, we included the first appli-
cation, to our knowledge, of high-throughput sequencing to pre-
cisely track genome-wide transcriptional effects of uncoupling
translation from transcription. Our results demonstrate that the
availability of the trace element selenium influences transcription
of
fdhF
Sec
in
T. primitia
through a translational-coupled mecha
-
nism involving the incorporation of the noncanonical selenocys-
teine amino acid and provide additional details on the effect.
Moreover, our genome-wide transcriptional data show that while
translation does indeed broadly influence transcription of func-
tional genes, consistent with the results of Proshkin et al. (6),
translation alone does not necessarily or uniformly govern tran-
scription. Specific post-initiation transcriptional responses dur-
ing periods of limited protein synthesis may therefore depend on
translational coupling (an environmentally determined condi-
tion) as well as on intrinsic sequence features within protein-
encoding genes (a selectable and heritable trait). This could pro-
vide bacterial cells a means to specifically govern transcription of
certain genes under nonpermissive or weakly permissive condi-
tions for protein synthesis.
RESULTS AND DISCUSSION
Translation through the SECIS element increases transcriptional
elongation of
fdhF
Sec
.
Homoacetogens, like the spirochete
T. primi
-
tia
, use formate dehydrogenase enzymes to perform a vital function
in the gut microbial communities of termites. These bacteria use the
acetyl coenzyme A (acetyl-CoA) pathway for lithotrophic metabo-
lism via the reduction of CO
2
with H
2
, producing acetate as the end
product of this anaerobic respiration (18–21). In some termites, ace-
tate generated through the acetyl-CoA pathway meets ca. one-fifth to
one-third of the insect’s energy demand (22). In return, homoaceto-
genic bacteria enjoy near-saturating levels of hydrogen in the termite
gut, which fuels their metabolism (23).
T. primitia
possesses two
paralogous isozymes of formate dehydrogenase (encoded by
fdhF
Sec
and
fdhF
Cys
).
fdhF
Sec
encodes an enzyme that uses a catalytic seleno
-
cysteine, whereas
fdhF
Cys
encodes a selenium-independent enzyme
that instead uses a catalytic cysteine, presumably for periods of sele-
nium scarcity (11). Homologs of both isozymes are pervasive in the
gut communities of a wide variety of termite species, suggesting a role
for selenium dynamics in shaping the evolution of termite gut ho-
moacetogens (11, 24–26).
We previously investigated the influence of selenium on the
transcription of
fdhF
Cys
and
fdhF
Sec
in
T. primitia
(11). Those
studies showed that selenium limitation increases overall tran-
script levels of
fdhF
Cys
but decreases transcript levels of
fdhF
Sec
only near the 3
=
end of the gene. The results are consistent with
fdhF
Sec
transcription being initiated but prematurely terminated
or paused under growth conditions in which selenium is scarce.
Here we used quantitative reverse transcription-PCR (qRT-PCR)
to map the locations in
fdhF
Sec
where premature termination/
pausing may occur. Seven forward and reverse primer pairs were
designed to generate a “sliding window” of ca. 100-bp amplicons
(Fig. 1A) to measure transcript levels over a range of nucleotide
positions near the beginning of
fdhF
Sec
in
T. primitia
cultures
amended with selenium and in control cultures that were not
amended with selenium. To eliminate amplification bias among
the different primer sets, the signal of each amplicon in control
cultures was divided by the signal from the corresponding ampli-
con in selenium-amended cultures. A plot of these fractions as a
function of nucleotide position produces a curve in which ordi-
nate values of
1 indicate that the level of transcript is higher in
selenium-amended cultures than in control cultures (Fig. 1B).
These results show that
fdhF
Sec
mRNA is prematurely terminated/
paused with greater frequency in control cultures than in
Matson et al.
2
®
mbio.asm.org
November/December 2013 Volume 4 Issue 6 e00869-13
mbio.asm.org
on January 23, 2014 - Published by
mbio.asm.org
Downloaded from
selenium-treated cultures over a range of nucleotide positions be-
tween 300 and 700 bp downstream from the start of the open
reading frame. The inflection point of the curve centers on ampli-
cons that include portions of the SECIS element located from 438
to 481 bp downstream from the start of
fdhF
Sec
.
A mechanism by which a deficit of selenium uncouples trans-
lation from transcription in
fdhF
Sec
can account for our results. In
the absence of selenium (and thus selenocysteine), translating ri-
bosomes would likely stall or terminate translation at the seleno-
cysteine codon (UGA), thus uncoupling translation from tran-
scription. The stem-loop mRNA secondary structure formed by
the SECIS element (see Fig. S3 in the supplemental material)
might then act as a barrier to transcription (analogous to stem-
loop structures involved in attenuation of, for example, the
trp
operon), causing premature termination/pausing of the
fdhF
Sec
transcript. The range of nucleotide positions over which termina-
tion occurs could reflect imprecise termination at the SECIS ele-
ment combined with the process of 3
=
-to-5
=
mRNA degradation.
Our results contrast previous studies performed with the
selenocysteine-encoding
fdhF
gene in
E. coli
in which selenium
limitation was not observed to influence transcription down-
stream from the selenocysteine codon (9, 27), a result that we
confirmed in
E. coli
using our own quantitative RT-PCR assays.
One possible explanation for the differences between the two spe-
cies is that the SECIS element in
E. coli
may not effectively inhibit
transcription. Indeed, the SECIS element of
fdhF
in
E. coli
has a
different structure than the SECIS element of
fdhF
Sec
in
T. primitia
(11). Subtle mutations in the
E. coli
SECIS element lead to very
different levels of selenocysteine incorporation and overall trans-
lation (28). Furthermore, previous research has shown that
E. coli
is able to (though infrequently) misincorporate cysteine at the
selenocysteine codon (9), which would allow some translation to
occur past the UGA codon even in the absence of selenium.
If coupling and uncoupling of translation controls the level of
mRNA produced downstream from the
fdhF
Sec
selenocysteine
codon in
T. primitia
, then independent factors that inhibit trans-
lation should counteract the effect of selenium on transcription.
Furthermore, a translational coupling-and-uncoupling mecha-
nism could also poise the transcriptional machinery to respond
rapidly to conditions of sudden selenium availability. We have
previously observed that transcription of the selenocysteine trans-
lation elongation factor SelB is constitutive in
T. primitia
and is
not influenced by selenium levels (11), meaning that if
selenocystyl-tRNA can be produced, separate
de novo
protein syn-
thesis should not be required to incorporate selenocysteine into
nascent polypeptides.
We tested these hypotheses using quantitative RT-PCR to
measure transcript levels upstream and downstream from the sel-
enocysteine codon under two different conditions in
T. primitia
cultures. The cultures were grown without selenium and were
allowed to enter logarithmic growth phase (approximately 5 days
postinoculation). Under these conditions, transcription of
fdhF
Sec
downstream from the selenocysteine codon is inhibited. One cul-
ture was subsequently amended with selenium, while the other
was similarly amended with selenium and concurrently treated
with tetracycline (10
gml
1
), a potent inhibitor of protein syn
-
thesis in spirochetes (29). At this tetracycline dose,
T. primitia
cells
did not divide, yet they remained motile for at least 72 h posttreat-
ment (see Fig. S4 and Movies S1 and S2 in the supplemental ma-
terial), indicating that the tetracycline treatment did not kill the
cells but effectively inhibited protein synthesis long past the time
period of the study.
The fraction of transcript produced downstream compared to
upstream of the selenocysteine codon demonstrated that the ad-
dition of selenium had a dramatic positive effect on transcript
levels downstream of the selenocysteine codon (Fig. 2A). After a
short lag, the
fdhF
Sec
transcript level downstream of the selenocys
-
teine codon in the culture treated only with selenium increased to
an approximately half-maximal level 15 min posttreatment and
reached a near-maximal level after 30 min. This transcriptional
response can be considered rapid compared to growth rate (30), as
the generation time of
T. primitia
cultures grown under these
conditions is more than 24 h (see Fig. S4 in the supplemental
material). In contrast, there was no increase in downstream
fdhF-
Sec
transcript levels in the culture that was treated with selenium
and tetracycline concurrently. These results show that while sele-
nium availability rapidly increases transcriptional productivity,
protein synthesis is ultimately required for downstream transcrip-
FIG 1
Influence of selenium on transcription of
fdhF
Sec
over gene position
and time. (A) Location of qRT-PCR primers used for measuring transcript
levels. “TGA” marks the position of the selenocysteine codon (amino acid
145). The SECIS element for selenocysteine incorporation occupies nucleotide
positions 3 to 46 bp downstream from this codon. Locations of amplicons
generated by primers used for mapping transcriptional termination/pausing
are indicated by horizontal bars. (B) Influence of selenium (50 nM sodium
selenate) on transcript levels over a range of nucleotide positions in
fdhF
Sec
.
Symbols indicate the relative signal intensities of measurements from controls
divided by those of treated cultures and correspond to the 3
=
positions of
amplicons indicated in panel A. The individual results of three biological rep-
licates are plotted; each data point is the average of duplicate qRT-PCR mea-
surements. The dashed line is the trend line through the approximate average
of these measurements. The third and fourth measurements (located between
400 and 600 bp downstream from the start of
fdhF
Sec
) include portions of the
SECIS element.
Translational Uncoupling Effect on Transcription
November/December 2013 Volume 4 Issue 6 e00869-13
®
mbio.asm.org
3
mbio.asm.org
on January 23, 2014 - Published by
mbio.asm.org
Downloaded from
tion of
fdhF
Sec
in
T. primitia
, a result that is consistent with the
conclusions of Proshkin et al. regarding the general requirement
of translation for transcriptional elongation (6).
We interpret our results to mean that a selenium deficit in
T. primitia
cells prevents downstream transcription of
fdhF
Sec
by
uncoupling translation from transcription at the selenocysteine
codon, a condition that is similarly achieved by stopping transla-
tion altogether. However, tetracycline did not markedly influence
the level of transcript produced upstream from the selenocysteine
codon. This was unexpected because, whereas a lack of selenium
would stop translation specifically at the selenocysteine codon,
tetracycline should prevent translation at or near the beginning of
the open reading frame. These results also suggest that selenium is
not responsible for inactivating potential RNases involved in de-
grading the 3
=
end of the transcript when selenium is not pro-
vided. Raw qRT-PCR data (see Fig. S5 in the supplemental mate-
rial) reveal that absolute levels of 5
=
transcription do not change in
cultures treated with selenium alone or concurrently treated with
selenium and tetracycline. Therefore, the rise in the fraction of
full-length
fdhF
Sec
transcript is due to production of mRNA
downstream from the selenocysteine codon. We observe no de-
crease in low levels of 3
=
transcript in cultures cotreated with sele-
nium and tetracycline. This probably reflects the fact that a mini-
mal but steady amount of 3
=
fdhF
Sec
mRNA is being produced by
T. primitia
under the conditions tested.
One possibility that could account for the selenium-induced
increase in 3
=
-to-5
=
transcript levels in Fig. 2A is that tetracycline
may have prevented
T. primitia
from translating a protein factor
necessary for a transcriptional response in
fdhF
Sec
to selenium. To
test this, a complementary experiment was performed wherein
tetracycline was added to
T. primitia
cells grown under conditions
that promote a high level of 3
=
fdhF
Sec
transcription (i.e.,
selenium-replete conditions) and suddenly forced into
translation-limiting conditions by the addition of tetracycline
(Fig. 2B). This allowed us to test whether or not a lack of transla-
tion directly arrests transcription of
fdhF
Sec
. We found that adding
tetracycline decreased the fraction of 3
=
fdhF
Sec
transcripts even
though selenium levels remained unchanged; however, the 3
=
fraction never reached the low level initially observed in Fig. 2A,
over the 6-h time period measured. As the abundance of full-
length transcripts depends on both production and degradation
of RNA, this decrease in 3
=
transcript levels is likely due to the role
of RNases in 3
=
mRNA degradation of older
fdhF
Sec
transcripts
that were transcribed under permissive conditions, while new
transcripts, which we presume are not transcribed beyond the
SECIS element, are no longer able to compete with RNase degra-
dation beyond this transcriptional elongation stopping point.
So do selenium limitation and exposure to tetracycline affect
transcription in comparable ways? To address this question, we
performed high-throughput RNA sequencing (RNA-Seq) of
T. primitia
cells during log-phase growth in three cultures initially
grown without selenium; one culture was subsequently treated
with selenium alone, one was treated with selenium plus tetracy-
cline, and a control culture received neither treatment. RNA was
extracted and processed from these cultures 6 h posttreatment, a
time point selected to allow sufficient time for transcriptional dif-
ferences between antibiotic-treated and nontreated cultures to ac-
cumulate (Fig. 2B). RNA-Seq was performed in short-read mode
(37 bp, nonpaired reads), allowing read depth to be used as a
measure of transcript level (31). This approach was used to eval-
uate the relationship between translation and transcription in
fdhF
Sec
by precisely mapping the location of transcriptional termi
-
nation under each of these conditions.
All samples produced comparable RNA-Seq data set libraries
(Table 1; also, see Materials and Methods). To limit the number of
perturbations due to sample handling, no attempt was made to
remove rRNA. We normalized genome-wide read depth in
selenium-treated and selenium- and tetracycline-treated samples
to that of the control sample based on small (ca. 10%) differences
in the read depth for rRNA transcripts; however, our results were
not markedly influenced by this normalization. To enhance detec-
tion of features along the genome, data were averaged over a mov-
ing average of 200-bp regions using the MATLAB computational
software package. While RNA-Seq data show that read depth dra-
FIG 2
General translation decoupling influences transcription of
fdhF
Sec
in a
manner analogous to that of selenium limitation. Levels of downstream and
upstream transcript in
fdhF
Sec
were measured by qRT-PCR. (A) Influence of
selenium on the transcriptional level upstream and downstream from the sel-
enocysteine codon (and SECIS element) as a function of time. Relative signal
intensities are reported as the fraction of signal downstream from the seleno-
cysteine codon divided by the signal upstream. Cultures were initially grown
without selenium. The arrow marks the time at which selenium was provided
to both cultures. At the same time point, tetracycline was administered to the
treatment culture but not the control. (B) The
T. primitia
culture was grown in
the presence of selenium (50 nM sodium selenite) for approximately 4 days
prior to tetracycline treatment. Unlike conditions under which selenium is not
initially added to
T. primitia
cultures, this growth condition promotes high
levels of transcription downstream from the selenocysteine codon. At time
zero, tetracycline (10
gml
1
) was administered to stop translation.
Matson et al.
4
®
mbio.asm.org
November/December 2013 Volume 4 Issue 6 e00869-13
mbio.asm.org
on January 23, 2014 - Published by
mbio.asm.org
Downloaded from
matically decreases over a similar range of nucleotide positions in
fdhF
Sec
in all three cultures, the culture treated with selenium
alone has a much greater read depth downstream of the seleno-
cysteine codon than either the selenium- and tetracycline-treated
or the untreated control culture (Fig. 3). This result is consistent
with qRT-PCR data. Apart from a slightly elevated read-depth
upstream from the selenocysteine codon in the selenium- and
tetracycline-treated sample, the transcriptional pattern resulting
from treating cells concurrently with tetracycline and selenium
was similar to the nontreated control. Thus, inhibiting protein
synthesis altogether in
T. primitia
cells indeed affects transcription
downstream from the selenocysteine codon in
fdhF
Sec
in a way
that is equivalent to the effect of selenium limitation.
The most simple and parsimonious explanation for these re-
sults involves a combination of translational uncoupling and tran-
scriptional inhibition by mRNA secondary structure. Our results
support the conclusion that uncoupling translation from tran-
scription alone is not sufficient to inhibit transcription, as the 5
=
transcript level of
fdhF
Sec
remains high in the face of translational
uncoupling. Instead, transcriptional truncation is likely the effect
of this uncoupling (either generally via the effect of tetracycline or
specifically via selenium limitation) combined with RNAP en-
countering a particular mRNA secondary structure that prema-
turely terminates or pauses transcription.
While many other mRNA structures are predicted to form in
fdhF
Sec
(32), the conspicuous location of mRNA truncation sug
-
gests that the SECIS element may act as a potent terminator of
transcription under all of the conditions tested (Fig. 3). In the case
of cultures treated with selenium alone, some premature tran-
scriptional termination or pausing near the SECIS element could
be due to inefficient translation of the selenocysteine codon under
our growth conditions. However, so long as translation is coupled
to transcription, RNAP can proceed with higher throughput and
thus synthesize greater amounts of full-length
fdhF
Sec
mRNA.
These results show that the trace element selenium can control
transcription through a translational coupling mechanism.
Far from being a phenomenon found in an auxiliary bacterial
gene, this mechanism occurs in an enzyme that is essential to the
central metabolism of
T. primitia
and important for the mutual-
istic relationship it shares with its host. As such, a transcription/
translation coupling mechanism that responds rapidly to sele-
nium availability could be highly advantageous in responding to
environmental selenium dynamics.
Selenium availability effects transcription of many
T. primi-
tia
genes.
We extended our observations on the effects of sele-
nium to genome-wide transcriptional changes that occur in
T. primitia
by mapping RNA-Seq reads to the closed genome
(NCBI reference sequence NC_015578.1). As we intentionally did
not enrich for mRNA in this study, our approach reduces the
number of genes that can be analyzed because the vast majority
(approximately 95%) of RNA-Seq reads map to rRNAs (Table 1).
Even with this limitation, these data allow us to reconstruct gen-
eral effects of limiting selenium availability on transcriptional pat-
terns in
T. primitia
cells (Table 2; also, see Table S2 in the supple-
mental material).
Selenium availability has a large influence on transcription in
T. primitia
. At least 16 genes are upregulated in response to ele-
TABLE 1
Overview of RNA-Seq output
Culture
No. of:
Total sequences
Mapped sequences
Sequences mapping to non-16S or 23S genes
Selenium treated
13,234,333
11,785,615
413,300
Selenium
tetracycline treated
12,158,875
10,990,439
489,708
Control
13,623,479
11,878,141
511,495
FIG 3
Transcriptional map of
fdhF
Sec
by RNA-Seq. Three cultures were
initially grown without selenium. One culture was subsequently amended with
selenium alone (50 nM sodium selenite) 6 h prior to sampling (blue line).
Another culture was subsequently amended with selenium and concurrently
with tetracycline 6 h prior to sampling (red line). The control culture received
neither treatment (gray line). The lines trace the moving average of Illumina
RNA sequencing read depth over a 200 bp sliding-window of nucleotide po-
sitions. The black arrow on the
x
axis indicates the
fdhF
Sec
open reading frame.
The approximate selenocysteine codon location is indicated (TGA).
TABLE 2
Genome-wide transcriptional changes in
T. primitia
a
Treatment
No. of genes
Upregulated
b
Downregulated
c
Selenium
16
98
Selenium plus
tetracycline
228
453
a
Out of a total of 3,838 predicted protein-encoding genes, 1,495 genes that met a cutoff
of
50 reads per kb for at least one of the conditions were analyzed.
b
Number of genes for which there is a
2-fold-higher transcript level compared to the
control.
c
Number of genes for which there is a
2-fold-lower transcript level compared to the
control.
Translational Uncoupling Effect on Transcription
November/December 2013 Volume 4 Issue 6 e00869-13
®
mbio.asm.org
5
mbio.asm.org
on January 23, 2014 - Published by
mbio.asm.org
Downloaded from
vated selenium levels. Upregulated genes include
fdh
Sec
, encoding
the selenium-dependent copy of formate dehydrogenase, and sev-
eral genes encoding components of hydrogenase enzymes that
putatively supply formate dehydrogenase enzymes with electrons
derived from hydrogen to carry out the reduction of CO
2
(33).
The overall pattern of upregulated genes is consistent with
T. primitia
cells responding to selenium by increasing their rate of
acetogenesis and growth enabled by the production of the seleno-
cysteine form of formate dehydrogenase, which is predicted to
have a higher catalytic rate than the selenium-independent form
of the enzyme (13). In addition, over 100 downregulated genes
were distributed throughout the genome of
T. primitia
under
high-selenium conditions (see Table S2 and Fig. S2A in the sup-
plemental material). Viewed another way, these genes are upregu-
lated upon removal of selenium in cultures of
T. primitia
cells
growing under selenium-replete conditions.
The large number of metabolic genes preferentially expressed
under low-selenium conditions was expected. As formate dehy-
drogenase is critical to the central metabolic pathway of
T. primi-
tia
, we interpret these results to mean that numerous genes are
needed to compensate for a reduced capacity for acetogenic me-
tabolism due to a deficiency in the preferred
fdhF
isoenzyme. Con-
sistent with this hypothesis was the finding that, in addition to
numerous hypothetical proteins of unknown function, genes re-
lated to motility, stress response, and acquisition of alternative
nutrient sources were found to be transcribed at lower levels when
the availability of selenium is high (see Table S2 in the supplemen-
tal material). These results suggest that life is more challenging for
T. primitia
without selenium than when selenium is available.
Translational inhibition has a broad and varied effect on
transcription.
We observed an unexpectedly wide variety of tran-
scriptional responses to globally inhibiting translation. Many
genes are transcribed at lower levels in tetracycline treated cells
(Table 2), an expected result based on the documented influence
of translation on transcription. As with
fdhF
Sec
, we observed many
genes for which tetracycline-mediated translation inhibition
causes premature truncation of transcription (for example, see
Fig. 4A to C). However, these genes are often part of multigene
operons and are located far downstream from putative promoters.
The majority of genes we analyzed are not markedly influenced
(
50% increase or decrease in transcript levels) by a lack of trans-
lation (Table 2; also see, for example, Fig. 4D to F). We also doc-
umented numerous genes for which inhibition of translation in-
creases transcriptional levels over the length of the genes (Table 2;
also see, for example, Fig. 4G to I). These observations are not
limited to individual genes but extend to genes in putative
T. primitia
operons where similar trends occur over several thou-
sand base pairs (Fig. S1). The list of genes that, on average, dem-
onstrate an increase or decrease in expression in response to tet-
racycline is quite large and is presented in Table S3 in the
supplemental material. The genes that are up- or downregulated
are distributed throughout the genome and are not localized to
any easily defined hot spots on the chromosome of
T. primitia
(see
Fig. S2B in the supplemental material).
It appears to be the case in
T. primitia
that translation has a
variable effect on transcription. In essence, translation may act as
a transcriptional regulatory governor that is capable of decreasing,
increasing, or not influencing transcriptional levels of specific
genes. Transcriptional responses of many
T. primitia
genes are
dependent on translation under conditions of limited protein syn-
thesis, but these effects may also involve mRNA secondary struc-
ture determined by the particular nucleotide sequence of the gene.
Such relationships between translation and transcription would
imply that mRNA secondary structure could help to control the
transcription of genes and would allow certain genes to be tran-
scribed even though translation is severely limited or has alto-
gether ceased. Transcriptional effects caused by the presence and
location of rare codons in some genes (34), as well as the place-
ment of Shine-Dalgarno-like sequences within an open reading
frame (35) and protein factors such as the NUS factors and Rho,
may allow additional layers of transcriptional responses to occur
during specific translational challenges. We envision a scenario in
which several of these factors act together in order to finely regu-
late the strength and length of transcription. Support for one such
cooperative interaction involving mRNA loops with RNA poly-
merase core enzyme and elongation factors was proposed recently
based on crystal structures of an RNAP “paused” transcriptional
complex (36). In such complexes, different NUS proteins are
thought to either stabilize or destabilize an RNA structure-
mediated clamp confirmation that can be paused or processive,
and the presence of hairpins is thought to stabilize the duration of
the pause (36).
A large subset of
T. primitia
genes are transcribed even when
protein synthesis is inhibited. While the transcription of some
genes is initiated at slightly higher levels, possibly leading to mul-
tiple RNAPs acting in consort and helping to facilitate transcrip-
tional processivity, this is not necessarily the case. Certain genes
may instead be preselected for preferential transcription under
conditions that severely limit protein synthesis by the intrinsic
nature of their nucleotide sequences. Perhaps it is ultimately a lack
of a particular secondary structure that allows some genes to be
highly transcribed during periods of inhibited protein synthesis.
Prior literature on attenuation-regulated genes espouses the view
that a specific subset of genes is selected to be preferentially tran-
scribed under conditions where translation is inhibited (1). It is
possible that the scope of gene transcription influenced by trans-
lation is even larger than envisioned and that the regulation is
ordered in strength from genes that are the most highly upregu-
lated through those that are mildly upregulated to those that are
not influenced by translation or are even repressed by it. Genes
that are upregulated in the absence of translation in
T. primitia
(Table S3) include many translation-related genes, such as ribo-
somal machinery, amino acid uptake, and classical attenuation-
prone tRNA and amino acid synthesis genes. In addition to these
targets, the transcription of several stress response genes is also
upregulated under translation-limiting conditions.
While mRNA degradation certainly plays a role in many of the
transcriptional patterns observed using qRT-PCR and RNA-Seq
approaches, we believe that changes in transcriptional processivity
account for the genome-wide changes in transcriptional patterns
that we observed, because numerous genes (and putative operons)
were not influenced by inhibiting translation. It could also be the
case that mRNA secondary structure serves to inhibit 3
=
RNA
degradation at certain locations within genes, but we believe that
such a phenomenon alone does not account for our results and
would provide no particular advantage to the cells.
Whatever the exact mechanisms that are responsible for these
responses may be, we hypothesize that the various links between
translation and transcription allow evolutionary processes to act
upon certain genes such that transcription is constant or even
Matson et al.
6
®
mbio.asm.org
November/December 2013 Volume 4 Issue 6 e00869-13
mbio.asm.org
on January 23, 2014 - Published by
mbio.asm.org
Downloaded from