In this article, we show that two transcription units sharing the same promoter, terminator and reporter open read-ing frame ORF, but differread-ing in the length of their 3¢-untranslate
Trang 1of gene length-dependent processes in yeast mRNA
biogenesis
Macarena Morillo-Huesca, Manuela Vanti and Sebastia´n Cha´vez
Departamento de Gene´tica, Universidad de Sevilla, Seville, Spain
Gene expression is a multistep process involving
tran-scriptional and post-transcriptional events RNA
polymerase II-dependent transcription starts by the
assembly of the pre-initiation complex (PIC), followed
by the initiation step After initiation, transcription
elongation is coupled with a set of RNA modifications
(capping, splicing and polyadenylation) occurring
along the transcription unit Transcription termination
is connected to the RNA cleavage required for
tran-script polyadenylation Formation of mRNP, the
mRNA–protein complex transportable to the
cyto-plasm, is also linked to transcription elongation [1]
The traditional view of transcription and RNA
pro-cessing as separate events has been replaced by mRNA
biogenesis as a new concept involving a complex net of functional interactions between RNA processing and the different steps of transcription [2,3]
Some elements of the gene expression machinery play a role at the initial steps of mRNA biogenesis whereas some others act all along the transcription unit Among the former, we find the general transcrip-tion factors involved in PIC assembly, initiatranscrip-tion and early elongation [4,5] The mechanisms of transcrip-tional regulation of gene expression that take place during these early events have been extensively studied and are fairly well understood [6,7] The latter set of elements is formed by those factors interacting with RNA polymerase II all along transcription elongation
Keywords
gene length; mRNA biogenesis; reporter
system; Saccharomyces cerevisiae;
transcription elongation
Correspondence
S Cha´vez, Departamento de Gene´tica,
Universidad de Sevilla, Facultad de Biologia,
Avda ⁄ Reina Mercedes, 6, E41012-Sevilla,
Spain
Fax: + 34 954557104
Tel: +34 954550920
E-mail: schavez@us.es
Enzymes
Acid phosphatase (EC 3.1.3.2)
(Received 26 September 2005, revised
14 December 2005, accepted 16 December
2005)
doi:10.1111/j.1742-4658.2005.05108.x
We have developed a simple reporter assay useful for detection and analysis
of mutations and agents influencing mRNA biogenesis in a gene length-dependent manner We have shown that two transcription units sharing the same promoter, terminator and open reading frame, but differing in the length of their 3¢-untranslated regions, are differentially influenced by muta-tions affecting factors that play a role in transcription elongation or RNA processing all along the transcription units In contrast, those mutations impairing the initial steps of transcription, but not affecting later steps of mRNA biogenesis, influence equally the expression of the reporters, inde-pendently of the length of their 3¢-untranslated regions The ratio between the product levels of the two transcription units is an optimal parameter with which to estimate the efficiency of gene length-dependent processes in mRNA biogenesis The presence of a phosphatase-encoding open reading frame in the two transcription units makes it very easy to calculate this ratio in any mutant or physiological condition Interestingly, using this assay, we have shown that mutations in components of the SAGA complex affect the level
of mRNA in a transcript length-dependent fashion, suggesting a role for SAGA in transcription elongation The use of this assay allows the identifica-tion and⁄ or characterization of new mutants and drugs affecting transcrip-tion elongatranscrip-tion and other related processes
Abbreviations
GLAM, gene length-dependent accumulation of mRNA; MPA, mycophenolic acid; ORF, open reading frame; PIC, pre-initiation complex; 3¢-UTR, 3¢-untranslated regions; SAGA, Spt-Ada-Gcn5-Acetyltransferase.
Trang 2[8,9] The relative contribution of this second kind of
element to gene expression is expected to depend on
the length of the transcription unit, it being greater in
longer genes than in shorter ones
The distinction between these two types of elements
is not always easy when analyzing a mutant or a
physiological agent affecting gene expression Nuclear
run-on allows measurement of transcription elongation
efficiency along the transcription unit [10], but it is
time-consuming and not easy to carry out in a high number
of samples In addition, the information obtained by
nuclear run-on shows the location of active polymerases,
but it gives no information about the quality of the
mRNA that is being synthesized This also happens
when RNA polymerase II location within the
transcrip-tion unit is analyzed by chromatin
immunoprecipita-tion; although, in this case, it is possible to distinguish
between those elements influencing transcription
elongation rate and those involved in processivity [11]
Other in vivo assays, such as sensitivity to 6-azauracil or
to mycophenolic acid, have been used in yeast to detect
transcription elongation defects, and they are easy to
perform [12]; however, those assays are too indirect to
obtain solid conclusions [13] In vitro assays are useful
for defining the role of a given element in a specific
step of gene expression (e.g transcription initiation,
transcription elongation or splicing), but they are also
time-consuming and not recommended as the first assay
with which to classify a mutant
The best assay for a rapid evaluation of gene
expres-sion is the use of a reporter system Available reporter
systems allow detection of defects in gene expression,
but cannot distinguish between an impairment of the
initial steps of transcription, including promoter
regu-lation, and an effect on the subsequent events of
mRNA biogenesis
We have shown elsewhere that expression of long
transcription units in Saccharomyces cerevisiae is more
sensitive to mutations affecting the
Tho2-Hpr1-Mft1-Thp1 (THO) complex [14], connected to transcription
elongation and mRNP formation [15] Taking the
dependence on the length of the transcription unit as a
criterion, we have developed a reporter assay useful
for detecting mutations and agents influencing mRNA
biogenesis all along the transcription unit In this
article, we show that two transcription units sharing
the same promoter, terminator and reporter open
read-ing frame (ORF), but differread-ing in the length of their
3¢-untranslated regions (3¢-UTR), are differently
influ-enced by mutations affecting factors that play a role in
mRNA biogenesis all along transcription elongation
In contrast, those mutations exclusively impairing the
initial steps of transcription influence equally the
expression of the reporters, independently of the length
of their 3¢-UTR
Results
An in vivo assay to measure gene length-dependent efficiency of mRNA accumulation
We constructed several plasmids containing the PHO5 coding region transcribed under the control of the GAL1 promoter, but differing in the length of their 3¢-UTR To increase the length of the 3¢-UTR we inserted Escherichia coli lacZ (either a short fragment
or the entire gene in both orientations) or its eukaryotic homolog Kluyveromyces lactis LAC4 These two sequences are equally large but differ in G + C and chromatin organization (see Discussion) Figure 1(A) shows the relative length of the 3¢-UTR of every tran-scription unit used in this work
The ability of the different transcription units to produce full-length mRNA was tested by performing northern analysis on cultures of the respective trans-formants grown in a galactose-containing medium In all cases, a unique transcript of the expected length was detected (Fig 1B) The intensity of the mRNA signals corresponding to the wild-type cells inversely correlated to the length of the transcription units (see lanes 1–5 in Fig 1B) This result does not necessarily mean that the shortest transcripts (PHO5 alone or PHO5::lacZD) accumulated at a higher level than the longest ones; length also might have influenced the yield of mRNA extraction from the cells and the effi-ciency of transference during blotting This is likely to
be the reason for the variability observed when the results of independent northern experiments are com-pared (see below) In any case, the quantification of the northern blots indicated that the three longest transcription units showed very similar levels of mRNA accumulation (Fig 1B), despite the sequence
of its 3¢-UTR being different (E coli lacZ in both ori-entations or K lactis LAC4)
We hypothesized that those mutations affecting transcription elongation all along the transcription unit should produce a negative length-effect on mRNA accumulation, which was stronger than the one that the wild type might exhibit In order to test this hypo-thesis, we chose the spt6–140 allele, a thermosensitive mutation affecting a bona fide transcription-elongation factor that colocalizes with RNA polymerase II along the transcription unit [16–18] As shown in Fig 1(B), even at permissive temperature, the signals of the three long transcripts were severely reduced in the mutant strain compared with the wild type, whereas the signals
Trang 3of the short transcripts suffered milder effects To
bet-ter estimate the effect of length, we calculated the
ratios between the mRNA signals of a strain
expres-sing the long transcription units vs the mRNA signal
of the same strain expressing just PHO5 (Fig 1C)
This long-⁄ short-transcript ratio was clearly lower in
spt6–140than in the wild type, even when the mRNA
containing the 0.3 kb fragment of lacZ was considered
the ‘long’ one However, this ratio was dramatically
reduced in the mutant when applied to the longest
transcription units, being around 5 times lower than in
the wild type (Fig 1C)
PHO5 encodes a periplasmic acid phosphatase, which is very easy to assay [19] The levels of acid phosphatase activity of an untransformed S cerevisiae strain in SC-galactose medium are almost undetecta-ble, as the endogenous PHO5 gene is repressed in media containing high levels of phosphate [20] As expected, the levels of phosphatase activity in strains lacking our reporter systems were extremely low (not shown) Therefore, the levels of acid phosphatase expressed by our transformants should reflect the amounts of their respective plasmid-encoded PHO5 mRNAs accumulated in the cell Nevertheless, in order
A
B
Fig 1 Influence of gene-length on mRNA accumulation (A) Transcription units used in this work, corresponding to plasmids pSCh202, pSCh212–18, pSCh212, pSCh211 and pSCh209-LAC4 (B) Northern blot showing the mRNA levels of the five transcription units described in (A) in a wild-type (MMY5.1) and an isogenic spt6–140 strain (MMY5.2) Values were normalized with respect to 25S rRNA Three independ-ent experimindepend-ents were averaged (C) Relative values of the mRNA levels shown in (B) with respect to the shortest transcript (PHO5 mRNA) (D) GLAM ratios of congenic wild-type and spt6–140 strains: relative levels of acid phosphatase activity expressed by the indicated transcrip-tion unit, with respect to the acid phosphatase activity from the shortest transcriptranscrip-tion unit (PHO5) Averages of four wild-type strains (MMY5.1 ⁄ 4 ⁄ 6 ⁄ 7) and four spt6–140 strains (MMY5.2 ⁄ 3 ⁄ 5 ⁄ 8) are shown For each strain, the average of at least three experiments was considered Error bars indicate standard errors.
Trang 4to reduce the error caused by the residual expression
of the endogenous PHO5, we subtracted the residual
acid phosphatase activities of the untransformed
strains to all the following results (see Experimental
procedures) We assayed acid phosphatase activity in
eight congenic strains transformed with our five
plas-mids: four of the strains being wild-type for SPT6 and
four of them having a spt6–140 allele The average of
the acid phosphatase activities was used to calculate
the ratios between those cells expressing the long
tran-scripts and those expressing the minimal PHO5
mRNA Figure 1(D) shows the calculated ratios When
we focused on the wild type, the comparison of the
acid phosphatase ratio calculated for PHO5::lacZD to
the ratios calculated for the three longest transcription
units did not indicate a significant influence of length
on Pho5 accumulation As previously suspected, this
result suggests that the apparent effect of length on
mRNA accumulation in the wild type may be the
con-sequence of a technical bias against long mRNAs
during northern experiments Nevertheless, and in
agreement with the mRNA ratios shown in Fig 1(C),
the acid phosphatase ratios of the spt6 mutant were
again clearly lower than those corresponding to the
wild type, with the difference being stronger when the
long transcript corresponded to any of the three
lon-gest 3¢-UTRs (Fig 1D) Moreover, the absolute values
of the acid phosphatase ratios are highly reproducible,
whereas the mRNA ratios are consistent within a
sin-gle northern experiment, but show a high variability
when comparing different experiments, probably due
to differences in mRNA extraction and⁄ or mRNA
transfer during blotting We concluded that
measure-ment of acid phosphatase activity was the best
estima-tion of the mRNA abundance in our systems, and
from here on we use the ratio of acid phosphatase
activities as an indicator of gene length-dependent
accumulation of mRNA (GLAM)
Reduced GLAM ratios in transcription-elongation
mutants
To further confirm that the GLAM ratio is a valid
parameter with which to detect defects in mRNA
bio-genesis, we extended this assay to a set of well-known
mutants affected in transcription elongation
Three spt16–197 strains, affected in one of the
sub-units of the yFACT complex [21], clearly showed
lower GLAM ratios than three congenic wild types
(Fig 2A) The mRNA ratios of a spt16–197 strain and
a congenic wild type confirmed the reliability of the
phosphatase results (Fig 2B) A similar pattern of
GLAM ratios was obtained when comparing an spt4D
mutant [17,22], lacking one of the subunits of the yD-SIF complex, with an isogenic wild type (Fig 2C) Since the length effect was especially clear when using the transcription units containing the longest 3¢-UTR,
we chose lacZ and GAL1pr::PHO5-LAC4 for the following assays The mRNA ratios cal-culated for these two transcription units in the spt4D strain and in the isogenic wild type confirmed again the validity of the GLAM ratio to predict elongation defects (Fig 2D)
Four additional mutants affected in transcription elongation showed decreased GLAM ratios compared with an isogenic wild type: rpb9D, a 6-azauracil-sensi-tive mutant lacking a subunit of RNA polymerase II [23]; leo1D and rtf1D, two mutants lacking subunits of the PAF-complex [24,25]; and elp3D, a mutant lacking the histone acetyltransferase subunit of the elongator [26] (Fig 2C) In all these cases, the GLAM ratios of the mutants were, at most, 50% of the wild-type ratios and, therefore, we consider this percentage as the threshold value to indicate a deficiency in gene length-dependent mRNA biogenesis
TFIIS, encoded by the DST1⁄ PPR2 gene in S cere-visiae, is a very well known transcription elongation factor involved in releasing RNA polymerase II from arrest sites by stimulating RNA cleavage [27] Although the role of TFIIS in elongation has always been connected to specific pause-sites, and it has never been shown to play a general role in RNA polymerase II-dependent transcription, we performed our assay in
a dst1D mutant The GLAM ratios in a dst1D mutant did not show a significant difference with respect to an isogenic wild-type strain when GAL1pr::PHO5-LAC4 was considered the long transcription unit, and only showed a weak decrease when GAL1pr::PHO5-lacZ was chosen (Fig 3A) We repeated the assays in the presence of sublethal doses of the NTP-depleting drug mycophenolic acid (MPA), a drug that has been repor-ted to enhance the transcriptional defects of dst1D mutants by reducing RNA polymerase II processivity [11] As is expected for an inhibitor of transcription elongation, the presence of 2 lgÆmL)1 MPA reduced the GLAM ratio of the wild-type strain (Fig 3A) Similar results were obtained with 6-azauracil (not shown), a drug that also causes depletion of the NTP pools, confirming the suitability of this assay for ana-lyzing elongation-inhibiting drugs However, MPA, even at 10 lgÆmL)1, did not decrease further the GLAM ratio of the dst1D mutant (Fig 3A)
As these results separate TFIIS from the other elon-gation factors studied in this work, we performed nor-thern blot experiments to confirm this negative result based on acid phosphatase data Figure 3(B) shows a
Trang 5representative northern experiment which illustrates
the absence of a dst1D effect on GLAM; neither
the PHO5-lacZ⁄ PHO5 mRNA nor the PHO5-LAC4 ⁄
PHO5ratio was affected by dst1D Moreover, the
pres-ence of MPA significantly reduced the mRNA levels
present in dst1D, but the effect on the long
transcrip-tion units was proportranscrip-tionally similar to the effect
caused on the minimal PHO5 mRNA, therefore
pro-ducing similar ratios in the wild type and in the
mutant strain (Fig 3C) Our results confirm that
TFIIS is not playing a general role in mRNA
biogen-esis all along the transcription unit
GLAM ratios are not affected by the impairment
of PIC assembly or transcription initiation With the exception of dst1D, all transcription elonga-tion mutants assayed so far showed reduced GLAM ratios One possible explanation for these results is an indirect effect on the GLAM ratios of any major impairment of transcription, regardless of the step of mRNA biogenesis where it occurs To rule out this possibility, we assayed a wide variety of mutants affec-ted in PIC assembly and transcription initiation We analyzed a TBP mutant [28]; a mot1 mutant, affected
A
B
C
D
E
Fig 2 Mutants affected in transcription elongation factors show reduced GLAM ratios (A) Averaged GLAM ratios (see legend of Fig 1) of three wild types (MMY11.3 ⁄ 8 ⁄ 12) and three congenic spt16–197 strains (MMY11.4 ⁄ 6 ⁄ 10) (B) Northern blot and mRNA-ratios (see legend
of Fig 1) of a wild-type (MMY11.3) and a congenic spt16–197 strain (MMY11.4) Notation of transcription units as in Fig 1(A) (C) GLAM ratios of the spt4D strain Y06986 and the isogenic wild-type BY4741 (D) Northern blot and mRNA ratios of Y06986 and BY4741 E GLAM ratios of rpb9D, leo1D, rtf1D, elp3D and an isogenic wild type (strains Y04437, Y02379, Y04611, Y02742 and BY4741).
Trang 6in one of the main TBP regulators [29]; a toa1 mutant,
affected in the large subunit of yTFIIA [29]; two sua7
mutants, affected in yTFIIB [30]; two tfa1 mutants,
affected in the large subunit of yTFIIE [31]; and two
mutants, srb10D and srb11D, lacking subunits of the
cyclin–kinase complex that negatively regulates
tran-scription initiation [32,33] None of them showed a
sig-nificant effect on the GLAM ratios, when compared
with the isogenic wild-type strains (Fig 4A–E) Only
toa1–18 showed a slightly reduced GLAM ratio when
PHO5-lacZ was used as the long transcript, but not
when PHO5-LAC4 was considered, excluding a general
effect of toa1–18 on GLAM (Fig 4B) Taking together
all the results shown in Fig 4, we concluded that the
GLAM ratios are not affected by alterations of PIC
assembly or transcription initiation
Modification of the chromatin structure is an
important requirement for transcription regulation and
PIC assembly at many promoters One of the main
factors involved in this process is the SAGA complex
[34], up to now mainly connected to the initial steps of
transcription We assayed our reporters in two mutants
affecting subunits of SAGA Both gcn5D, a mutant
lacking the histone acetyltransferase present in SAGA,
and spt3D showed reduced GLAM ratios (Fig 5A) As
this result suggests a role of SAGA in mRNA
biogen-esis all along the transcription unit, we performed
nor-thern blot experiments to confirm the acid phosphatase
data As shown in Fig 5(B), the accumulation of long
mRNAs was more sensitive to the gcn5D mutation
than the accumulation of the minimal PHO5 transcript
rendering, thus significantly lower ratios
One possible explanation for these results is an
indi-rect effect on the GLAM ratios by any mutation
gen-erating abnormal chromatin structures In order to test
this hypothesis, we analyzed a wide set of mutants
including deletions of histone genes like hta1htb1D [35]
or htz1D [36]; mutants affected in nucleosome
remode-ling like isw1D [37], chd1D [38] or swr1D [39]; and an
rpd3D mutant, lacking the main histone deacetylase involved in transcription [40] No significant decrease
of the GLAM ratios were observed in any mutant,
A
B
C
Fig 3 dst1D, even in the presence of mycophenolic acid, does not
show reduced GLAM ratios (A) GLAM ratios (see legend of Fig 1)
of dst1D and an isogenic DST1 (strains MMY9.2 and BY4741)
cul-tured in the absence or in the presence of sublethal amounts of
MPA (B) mRNA levels of the transcription units and strains
ana-lyzed in (A) cultured in the absence of MPA The relative values of
the mRNA levels with respect to the shortest transcript (PHO5
mRNA) are also shown Three independent experiments were
aver-aged Notation of transcription units as in Fig 1(A) (C) mRNA
lev-els of the transcription units and strains analyzed in (A) cultured in
the presence of MPA (10 lgÆmL)1) A representative experiment is
shown.
Trang 7suggesting that the detected effect of the SAGA mutants on the GLAM ratio would not be due to an indirect influence of altered chromatin structure but to
a sustained role of SAGA after the initial steps of transcription (Fig 5C–D)
Influence of mRNA processing on the GLAM ratios
Transcription elongation is coupled with mRNA cessing, since all RNA modifications leading to pro-duce a mature exportable mRNA take place or start during elongation We decided to analyze a set of mutations affecting proteins that play a post-transcrip-tional role in gene expression We included mft1D and thp2D, two mutants lacking subunits of the THO com-plex [41]; ref2D and syc1D, involved in the 3¢ cleavage previous to polyadenylation and termination [42,43]; and three mutants, cdc40D, cus2D and cwc15D, lacking proteins connected to RNA splicing [44–46] As expec-ted from the reporexpec-ted requirement of the THO com-plex for the expression of long genes [14], the GLAM ratios of the two THO mutants were dramatically reduced when compared with the wild type (Fig 6) The absence of effect of cdc40D and cus2D on the GLAM ratios was also expected, as no intron is located in the transcription units used in this study However, cwc15D, another mutant connected to spli-cing, did show a significant reduction (Fig 6) ref2D also showed low GLAM ratios, whereas syc1D, another mutant connected to 3¢ cleavage did not (Fig 6) We conclude that only a subset of mRNA-processing functions influences mRNA biogenesis in a gene length-dependent manner
Discussion
Some factors required for mRNA biogenesis play their role during PIC assembly, transcription initiation and early elongation, whereas some others functionally interact with Pol II all along transcription elongation
A
B
C
D
E
Fig 4 Mutants affected in PIC assembly and transcription initiation
do not show reduced GLAM ratios GLAM ratios (see legend of Fig 1) of the following mutants and their corresponding wild types: (A) tbp1-P65S and an isogenic TBP1 (strains YAK293 and YAK289) (B) mot1–1, toa1–18 and a wild type with the same genetic back-ground (strains FY1214, JMY498 and FY98) (C) L50D, sua7-K205E and an isogenic SUA7 (strains FP177, FP207 and FP142) (D) tfa1-T218D, tfa1-C127F and an isogenic TFA1 (strains YSB326, YSB331 and YSB324) (E) srb10D, srb11D and an isogenic wild type (strains SLY7, SLY107 and SLY3).
Trang 8In order to develop an easy test for detecting muta-tions or drugs that influence mRNA biogenesis all along the transcription unit, we designed a novel two-reporter assay based on the PHO5 gene We hypo-thesized that gene length is the key element that distin-guishes between factors involved in the initial steps of transcription and factors influencing mRNA biogenesis all along the transcription unit, as it is well established for the well-known elongation factor ELL in Dro-sophila cells [47] We supposed that long transcription units would be more strongly impaired by mutations affecting this second kind of factors than shorter ones, whereas those mutations causing dysfunction of a gen-eral factor only involved in the early steps of transcrip-tion would affect equally all transcriptranscrip-tion units, regardless of their length In order to quantify the results of the assay, we have defined gene length-dependent efficiency of mRNA accumulation as the levels of a long mRNA encoding PHO5, divided by the levels of the minimal PHO5 mRNA, when both
A
B
C
D
Fig 5 Mutants affected in SAGA, but not other chromatin-related mutants, show reduced GLAM ratios (A) GLAM ratios (see legend
of Fig 1) of gcn5D, spt3D and an isogenic wild type (strains Y07285, Y04228 and BY4741) (B) Northern blot showing the mRNA levels of the transcription units and strains (gcn5D and wild type) analyzed in (A) The relative values of the mRNA levels with respect to the shortest transcript (PHO5 mRNA) are also shown Notation of transcription units as in Fig 1(A) (C) GLAM ratios of hta1htb1D and an isogenic wild type (strains FY710 and FY120) (D) GLAM ratios of htz1D, swr1D, isw1D, chd1D, rpd3D and an isogenic wild type (strains Y01703, Y03693, Y03385, Y06160, Y01114 and BY4741).
Fig 6 Influence of mutations affecting mRNA processing on the GLAM ratios GLAM ratios (see legend of Fig 1) of mft1D, thp2D, cdc40D, cus2D, cwc15D, ref2D, syc1D and an isogenic wild type (strains Y00508, Y02861, Y04201, Y01158, Y03521, Y03554, Y02435 and BY4741).
Trang 9are transcribed from the same promoter We have
shown that the so-defined ratio can be estimated by
assaying the acid phosphatase activity encoded by
PHO5 (GLAM ratio), and that the ratios obtained
from acid phosphatase assays are in fact more
repro-ducible than those directly calculated from northern
experiments As all transcription units used in this
assay express an identical Pho5 protein, it is unlikely
that mutations or drugs affecting translation, protein
stability or other post-translational processes, may
influence the GLAM ratios
We tested the consistency of the GLAM ratios by
performing the assay in a wide set of previously
char-acterized mutants involved in mRNA biogenesis The
GLAM ratios were always calculated for two different
long transcription units: the one containing E coli
lacZ (GAL1pr::PHO5-lacZ) and the other containing
K lactis LAC4 (GAL1pr::PHO5-LAC4) These two
genes share the same length, but display a very
differ-ent G + C contdiffer-ent Their chromatin structure in
S cerevisiaeis also completely different: random
nucleo-some positioning in lacZ [14] but translationally
posi-tioned nucleosomes in LAC4 (S Jimeno-Gonza´lez,
P M Alepuz and S Cha´vez, unpublished) We
consid-ered that a mutant showing similar GLAM ratios with
both long transcription units would indicate a gene
length-dependent effect, whereas a mutant exhibiting
differences between the GLAM ratios calculated with
each long transcript might involve sequence-dependent
or chromatin-dependent phenomena Among all the
mutants analyzed in this study, only toa1–18 affecting
TFIIA showed a significant difference between the
GLAM ratio calculated with GAL1pr::PHO5-LAC4
and the one calculated with GAL1pr::PHO5-lacZ We
did not find an explanation for this result, unless a
connection exists between TFIIA and the chromatin
organization of the transcribed region In all other
strains tested, the two values were not significantly
dif-ferent, although in most cases the GLAM ratios
calcu-lated with GAL1pr::PHO5-lacZ were slightly higher
than those calculated with GAL1pr::PHO5-LAC4
Mutations affecting transcription elongation factors,
such as SPt6, yFACT, yDSIF, and the PAF1 complex,
as well as the elongator, clearly showed lower GLAM
ratios than their isogenic wild types In contrast, those
mutations described to affect factors involved in PIC
assembly or transcription initiation, like TBP, Mot1,
TFIIB, TFIIE or Srb10-Srb11, show very similar
GLAM ratios compared with their corresponding
wild-types, with SAGA mutations being the only
exception (discussed below) In all these mutants, the
resulting GLAM ratios for two long transcription units
(GAL1pr::PHO5-lacZ and GAL1pr::PHO5-LAC4)
were very similar Moreover, the presence of sublethal doses of nucleotide-depleting drugs affecting transcrip-tion elongatranscrip-tion also reduced the GLAM ratios We conclude that these general results are solid enough to validate the novel two-reporter assay as a useful tool with which to detect transcription elongation defects Mutations or drugs can affect two different aspects
of transcription elongation: elongation rate and proces-sivity A recent study by Mason and Struhl [11] has shown that none of the many putative elongation fac-tors that they tested affect the elongation rate, although mutations in the THO complex and Spt4 sig-nificantly reduce processivity Our assay is based on the comparison of transcription units of different length, which makes it an optimal method with which
to detect processivity defects; consequently thp2, mft1D and spt4D show the lowest GLAM ratios Mason and Struhl also showed that those elements affecting elon-gation rate, such as 6-azauracil and MPA, simulta-neously reduce processivity [11]; the decrease of the GLAM ratios in response to these two NTP-depleting drugs indicates that our in vivo assay can detect all the elongation defects detected by RNA polymerase II-dependent chromatin immunoprecipitation Moreover, some putative elongation mutants that did not show reduced processivity in the study by Mason and Struhl, such as elp3D, rtf1D and leo1D [11], did show signifi-cantly low GLAM ratios, suggesting that our in vivo assay displays a high sensitivity to detect elongation defects Finally, mutations in SPt6 and SPt16 that can-not be analyzed by RNA polymerase II-dependent chromatin immunoprecipitation due to technical limi-tations of that assay [11], show reduced GLAM ratios, supporting a contribution of FACT and SPt6 to proc-essivity We conclude that the new in vivo assay des-cribed in this study is a convenient complementary tool with which to analyze transcription elongation The only transcription elongation factor tested whose mutation did not produce lower GLAM ratios than its isogenic wild type was TFIIS When either cal-culated with the assayed acid phosphatase activities or inferred from northern experiments, the ratios of a dst1D mutant were not significantly low Moreover, the presence of sublethal doses of nucleotide-depleting drugs, like 6-azauracil or MPA, reduced the GLAM ratios of a dst1D mutant; however, it also reduced the GLAM ratios of an isogenic wild type accordingly This differentiated behavior of dst1D separates TFIIS from the other transcription elongation factors tested
in this work The most logical explanation for this phenomenon is that, as has been recently suggested [11], TFIIS does not play a relevant role in elongation all along the transcription unit, or at least along the
Trang 10transcription units used in this assay Alternately,
TFIIS may play a general function, but only during
early elongation If this were the case, both long and
short transcription units would be equally influenced
by the absence of TFIIS, and as a result the GLAM
ratio would be unaffected A role of TFIIS centered in
early elongation has been suggested by the genetic and
physical interactions of TFIIS [48,49] A relevant role
of TFIIS in early elongation has been also found in
Drosophila heat-shock genes [50] Moreover, a role of
TFIIS in activating the GAL1 promoter has been
recently demonstrated [51] An elongation role of
TFIIS in vivo, in positions far from the promoter, has
only been shown when an artificial arrest site was
introduced, and even in this case the influence of
TFIIS depended on the degree of transcriptional
acti-vation [52] The present results confirm the difficulties
involved in studying the dst1D mutant in vivo, an
observation already reported elsewhere [11]
The SAGA complex is one of the main elements
that mediate in activation of gene expression; it acts
through its ability to interact with gene-specific factors
and to stimulate PIC assembly Surprisingly, gcn5D
and spt3D mutants, both lacking subunits of SAGA,
show reduced GLAM ratios We consider it unlikely
that these gene length-dependent effects of the SAGA
mutants take place at the level of PIC assembly, since
all transcription units used to calculate the GLAM
ratios share the same promoter We have also ruled
out an indirect effect of SAGA mutations on the
GLAM ratios provoked by overall effects on
chroma-tin structure, just as other mutations affecchroma-ting
struc-tural elements of chromatin (hta1htb1D, htz1D) or
chromatin remodeling (swr1D, isw1D, chd1D, rpd3D)
show similar GLAM ratios to their isogenic wild types
We conclude that SAGA might play an additional role
after transcription initiation all along the transcription
unit The genetic interactions of GCN5 with the
elon-gator [53], the sensitivity of several SAGA mutants to
mycophenolic acid [54] and the genetic interactions
between genes encoding SAGA subunits and
elonga-tion factors [55,56] also suggest a role for SAGA
dur-ing transcription elongation Alternatively, the absence
of SAGA might affect the recruitment of other factors
required for postinitiation events in mRNA biogenesis
As discussed above, mutants affected in factors
required for transcription elongation all along the
transcription unit show reduced GLAM ratios
However, transcription elongation is not the only
gene-length dependent process in mRNA biogenesis
Formation of the mRNP complex, mRNA export or
splicing are other events that may be gene
length-dependent Mutants lacking subunits of the THO
complex, involved in mRNP formation, show the lowest GLAM ratios measured in this work This is not the case in other RNA processing mutants that have been analyzed in this work In agreement with our results, an important part of the RNA-process-ing machinery does not significantly influence mRNA accumulation [15] However, a part of the RNA pro-cessing machinery physically interacts with elongating Pol II; as a consequence, the absence of elements involved in RNA processing may indirectly affect transcription elongation [57] This might be the explanation of the low GLAM ratios shown by cwc15D and ref2D, lacking proteins connected to splicing and 3¢ cleavage, respectively In contrast to the consistency of the results obtained using our reporter assay with well-known mutants affected in transcription initiation or elongation, the behavior of mutations affecting mRNA processing is heterogene-ous Additional analyses are required to understand more fully the effect of these mutations on gene length-dependent accumulation of mRNA
Experimental procedures
Materials
Suppliers are indicated below at first mention, except for chemical reagents, which were purchased from Sigma (St Louis, MO, USA)
Yeast strains, plasmids and media
Yeast strains used are described in Table 1 All MMY strains were constructed by standard genetic methods of tetrad analysis or transformation [58] MMY5 strains are congenic and were generated by crossing FY120 and FY137 AGY1–10 A strain was obtained by crossing FY348 and W303-ZT; further crossing of FY120 and AGY-10A rendered all MMY11 strains MMY9.2 was obtained by sporulating Y24411
All plasmids used are mono-copy CEN-based and are lis-ted in Table 2 Cells were grown in yeast extract–peptone medium or in synthetic complete medium (DIFCO, Detroit,
MI, USA), with 2% glucose or 2% galactose, at 30C [58]
Acid phosphatase assays
Yeast cells with the appropriated plasmids were grown on selective synthetic medium lacking uracil with 2% galactose and collected when cultures reached an optical density at
600 nm (OD 600) of 0.8–1 Acid phosphatase activity of intact cells was assayed as described [19] The acid phos-phatase activities of the transformants were corrected by