The EMBO Journal Vol.5 no.ll pp.2995-3000, 1986
`
`Functional dissection of Escherichia coli promoters: information in
`the transcribed region is involved in late steps of the overall
`process
`
`Wolfgang Kammerer, Ulrich Deuschle, Reiner Gentz1
`and Hermann Bujard
`
`Zentrum fiir Molekulare Biologic, Universitat Heidelberg, P.O. Box
`106249, D—6900 Heidelberg, FRG, and ‘Central Research Units,
`F.Hoffmann-La Roche and Co. AG, CH4002 Basel, Switzerland
`
`Communicated by H .Bujard
`
`After binding to a promoter Escherichia coli RNA polymerase
`is in contact with a region of about 70 bp. Around 20 bp of
`this sequence are transcribed. Information encoded within
`this transcribed region is involved in late steps of the func-
`tional program of a promoter. By changing such ‘down-
`stream’ sequences promoter strength in viva can be varied
`more than 10-fold. By contrast, information for early steps
`of the promoter program such as recognition by the enzyme
`and formation of a stable complex resides in a central core
`region of about 35 bp. Our data show that the strength of
`a promoter can be limited at different levels of the overall
`process. Consequently promoters of identical strength can
`exhibit different structures due to an alternate optimization
`of their program.
`Key words: E. coli promoters/signal elements/functional pro-
`grammes
`
`Introduction
`
`Prokaryotic promoters encode a complex program whose ultimate
`goal is the release of a transcriptional elongation complex some-
`times predisposed to interact differentially with signals such as
`temiinators (Grayhack et al. , 1985) located distal to the promoter.
`In a simplified model the program of an unregulated Escherichia
`coli promoter can be subdivided into four major steps: (i) recog-
`nition of the sequence by RNA polymerase; (ii) isomerization
`of the initial complex into a conformation capable of initiation;
`(iii) initiation of RNA synthesis; and (iv) transition into an elong-
`ation complex and promoter clearance.
`In principle each of these steps can be rate limiting for the
`overall function of a promoter. Thus, promoters of identical
`strength may differ in their structure due to alternate functional
`optimizations (Bujard, 1980; Deuschle et al., 1986). Here we
`report the identification of structural elements which are respon-
`sible for partial functions of the overall process. Modification
`of such elements has allowed us to alter the properties of pro-
`moter sequences in a predictable manner by shifting the rate-
`limiting event to a different step of the overall process.
`One promoter analyzed here (PN25) is a typical representative
`of promoters found in the ‘early’ expression class of coliphage
`T5 (Gentz and Bujard, 1985). These promoters belong to the most
`efficient transcriptional initiation signals identified so far (Deu-
`schle et al., 1986). Their sequences show homologies not only
`within the region commonly considered to be essential for pro-
`moter function (between +l and -36, +1 being the first nucleo-
`tide transcribed), but also around position -43 and between +1
`and +20 (Figure 1). A second promoter included in this study
`
`© IRL Press Limited, Oxford, England
`
`is Pm", a sequence synthesized by Dobrynin et al. (1980) accord-
`ing to a consensus sequence proposed by Scherer et al. (1978).
`In vitro both promoters are readily recognized by RNA polymer-
`ase and form stable complexes with the enzyme. However,
`whereas PN25 initiates efficient RNA synthesis in vivo and in
`vitro, Pam is a rather poor promoter in both environments (Deu-
`schle et al., 1986). By exchanging defined sequence elements
`both promoters can be converted into signals closely resembling
`each other in their in vivo and in vitro properties.
`
`Results
`
`Experimental strategy
`
`Based on the analysis of conserved sequences (Gentz and Bujard,
`1985) and on footprint experiments (U.Peschke, unpublished
`results) of promoters from coliphage T5 we define a promoter
`as a sequence extending from position +20 to -50 and subdivide
`it rather arbitrarily into a ‘core’, an ‘upstream’ (USR) and a
`‘downstream’ (DSR) region. Typical features for a DSR of some
`T5 promoters such as PN25 are the conserved pentamer around
`+7 and a stretch of purines between +9 and +18. The prominent
`motive of a USR is a block of As centered around position -43
`(Figure 1). By contrast the design of Pa," (Figure 1) was based
`exclusively on homologies found within the core region. Its DSR
`and USR are fortuitously dependent on the site of integration,
`or in the case of the DSR on sequences designed to function as
`translational start signals.
`To probe a possible role of DSRs in promoter function the
`core/USR of both PN25 and Pm" were fused with various DSRs
`and the resulting sequences were studied in vitro and in vivo.
`Two DSRs were synthesized, the ‘anti ’— and the ‘pex’—DSR.
`Based on a consensus sequence of six ‘early’ T5 promoters (Gentz
`and Bujard, 1985), the anti—sequence was consmicted by inserting
`C for A, T for G and vice versa but avoiding runs of more than
`three Gs or Cs. The pex—DSR is identical with the corresponding
`sequence of PN25 except that the pentameric sequence around
`position +7 is exchanged (GGGTC replaces TTTGA). The two
`synthetic sequences and several naturally occurring DSRs were
`used to construct the promoters depicted in Figure 1.
`Promoter constructs
`
`The synthetic DSRs anti and pex are 23—bp and 24-bp long oligo—
`nucleotides respectively flanked by a Hinfl and a BamHI site.
`Using the Hinfl cleavage site centered around position -3 of
`PN25 fusions PN25/a,,,,« and PN25/pex were obtained. A third con-
`struct in which the lac operator was joined to the PN25 core
`region resulting in PN25/la, has been described previously (Stuc-
`ber et al., 1984). For fusing the various DSRs to the core se-
`quence of Pam the Rsal site of this promoter located around
`position -3 was converted into a Hinfl site. The resulting pro-
`moter sequences such as Pc,,,,/pa, are homologous to their PN25
`counterparts (e. g. PN25/pex) up to position -5. In addition the
`Rsal site of Pam was directly used for fusions with the DSRs
`of PUP (Yanofsky et al., 1981), and PD/E20 (Gentz and Bujard,
`1985) since these promoters carry conveniently located Rsal sites
`Mylan v. Genentech
`Mylan v. Genentech
`IPR2016-00710
`IPR2016-00710 2995
`Genentech Exhibit 2074
`
`Genentech Exhibit 2074
`
`

`
`W.Kammerer et al.
`
`PN25
`
`Pm
`
`+20
`+10
`+1
`-10
`-20
`-30
`-40
`-50
`I
`I
`I
`I
`I
`I
`I
`I
`TcAT5_I@\rTTAcA5tsAAAATIITtcT AGTT‘C ATAAAEGAGAGGAGTT
`ATAAAGGGTCGAGAAGAGTT
`
`ATCCGGAATCCTCTTCCCGG
`
`AAATTGTGAGCGGATAACAA
`
`AATTCACCGTCGTTTTTTAAGCTTGGCGG GETTCC ATAAGGAGGTGGATCCGGCA
`GEKTTE ATAAATTTGAGAGAGGAGTT
`ATAAAGGGTCGAGAGGAGTT
`
`ATCCGGAATCCTCTTCCCGG
`
`GGTACCCAGTTCGATGAGAGCGATAAC
`
`GGTACGCAAGTTCACGTAAAAAGGGTA
`
`DSR
`
`~25
`pex
`
`anti
`
`lac
`
`con
`~25
`pex
`
`anti
`
`0/620
`
`crp
`
`o——usR————n-4-— CORE %—————u:—osR?—+
`
`Fig. l. Nucleotide sequences of the promoters studied. The sequence of coliphage T5 promoter PN25 comprising 70 bp from position +20 to -50 is shown in
`the upper part. The USR, core and DSR are delineated and highly conserved sequences are boxed. The conserved pentamer and the purine-rich sequence
`typical for a DSR of ‘early’ T5 promoters as well as the block of As around -43 are underlined. The starting nucleotide for RNA synthesis at +1 has been
`determined for PN25 (Stiiber, 1980). The Hinfl cleavage site is overlined. The lower pan shows the sequence of Pm" (consensus sequence between +1 and
`-40; the Rsal and Hinfl sites are overlined) and DSRs which were fused to the Pa," core via the Rsal site (D/E20 and trp) or after changing the sequence of
`Pm, via the Hinfl site (N25, pex, arm). In some of our constructs a spontaneous change from G to A at position + l5 of PN25 has occurred. However, this
`base change did not affect the parameters examined in this study.
`
`between their core and DSR’s. The distance between the -10
`
`region and the starting nucleotide (+ 1) of the latter two promoters
`(Pm,/,,p and Pm"/D/E20) is, however, increased by one nucleotide
`compared with the other constructs (Figure 1).
`All the promoter sequences were flanked downstream by
`BamHI and upstream by EcoRI or )0toI cleavage sites respect-
`ively. This allowed their oriented integration into the pDSl/to2
`vector.
`
`Recognition ofpromoters with altered DSR by E. coli RNA pol-
`ymerase
`
`Stable complexes between RNA polymerase and promoters are
`formed in vitro at distinct rates which can vary at least 50-fold
`(Brunner, 1986). Both promoters PM5 and Pam bind RNA pol-
`ymerase with rates above 5 X 107 M“s“ and form stable com-
`plexes with the enzyme (Brunner, 1986; Karmnerer, 1986). To
`examine whether information within the DSR is contributing to
`the rate of complex formation, mixtures of fragments carrying
`the various promoter constructs were exposed to increasing but
`limiting amounts of RNA polymerase and the resulting complexes
`were monitored by adsorption onto nitroceflulose filters. As seen
`in Figure 2 replacement of the original DSRs by various se-
`quences including the anti-DSR affects neither the rate of com-
`plex formation of RNA polymerase to PN25 nor to Pa," derived
`sequences. These findings are supported by the experiment de-
`picted in Figure 3. Here we have compared the interaction of
`RNA polymerase with the intact sequence of Pm, and with a
`version of this promoter truncated at position -4. Again the
`enzyme binds with comparable rates to both sequences. We
`therefore conclude that the information required for promoter
`recognition and for the fomiation of stable complexes must reside
`upstream of position -4.
`
`The eflect of dyfferent downstream sequences on RNA synthesis
`in vitro
`
`The efficiency of a promoter in vitro depends strongly on assay
`conditions. Therefore in experiments described here promoters
`2996
`
`were either directly compared with each other under competitive
`conditions, or their strength was determined in relation to an
`internal standard. In a first set of experiments stochiometric mix-
`tures of fragments carrying the various promoters were used as
`templates to produce ‘run off’ transcripts of different sizes at RNA
`polymerase concentrations of about two enzyme molecules per
`promoter. When PN25 was compared with PN25/,,,,,; a most strik-
`ing result was obtained. As seen in Figure 4 (lane 12) PN25/0,,”
`is a much less efficient promoter than PN25. Replacing the anti-
`DSR by the lac operator results in PN25/lac, a promoter of inter-
`mediate strength. By contrast, Pc,,,,, a poor promoter under com-
`petitive conditions, increases in strength if its core sequence is
`combined with the DSRs of PN25 or PD/E20 (Figure 4, lanes
`6-8). A minor increase in promoter strength is observed when
`the DSR of P
`is fused to Pm" (Pm,/,,p). The fusion of the anti-
`DSR to Pa,"
`C0,,/a,,,;) diminishes even the low activity of this
`promoter.
`Quantitative data for some of the promoter constructs were
`obtained by determining the in vitro RNA obtained from super-
`coiled templates following the procedure of Deuschle et al.
`(1986). The results (Table I) show that the replacement of the
`DSR of PN25 by the synthetic anti—DSR reduces the promoter
`strength 10-fold. Similarly, Pm, loses efficiency upon transition
`to Peon/and (Figure 4). However, if Peon/mi is compared with
`PCMM5 a 4- to 5-fold increase in promoter efficiency is observed
`(Figure 4 and Table 1). Thus, information relevant for promoter
`strength in vitro is encoded downstream of the transcriptional
`start site.
`
`The influence of DSRs on the in vivo activity of promoters
`
`Several promoter constructs (Table I) integrated in plasmid pDSl/
`t,,2 were transformed into E. coli C600 cells. RNA of logarith-
`mically growing cultures (ODWO = 0.7) was pulse-labelled for
`1 min with [3H]uridine and quantified according to Deuschle et
`al. (1986). The results summarized in Table I show that also in
`vivo the replacement of the natural DSR by the anti-DSR reduces
`
`

`
`Functional dissection of Escherichia coli promoters
`
`PROMOTER
`
`CORE
`I
`1 N25
`
`COD
`
`fl DSR
`
`P
`
`P
`
`3/ fir‘
`
` : '-1"
`:7
`-'*
`I
`/ I
`-50 -5 +125
`
`1
`
`2
`
`3
`
`A M
`
`Fig. 3. Sequences essential for specific RNA polymerase binding. A 32P
`end-labelled DNA fragment of 700 bp carrying both Pb,“ and Pm, was
`cleaved at position -4 and -50 of Pa," by Rsal or EcoRI respectively (left
`part of the figure). Mixtures of these fragments (lane 4) were exposed to
`increasing amounts of RNAP. The enzyme DNA complexes were collected
`on nitrocellulose filters and analyzed by PAGE (6% polyacrylamide, 8 M
`urea,
`1 mA/cm, 1.5 h) and autoradiography. The RNAP/promoter ratios
`were 0.25, 0.5 and 1.5, for lanes 1-3 respectively. M contains size
`markers of 622 and 504 bp in length.
`
`1
`
`1
`l
`
`11
`
`1
`
`PRdM0TERiii1_fi
`CORE
`DSR
`1
`
`I N25 1
`
`con
`
`O
`000-0
`I
`
`-21.2
`
`.<~NZ5
`
`-180
`
`I .000.
`
`C O
`
`F
`
`DSR
`
`D/E20->
`
`N25->
`
`..
`
`_ C 0 <—lc1c
`-
`<—unti
`
`.
`
`-122
`
`frp
`
`g
`
`anti-— .
`COD->.
`
`.. ..
`
`123z.567aM9 1011121
`
`g
`
`Fig. 4. The strength of hybrid promoters in vitro. Pm" and PN25 as well as
`various core/DSR combinations were transcribed in vitro individually (lanes
`1-5 and 9-11) and in assays containing stoichiometric amounts of
`fragments carrying the respective promoters (lanes 6-8 and 12). The
`fragments were sized in such a way that run-off transcripts of different
`lengths are obtained. The [oz-"Pl-UTP labelled RNAs were analyzed by
`PAGE (4% polyacrylamide, 8 M urea, 0.5 mA/cm, 2 h) and
`autoradiography. The left part of the figure shows transcripts from various
`PC,,,,,D5R constructs as indicated. Lanes 6-8 contain mixtures of transcripts
`obtained with Pm" derivatives at 200 (lanes 6 and 7) and 300 mM (lane 8)
`KCl respectively. The analysis of PN25 and two of its derivatives is shown
`in the right part of the figure. With the exception of the assay analyzed in
`lane 8 all experiments were carried out in 200 mM KCI. M denotes a size
`marker (Hpall digest of pBR322, the length in bp for some fragments is
`given on the right side).
`
`versa, with identical DSRs and differing core and USR sequences.
`Thus, a mixture of fragments containing PN25 and PN25/am; as
`well as Placy;/5 and Pm; was transcribed in vitro and the run-
`
`2997
`
`frp+
`
`con+
`N25->
`
`ll
`
`——unt1
`
`J
`
`-—lc1c
`
`._N25
`
`4
`
`1
`
`2
`
`3
`
`C M C
`
`4
`
`5 Q]
`
`Fig. 2. The signal strength of PN25, Pm and hybrid sequences. DNA
`mixtures containing stoichiotnetric amounts of promoter carrying fragments
`(lanes C) were exposed to increasing but limiting amounts of RNA
`polymerase under standard conditions. The enzyme/DNA complexes retained
`on nitrocellulose filters were analyzed by PAGE (6%, 8 V/cm, 4 h, stained
`with ethidium bromide). The left part of the gel shows the analysis of
`sequences composed of the Pa," core region and DSRs as indicated. A
`corresponding experiment for PN25 sequences is shown at the right pan of
`the gel. The actual RNAP/promoter ratios were 0.3, 1.0 and 3.0 in lanes
`1-3 and 1.0, 0.3 and 0.1 in lanes 4-6 respectively. The position of
`fragments carrying the B-lactamase promoter are indicated (5). In lanes C
`promoter—free fragments which are not bound by RNAP can be identified.
`M denotes a size marker (Hinfl digest of pBR322 with fragments between
`153 and 517 bp in length).
`
`the strength of PN25 by a factor of 10. The lac operator (PN25/kw)
`though clearly better than PN25/mm still diminishes the efficien-
`cy of PN25 about 3-fold. Pc,,,,, a rather inefficient promoter in
`vivo, loses strength when fused to the anti sequence (Pm;/,,,,,,-).
`However, the combination Pam/N25 is a more than 10-fold bet-
`ter promoter than Pc,,,,/a,,,,- and reaches the strength of PN25.
`After placing the anti—DSR distal to a promoter we have found
`no effect on transcription (data not shown) demonstrating that
`it does not cause termination of an elongating complex but instead
`acts in concert with other promoter functions.
`To examine whether the conserved pentamer TI'I‘GA within
`the DSRs of certain T5 promoters is an essential element the pear-
`DSR was fused to several core promoter sequences. As seen
`in Table I replacement of the natural pentameric sequence by
`GGGTC reduces the strength of both PN25 and Pa,"/N25 by less
`than 30%. Thus, this pentamer is not the sole contributor to the
`effect observed with the N25 downstream sequence.
`DSRs act in late stages of promoter fimction
`The above data show that signal elements involved in promoter
`recognition and formation of a stable RNAP/promoter complex
`are located upstream of position -4. DSRs must therefore con-
`tribute to the later steps of the overall process. We have therefore
`examined the in vitro RNA synthesis initiated by promoters with
`identical core and USR but different DSR sequences and vice
`
`

`
`W.Kammerer et al.
`
`Table I. Promoter strength in vivo and in vitro of PN25, Pam and their derivatives
`
`Promoter
`Core
`
`N25
`
`can
`
`DSR
`
`N25
`anti
`
`pex
`[ac
`con
`anti
`N25
`
`pex
`D/E20
`
`trp
`
`Relative promoter strength
`In vitro
`In vivo
`
`18
`2
`
`—
`—
`—
`3
`12
`
`—
`—
`
`—
`
`25
`2.7
`
`15
`8
`4
`1.8
`25
`
`16
`13
`
`8
`
`The relative promoter strengths were detennined according to Deuschle et
`al. (1986). The leftmost column indicates the core and USR sequences as
`shown in Figure 1. The various sequence combinations are indicated by the
`‘DSR’ column. Both in vitro and in viva promoter strengths are related to
`the promoter of the B-lactamase (bla) and are given in ‘PW-units’ (Deuschle
`et al., 1986). The maximal deviations are around :1: 10%.
`
`1547
`
`865 -
`
`685
`
`__
`
`P
`J lctcUV5
`-— - <-/PN25/anti
`
`‘\ N25
`Ptucl
`
`RNAP
`M 30 751.8 02 0.04 ?———PROMOTER
`
`Fig. 5. Dependence of the relative promoter strength on the concentration of
`RNA polymerase. Fragments carrying various promoters were prepared to
`give run-off transcripts of different size. For transcription these fragments
`were mixed in molar ratios of 2:2:l:l for PUV5, PN25,a,,,,-, PN25 and Pm,
`respectively. These mixtures were transcribed in the presence of
`[oz-"Pl-UTP and 200 mM KCI at the RNAP/promoter ratios indicated.
`Aliquots of the various assays were applied to the gel to give roughly
`constant amounts of PN25,a,,,,- transcripts after separation in PAGE (4%
`polyacrylamide, 8 M urea, 0.6 mA/cm, 2.5 h) and autoradiography. The
`positions of the various transcripts and the length of the size markers (M)
`are indicated.
`
`off transcripts were analyzed. As seen in Figure 5 at very low
`enzyme concentrations (lane 5) PN25 and PN25/am produce com-
`parable amounts of RNA. Under these conditions the complex
`formation is limiting and since both promoters compete equally
`well for the enzyme (Figure 2) and also form stable complexes
`with the enzyme, differences at later steps of the process have
`little impact. At higher enzyme concentrations (lanes 1-4), how-
`ever, the amount of RNA produced from PN25/am is clearly
`limited by the function of the anti-DSR. Thus, the difference in
`promoter strength of these two promoters as seen at higher en-
`zyme levels (lanes 1 -4) is due to events controfled by the DSR.
`A complementary observation is made with Pm, and Placw/5.
`Both promoters have identical DSRs (Pm, is a hybrid sequence
`between Pm, and Plac; Amann et al., 1983); however, Pm, binds
`
`2998
`
`RNA polymerase five times more efficiently than PWUV5 (Kam-
`merer, 1986). Since these two promoters should not differ in pro-
`cesses directed by their DSRs the difference seen in promoter
`strength should be comparable with the different rates of com-
`plex formation. This becomes obvious when the concentration
`of RNA polymerase is lowered (Figure 5, lanes 1 -3): the amount
`of RNA synthesized from Placw/5 decreases strongly followed
`by Pm, specified transcripts. At very low enzyme concentrations
`both promoters are competed out by PN25 and PN25/,,,,,,-. At con-
`ditions of excess RNAP, Placyg/5 and PW, still differ somewhat
`in their in vitro strength — indicating that the rate of complex
`formation is not the only parameter which determines the func-
`tional difference of these two promoters. Thus,
`the relative
`strength of promoters in vitro also depends strongly upon the
`concentration of RNA polymerase.
`
`Discussion
`
`It has generally been accepted that the information essential for
`the function of unregulated E. coli promoters is stored within
`about 35 bp, spanning from position +1, the starting position
`for RNA synthesis, to about position -35. It is this region where
`the most striking sequence homologies among promoters are
`found and where the overwhelming number of promoter mu-
`tations were mapped. Several lines of evidence, however, sug-
`gest that sequence information flanking this region may be
`important for promoter fimction as well. (i) When bound to a
`promoter, E. coli RNA polymerase covers close to 70 bp
`(Schmitz and Galas, 1979; Siebenlist et al., 1980). Promoter
`function might therefore not be independent of contacts between
`the enzyme and the sequences outside of the 35—bp region. (ii)
`The sigma subunit of the enzyme is only released after a sequence
`of 8 —11 nucleotides has been transcribed (Hansen and McClure,
`1980; Straney and Crothers, 1985). Information encoded in this
`region may participate in this process. (iii) Conserved sequences
`upstream and downstream of the 35—bp core region are found
`in some strong promoters (Bujard, 1980; Bujard et al., 1983;
`Gentz and Bujard, 1985).
`In this study we have defined a promoter as a 70—bp sequence
`containing a ‘core’ of 35 bp, a USR and a DSR of 15 and 20 bp
`respectively (Figure 1) and have examined the potential func-
`tion of the region downstream of position +1.
`
`DSRs can influence promoter strength
`Based on a prototype DSR typical for the sequence between +1
`and +20 of some phage T5 promoters such as PN25 (Gentz and
`Bujard, 1986) we have synthesized an anti—DSR. This sequence
`was fused to the core region of PN25 and Pa,” (Figure 1). In both
`cases the in vitro and the in vivo promoter strength was signifi-
`cantly reduced (Figure 4, Table I). By contrast, when the anti-
`DSR of Peon/and was replaced by the original DSR of PN25 yield-
`ing Pm,/N25 the promoter activity in viva was raised 14-fold
`(Table 1). Similarly, when Pm, was combined with the DSR of
`PD/E20, another T5 promoter, the strength was again increased
`in vivo and in vitro (Figure 4, Table I). DSRs of other promoters
`like Pm and PW, caused intermediate effects: they decreased the
`strength of PN25 but increased the activity of Pam (Table I, Fig-
`ure 4). Thus, by changing the sequence within the first 20 bp
`of the transcribed region the in vivo activity of promoters Pam
`and PN25 was varied up to 14-fold. We have attempted to ident-
`ify a signal within the DSR of the T5 promoters and have there-
`fore exchanged the conserved pentameric sequence 'I‘TTGA
`within the N25—DSR. The resulting promoter construct PN25/pex
`(Figure 1) showed a reduced activity (Table I) which, however,
`
`

`
`demonstrates that this sequence is not the only information in-
`volved in the observed downstream effects. Nevertheless it is
`
`obvious that the sequence of the PN25 DSR must contain infor-
`mation which can increase decisively the strength of at least some
`promoters.
`
`DSRs encode ‘late’ promoter functions
`To examine whether DSRs are involved in the recognition pro-
`cess we have compared the rate of complex formation between
`RNA polymerase and various promoter constructs. In exper-
`iments as depicted in Figure 2 it was shown that information
`within the downstream region does not contribute to the recogni-
`tion process. In fact, as demonstrated in Figures 2 and 3 the se-
`quence elements essential for recognition must reside upstream
`of position -5. Similarly, no effect on promoter recognition was
`observed when sequences upstream of position — 36 were remov-
`ed (Karnrnerer, 1986).
`From these results we conclude that the promoter region ex-
`tending between +2 and -36, which we define as ‘core’ (Figure
`1), contains the essential elements for the early steps of the in-
`teraction between a promoter and RNA polymerase; these in-
`clude recognition of a sequence by the enzyme and isomerization
`of the initial complex into a state capable of starting RNA syn-
`thesis. DSRs must therefore contain signal functions required at
`a later stage of the overall process.
`These conclusions are supported by experiments as depicted
`in Figure 5. At limiting RNA polymerase concentrations PN25
`and PN25/a,,,i which both compete equally well for the enzyme
`produce similar amounts of transcripts, i.e. PN25 equals PN25/am,-.
`At non—limiting enzyme concentrations, however, the overall
`efficiency of the two promoters is detennined by the function
`of their DSRs, i.e. PN25 > PN25/,,,,,,-. The same experiment
`shows, furthermore, that the strength of Pkwy;/5 and Pm, which
`both have identical DSRs correlates with the rate of complex for-
`mation between enzyme and promoter. Thus, the four promoter
`sequences differ in strength for different reasons. Their sequences
`are the result of diverse optimization processes and in vitro their
`functional hierarchy is influenced by the concentration of RNAP:
`at high enzyme to promoter ratios PN25 > Pm, > PWUV5 >
`PN25/0,", whereas at low enzyme concentrations PN25 ~ PN25/mm
`> Ptacl > PlacUV5-
`Conclusions and implications
`Are these latter considerations relevant for the situation in vivo?
`
`Our data show that promoters which bind RNAP with high for-
`ward rate constants (e. g. PN25 or Pam) can lose efficiency when
`combined with DSRs which apparently slow down later steps
`of the overall process (Kammerer, 1986). This suggests that high
`forward rate constants and efficient DSRs may be prerequisites
`for an optimal promoter. On the other hand some of the strongest
`promoters identified so far (PA 1 and PD/E20 from phage T7 and
`T5 respectively; Deuschle et al., 1986) bind RNAP five times
`less efficiently than PN25 (Kammerer, 1986), indicating that op-
`timal function of a promoter in vivo does not necessarily depend
`on a highly effective recognition of the sequence. By contrast,
`for Pm] and Placw/5 the in vivo strength (5—fold difference;
`Deuschle et al., 1986) correlates well with the rate of complex
`formation between RNAP and promoter found in vitro (Kam-
`merer, 1986). This suggests strongly that it is indeed the recog-
`nition by the enzyme which causes the difference in strength
`between these two promoters.
`How can these apparent contradictions be resolved? We pro-
`pose that promoters with a rather low forward rate constant for
`RNAP binding (<5 X 10° M“s“) are limited by processes
`
`Functional dissection of Escherichia coli promoters
`
`such as recognition of the sequence and formation of a stable
`enzyme/ promoter complex. For such sequences RNA chain in-
`itiation and promoter clearance can be slow as long as the com-
`plex is stable enough to ensure the start of each enzyme bound.
`Mutations within the DSR of such promoters would in general
`not be detected. Above a certain rate of complex formation
`(> 107 M”‘s“), however, later steps of the overall process such
`as initiation and promoter clearance can become rate limiting and
`consequently information contained within the DSR will be rele-
`vant. Both genetically well—studied promoter systems P,,p and
`P1“ are recognized by RNAP rather inefficiently (unpublished
`results). The discovery of promoter mutations within the DSR
`is therefore unlikely, although such mutations were reported for
`Plac (Maquat et al., 1980). However, by encoding the 5’ ter-
`minal 20 nucleotides of a mRNA, a DSR usuaHy also contains
`information relevant for translation (Shine and Dalgamo, 1975)
`and for mRNA stability (Yamarnoto and Irnmamoto, 1975).
`Rigorous proof of a promoter mutation in this region has therefore
`to be provided at the level of RNA synthesis.
`As mentioned above some of the strongest promoters in vivo
`(PA; and PD/E20) are recognized five times less efficiently by
`RNAP than, for example, PH207 or PN25. Why then did pro-
`moters with such high forward rate constants evolve? The pro-
`moter strengths determined in vivo were all obtained from fast
`growing cells in mid log phase (Deuschle et al. , 1986). The con-
`centration of RNAP at this stage of growth may not favor pro-
`moters with optimal recognition properties, and promoters like
`some of coliphage T5 may be optimized for conditions where
`stringent competition for the enzyme is required. In this context
`it appears intriguing to us that promoters which are optimized
`according to different principles can form different functional
`hierarchies depending on the concentration of active polymerase.
`Thus, controlling for example the concentration of the 0-subunit
`at different growth conditions could profoundly change the pattern
`of mRNA abundance thereby affecting the physiological state of
`the cell.
`
`Materials and methods
`
`Plasmids and DNA sequences
`The plasmids of the pDS system, their nomenclature and their preparation were
`described previously (Stiiber and Bujard, 1982; Deuschle er al., 1986). The cloning
`and characterization of the promoters PN25, PD,E20, PMCUV5, Pm, and Pam has
`been described in detail previously (Deuschle et al., 1986). Promoter Pm, was
`isolated from plasmid ptrpH1 (Amann et al., 1983). Oligonucleotides were syn-
`thesized using the triester method and purified by gel electrophoresis in 8 M urea.
`The sequences of all promoters and their derivatives were verified by dideoxy
`sequencing (Sanger et al., 1977).
`Preparation of in vitro RNA
`The standard assay for ‘run-off‘ transcripts contained in a volume of 50 ul was:
`20 mM Tris/HCI pH 8; 10 mM MgCl,; 5% glycerol, 1 mM DTT, 200 mM KCl;
`300 p.M each of ATP and GTP; 150 p.M CTP; 50 [l.M UTP combined with
`30-150 nM [oz-”P]UTP (3000 Ci/mmol).
`After pre-incubation of 0.2 pmol template per promoter fragment in the reaction
`mixture at 37°C for l min transcription was started by the addition of 1 pmol
`RNAP. The concentration of UTP was raised to 1 mM after 2 min and the incu-
`bation was continued for another minute. Samples were prepared for electrophoresis
`(8 M urea, 4% polyacrylarnide) by mixing 1/10 of the assay with 5 pl sample
`buffer (95% fonnarnide, 1 X TBE containing bromophenol blue and xylenexy—
`anol FF). Autoradiograms were quantitatively evaluated using a densitometer.
`Transcripts from supercoiled templates were obtained in an assay which was
`modified as follows: the volume and the amount of template were doubled and
`the ratio of RNAP:promoter was 50:1.
`Quantitation of RNA
`For detemiing promoter strengths in vitro or in vivo RNA was labelled, isolated
`and subjected to hybridization according to Deuschle et al. (1986). In this method
`the promoter under investigation transcribes the coding sequence of the dihydro-
`
`2999
`
`

`
`W.Kammerer et al.
`
`folate reductase of the mouse (dhfr). The dhfr-specific RNA is compared with
`an internal standard, the 13-lactamase (bla) specific RNA which is transcribed
`from the same plasmid but under the control of PM.
`Analysis of promoter/RNAP complexes by adsorption to nitrocellulose
`DNA fragment mixtures (0.25 prnol per fragment) in a volume of 200 p.1 containing
`120 mM KC]; 20 mM Tris—HCl pH 8.0; 10 mM MgCl,, 5% glycerol and 1 mM
`DTT as well as a small amount of a “P-labelled promoter containing fragment
`were incubated at 37°C for 2 min. One aliquot (50 ul) was removed and stored
`on ice as control, before the mixture was divided into 50-11.1 portions to which
`different dilutions of RNA? in 50 ul of assay buffer and pre-warmed to 37°C
`were added. After 5 min at 37°C competitor DNA was added (0.5-2 ug of single-
`stranded fd DNA per assay in binding buffer without KCl) and incubation was
`continued for another 5 min before the mixture was filtered through nitrocellulose
`(4.5 um pore size, Schleicher and Schiill) pre-equilibrated with binding buffer
`without KCl. The filters were rinsed twice with 200 pl of binding buffer con-
`taining 60 mM KCl and the adsorbed fragments were eluted with three 50-p.l
`portions of 10 mM Tris-HCI pH 8.0;
`1 mM EDTA; 0.1% SDS. Complete
`removal of DNA from the filter was examined by monitoring the radioactivity
`of the labelled fragment. The DNA was precipitated by ethanol and the redissolved
`pellet was analyzed by PAGE and ethidium bromide staining, or by autoradi-
`ography of the dried gels with Kodak X-ray film.
`
`Acknowledgements
`We thank Dr W.Bannwarth for oligonucleotide synthesis. This work was sup-
`ported in part by grants Bu 338/ 12-15 from the Deutsche Forschungsgemeinschaft
`and by the Fonds der Chemischen Industrie Deutschlands.
`
`References
`Amarm,E., Brosius,J. and Ptashne,M. (1983) Gene, 25, 167-178.
`Brunner,M. (1986) Thesis, University of Heidelberg, FRG.
`Bujard,H. (1980) Trends Biochem. Sci., 5, 274-278.
`Bujard,H., Baldari,C., Brunner,M., Deuschle,U., Gentz,R., Hughes,J., Kam-
`merer,W. and Stiiber,D. (1983) In Papas,T.S., Rosenberg,M. and Chirikjian,
`J .G. (eds), Gene Amplification and Analysis, Expression of Cloned Genes in
`Procaryotic and Eucaryotic Cells. Elsevier, Amsterdam, Vol. 3, pp. 65 -87.
`Deuschle,U., Kammerer,W., Gentz,R. and