The EMBO Journal Vol.5 no.ll pp.2987—2994, 1986
`
`Promoters of Escherichia coli: a hierarchy of in vivo strength
`indicates alternate structures
`
`Ulrich Deuschle‘, Wolfgang Kammerer‘, Reiner Gentz
`and Hermann Bujard‘
`Central Research Units, F.Hoffmann—La Roche & Co. AG, CH-4002 Basel,
`Switzerland, and ‘Zentrum fiir Molekulare Biologie, Universitat Heidelberg,
`PO Box 10 62 49, 6900 Heidelberg, FRG
`
`Communicated by H.Bujard
`
`The strength in vivo of 14 promoters was determined in a
`system which permits the quantitation of RNA synthesis with
`high accuracy. Up to 75-fold differences in promoter strength
`were measured and the most efficient signals are promoters
`from coliphages T7 and T5. Their activity approaches the
`strength of frilly induced promoters of the rRNA operons
`which may be close to the functional optimum of a single se-
`quence. By contrast, a synthetic ‘consensus promoter’ belongs
`to the less efficient signals. Our data show that optimal pro-
`moter function can be achieved by alternate structures and
`strongly suggest that information outside of the ‘classical’ pro-
`moter region contributes to promoter activity.
`Key words: E. coli promoters/in vivo strength/altemate structures
`
`Introduction
`
`Despite a wealth of information on the structure of Escherichia
`coli promoters, the topography of RNA polymerase/promoter
`complexes and on the processes goveming the onset of specific
`transcription (for review, see Rosenberg and Court, 1979), our
`understanding of structure/function relationships of E. coli pro-
`moters still remains unsatisfactory. The steadily increasing
`number of elucidated promoter sequences has allowed the ex-
`tensive study of structural homologies and refined ‘consensus se-
`quences’ were proposed (for review, see Hawley and McClure,
`1983). So far, however, such model sequences were of little
`predictive value, whenever fimctional parameters of an individual
`promoter were to be derived from structural information alone.
`Quantitative infomration on the function of defined promoter se-
`quences is obviously required for a better understanding of how
`the complex functional programme of a promoter is stored in
`a DNA sequence.
`Here we describe an experimental system for the accurate
`determination of promoter strength in vivo and in vitro. Promoter
`activities are measured by monitoring RNA synthesis in relation
`to an internal standard. Thus, the results are independent of trans-
`lational effects and of gene dosage. Differences in mRNA half-
`life can be taken into account. Fourteen promoters were char-
`acterized. The up to 75-fold differences in promoter strength
`found in vivo yielded a functional hierarchy of sequences which
`permits novel correlations between structural and functional
`parameters. Evidence for a hypothesis derived from these and
`other data (Gentz and Bujard, 1985) is also presented in the ac-
`companying publication (Kamrnerer et al. , 1986), which shows
`that sequence elements located in the transcribed region can
`significantly contribute to promoter strength.
`
`© IRL Press Limited, Oxford, England
`
`Results
`
`Experimental strategy
`To detemrine accurately the efficiency of transcriptional signals
`we have developed a system in which the RNA produced under
`the control of such signals is quantified in relation to an internal
`standard. The system outlined in Figure 1 consists of: (i) plasmid
`pDS2 carrying the 6—lactamase (bla) transcriptional unit as an
`internal standard as well as indicator regions (the sequences for
`dihydrofolate reductase of the mouse, dhfi, and of the chloram-
`phenicol acetyltransferase, cat) which can be brought under the
`control of transcriptional signals in various ways, and (ii) three
`coliphage M13 derivatives containing sequences complementary
`to coding regions of the internal standard (bla) or the indicator
`regions (dhfr, cat), respectively. Radioactively labelled RNA
`from these regions is hybridized in solution against the single-
`stranded (ss) DNA probes of the three M13 recombinants. Ad-
`sorption of the hybrids to nitrocellulose via their ssDNA portion
`allows a rapid quantitation of the hybridized RNA. Analysis of
`the S1-resistant material of such hybrids permits the characteriza-
`tion of the transcription products.
`Promoters of interest are integrated in front of the dhfr/cat in-
`dicator region and the amount of RNA specific for dhfr and/or
`cat sequences is compared with the transcripts produced in the
`bla region. Consequently, all promoters are characterized with
`respect to the B-lactamase promoter (Pbla) and promoter strengths
`are given in ‘Pbla-units’.
`Three more aspects were considered. Firstly, intense transcrip-
`tion interferes with plasmid replication (Stiiber and Bujard, 1982;
`Bujard et al. , 1985). Insertion of an efficient terminator at site
`1 or 2 of pDS2 (Figure 1) is therefore essential for the stable
`integration of strong promoters. Secondly, fragments carrying
`the promoter of interest may also initiate transcription in the op-
`posite direction, i.e. into the bla region, thereby increasing the
`level of expression of the internal standard. Such activities can
`be picked up by analysing the S1-resistant RNA/DNA hybrids
`(Figure 4). The integration of an additional terminator between
`Pbla and the promoter insertion site (site 3 in Figure 1) resolves
`this problem. Finally, the half-life times of in vivo RNA can vary
`significantly and are often influenced by the structure of the 5’
`region of the molecule (Yamarnoto and Imrnamoto, 1975; Belasco
`et al. , 1986). Since different promoters yield RNAs with differ-
`ing 5’ sequences it is necessary also to monitor the stability of
`the various in vivo transcripts.
`
`Integration of promoters into the pDS2 system
`The promoter-carrying fragments surmnarized in Table I were
`integrated into the polylinker of pDS2/tol (Figure 1). Deleterious
`overproduction of DHFR or CAT protein was avoided by fus-
`ing the dhfr sequence to the promoter fragment out—of—fiame with
`potentially efficient ribosomal binding sites and by reducing tran-
`scription into the cat gene at site 1 (Figure 1) with terminator
`to of phage lambda. Plasmids containing strong promoters in-
`serted in the proper orientation were selected by plating the trans-
`
`Mylan v. Genentech
`Mylan v. Genentech
`IPR2016-00710
`IPR2016-00710 2937
`Genentech Exhibit 2075
`Genentech Exhibit 2075
`
`

`
`U.Deuschle et al.
`
`site 2
`
`>002
`\l/ ori
`
`‘ii-ansg-‘pfim
`
`labeled
`RNA
`
`Mcofrilter birding
`S1-Mapping
`
`Fig. 1. Experimental principle. Promoters integrated into the polylinker site of pDS2 (a derivative of pDSl, Stiiber and Bujard, 1982) control the transcription
`of two indicator sequences, dhfr and cat. Sites where terrninators can be inserted allow the prevention of clockwise readthrough transcription into the
`replication region (sites 1, 2) as well as safeguarding the bla region (site 3) which is used as an internal transcriptional standard unit. The directions in which
`transcription is irritated by PM, and by the integrated promoters (P I) are indicated by arrows. Translational initiation signals are denoted as . X, E and B
`designate }0ioI, EcoRI and BamHI cleavage sites respectively. Labelled RNA transcribed frorri these plasmids is analysed by hybridization agaiiist three
`M13-derived ssDNA probes which carry a portion of the bla gene, the coding sequence of DHFR or of CAT, respectively. Analysis of the hybrids by
`nitrocellulose adsorption or electrophoresis of the Sl-resistant material permits a quantitative and a qualitative characterization of the three RNA species.
`Transcripts originating within the cloned promoter fragment and directed towards the bla region are identified by the Ml3mp9 bla probe which extends up to
`the EcoRI site. In such cases two species of S1-resistant hybrids are obtained (Figure 4). Insertion of a terminator at site 3 prevents‘ interference of such
`transcription with the bla standard unit.
`
`Table I. Promoter canying fragments integrated in pDS1/tol
`
`Table II. Half-life of mRNAs: influence of the 5' region
`
`Promoter
`
`Fragment
`size
`(in bp)
`
`Integration sites
`
`Transcript length
`up to the coding
`sequence of DHF
`
`Half-life of transcripts (min)
`Promoter
`(in pDSl/tol)
`dh
`Cat
`Ma
`fr
`(Pm directed)
`
`
`157
`‘32
`‘O6
`81
`'06
`95
`76
`64
`135
`
`2'2
`47'
`254
`233
`220
`330
`143
`154
`278
`
`Em“
`PHW
`ECO“
`PD/E20
`“'01
`PN25
`Er:oRI
`P025
`E60“
`P15
`[scam/3amH[
`inA1
`A _EwR1/Bamfll
`1:“
`ECORI/Bamm
`PM
`E R]
`P
`C0
`L
`92
`M01
`[29
`PM
`47
`Em“/BamHl
`9'
`PWUV5
`63
`EcoRI
`285
`Pm,
`23
`56””/B“’"HI
`61
`Peon
`Thirteen promoters were integrated into pDSl derivatives using the restric-
`tion sites indicated. Of these promoters, nine are coliphage promoters,
`namely PHzo1, Pm, 13,5, PD,E2., and P525 from phage T5; PM, PA, and
`PAS from Phase T7 {Ind PL from Phage_ |ambd3- Plac and PlacUV5 art? PW
`moters of the E. coli lac operon; Pm! is a trp/|acUV5 promoter hybrid and
`Pm, is a promoter synthesized according to a consensus sequence. The exact
`position of a promoter can be derived from the distance between the first
`transcribed nucleotide and the A of the initiation codon of the dhfr se-
`q“°"°°-
`formed bacteria onto plates containing 30 pg/ml chloramphenicol.
`Due to the lealdness of terminator to, resistance of this magnitude
`is indicative of strong promoters.
`
`Phage M13 de"l'Vatl‘Ve-9 f0’ ¢IW"”fi’i"8 RNA by liquid
`hybfldizflfion
`‘
`Hybridization probes for the bla- and the dhfr-specific RNA were
`obtained by integrating the 752-bp PstI—EcoRI fragment of
`pBR322 and the 672-bp BamHI-—HindI[I fragment of pDSl in-
`
`2988
`
`1.6 :1: 0.4
`0.7 :1: 0.1
`0.6 i: 0.1
`Pm,
`9,5
`1.2 -.l: 0.1
`1.4 i 0.2
`1.8 :l: 0.3
`
`Pmuv ,
`1.4 :l: 0.3
`1.5 e 0.2
`1.4 :l: 0.4
`
`RNA was labelled in vivo with [3H]uridine for 1 min before rifampicin was
`added to the culture (200 iig/ml) and samples were withdrawn at defined
`times
`1:0, 2.0,. 3.0, 5 and I0 mug). RN:
`isolatggeand quantified
`by hybridization against the proper M1 mp9 envatives.
`errors given
`are maximal deviations from the mean of four independent measurements.
`Whereas the dhfr- and car—specific RNA is synthesized under the control of
`the promoters indicated, the bla-specific transcripts are always initiated by
`PM (rightmost column).
`
`to the properly cleaved Ml3mp9. As cat-specific probe the
`865-bp HindIII —)G7aI fragment of pDS1 was integrated into the
`Hindlll — HincH cleaved M13mp8 after the ends of the Xbal
`cleavage site had been filled in. The 752-bp fragment contains
`part of the bla region (570 bp) and a stretch of 200 bp upstream
`of the Ma promoten This region is convenient for identifying
`.
`.
`.
`.
`.
`.
`tranScnpts_ whlch ongmate wlthm the ‘_:l°ned_Promoter fragment
`and are directed towards the bl“ region (Figure 4)-
`Examination of RNA stability in vivo
`;sld(:¢fch[;)s:tinn;lrU‘I11327;n,II.i'erE%'realr:be'lIiIl1 _3n.I£'S‘
`of 5 min and Ion er
`ield vgalu’es deséfibin essential]
`_
`g
`y
`.
`.
`. g .
`.
`y
`y
`state whereas short labelling periods are indicative for newly syn-
`thesized RNA and reflect the rate of RNA synthesis. Since dhfr-
`specific RNAs initiated at different promoters have different 5’
`regions, their stability in vivo cannot be predicted. Therefore
`RNA was labelled in vivo for 5 and 1 min, respectively and the
`ratio of bla- to dhfr-specific RNA was determined. Except for
`
`

`
`Alternate structures for E. coli promoters
`
`lucUV5
`
`CC\l
`LIJ\
`
`D
`
`Fig. 2. Direct visualization of in vivo RNA synthesized under the control of various promoters. Aliquots of 3H-labelled total in vivo RNA were separated in
`4% polyacrylamide/8 M urea gels (20 V/cm;
`l X TBE). The major RNA species seen around position 865 nucleotides are transcripts initiated by the
`promoters indicated and terminated at site I by to. The differences in abundance of these transcripts correlate with the promoter strengths summarized in
`Table Ill. The size differences of the transcripts are due to the differing locations of the promoters within the cloned fragment. The position of the rRNAs are
`indicated (4), the rightmostilane contains RNA prepared from a promoter-less plasmid construct. M denotes size markers.
`
`promoter PH“, differences in the relative abundance of the two
`RNAs was 10% or less (data not shown). In the case of PHZO7,
`however, the apparent promoter strength increased from 20 to
`30 Pb,a—units upon reduction of the labelling time from 5 to 1
`min. The half-life time of the bla-, dhfr- and cat-specific RNA
`was therefore determined with plasmids containing the promoters
`PHM, P55 and Placuvs, respectively. Whereas the stability of the
`transcripts initiated by promoters P15 and PMCUV5 were found to
`be indistinguishable from that of the bla transcript, the RNA
`originating from PH“, is distinctly less stable exhibiting a half-
`life time of only 36 s (Table H).
`
`Determination of in vivo promoter strength
`
`Various promoters integrated into pDSl or pDS2 derivatives were
`transferred into either E. coli C600 or M15. For each promoter
`three independent cultures were grown up and labelled RNA was
`extracted. Each RNA preparation was directly analysed by elec-
`trophoresis (Figure 2) and aliquots were hybridized to the respec-
`tive ssDNA probes, M13mp9 bla, M13mp9 dhfr, M13mp8 cat.
`The RNA/DNA hybrids were then analysed by nitrocellulose ad-
`sorption as well as by electrophoresis of the Sl—resistant material.
`Results of a typical experiment in which four different promoters
`were examined are shown in Figures 3 and 4. The reproducibility
`of the nitrocellulose adsorption method is demonstrated in Figure
`3 where a single RNA preparation has been analysed in duplicate
`
`for each promoter. For low background values it is essential that
`the RNA preparation is virtually free of cellular DNA. The varia-
`tion in the absolute amount of bla—specific RNA is a reflection
`of gene dosage due to changes in plasmid copy number (Stiiber
`and Bujard, 1982). The analyses of the same RNA/DNA hybrids
`after nuclease S1 digestion are shown in Figure 4. Although this
`latter procedure permits only a rough estimate of the relative
`abundance of the different RNA species it is an important con-
`trol for the number of promoters within a cloned fragment. Thus,
`the fragment carrying promoter PH”, directs transcription also
`towards the bla region, as revealed by a second RNA species
`of ~ 800 nucleotides in length (Figure 4). The efficiency of this
`second promoter, which is ~60% of Pbla, was taken into ac-
`count when calculating the strength of PHW. A similar transcript
`was observed for the fragment containing PA, of phage T7 and,
`though in much lower abundance, for Pm}. These promoters
`were therefore analysed in a plasmid containing terminator T1
`at site 3 (Figure 1). As expected, the apparent promoter strength
`was increased for PH”, and PM.
`To examine the reliability of Pbla as an internal standard we
`replaced this promoter by Placuvs and detemiined the hierarchy
`of four promoters with respect to this standard. As seen in Table
`HI there is full agreement between the data derived from both
`sets of experiments. Furthermore as can be seen in Figure 3,
`the amount of Pb1a—specific RNA is not
`influenced by the
`
`2989
`
`

`
`U.Deuschle et al.
`
`
`ss- DNA
`
`RNA/DNA'hybrids retained on nitrocellulose
`
`probe
`(M13 derived)
`
`Autorndiogrum
`
`c p in
`
`
`
`r
`
`bla
`mp9
`—
`l
`
`751.
`
`761
`
`826
`
`829
`
`573
`
`563
`
`712
`
`136
`
`mp9
`
`rim
`
`mp8
`
`cut
`“
`
`mp9
`
`mp8
`
`]___j___
`
`Q Q Q C
`
`25255
`
`39031.
`
`5390
`
`31.531.
`
`0 O O 0
`
`21715
`
`37563
`
`51.19
`
`35207
`
`1% O
`
`1- O
`
`O
`
`O
`
`881
`
`889
`
`12
`
`11
`13
`
`9
`
`21.37
`
`2361.
`
`13
`
`12
`11
`
`15
`
`191
`
`1.58
`
`11
`
`12
`9
`
`8
`
`5161.
`
`5201.
`
`11.
`
`12
`12
`
`10
`
`532
`
`51.1
`
`126
`
`118
`
`81.
`
`18
`
`18
`
`20
`19
`
`23
`
`..
`
`4
`
`._.
`
`Fig. 3. Quantitation of in vivo RNA. RNA labelled in vivo with [3H]uridine was hybridized to single-stranded M13-derived DNA probes which contain bIa—.
`dhfr— and cat—specific sequences, respectively (mp9/bla, mp9/dhfr, mp8/cat). The RNA/DNA hybrids were adsorbed to nitrocellulose in a ‘minifold‘ filtration
`unit. The left pan of the figure displays an autoradiogram of a nitrocellulose sheet in which RNA synthesized from promoters PMS, P15, PD,E2o and PH”,
`integrated in pDSl/tol was analysed in duplicate. The leftmost column shows the DNA probes used; mp9 and mp8 are control DNAs without any plasmid-
`specified sequence. After dissection of the nitrocellulose filter, the radioactivity of the individual probes was detennined (right part of the figure). The
`difference in dhfr- and cat-specific RNA refiects the efficiency of tenninator to whereas the difference in dhfr- and bla—specific RNA reflects the relative
`efficiency of promoters. The variation in absolute counts of the bla—specific RNA is due to differences in plasmid copy number which changes significantly
`during cellular growth (Stiiber and Bujard, 1982).
`
`simultaneous presence of a strong promoter in the plasmid.
`In summary, the relative strengths of 14 promoters were deter-
`mined as exemplified for the four promoters in Figures 3 and
`4 and the results are summarized in Table III. Wherever
`necessary, corrections for additional counter-clockwise transcrip—
`tion (into the bla region) and mRNA half-life time were perfonn-
`ed. These promoter strengths correlate well with the abundance
`of dhfr-specific transcripts in the electrophoretic pattern of total
`in vivo RNA as depicted in Figure 2.
`Comparison ofrRNA synthesis with PN25—directed transcription
`in vivo
`Up to 23% of the newly synthesized RNA in E. coli can be dhfr-
`specific if a strong promoter is carried by pDS2. It was therefore
`of interest to estimate the amount of rRNA synthesized under
`.
`.
`.
`the same conditions and to compare the efficiency of our cloned
`promoters with promoters of the rRNA operon. In analyses as
`depicted in Figure 5, the mass ratio between rRNA and dhfr—
`specific RNA initiated by PN-25 was found to be 4: 1. On a molar
`basis this corresponds to two dhfr-specific transcripts per rRNA
`transcript. At the generation time prevailing during our ex-
`periments (1.5 per h) an E. coli cell contains 2.5 genomes (Sar-
`mientos and Cashel, 1983) and consequenfly 17 -18 rRNA
`operons. Due to the copy number of pDS2 (Stiiber and Bujard,
`1982) the PN25 —dhfr transcription unit is present ~ 80 times in
`such cells. Thus, ribosomal promoters initiate transcription under
`these conditions ~2.5 times more efficiently than PN25.
`
`Discussion
`
`The goal of this study was to accurately measure the in vivo
`strength of a group of well defined promoter sequences and to
`
`2990
`
`Table III. The strength of cloned promoters in vivo
`
`promote,
`
`Pm,
`PD/F10
`Pms
`£025
`P”
`P“
`13::
`PL
`Pm
`P
`P:’°l”V5
`PC:
`pm
`
`Relative etrenfl
`P “ms
`M“
`55 :1: 4“
`56 i 3
`30 i 5
`13 * é
`76 i 9
`20 : 4
`22 i 3
`37 i 7b
`5.7 :i: 0.5
`3 3 i 0 3
`i7 i 2‘
`4 i 0.2
`1
`
`“ms
`
`P
`
`"wvs
`
`12 i 06b
`2.1 3: 0.2
`1 4 i 0 2
`5:2 i 0:4
`
`“Value corrected for short half-life of RNA.
`
`“PL was examined in E. coli C600 since it is suppressed in our
`DZ29l—strain. The relative promoter strengths were calculated from
`hybridization experiments as exemplified in Figure 4. Depending on whether
`Pm or Pmuve were used to initiate transcription of the bla region, promoter
`strengths were calculated in Pm or Placuvs units, respectively.
`
`attempt an interpretation of sequence data based on functional
`infomiation. A prerequisite for this goal was a reliable procedure
`for quantifying promoter strength. With the construction of
`plasmid pDS2 and three phage M13 derivatives (Figure 1) a
`system was established which fulfilled all requirements: (i) pro-
`
`

`
`Alternate structures for E. coli promoters
`
`I
`I
`163 235
`
`I I
`
`dhfr
`
`Fig. 5. Comparison of Pms-directed RNA synthesis with the activity of
`rRNA operons. Cells harbouring a pDS2 derivative containing PMS and to
`in site 1 were pulse labelled with [3H]uridine for l min under standard
`conditions. The total RNA was extracted and separated by PAGE (4%
`polyacrylamide, 8 M urea). Microdensitometric evaluation of the
`autoradiogram revealed the ratio of rRNA to dhfr-specific RNA synthesized,
`which was 4.321 by mass.
`
`influence of 5’ regions on mRNA half—1ife has to be examined,
`as demonstrated for PHZO7 (Table H). Finally, we have observ-
`ed that the activity of some promoters, for example P15, can vary
`significantly depending on the stage of growth and that varia-
`tions of up to :l: 25 % are found when different host cells are us-
`ed. We have therefore limited our studies to two E. coli strains
`
`(C 600 and DZ 291) and to logarithmically growing cultures at
`ODM = 0.6. By observing the above precautions relative pro-
`moter strength can be determined within one host strain to an
`accuracy of :1: 15% ('I‘able HI).
`
`The hierarchy of promoter strength
`The 14 promoters examined can be arranged in a hierarchy ac-
`cording to their activities in vivo under defined conditions (Table
`IH). Using Pbla as standard, the promoter strengths span almost
`two orders of magnitude. The strongest promoter identified is
`PA, which is ~ 76 times Pbla. It is followed by a group of phage
`promoters (PD,F2°, P5207, PL, PN-25) whose activity in vivo is bet-
`ween 30 and 60 units. At the lower end of the hierarchy one
`finds the fully induced Plac, Placuvs as well as Peon. Pm; takes
`an intermediate position; if compared with PMCUV5, one of its
`parent sequences, it is five times more efficient.
`Is there a way to fit this hierarchy into an absolute scale of
`promoter strength? In fast growing cells the operons producing
`rRNA belong to the most intensely transcribed regions. The rate
`of chain initiation per second is estimated to be ~ 1. This is close
`to the maximum which can be expected if a rate of chain elonga-
`tion of 50 nucleotides/s and a space requirement of 50 bp per
`RNA polymerase molecule is assumed. As shown for the rmB
`operon it is a single promoter, namely P,, which is capable of
`directing such intense transcription (Sarmientos and Cashel,
`1983). Our estimates indicate (Figure 5) that PN25 can reach 40%
`of the activity of a fully induced rRNA promoter. This suggests
`that promoters like PM, PH”, and PD/E20 can initiate RNA syn-
`thesis with nearly maximal rates. We are presently examining
`2991
`
`pDS1/‘ro‘l*
`
`M
`
`PH207
`Fiiszo
`PJ5
`PNZS
`1l2I311]2]3l‘|l2l3l’lI2I3
`
`15t.7——
`355—¢
`753/-_-Q
`635
`540"
`
`" -
`——
`
`'
`
`:":’.
`
`p
`
`-
`
`:
`
`_/cut
`"' —dhfr
`—blu
`
`,
`
`ti
`
`ii
`
`1
`
`
`
`Fig. 4. Electrophoretic analysis of S1-resistant hybrids obtained with in vivo
`RNA. The RNA/DNA hybrids analysed are aliquots of the preparations
`shown in Figure 3. For each promoter (PN25, P15, PD,Ezo and Pflm) the
`S1—resistant bla-, dhfr— and cat—specific RNA/DNA hybrids (lanes 1, 2 and
`3, respectively) are seen in the autoradiogram of a polyacrylamide gel. The
`difference in the abundance between bla- and dhfr-specific hybrids reflects
`the difference in efficiency of these two promoters, whereas the effect of
`terminator to can be seen by comparing dhfr-specific with cat-specific
`hybrids. With Pmo, two transcripts are observed in lane 1, of which the
`lower one represents RNA initiated from Pm. The upper band around
`position 750 nucleotides is due to a promoter within the cloned fragment
`which initates transcription towards the bla region. The positions of bla-,
`dhfr— and cat-specific hybrids are indicated; M denotes size markers.
`
`moter activity is measured at the RNA level independent of any
`translational effects; (ii) an internal standard compensates for
`physiological variations such as gene dosage, etc; (iii) differences
`in RNA stability can easily be detected and properly corrected;
`(iv) the quantitation of RNA by hybridization in solution followed
`by adsorption to nitrocellulose is rapid and highly reproducible.
`The data obtained allow for the first time arrangement of 14
`promoters utilized by E. coli RNA polymerase into a hierarchy,
`based on their primary function namely, to commence RNA syn-
`thesis in viva.
`
`Determination of promoter strength
`
`The accuracy and reproducibility of the hybridization procedure
`described here is demonstrated in Figure 3. For individual in
`vivo RNA preparations isolated from the same culture the stan-
`dard deviations are < 3%. larger deviations may arise for several
`reasons: (i) the RNA preparation still contains DNA; (ii) dhfr
`transcripts with different 5’ and/or 3’ ends are compared; (iii)
`promoters are compared at different physiological states of the
`cell, or in different strains.
`Following the protocol described, highly purified RNA is ob-
`tained yielding results as shown in Figure 3. The 3’ end of dhfr
`transcripts can be kept invariable by utilizing the same terminator
`at the same site (in our study to at site 1). However, the potential
`
`

`
`PD/E20
`PH207
`PN25
`P625
`PJ5
`
`PA1
`PA9
`PA3
`
`PL
`
`Plac
`Placuvs
`Ptacl
`Pcon
`
`Pbla
`
`U.Deuschle et al.
`
`a.
`
`T5
`
`T7
`
`A,
`
`b.
`
`c
`
`d.
`
`
`
`TTAGGCACCCCAGGCTTTACA TTTATGCTTCCGGCTGG
`CTAGGCACCCCAGG TTTACA TTTATGCTTCCGGCTG
`TTCTGAAATGAGCTE TGACA
`ATTAATCATCGGCTC
`AATTCACCGTCGTT TTGACA TTTTTAAGCTTGGCGG
`
`TTTTTTCTAAATAC TTCAAA TATGTATCCGCTCATGA'
`
`GTGTGQAATTGTGAGCGGATAACAAT
`GTGTGQAATTGTGAGCGGATAACAAT
`GTGTGQAATTGTGAGCGGATAACAAT
`GGTACQAIAAGGAGGTGGATCCGGCA
`
`'ACCCIQATAAATGCTTCAATAATAT
`
`T5"ear1y" —-—aAAAAa--—--TTGCTa ———————————————— --TATAAT—--—TCAT-—-TTGA ---Pu-----
`
`PL
`
`----CTCTGG———--TTGACA ---------------- -—GATACT--———CAT—--CAGGACGCACTGACC
`
`H/MCC
`
`-——-—a ————— ——tcTTGACa——-t — - - - - - - -—t—tg—TAtAaT—-—--cat ———————————————— --
`
`Fig. 6. Promoter sequences. (a) The sequences of the 14 promoters studied are aligned with respect to the first T of the -33 hexarner and the last T of the
`-10 hexamer. The conserved regions around -33 and -10 are boxed, the A clusters around -43 and the first nucleotide transcribed (+1) are underlined.
`The motifs typical for the downstream region of ‘early’ T5 promoters, TTGA around +7 and the run of purines are indicated within the sequences of Pmo,
`and PNZ5. (b,c) Sequence elements of two prototype promoters. The features characterizing an ‘early’ T5 promoter were derived from six such signals (Gentz
`and Bujard, 1985). In contrast, corresponding sequence elements of PL, a promoter comparable in strength to an ‘early’ T5 promoter, is shown in c. Whereas
`the AT content is -80% for a typical ‘early’ T5 promoter, it is only 47% for PL. (d) Essential features of a consensus sequence derived from ~ 150
`different promoters (Hawly and McClure, 1983).
`
`whether this upper limit can indeed be reached by a single pro-
`moter sequence like PA, or whether tandem arrangements of pro-
`moters, as found within the early region of phage T7 and T5,
`are required. In any case our data strongly suggest that some
`of the promoters studied here encode a program which is close
`to the functional optimum of a single sequence.
`
`Structural considerations
`
`The information available now allows us to examine our collec-
`
`tion of promoter sequences with respect to functional properties.
`If we define a strong promoter as a sequence having at least 20
`times the activity of Pbla in vivo, then, with the exception of P15
`all phage promoters of this collection are strong promoters (Table
`HI). None of these strong promoters contains simultaneously both
`of the so-called ‘canonical’ hexamers around position — 10
`(TATAAT) and -33 (TTGACA), respectively (Figure 6). All
`phage promoters — with the exception of PL — are rich in AT
`and contain striking blocks of AT pairs around position -43.
`Promoters PH”, and PNZ5 which control ‘early’ genes of phage
`T5 show, in addition, a structure typical for this class of signals
`(Gentz and Bujard, 1985) namely the tetrameric sequence TTGA
`around +7 and a stretch of purines spanning from around +9
`to +20. We have derived a consensus sequence from six ‘early’
`2992
`
`T5 promoters, and the most prominent sequence elements are
`given in Figure 6. There are significant differences between a
`typical ‘early’ T5 promoter and a consensus sequence, as describ-
`ed by Hawley and McClure (1983). The most striking features
`are the homologies outside of the ‘classical’ (position +1 and
`-35, Figure 6) promoter region which strongly suggests con-
`tributions of these regions to promoter function. In the accom-
`panying paper (Karnrnerer et al. , 1986) we present evidence that
`indeed information encoded in one of these regions can contribute
`significantly to promoter activity.
`Most interesting is the comparison between PL from phage
`lambda and a typical ‘early’ promoter from phage T5 like PHZW.
`In vivo both promoters belong to the most efficient signals with
`very similar overall strength (Table III). Nevertheless, they dif-
`fer clearly in their structure. (i) The AT-content (between -50
`and +20) is 81% for P520, but only 47% for PL. (ii) The most
`conserved hexameric sequence around -10 is ‘consensus‘—like
`for PH207 but shows two deviations in PL. Similarly, Pm." dif-
`fers from the ‘consensus’ around -33 exhibiting a sequence fre-
`quently found with strong T5 promoters (TTGCT) whereas PL
`is ‘consensus’-like in this region. (iii) Around -43 Pflm, like
`many strong promoters, has a block of AT pairs; PL, however,
`is GC rich (Figure 6). (iv) There are no similarities between the
`
`-50
`
`-43
`
`-33
`
`-7
`
`+1
`
`+7
`
`+20
`
`AACTGQAAAAAIAGT TGACA CCCTAGCCGATAGGCTT AAGAT»TACCQAGTTCGATGAGAGCGATAAC
`TTTTAAAAAATTCATTTGCTA AACGCTTCAAATTCTCG ATAAT TACTTQAIAAAHIEH AAACAAAAA
`TCATAAAAAATTTA TTGCTT TCAGGAAAATTTTTCTG ATAAT GATTCATAAAT
`GAGAGGAGTT
`TGAAAAATAAAATTCTTGATA AAATTTTCCAATACTAT ATAAT TTGTIATTAAAGAGGAGAAATTAAC
`ATATAAAAACCGTTATTGACA CAGGTGGAAATTTAGAA ATACT'TTAGIAAACCTAATGGATCGACCTT
`
`ACAGCCAICGAGAGGGACACGGCGA
`TTATCAAAAAGAGT TTGACT TAAAGTCTAACCTATA GATAC
`CACGAAAAACAGGT TTGACAACATGAAGTAACATGCAE AAGA *CAAATCQCTAGGTAACACTAGCAGC
`GGTGAAACAAAACG&TTGACA ACATGAAGTAAACACG
`ACGATGTACCACAIGAAACGACAGTGAGTCA
`
`TTATCTCTGGCGGTGTTGACA TAAATACCACTGGCGG GATAC GAGCAQATCAGCAGGACGCACTGACC
`
`
`
`

`
`two promoters downstream of +1.
`Furthermore, PL and a typical ‘early’ T5 promoter like PN25
`differ drastically in their rate of complex formation with E. coli
`RNA polymerase as well as in their in vitro strength if compared
`under competitive conditions and the same holds when PA, is
`compared with PN25 (von Gabain and Bujard, 1979). In both
`cases PN25 readily out-competes PL and PM.
`Complex signals like promoters encode a program for a multi-
`step process. Such a process can be lirriited at various levels
`(Kammerer et al. , 1986) and sequences optimized in different
`ways can result in signals with identical overall properties, i.e.
`promoter strength. Therefore, unlike, for example operator/
`repressor systems where a protein has merely to select and to
`bind a specific sequence, a unique consensus sequence describ-
`ing the functional programs of promoters should not exist. The
`promoters PHZM, PA, and PL resemble each other closely in their
`overall function in vivo, i.e. they commence RNA synthesis with
`close to maximal rates. Nevertheless, they differ clearly in their
`sequences and in individual parameters of promoter function.
`Together with the results reported in the accompanying publica-
`tion (Karnmerer et al. , 1986) this supports the hypothesis pro-
`posed earlier (Bujard, 1980), namely that optimal function of
`complex signals like promoters can be encoded in alternate but
`equivalent sequences.
`
`Materials and methods
`Nucleic acids
`
`Ml3mp8 and M13mp9 (Messing and Vieira, 1982) were obtained from R.Cor-
`tese, EMBL Heidelberg. The pDSl cloning system and its nomenclature has been
`described in detail previously (Stiiber and Bujard, 1982). An earlier version of
`M13mp9 contained a portion of pBR322 DNA (position 2348 -2718, Sutcliffe,
`1979). This sequence was removed since it increased the background in hybridiza-
`tion experiments. Phage M13 derivatives were grown in E. coli 71-18 cells and
`isolated as described by Herrmarm et al. (1980). Phage DNA was extracted by
`hot phenol (65 °C) followed by chloroform/isoamylalcohol (24: 1) extraction before
`it was ethanol precipitated and redissolved in hybridization buffer. Plasmids as
`well as RF-DNA from M13 phages were prepared as described before (Stiiber
`and Bujard, 1982), or by alkaline extraction (Bimboim and Doly, 1979). Various
`DNA sequences were cloned using standard procedures (Maniatis et al. , 1982).
`E. coli was transfomied with DNA according to Monison (1979). All DNA con-
`structs were verified by sequence analysis.
`Promoters.The identification and characterization of promoter sequences of phage
`T5 (Pmm, PD,E,,,, PNZ5, P525, P15) has been reported (Gentz and Bujard, 1985).
`The major ‘early’ promoters of phage T7 (PM, PM, PM) were cloned as frag-
`ments delineated within the T7 genome by the nucleotide positions 258-543,
`547-651 and 657-763, respectively (Dunn and Studier, 1983). Promoter PL
`from phage lambda was isolated as an Haelll fragment from plasmid pPLC28
`(Remault et al. , 1981). Pm, was transferred from plasmid ptac 12 (Amrrian et
`al., 1983) to the pDSl system as a Pstl/}0ioI fragment after insertion of H101
`linkers at the Pvull cleavage site downstream of the promoter. Pm was recovered
`from p