Downloaded from
`
`http://mmbr.asm.org/
`
` on October 5, 2015 by guest
`
`MICROBIOLOGICAL REVIEWS, Sept. 1996, p. 512–538
`0146-0749/96/$04.00⫹0
`Copyright 䉷 1996, American Society for Microbiology
`
`Vol. 60, No. 3
`
`Strategies for Achieving High-Level Expression
`of Genes in Escherichia coli†
`SAVVAS C. MAKRIDES*
`Department of Molecular Biology, T Cell Sciences, Inc., Needham,
`Massachusetts 02194
`
`INTRODUCTION .......................................................................................................................................................512
`CONFIGURATION OF EFFICIENT EXPRESSION VECTORS........................................................................513
`TRANSCRIPTIONAL REGULATION.....................................................................................................................513
`Promoters.................................................................................................................................................................513
`Transcriptional Terminators.................................................................................................................................515
`Transcriptional Antiterminators...........................................................................................................................515
`Tightly Regulated Expression Systems ................................................................................................................516
`TRANSLATIONAL REGULATION .........................................................................................................................516
`mRNA Translational Initiation.............................................................................................................................516
`Translational Enhancers........................................................................................................................................517
`mRNA Stability .......................................................................................................................................................517
`Translational Termination ....................................................................................................................................518
`PROTEIN TARGETING............................................................................................................................................518
`Cytoplasmic Expression .........................................................................................................................................518
`Periplasmic Expression..........................................................................................................................................520
`Extracellular Secretion...........................................................................................................................................521
`FUSION PROTEINS..................................................................................................................................................521
`MOLECULAR CHAPERONES ................................................................................................................................522
`CODON USAGE .........................................................................................................................................................524
`PROTEIN DEGRADATION......................................................................................................................................524
`FERMENTATION CONDITIONS ...........................................................................................................................525
`CONCLUSIONS AND FUTURE DIRECTIONS....................................................................................................526
`ACKNOWLEDGMENTS ...........................................................................................................................................527
`REFERENCES ............................................................................................................................................................527
`
`INTRODUCTION
`
`The choice of an expression system for the high-level pro-
`duction of recombinant proteins depends on many factors.
`These include cell growth characteristics, expression levels,
`intracellular and extracellular expression, posttranslational
`modifications, and biological activity of the protein of interest,
`as well as regulatory issues in the production of therapeutic
`proteins (191, 254). In addition, the selection of a particular
`expression system requires a cost breakdown in terms of pro-
`cess, design, and other economic considerations. The relative
`merits of bacterial, yeast, insect, and mammalian expression
`systems have been examined in detail in an excellent review by
`Marino (362). In addition, Datar et al. (121) have analyzed the
`economic issues associated with protein production in bacterial
`and mammalian cells.
`The many advantages of Escherichia coli have ensured that it
`remains a valuable organism for the high-level production of
`recombinant proteins (177a, 197, 254, 362, 406, 426, 510).
`However, in spite of the extensive knowledge on the genetics
`and molecular biology of E. coli, not every gene can be ex-
`pressed efficiently in this organism. This may be due to the
`unique and subtle structural features of the gene sequence, the
`
`* Mailing address: Department of Molecular Biology, T Cell Sci-
`ences, Inc., 119 4th Ave., Needham, MA 02194.
`† This review is dedicated to the memory of William John Steele, an
`inspired scientist, a great man, mentor, and friend, who died on 8
`December 1995. The world is a better place because of him.
`
`stability and translational efficiency of mRNA, the ease of
`protein folding, degradation of the protein by host cell pro-
`teases, major differences in codon usage between the foreign
`gene and native E. coli, and the potential toxicity of the protein
`to the host. Fortunately, some empirical “rules” that can guide
`the design of expression systems and limit the unpredictability
`of this operation in E. coli have emerged. The major drawbacks
`of E. coli as an expression system include the inability to per-
`form many of the posttranslational modifications found in eu-
`karyotic proteins, the lack of a secretion mechanism for the
`efficient release of protein into the culture medium, and the
`limited ability to facilitate extensive disulfide bond formation.
`On the other hand, many eukaryotic proteins retain their full
`biological activity in a nonglycosylated form and therefore can
`be produced in E. coli (see, e.g., references 170, 342, and 486).
`In addition, some progress has been made in the areas of
`extracellular secretion and disulfide bond formation, and these
`will be examined.
`The objectives of this review are to integrate the extensive
`published literature on gene expression in E. coli, to focus on
`expression systems and experimental approaches useful for the
`overproduction of proteins, and to review recent progress in
`this field. Areas that have been covered in detail in recent
`reviews are included in abbreviated form in order to present
`their key conclusions and to serve as a source for further
`reading. As a matter of definition, the terms “periplasmic ex-
`pression” and “extracellular secretion” will be used to refer to
`the targeting of protein to the periplasm and the culture me-
`dium, respectively, to avoid confusion.
`
`512
`
`BEQ 1023
`Page 1
`
`

`
`VOL. 60, 1996
`
`HIGH-LEVEL GENE EXPRESSION IN E. COLI
`
`513
`
`Downloaded from
`
`http://mmbr.asm.org/
`
` on October 5, 2015 by guest
`
`FIG. 1. Schematic presentation of the salient features and sequence elements of a prokaryotic expression vector. Shown as an example is the hybrid tac promoter
`(P) consisting of the ⫺35 and ⫺10 sequences, which are separated by a 17-base spacer. The arrow indicates the direction of transcription. The RBS consists of the SD
`sequence followed by an A⫹T-rich translational spacer that has an optimal length of approximately 8 bases. The SD sequence interacts with the 3⬘ end of the 16S rRNA
`during translational initiation, as shown. The three start codons are shown, along with the frequency of their usage in E. coli. Among the three stop codons, UAA
`followed by U is the most efficient translational termination sequence in E. coli. The repressor is encoded by a regulatory gene (R), which may be present on the vector
`itself or may be integrated in the host chromosome, and it modulates the activity of the promoter. The transcription terminator (TT) serves to stabilize the mRNA and
`the vector, as explained in the text. In addition, an antibiotic resistance gene, e.g., for tetracycline, facilitates phenotypic selection of the vector, and the origin of
`replication (Ori) determines the vector copy number. The various features are not drawn to scale.
`
`CONFIGURATION OF EFFICIENT
`EXPRESSION VECTORS
`The construction of an expression plasmid requires several
`elements whose configuration must be carefully considered to
`ensure the highest levels of protein synthesis (22, 64, 120, 142,
`355, 538, 612). The essential architecture of an E. coli expres-
`sion vector is shown in Fig. 1. The promoter is positioned
`approximately 10 to 100 bp upstream of the ribosome-binding
`site (RBS) and is under the control of a regulatory gene, which
`may be present on the vector itself or integrated in the host
`chromosome. Promoters of E. coli consist of a hexanucleotide
`sequence located approximately 35 bp upstream of the tran-
`scription initiation base (⫺35 region) separated by a short
`spacer from another hexanucleotide sequence (⫺10 region)
`(174, 232, 236, 344, 465). There are many promoters available
`for gene expression in E. coli, including those derived from
`gram-positive bacteria and bacteriophages (Table 1). A useful
`promoter exhibits several desirable features: it is strong, it has
`a low basal expression level (i.e., it is tightly regulated), it is
`easily transferable to other E. coli strains to facilitate testing of
`a large number of strains for protein yields, and its induction is
`simple and cost-effective (612).
`Downstream of the promoter is the RBS, which spans a
`region of approximately 54 nucleotides bound by positions ⫺35
`(⫾2) and ⫹19 to ⫹22 of the mRNA coding sequence (269).
`The Shine-Dalgarno (SD) site (514, 515) interacts with the 3⬘
`end of 16S rRNA during translation initiation (133, 532). The
`distance between the SD site and the start codon ranges from
`5 to 13 bases (93), and the sequence of this region should
`eliminate the potential of secondary-structure formation in the
`mRNA transcript, which can reduce the efficiency of transla-
`tion initiation (198, 229). Both 5⬘ and 3⬘ regions of the RBS
`exhibit a bias toward a high adenine content (140, 499, 502).
`The transcription terminator is located downstream of the
`coding sequence and serves both as a signal to terminate tran-
`scription (465) and as a protective element composed of stem-
`loop structures, protecting the mRNA from exonucleolytic
`degradation and extending the mRNA half-life (35, 37, 147,
`227, 249, 597).
`In addition to the above elements that have a direct impact
`on the efficiency of gene expression, vectors contain a gene that
`confers antibiotic resistance on the host to aid in plasmid
`selection and propagation. Ampicillin is commonly used for
`this purpose; however, for the production of human therapeu-
`
`tic proteins, other antibiotic resistance markers are preferable
`to avoid the potential of human allergic reactions (42). Finally,
`the copy number of plasmids is determined by the origin of
`replication. In specific cases, the use of runaway replicons
`results in massive amplification of plasmid copy number con-
`comitant with higher yields of plasmid-encoded protein (387,
`415). In other cases, however, there appeared to be no advan-
`tage in using higher-copy-number plasmids over pBR322-
`based vectors (612). Furthermore, Vasquez et al. (572) re-
`ported that
`increasing the copy number of
`the plasmid
`decreased the production of trypsin in E. coli and Minas and
`Bailey (379) found that the presence of strong promoters on
`high-copy-number plasmids severely impaired cell viability.
`
`TRANSCRIPTIONAL REGULATION
`
`Promoters
`A promoter for use in E. coli (Table 1) should have certain
`characteristics to render it suitable for high-level protein syn-
`thesis (207, 612). First, it must be strong, resulting in the
`accumulation of protein making up 10 to 30% or more of the
`total cellular protein.
`Second, it should exhibit a minimal level of basal transcrip-
`tional activity. Large-scale gene expression preferably employs
`cell growth to high density and minimal promoter activity,
`followed by induction or derepression of the promoter. The
`tight regulation of a promoter is essential for the synthesis of
`proteins which may be detrimental to the host cell (see, e.g.,
`references 68, 137, 544, 563, and 599). For example, the toxic
`rotavirus VP7 protein effectively kills cells and must be pro-
`duced under tightly regulated conditions (592). However, in
`some cases, promoter stringency is inconsequential, because
`even the smallest amount of gene product drastically curtails
`bacterial survival because of its severe toxicity (615). For ex-
`ample, molecules that inactivate ribosomes or destroy the
`membrane potential would be lethal. Toxicity to the host is not
`restricted to foreign genes but may also result from the over-
`expression of certain native genes, such as the traT gene, which
`encodes an outer membrane lipoprotein (423), the EcoRI re-
`striction endonuclease in the absence of the corresponding
`protective EcoRI modification methylase (423), and the lon
`gene (558). Furthermore, incompletely repressed expression
`systems can cause plasmid instability, a decrease in cell growth
`rate, and loss of recombinant protein production (40, 98, 374).
`
`BEQ 1023
`Page 2
`
`

`
`514
`
`MAKRIDES
`
`MICROBIOL. REV.
`
`TABLE 1. Promoters used for the high-level expression of genes in E. coli
`
`Promoter (source)
`
`Regulation
`
`Induction
`
`Reference(s)
`
`lac (E. coli)
`
`trp (E. coli)
`lpp (E. coli)
`phoA (E. coli)
`recA (E. coli)
`araBAD (E. coli)
`proU (E. coli)
`cst-1 (E. coli)
`tetA (E. coli)
`cadA (E. coli)
`nar (E. coli)
`tac, hybrid (E. coli)
`
`trc, hybrid (E. coli)
`
`lacI, lacIq
`lacI(Ts),a lacIq(Ts)a
`lacI(Ts)b
`
`phoB (positive), phoR (negative)
`lexA
`araC
`
`cadR
`fnr (FNR, NARL)
`lacI, lacIq
`lacId
`lacI, lacIq
`lacI(Ts),a lacIq(Ts)a
`lacI
`lacI, lacIq
`
`␭cIts857
`
`␭cIts857
`␭cIts857
`lacIq
`␭cIts857, lacIq
`lacIq
`lacIq, lacI
`
`lacIq
`
`Downloaded from
`
`http://mmbr.asm.org/
`
` on October 5, 2015 by guest
`
`IPTG
`Thermal
`Thermal
`Trp starvation, indole acrylic acid
`IPTG, lactosec
`Phosphate starvation
`Nalidixic acid
`L-Arabinose
`Osmolarity
`Glucose starvation
`Tetracycline
`pH
`Anaerobic conditions, nitrate ion
`IPTG
`Thermal
`IPTG
`Thermal
`IPTG
`IPTG
`
`Thermal
`Reduced temperature (⬍20⬚C)
`Reduced temperature (⬍20⬚C)
`Thermal
`Thermal
`IPTG
`Thermal, IPTG
`IPTG
`IPTG
`T4 infection
`IPTG
`Oxygen, cAMP-CAPe
`
`17, 18, 221, 460, 610
`234
`604
`365, 470, 549, 612
`128a, 142, 185, 275, 401
`84, 274, 291, 306, 382, 562
`145, 260, 428, 516
`554
`247
`564
`125, 523
`102, 480, 561
`335
`7, 123, 471
`603
`65
`4, 9
`261, 263
`186
`366
`43, 80, 129, 130, 240, 454
`187, 433
`187, 206, 433, 551
`150, 493
`537, 548
`141, 190, 239
`375
`190, 605
`71, 390
`143, 210
`605
`304, 305
`1,256, 349
`
`lpp-lac, hybrid (E. coli)
`Psyn, synthetic (E. coli)
`Starvation promoters (E. coli)
`pL (␭)
`pL-9G-50, mutant (␭)
`cspA (E. coli)
`pR, pL, tandem (␭)
`T7 (T7)
`T7-lac operator (T7)
`␭pL, pT7, tandem (␭, T7)
`T3-lac operator (T3)
`T5-lac operator (T5)
`T4 gene 32 (T4)
`nprM-lac operator (Bacillus spp.)
`VHb (Vitreoscilla spp.)
`Protein A (Staphylococcus aureus)
`a lacI gene with single mutation, Gly-187 3 Ser (72).
`b lacI gene with three mutations, Ala-241 3 Thr, Gly-265 3 Asp, and Ser-300 3 Asn (604).
`c The constitutive lpp promoter (Plpp) was converted into an inducible promoter by insertion of the lacUV5 promoter/operator region downstream of Plpp. Thus,
`expression occurs only in the presence of a lac inducer (142).
`d Wild-type lacI gene.
`e cAMP-CAP, cyclic AMP-catabolite activator protein.
`
`Lanzer and Bujard carried out extensive studies on the com-
`monly used lac-based promoter-operator systems and demon-
`strated up to 70-fold differences in the level of repression when
`the operator was placed in different positions within the pro-
`moter sequence (328). Thus, when the 17-bp operator was
`placed between the ⫺10 and ⫺35 hexameric regions, a 50- to
`70-fold-greater repression was caused than when the operator
`was placed either upstream of the ⫺35 region or downstream
`of the ⫺10 site (328).
`A third important characteristic of a promoter is its induc-
`ibility in a simple and cost-effective manner. The most widely
`used promoters for large-scale protein production use thermal
`induction (␭ pL) or chemical inducers (trp) (Table 1). The
`isopropyl-␤-D-thiogalactopyranoside (IPTG)-inducible hybrid
`promoters tac (123) or trc (65) are powerful and widely used
`for basic research. However, the use of IPTG for the large-
`scale production of human therapeutic proteins is undesirable
`because of its toxicity (159) and cost. These drawbacks of IPTG
`have until now precluded the use of the tac or trc promoter
`from the production of human therapeutic proteins and ren-
`dered the large-scale expression of proteins for basic research
`prohibitively expensive. The availability of a mutant lacI(Ts)
`gene that encodes a thermosensitive lac repressor (72) now
`permits the thermal induction of these promoters (4, 9, 234). In
`addition, the new vectors exhibit tight regulation of the trc
`
`promoter at 30⬚C (9). Two different lac repressor mutants that
`are thermosensitive (586, 604) as well as IPTG inducible (586)
`have recently been described. Although the wild-type lacI gene
`can be thermally induced (602, 603), this system is not tightly
`regulated and cannot be used in lacIq strains, since a temper-
`ature shift does not override the tight repression caused by the
`overproduction of the lac repressor (603). Thus, this system is
`limited to the production of some proteins that are not detri-
`mental to the host cell.
`Cold-responsive promoters, although much less extensively
`studied than many of the other promoters included here, have
`been shown to facilitate efficient gene expression at reduced
`temperatures. The activity of the phage ␭ pL promoter was
`highest at 20⬚C and declined as the temperature was raised
`(187). This cold response of the pL promoter is positively
`regulated by the E. coli integration host factor, a sequence-
`specific, multifunctional protein that binds and bends DNA
`(164, 165, 188). The promoter of the major cold shock gene
`cspA (206, 551) was similarly demonstrated to be active at
`reduced temperatures (187). Molecular dissection of the cspA
`and pL promoters led to the identification of specific DNA
`regions involved in the enhancement of transcription at lower
`temperatures; this has allowed the development of pL deriva-
`tives that are highly active at temperatures below 20⬚C (433).
`The rationale behind the use of cold-responsive promoters for
`
`BEQ 1023
`Page 3
`
`

`
`Downloaded from
`
`http://mmbr.asm.org/
`
` on October 5, 2015 by guest
`
`VOL. 60, 1996
`
`HIGH-LEVEL GENE EXPRESSION IN E. COLI
`
`515
`
`gene expression is based on the proposition that the rate of
`protein folding will be only slightly affected at about 15 to 20⬚C,
`whereas the rates of transcription and translation, being bio-
`chemical reactions, will be substantially decreased. This, in
`turn, will provide sufficient time for protein refolding, yielding
`active proteins and avoiding the formation of inactive protein
`aggregates, i.e., inclusion bodies, without reducing the final
`yield of the target protein (433). It would be interesting to
`compare the transcriptional activities of other promoters de-
`rived from cold shock genes (288, 402).
`Other promoters that have been characterized recently (Ta-
`ble 1) possess attractive features and should provide additional
`options for high-level gene expression systems. For example,
`the pH promoter (102, 561) is very strong: recombinant pro-
`teins are produced at levels of up to 40 to 50% of the total
`cellular protein (480). This expression level, however, will
`probably vary for different genes, because protein synthesis
`depends on translational efficiency as well as promoter
`strength.
`E. coli promoters are usually considered in terms of a core
`region composed of the ⫺10 and ⫺35 hexameric sequences
`including a 15- to 19-bp spacer between the two hexamers
`(344). However, it has been proposed that elements outside
`the core region stimulate promoter activity (134). Many studies
`have demonstrated that sequences upstream of the core pro-
`moter increase the rate of transcription initiation in vivo (172,
`213, 264, 290, 618). Gourse and colleagues have shown that a
`DNA sequence, the UP element, located upstream of the ⫺35
`region of the E. coli rRNA promoter rrnB P1, stimulates tran-
`scription by a factor of 30 in vitro and in vivo (290, 453, 468).
`The UP element functions as an independent promoter mod-
`ule because when it is fused to other promoters such as lacUV5,
`it stimulates transcription (453, 468). Upstream activation in E.
`coli and other organisms has been reviewed in detail (110). The
`ability of the UP element to act as a transcriptional enhancer
`when fused to heterologous promoters may be of general util-
`ity in high-level expression systems.
`Although the extraordinary strength of the rRNA promoters
`P1 and P2 is well documented (173, 414), these promoters have
`not been exploited for the high-level production of proteins in
`E. coli, mainly because their regulation is more difficult. The in
`vivo synthesis of rRNA is subject to growth rate control (213),
`and P1 and P2 are active during periods of rapid cell growth
`and are downregulated when cells are in the stationary phase
`of growth. Therefore, the rRNA promoters would be contin-
`uously active or “leaky” during the preinduction phase. In vivo
`P2 is the weaker, less inducible promoter in rapidly growing
`cells. However, when uncoupled from P1, the P2 promoter
`shows increased activity (up to 70% of that of P1) and becomes
`sensitive to the stringent response, indicating that in its native
`tandem context, P2 is partially occluded (173, 289). Brosius
`and Holy (66) inserted the lac operator sequence downstream
`of the rrnB rRNA P2 promoter and achieved repression of P2
`in strains harboring the lacIq gene. Transcriptional activity was
`measured by the production of chloramphenicol acetyltrans-
`ferase and by the expression of the 4.5S RNA. However, the P2
`construction was only half as active as the tac promoter, and
`furthermore, when the rrnB P1 promoter was placed upstream
`of the P2 promoter, transcriptional repression was incomplete
`(66).
`It is tempting to speculate that rRNA promoters could be
`tightly regulated by using the concept of inverted promoters
`(see the section on tightly regulated expression systems, be-
`low). Thus, a rRNA promoter could be cloned upstream of the
`gene of interest but in the opposite transcriptional direction.
`The use of ␭ integration sites and a regulated ␭ integrase
`
`would facilitate the inversion of the promoter for induction,
`and the presence of strong transcription terminators upstream
`of the highly active promoter would prevent destabilization of
`the vector during the preinduction phase.
`
`Transcriptional Terminators
`In prokaryotes, transcription termination is effected by two
`different types of mechanisms: Rho-dependent transcription
`termination depends on the hexameric protein rho, which
`causes the release of the nascent RNA transcript from the
`template. In contrast, rho-independent termination depends
`on signals encoded in the template, specifically, a region of
`dyad symmetry that encodes a hairpin or stem-loop structure in
`the nascent RNA and a second region that is rich in dA and dT
`and is located 4 to 9 bp distal to the dyadic sequence (83, 122,
`439, 455, 456, 465, 594, 609). Although often overlooked in the
`construction of expression plasmids, efficient transcription ter-
`minators are indispensable elements of expression vectors, be-
`cause they serve several important functions. Transcription
`through a promoter may inhibit its function, a phenomenon
`known as promoter occlusion (5). This interference can be
`prevented by the proper placement of a transcription termina-
`tor downstream of the coding sequence to prevent continued
`transcription through another promoter. Similarly, a transcrip-
`tion terminator placed upstream of the promoter that drives
`expression of the gene of interest minimizes background tran-
`scription (413). It is also known that transcription from strong
`promoters can destabilize plasmids as a result of overproduc-
`tion of the ROP protein involved in the control of plasmid copy
`number as a result of transcriptional readthrough into the
`replication region (539). In addition, transcription terminators
`enhance mRNA stability (237, 404, 597) and can substantially
`increase the level of protein production (237, 572). Particularly
`effective are the two tandem transcription terminators T1 and
`T2, derived from the rrnB rRNA operon of E. coli (67), but
`many other sequences are also quite effective.
`
`Transcriptional Antiterminators
`In bacteria, many operons involved in amino acid biosynthe-
`sis contain transcriptional attenuators at the 5⬘ end of the first
`structural gene. The attenuators are regulated by the amino
`acid products of the particular operon. Thus, the availability of
`the cognate charged tRNA leads to the formation of a second-
`ary structure in the nascent transcript followed by ribosome
`stalling. In the absence of the cognate charged tRNA, an an-
`titerminator structure which prevents formation of the RNA
`hairpin in the terminator and prevents transcriptional termi-
`nation is formed (325). The antiterminator element that en-
`ables RNA polymerase to override a rho-dependent termina-
`tor in the ribosomal RNA operons has been identified and is
`referred to as boxA (41, 341). Transcriptional antitermination
`is a remarkably complex process that involves many known and
`as yet unidentified host factors. This topic has been covered in
`great detail in two excellent recent reviews (110, 456). Here, we
`will briefly consider the use of antitermination elements that
`are useful in the expression of heterologous genes in E. coli.
`One of the more powerful and widely used expression sys-
`tems in E. coli makes use of the phage T7 late promoter (537,
`548). The activity of this system depends on a transcription unit
`that supplies the T7 RNA polymerase, whose tight repression
`is essential to avoid leakiness of the T7 promoter. Several
`approaches have been used to regulate the expression of the
`T7 polymerase, and each has its own unique disadvantages
`(374). Mertens et al. (374) addressed this problem by con-
`structing a reversibly attenuated T7 RNA polymerase expres-
`
`BEQ 1023
`Page 4
`
`

`
`Downloaded from
`
`http://mmbr.asm.org/
`
` on October 5, 2015 by guest
`
`516
`
`MAKRIDES
`
`MICROBIOL. REV.
`
`sion cassette based on ␭ pL regulation. Thus, the basal expres-
`sion level of the T7 polymerase was attenuated by inserting
`three tandemly arranged transcription terminators between
`the promoter and the gene encoding the T7 polymerase. For
`induction, the phage ␭-derived nutL-dependent antitermina-
`tion function was also incorporated to override the transcrip-
`tion block. Alternatively, an IPTG-inducible promoter was
`similarly used, allowing conditional reversion of attenuation
`upon induction (374).
`The transcriptional antitermination region from the E. coli
`rrnB rRNA operon has been used in the expression vector
`pSE420, which utilizes the trc promoter (64). The rationale in
`this case was to facilitate transcription through areas of severe
`secondary structure, thus reducing the possibility of premature
`transcription termination by the host RNA polymerase. In this
`case, however, the presence of the rrnB antiterminator is ap-
`parently ineffective (64a).
`
`Tightly Regulated Expression Systems
`The advantages of tightly regulated promoters (see the sec-
`tion on promoters, above) have led to the design of many
`ingenious and highly repressible expression systems that are
`particularly useful for the expression of genes whose products
`are detrimental to host growth. The various approaches in-
`clude the use of a “plating” method (544), the increase of the
`repressor-to-operator ratio (9, 391),
`induction by infection
`with mutant phage (68, 137), attenuation of promoter strength
`on high-copy-number vectors (587), the use of transcription
`terminators (374, 375, 413) in combination with antitermina-
`tors (374), the use of an inducible promoter within a copy-
`number-controllable plasmid (558), “cross-regulation” systems
`(97, 98), cotransformation of plasmids utilizing the SP6 RNA
`polymerase (473), and the use of antisense RNA complemen-
`tary to the mRNA of the cloned gene (423). Finally, one
`elegant approach involves the principle of invertible promot-
`ers: the promoter, flanked by two ␭ integration sites, faces in
`the direction opposite that of the gene to be expressed and is
`inverted only by inducing site-specific genetic recombination
`mediated by the ␭ integrase (16, 21, 235, 441, 599).
`The above systems have advantages as well as disadvantages,
`depending on their intended use. Thus, methods that rely on
`solid media cannot easily be used for large-scale expression.
`High-level repressor systems often cause a substantial decrease
`in protein yield (9, 531), thus necessitating optimization of the
`repressor-to-operator ratio (234). Induction mediated by ␭
`phage adds further complexity to the system. The use of in-
`verted promoter circuits involves complex vector construc-
`tions. Although most of the above systems have not yet been
`used for the high-level production of proteins on a large scale,
`they nevertheless provide important tools for the armamentar-
`ium of gene expression.
`
`TRANSLATIONAL REGULATION
`
`mRNA Translational Initiation
`The extensive knowledge of the transcriptional process has
`allowed the use of prokaryotic promoters in cassette fashion,
`unaffected by the surrounding nucleotide context (232, 236,
`317, 344). However, the determinants of protein synthesis ini-
`tiation have been more difficult to decipher; this is not surpris-
`ing, considering the complexity of this process (224, 579). It is
`now clear that the wide range of efficiencies in the translation
`of different mRNAs is predominantly due to the unique struc-
`tural features at the 5⬘ end of each mRNA species. Thus, in
`
`contrast to the portable promoters, no universal sequence for
`the efficient initiation of translation has been devised. How-
`ever, progress in this aspect of gene expression in E. coli has
`been strong, and general “guidelines” have emerged (131, 133,
`196, 198, 218, 368, 369, 458, 579, 590).
`The translational initiation region of most sequenced E. coli
`genes (91%) contains the initiation codon AUG. GUG is used
`by about 8% of the genes, and UUG is rarely used as a start
`site (1%) (218, 224, 535). In one case, AUU is used as the start
`codon for infC (75). This codon is required for the autogenous
`regulation of infC. The translational efficiency of the initiation
`codons in E. coli has been examined. AUG is the preferred
`codon by two- to threefold, and GUG is only slightly better
`than UUG (458, 573).
`Shine and Dalgarno (514, 515) identified a sequence in the
`RBS of bacteriophage mRNAs and proposed that this region,
`subsequently called the Shine-Dalgarno (SD) site, interacts
`with the complementary 3⬘ end of 16S rRNA during transla-
`tion initiation. This was confirmed by Steitz and Jakes (532).
`The spacing between the SD site and the initiating AUG codon
`can vary from 5 to 13 nucleotides, and it influences the effi-
`ciency of translational initiation (196). Extensive studies have
`been carried out to determine the optimal nucleotide sequence
`of the SD region, as well as the most effective spacing between
`the SD site and the start codon (28, 93, 131, 593). Ringquist et
`al. (458) examined the translational roles of the RBS and
`reached the following conclusions. (i) The SD sequence UAA
`GGAGG enables three- to sixfold-higher protein production
`than AAGGA for every spacing. (ii) For each SD sequence,
`there is an optimal although relatively broad spacing of 5 to 7
`nucleotides for AAGGA and 4 to 8 nucleotides for UAAGG
`AGG. (iii) For each SD sequence, there is a minimum spacing
`required for translation; for AAGGA, this minimum spacing is
`5 nucleotides, and for UAAGGAGG, it is 3 to 4 nucleotides.
`These spacings suggest that there is a precise physical relation-
`ship between the 3⬘ end of 16S rRNA and the anticodon of the
`fMe