Escherichia coli Recombinant DNA Technology
Chapter
108
JAMES R. SWARTZ
In 1973, the first gene was cloned; in 1977, the first recombinant DNA protein was produced. These events have given birth to a new industry driven by exciting applications and fueled by a tremendous body of knowledge accumulated from decades of research. Escherichia coli, one of the most intensively studied organisms, has played a dominant role. It was the demonstration organism for most of the significant advances, and it continues to play a major role in commercial applications. This chapter provides a brief history as well as a current summary of this technology.
By design, this chapter is somewhat unusual for this compendium. It is intended to cover the commercial applications of science rather than science itself. However, the application of scientific knowledge to useful purposes often generates new knowledge; some of it quite basic. This is certainly the case for recombinant DNA (rDNA) biotechnology, and this chapter provides many examples.
Because of the focus of this chapter on applied science, many patents, in addition to journal articles and book chapters, are referenced. The patent references identify intellectual property and expand the breadth of coverage. However, this chapter is not and cannot be a complete account of all applications and all useful science and technology related to E. coli rDNA technology. I have attempted to describe and provide examples of the most prominent applications and the most basic and useful science and technology. Much of the chapter deals with the production of rDNA proteins as products, mostly pharmaceuticals. This application is, to date, the most commercially important one, and it is hoped that this focus will adequately convey the science and technology which have allowed E. coli to deliver so many benefits through rDNA technology.
It is difficult to say when the history of rDNA technology really began—perhaps in 1944 with the discovery by Avery and colleagues that DNA was the "transforming principle," perhaps with the elucidation of the DNA structure in 1953 by Watson and Crick and the deciphering of the genetic code in the mid-1960s, perhaps with the description of restriction endonucleases by Arber in 1962. However, the potential power of rDNA biotechnology did not become apparent until the first cloning of a gene in 1973 in the laboratories of Boyer and Cohen (40, 51, 153; S. N. Cohen and H. W. Boyer, U.S. patent 4,468,464, August, 1984). Those initial results have formed the basis of a multibillion dollar industry with the potential to affect all of our lives.
Other key events are listed in Table 1. The first recombinant protein to be produced in the laboratory was somatostatin (103). It was produced in the cytoplasm of E. coli as part of a larger fusion protein. The mature polypeptide was then released by cyanogen bromide to produce a molecule recognizable by antisomatostatin antibodies and with the ability to inhibit the release of growth hormone from rat pituitary cells. Although this was a clear demonstration of the technology’s potential, it was not a commercial success. Insulin, the next protein to be expressed, was. The A and B chains of human insulin were separately produced in fusion proteins (84) and were released by cyanogen bromide cleavage. The two chains were then assembled into active human insulin by an in vitro folding reaction. This was a major accomplishment in molecular biology and in protein biochemistry.
Table 1Key events in the history of E. coli rDNA technology |
The first mature protein to be produced was human growth hormone (83). Although the recombinant protein contained an extra methionine residue at its N terminus, it proved to be fully active after extraction, solubilization, and a gentle in vitro folding reaction. These accomplishments set the stage for a rapid series of successes in expressing heterologous proteins in the cytoplasm of E. coli.
However, rDNA proteins produced cytoplasmically often accumulate as insoluble inclusion bodies (refractile particles). They also often have an extra methionine at the N terminus. To avoid these limitations, researchers began to investigate the secretion of heterologous proteins into the periplasmic space. The first attempts also used fusion proteins (188, 203; T. J. Silhavy, H. A. Shuman, J. Beckwith, and M. Schwartz, U.S. patent 4,336,336, June, 1982; W. Gilbert, S. A. Broome, L. J. Villa-Komaroff, and A. A. Efstratiadis, European patent 6694, January, 1980), but subsequent work demonstrated that mature proteins could be transported to and recovered from the periplasmic space (88, 98; W. Gilbert and K. Talmadge, European patent 38,182, October, 1981; G. L. Gray and H. L. Heyneker, European patent 0,127,305, May, 1984). Although there are many examples of heterologous proteins which fold properly in the periplasmic space (41, 99, 161, 212), secreted proteins often are deposited as periplasmic inclusion bodies (30, 218). Thus, the history of E. coli rDNA technology has continually been marked by the need to solubilize and fold aggregated proteins. As will be described later, a significant degree of success has been achieved.
In 1985, Smith reported the first example of phage display technology (195). This technique uses bacteriophage to display a family of closely related polypeptides by fusing the structural genes for the desired polypeptides to the gene for a bacteriophage coat protein. The affinity of the desired protein for its ligand, receptor, substrate, or antibody can be used to separate the phage containing the desired gene from the population. This technique has been further developed (18, 140, 144) into a powerful screening technique to identify rare proteins or peptides in large libraries and to simultaneously isolate the corresponding gene.
The history of E. coli rDNA technology would not be complete without mentioning a companion and often competitive technology, the expression of heterologous proteins in mammalian cells. Two of the protein pharmaceuticals targeted in the early 1980s, erythropoietin and tissue plasminogen activator, are relatively complex, glycosylated proteins. Attempts to produce active erythropoietin and tissue plasminogen activator with E. coli were not successful, and it became apparent that the glycosylation chains conferred desirable biochemical properties (63, 85). Since then, significant investment has been made in mammalian cell technology for the production of heterologous proteins. Mammalian cell culture, primarily using immortalized Chinese hamster ovary cells, has become a successful technology. Although these developments have limited the use of E. coli in making high-value protein pharmaceuticals, E. coli continues to be an important producer of recombinant proteins. (In fact, it is interesting that a tissue plasminogen activator analog which may have more desirable qualities is now being produced by E. coli and is in the latter stages of development [135].)
In the Laboratory.
E. coli-based rDNA technology has provided and continues to provide the foundation for the entire rDNA industry, especially with respect to basic laboratory techniques. Whether applied to basic research or to the development of commercial products, most laboratory techniques depend on E. coli technology (176). DNA libraries are generally constructed in E. coli. Most DNA manipulations are done with E. coli plasmids. DNA sequencing generally uses E. coli to produce DNA of the proper size and composition for efficient sequencing. Protein production is generally first tested in E. coli, and the organism provides an excellent expression system for protein design and screening (48), especially with phage display technology (18, 136). Finally, E. coli extracts can be used effectively for the in vitro synthesis of both natural and "unnatural" (i.e., with mutated amino acid sequences or with modified amino acids) proteins (65).
Pathway Engineering.
In the late 1970s and early 1980s, as the power of rDNA technology became apparent, many applications were envisioned. Because of their high unit value, pharmaceutical proteins became the primary focus. However, several early companies, including Amgen, Cetus, and Engenics, began projects in which the rDNA protein was not the product. Instead, the rDNA proteins were intended to either amplify or alter existing biosynthetic pathways to overexpress existing or new metabolites. This type of endeavor became known as pathway engineering or, more generally, metabolic engineering (9). More recently, these approaches have been applied to enhancing the ability of E. coli to overexpress heterologous proteins (7, 47, 57, 162) and to designing microorganisms to biodegrade pollutants (217).
Two of the first examples of rDNA pathway engineering in E. coli addressed the production of tryptophan (2, 3) and indigo (69). In the first case, endogenous, rate-limiting enzymes were modified and overexpressed to increase the synthetic rate of an existing biosynthetic pathway. In the second example, overexpression of an enzyme from a different organism created a new pathway which produces indigo, a useful product which otherwise is not formed.
Another interesting example of pathway engineering is the production of ethanol by E. coli (5, 32, 102). Here the basic glycolytic pathway is modified for ethanol production by overexpressing an efficient alcohol dehydrogenase and a pyruvate carboxylase with a high affinity for pyruvate. Together, these modifications allow the efficient conversion of cellulose hydrolysates into ethanol.
In general, pathway engineering has been applied to the production of small biomolecules which have relatively low unit value. Because of correspondingly large production volumes, large capital investments are required. Often the competing production technology is highly evolved. The new technology must deliver high specific productivities and high product yields from inexpensive, readily available substrates, and the process must be stable and easily controlled. Also, the market must be reasonably stable. To date, rDNA technology has made only modest inroads into this very competitive business. However, early problems such as plasmic instability (101) have motivated the development of new techniques (191, 229) and new approaches (204). Also, there is now an increased understanding that the supporting metabolic processes, as well as the target pathway (9), may need to be augmented. This realization, in fact, is motivating the search for a better understanding of metabolic control mechanisms (128). As this knowledge base grows, it is likely that there will begin to be successful examples of production processes for small biomolecules which are based on E. coli rDNA pathway engineering.
Recombinant Protein Production.
This is the most direct and, to date, most successful application of rDNA technology in E. coli. As can be seen in Table 2, E. coli rDNA technology has fostered a significant new industry. The field is dominated by human pharmaceuticals, a situation that is unlikely to change in the near future. Four of the first five products have become "blockbuster" products. Many more E. coli-produced, protein pharmaceuticals are now in development.
Table 2Sales of rDNA proteins in 1993 |
As important as the revenues are to companies and investors, a more dramatic value is realized by those who enjoy the benefits brought by these products. Diabetics now no longer need to fear producing antibodies to animal insulins. Growth hormone-deficient children can avoid the pains of dwarfism without the fear of contracting Kreutzfeld-Jacob syndrome. Children with chronic granulomatous disease can hope for a normal life thanks to the benefits of gamma interferon therapy. Thousands who must endure chemotherapy or radiation therapy for cancer can expect more rapid recoveries with fewer infections because of granulocyte colony-stimulating factor. These products are bringing hope and life itself to millions.
It is interesting to examine the biochemistry of these products in more detail. Although insulin is dimeric, in general the early products were small proteins with relatively simple tertiary structures. In most cases, production depended on in vitro folding. In many cases, it still does. However, E. coli rDNA technology is no longer limited to simple, small proteins.
In many cases, in vivo folding is possible. Perhaps the most dramatic example is the cytoplasmic production of properly folded, active human hemoglobin (96). Coexpression of the α- and β-globin polypeptides results in an assembled, soluble tetrameric protein of 50 kDa containing the required heme group. More frequently, it is necessary to transport the protein out of the cytoplasm in order to obtain in vivo folding. This is the case, for example, for Fab antibody fragments which fold properly and form several essential disulfide bonds when secreted into the periplasmic space (24, 165). There are now many examples of successful periplasmic folding of secreted rDNA proteins (99, 104, 117, 220). In addition, improved technology has been developed for in vitro protein folding, when required (50, 174).
These accumulated technologies have greatly expanded the range of proteins that can be successfully produced in E. coli. The technology can be applied to the production of human and animal vaccines, animal pharmaceuticals, enzymes, bioadhesives, and biomaterials (82). In the early days of E. coli rDNA protein production, it was not uncommon for protein production to cost thousands of dollars per gram of active protein. Now, a product such as bovine growth hormone is being produced by E. coli technology for a price-sensitive market. As judged from selling prices, it appears that production costs may be less than $10/g. These low costs require extensive process development and the economies of large-scale production, but we now know that they are feasible. Thus, E. coli rDNA protein production technology has progressed to an amazing extent since it produced the first rDNA protein. Several rapidly developing technologies are in competition. These include mammalian cell, insect cell, and animal and plant production of recombinant proteins. It is likely that each will be important, and it will be interesting to see what role evolves for each over time. In many ways, the E. coli technology is the most developed. Although it is certainly not the answer for all applications, it appears likely that E. coli technology will continue to be of major importance for the production of recombinant proteins.
The best answer to this question lies in the preceding discussion and in Table 2. Nonetheless, it is useful to reflect on both the strengths and limitations of the technology.
Since E. coli has been the dominant microbiological research target for biochemists, geneticists, and microbial physiologists for decades, it was logical that it would be the first target for molecular biologists. Because the first rDNA work as well as the first applications utilized E. coli, approaches based on this organism established a considerable head start in accumulating relevant knowledge and demonstrated utility. However, the real strength of E. coli rDNA technology can be found in its versatility and its potential for enabling the production of proteins and efficient metabolic pathways at low cost.
Versatility.
E. coli is clearly the organism with which we can most effectively combine an extensive base of genetic and physiologic knowledge, a well-developed ability to quickly alter the organism, and the ability to quickly assess the consequences of these alterations.
As this compendium shows, a tremendous knowledge base for E. coli has accumulated after decades of intensive study. With this knowledge has come the ability to quickly and precisely modify the chromosome (91, 168) and to implement plasmid-based modifications (76, 176). Because of the organism’s fast growth rate and ease of culture, we can quickly assess the consequences of any changes. Alteration of the chromosome has allowed the implementation of such improvements as a reduction in protease activity (13), the avoidance of amino acid analog incorporation (27), alteration of promoter control (S. Bass and J. R. Swartz, U.S. patent 5,304,472, April, 1994), stabilization of plasmid copy number (191), alteration of basic metabolic carbon flow (47), and the formation of intracellular disulfide bonds (58), to name just a few examples.
Modification of the expression vector is rapid and simple, and the desired number of gene copies can be quickly and reliably achieved (12, 182). This capability allows the rapid characterization of a variety of changes affecting transcription, translation, protein export, and even protein folding (221).
With the combination of the strong knowledge base, ease of manipulation, and rapid assessment of effects, E. coli technology provides an unparalleled versatility in optimizing rDNA protein production or enhanced metabolic function.
Potential for Low-Cost Production.
To achieve low-cost production, several factors must come together. We need high-volume production (low capital costs), reasonable raw material costs (inexpensive substrates and high product yields), inexpensive isolation and purification procedures, and reliable process performance.
E. coli has favorable intrinsic characteristics relevant to each of these requirements. It has a rapid growth rate (at least an order of magnitude greater than rates of mammalian cells) and the corresponding ability to rapidly produce heterologous proteins as well as to rapidly metabolize substrates for small molecule production. E. coli can be grown to very high cell densities, often exceeding 50 g (dry weight)/liter (226, 230). rDNA proteins can be accumulated at levels up to 50% of total cell protein (112, 184), and poly-β-hydroxybutyrate has been accumulated at up to 80% of the cell dry weight (193). E. coli has evolved to survive a wide variety of environmental conditions and does not require expensive medium components. The simplicity of the medium and the high metabolic rates also encourage the use of computer control for reliable process performance. These capabilities allow potential productivities and product yields which are significantly higher than those achieved with other expression systems.
The expense involved in extracting, folding (if necessary), and purifying the rDNA proteins is often cited as a disadvantage for E. coli. Indeed, that can be true. However, as new methods are being developed for protein isolation and folding (94, 174, 199), this factor is becoming less important. Indeed, the rDNA proteins which have required low-cost production—insulin, human hemoglobin, bovine growth hormone, chymosin, and insulin-like growth factor I (IGF-I)—have all been manufactured by using E. coli technology. This is true despite the fact that the four administered parenterally must be highly purified to avoid possible adverse immunologic reactions and that four of the five require in vitro folding.
Even in the very competitive area of small-biomolecule production, recent advances in metabolic pathway engineering (162) are very promising. Again, the rapid metabolic rates and the tremendous knowledge base coupled with the ability to quickly improve the organism appear to offer significant opportunities for low-cost production.
Limitations.
The most obvious current limitation for E. coli is its inability to glycosylate proteins. The inability to properly fold many rDNA proteins in vivo can also limit its usefulness for some applications. Although the endotoxic lipopolysaccharide of E. coli is sometimes mentioned as a disadvantage, in practice it has not been a serious problem. Finally, for many bioconversions and metabolic engineering applications, E. coli may be too sensitive to the toxicity of reagents or products as well as to hostile environments. Obviously, E. coli will not be the answer for all applications.
The production of rDNA pharmaceutical proteins, the most developed of the E. coli rDNA technologies, will be used here as a vehicle to summarize the general technology. In the time since the first rDNA protein was expressed, a tremendous literature has developed to describe relevant science and technology. Tasks that were once difficult, unreliable, and tedious can now be done easily and quickly with commercially available kits and reagents. Although a significant portion of relevant knowledge is protected as trade secrets, an even larger portion has been made public through patents or through open publication by academic and industrial scientists.
Although it is possible to produce recombinant proteins from genes integrated into the E. coli chromosome, the vast majority of applications utilize plasmid-based expression. The plasmid must contain three key elements to be useful: (i) an origin of replication to allow autonomous plasmid replication, (ii) an element providing selective pressure to allow cell transformation and plasmid retention, and (iii) a convenient locus or loci for promoter and gene insertion. A variety of plasmids are now available commercially that allow convenient insertion and expression of foreign DNA.
Origin of Replication.
The most commonly used origin of replication is that derived from the plasmid ColE1. It is used in pBR322 and its derivatives and in the pUC plasmids (166). Each plasmid is maintained within a characteristic range of number of copies per cell according to the interaction between two plasmid-encoded RNA molecules, RNA I and RNA II (123). RNA I binds to RNA II, inhibiting the maturation of RNA II into a form which is required as a template for plasmid replication. Stronger RNA I-RNA II binding therefore results in lower copy number. This interaction is enhanced by a plasmid-expressed protein called Rop or Rom and is reduced by RNase E cleavage of RNA I (130). For example, pBR322 is normally present at 15 to 20 copies per cell, but when the rop gene is deleted as in the pUC plasmids, the copy number increases to 500 to 700 (176). The RNA I-RNA II interaction is also responsible for plasmid incompatibility.
Because the plasmid replication reactions are now relatively well understood, plasmid copy number can be easily modified (154, 182). However, more is not necessarily better. High plasmid copy number can negatively affect host cell metabolism (10). The transcription rate for the desired mRNA is a function of both gene dosage and promoter function. Use of a strong, controllable promoter is normally better for achieving high transcription rates than the use of high-copy-number plasmids.
Alternatively, a so-called runaway plasmid can be used (125, 170; B. E. Uhlen, K. Nordstrom, S. Molin, and P. Gustaffson, U.S. patent 4,487,835, December, 1984). The plasmid is maintained at low copy number during cell growth and is induced to increase copy number at the time of product induction. Although impressive results have been obtained with this approach (71), such complicated regulation is usually not necessary to obtain high product yields.
Selection for Plasmid Introduction and Retention.
The plasmid must be designed to provide a growth advantage to the cell under certain conditions in order to select plasmid-bearing cells. Such positive selection may also be necessary to counteract any disadvantage conferred by plasmid presence or by plasmid-driven protein expression. Although the presence of plasmids similar to pBR322 (15 to 20 copies per cell) does not appear to significantly reduce the cell growth rate, high-copy-number plasmids and plasmids expressing significant levels of recombinant proteins do place a significant burden on cell metabolim (23, 26, 44, 160). Therefore, it is advantageous to maintain a selective pressure for plasmid retention and also to use a controllable promoter to minimize the disadvantage of plasmid presence during cell growth.
The standard method for plasmid selection is to use an ample, pBR322 encodes for resistance both to β-lactam antibiotics such as ampicillin and to tetracycline (29). Resistance to other antibiotics, such as chloramphenicol, can also be used (28). While this approach has been successful for years, it should be used carefully. For example, since the mechanism for resistance to β-lactam antibiotics is their degradation, depletion of the antibiotic may remove selective pressure during longer fermentation processes (110, 164). Also, β-lactam antibiotics are not allowed in the production of human pharmaceuticals. Although tetracycline is allowed and is not degraded, the tetracycline resistance mechanism may affect cellular metabolism, with a possible reduction in product yields. The tetracycline resistance conferred by pBR322 and similar plasmids is effected by a plasmid-encoded membrane protein which expels tetracycline from the cytoplasm (142). This resistance gene is classified as a class C gene and thus differs from the more thoroughly studied class B resistance gene in the Tn10 transposon (146). The Tn10-associated resistance is conferred by a proton antiporter which expels a Mg2+-tetracycline complex (223, 224). Unfortunately, we do not yet know whether the class C tetracycline resistance of pBR322 uses the same mechanism. It is known, however, that the pBR322-associated tetracycline resistance can render the cell more sensitive to aminoglycoside antibiotics (146) and better able to transport potassium (62). It is likely that tetracycline resistance has other effects as well, some of which may affect recombinant protein expression and secretion.
Because antibiotic resistance is not always a satisfactory selection, other methods have been devised. These include insertion of mini-F DNA from F factor (228, 229), insertion of the partition locus from plasmid pSC101 (143, 192), use of the parB locus of plasmid R1 (79), use of a plasmid-borne repressor of a chromosomally inserted lethal gene (173; C. L. Hershberger and P. R. Rosteck, Jr., U.S. patent 4,650,761, March, 1987), and use of a plasmid-borne valS gene in a temperature-sensitive valS host (191), among others. Each of these methods offers certain advantages, but plasmid stability requires more than plasmid retention (68, 101). If plasmid presence confers a sufficient disadvantage, the plasmid may become genetically modified. In these cases, it is more effective to lessen the negative impact of plasmid presence (for example, by using tightly controlled promoters on lower-copy-number plasmids) rather than to introduce stronger selection for plasmid retention. Nonetheless, there is still room for developing improved means of plasmid selection and retention. An effective method should allow inexpensive plasmid retention for large-scale production and should not adversely affect metabolic processes related to protein production.
If the host cell and the plasmid can be classified as the rDNA infrastructure, then the protein expression elements—the promoter, ribosome binding site (RBS), secretion signal sequence, structural gene, and transcription terminator—embody the core of rDNA technology. Although there is still significant room for improvement, a strong body of technology has been assembled for optimizing these components. The field has been reviewed by Glick and Pasternak (82), Shatzman (185), Balbas and Bolivar (12), and more recently Georgiou (76).
Elements Related to Transcription.
The promoter is one of the most important elements for recombinant protein production. Ideally, an effective promoter should be regulated for minimal expression during growth and for rapid transcription after induction. The induction signal should be easily applied and inexpensive and should not adversely affect other metabolic processes. Occasionally, it also is useful to modulate promoter induction in order to optimize expression rates for improved protein folding and secretion (30, 120). The promoters normally used are the lacUV5, trp, tac, lpp-lac, lambda p R, and phoA promoters. The T7 promoter offers an interesting alternative; its induction is effected by expressing a new RNA polymerase rather than by directly inducing or derepressing the promoter.
The lacUV5, tac, and lpp-lac promoters vary in the efficiency of the RNA polymerase binding site but share the same operator region. All are inducible by galactosides, most commonly by isopropyl-β-d-thiogalactopyranoside (IPTG) or by lactose. IPTG is a strong, gratuitous inducer, but it is relatively expensive for commercial application and may adversely affect cellular metabolism (122). Lactose is less expensive but requires β-galactosidase activity for conversion to allolactose, the true inducer (17). It is important to realize that overexpression of the lac repressor (lacI q) is necessary to avoid premature expression from multicopy plasmids and that induction of the lactose permease may adversely affect the proton motive force across the cytoplasmic membrane (1, 64). Despite these limitations, the lac-based promoters are often used very successfully. They are strong, are easily induced, and offer the rare capability of being effectively modulated at various inducer concentrations (30).
The trp promoter is also commonly used but is somewhat more difficult to control (225). Induction normally requires tryptophan depletion but is often premature, and control may require tryptophan addition, overexpression of the trp repressor, or both. The regulation may also be complicated by induction of tryptophanase activity, which then rapidly degrades the inhibitor. Addition of indole-3-acrylic acid provides strong induction. Although sometimes problematic, the trp promoter often allows effective accumulation of rDNA proteins which are not toxic to cell metabolism (184).
The lambda p R promoter is a strong promoter which is controlled by a heat-sensitive repressor (134). The p L promoter is similar. These promoters are normally induced by raising the culture temperature to 42°C to denature the repressor. While this is effective in inducing strong protein expression, the temperature increase may adversely affect protein folding and will also induce heat shock proteins which may degrade rDNA proteins. The phoA promoter is a strong promoter which is easily induced by phosphate limitation (214). Although phosphate limitation may also induce protease expression (198) and cause depletion of ribosome pools (55), a specific mutation in the phoS protein which senses periplasmic phosphate concentrations can allow promoter induction without complete phosphate starvation (Bass and Swartz, U.S. patent 5,304,472).
A number of other promoters have also been developed for expression of rDNA proteins. Some combine different features of several systems. For example, Studier et al. (201) have placed the T7 RNA polyerase under control of the lac promoter with the T7 promoter placed in front of the structural gene for the desired protein. IPTG induction then initiates rapid and preferential transcription of the desired gene.
The amount of effort expended on the development of controllable promoters attests to their importance. However, there is still room for improvement, especially for large-scale production of proteins inhibitory to growth as well as for systems in which modulated synthesis is beneficial. The induction mechanism should be inexpensive, easily and rapidly applied, and without effect on other cellular processes.
The other element related to transcription is the transcription terminator. The terminator stops the RNA polymerase from extending the message beyond the desired gene. Several have been studied. They often depend on RNA secondary structure and/or helper proteins (45, 141). Although not necessary, they may be beneficial, especially if continued transcription of downstream genes negatively affects other plasmid properties such as copy number or drug resistance (12).
Factors Affecting Translation.
Translation of the mRNA by the ribosome is a complex and important step for rDNA protein production. Efficient ribosomal binding to the message is critical. Codon selection for the structural gene may affect translation efficiency, especially for the section encoding the N terminus, and may also affect translational accuracy. Finally, an effective translational stop signal is required.
Ringquist et al. offer recommendations for efficient ribosomal binding (171). A strong RBS (UAAGGAGG) and an optimal length of approximately nine A-rich nucleotides between the RBS and the AUG initiation codon are beneficial. It is also important to minimize possible mRNA secondary structures near the RBS (59, 177). Surprisingly, the nucleotide sequence which encodes the N terminus of the protein can have a profound effect on the efficiency of translation initiation, apparently well beyond the effect predicted by mRNA secondary structure. Devlin et al. increased expression of human granulocyte colony-stimulating factor from undetectable levels to 17% of soluble protein by altering all possible G and C residues at the 5' end of the gene to A and T (60). Reducing the G-C content was also beneficial for a bovine growth hormone fusion protein (98). Possibly the most dramatic example was presented by Seow et al. (184), who increased human tumor necrosis factor beta accumulation from undetectable levels to 34% of cellular protein by altering several N-terminal codons and using codons preferred for E. coli genes.
In general, codon usage by itself has, at most, a modest effect on rDNA protein production. Although Sorensen et al. showed that rare codons may decrease the rate of translation by up to sixfold (197), the results of Ernst and Kawashima suggest that this has little effect on protein production (70). In contrast, codon usage can affect translational fidelity. Seetharam et al. showed that the use of the rare AGA codon for arginine resulted in a significant substitution of lysine for arginine in rapidly translated IGF-I (181). This mistranslation was abolished by using CGT, a preferred codon for E. coli. Finally, translation termination seems to be favored by using the stop codon UAA (12), the codon most commonly used for highly expressed E. coli genes.
The specialized ribosome approach of Hui and deBoer also deserves mention. They designed a new RBS as well as a mutated ribosome to recognize it (100). When the mutated ribosome is expressed, it efficiently recognizes only the mutated RBS and thereby provides selective tanslation of the desired protein. This approach has recently been reviewed by Leipold and Dhurjati (127).
Secretion Signals.
The export of rDNA proteins into the periplasm or outside of the cell offers several potential advantages: a high probability of having the correct N terminus, an improved probability of obtaining proper folding of disulfide-containing proteins upon export to the higher redox potential of the periplasm, protection from the proteases that reside in the cytoplasm, and the possibility of a more selective protein isolation. In almost all cases, a secretion signal sequence is required at the N terminus of the nascent polypeptide to allow export of the nascent polypeptide across the cytoplasmic membrane. Izard and Kendall have recently reviewed the field (105).
The choice of signal sequence can sometimes be critical for rDNA protein export. Voss et al. tested eight different signal sequences for the export of human alpha 2c interferon (212). Only the STII signal sequence derived from the heat-stable enterotoxin of E. coli allowed detectable export. There are three important portions of the signal sequence: the net positive charge near the N terminus, the central hydrophobic region, and the C-terminal region recognized by leader peptidase (105). The signal sequences can vary significantly in length, but most have about 24 amino acids. Interestingly, the primary sequence of the protein to be translocated may also affect export. A particularly interesting experiment showed that alteration of the N-terminal region of alkaline phosphatase resulted in impaired export (175). Export could be restored, however, by increasing the hydrophobicity of the central region of the signal sequence. The authors suggest that increased signal sequence hydrophobicity may offer a general advantage for the export of foreign proteins in E. coli.
The E. coli host organism provides many essential functions for the production of rDNA proteins. Its importance for process development and optimization rivals that of the expression vector. Nevertheless, the production host has received less attention. For efficient production of rDNA proteins, the host must perform beyond its evolved capabilities. It must supply the protein synthetic machinery as well as the building blocks and energy required to rapidly express a new, foreign protein. It also must be able to store significant quantities of that foreign protein. In addition, we often want the cell to efficiently export and fold foreign proteins through a complicated process specifically evolved for native proteins. We also want the cell to suspend its highly evolved mechanism for dealing with unwanted proteins, namely, proteolysis. Clearly, we are asking a lot, and it is a testament to the versatility of the organism that it has been able to deliver significant successes without major modifications. However, over the years, improvements and increased understanding have been gained.
Throughout the development of E. coli rDNA technology, the most commonly used strains have been E. coli K-12 derivatives. K-12 strains have been used in laboratory research since 1947, when Tatum and Lederberg discovered genetic recombination by using a K-12 strain. Thereafter, K-12 strains were favored by geneticists, and K-12 derivatives were used for the first recombinant DNA experiments (40, 51, 153).
However, E. coli B has been favored by E. coli physiologists (156) because of its faster growth rate in minimal medium and the collection of E. coli B bacteriophages which can be used as tools for genetic manipulation. As the rDNA technology developed, the molecular biologists continued to use K-12 strains as hosts. The experience gained with K-12 caused E. coli B and other strains to fade into the background. This trend was strengthened by the National Institutes of Health guidelines for work with recombinant organisms. The safety of K-12 strains was more actively investigated, and these strains were then given preferential treatment by the guidelines. Large-scale work requires approval by local institutional biosafety committees, which may be reluctant to approve other strains without the same level of safety information that exists for K-12.
There are occasional reports that other strains may offer advantages (201). For example, in addition to exhibiting improved growth on minimal medium, E. coli B does not suffer from valine toxicity (210) and is naturally deficient in the lon and OmpT proteases (163, 201). However, defined mutations can be introduced into the K-12 strains to avoid both valine toxicity and expression of proteases. K-12 strains are likely to continue to be dominant.
What Host Attributes Do We Want?
It will be advantageous for the host to have (i) the ability to provide a surplus of protein synthetic apparatus (building blocks and energy), particularly during the early stationary phase, (ii) precise control over the recombinant promoter function, (iii) a deficiency of harmful protease(s), (iv) resistance to bacteriophage, (v) an enhanced ability to make large quantities of recombinant protein with an accurate primary sequence, (vi) an adequate supply of helper proteins for efficient protein secretion and folding, (vii) improved in vivo protein folding or inclusion body formation, (viii) improved product release, either during or after the fermentation, and (ix) the ability to provide cell extracts which are amenable to efficient folding and purification of the product.
The host can be modified to enhance each of these attributes, either by chromosomal changes or by plasmid-based overexpression of chromosomal genes. A complete survey is beyond the scope of this chapter. Instead, examples will be given for improving basic metabolism, promoter control, and product stability.
Improved Basic Metabolism.
The effective production of high levels of recombinant proteins places significant demands on the production organism. Efficient production is ideally achieved by growing the cell culture quickly to a high cell density, e.g., 30 to 50 g (dry weight)/liter or higher, and then inducing product formation (226). In almost all fermentors, the high specific metabolic capacity of E. coli causes the maximum oxygen assimilation rate to exceed the fermentor’s oxygen delivery capacity at a cell density of approximately 10 to 20 g/liter. To avoid oxygen depletion thereafter, it is common to limit the carbon and energy source, normally glucose. As the cell mass increases to higher levels, this limitation results in lower specific growth rates. The culture therefore enters what may be termed an early stationary phase, as illustrated in Fig. 1. The demands on the organism are increased when the promoter is induced, initiating the metabolic demands of rDNA protein expression. If the rDNA promoter is induced by limiting nutrients such as phosphate or tryptophan or by increasing the temperature, the cell’s metabolic condition is further changed. Finally, the rDNA protein may be toxic. Although careful process optimization and control are important in limiting the metabolic demands, process performance can also be improved by using an organism better able to support high-level rDNA protein production.
One complication of high-density growth and expression is the accumulation of acetate to toxic concentrations when glucose is the carbon and energy source. Continuous culture experiments showed that acetate is formed under aerobic conditions even when glucose is the limiting nutrient (66, 147, 169). Evidently, the rate of pyruvate formation exceeds either the amphibolic (66) or respiration (6) capacity of the organism when the growth rate exceeds 60% of its maximum. Thus, acetate accumulation is difficult to avoid in a batch fermentation. Acetate can begin to inhibit rDNA protein production at approximately 40 mM and growth at 100 mM (106, 133). Even if the accumulated acetate does not reach these levels, it is rapidly assimilated when glucose sufficiently limits growth (Fig. 1). Since acetate is taken up in its protonated form, the resulting pH rise may disrupt metabolism.
A number of researchers have tested host modifications to limit acetate formation by either slowing the rate of pyruvate formation or channeling excess pyruvate into less harmful compounds. Chou et al. (47) introduced a mutation in the ptsG gene, which encodes enzyme II of the glucose phosphotransferase system. The mutant produced less acetate and more recombinant protein in batch culture. Interestingly, although its growth rate was reduced in a defined medium, the growth rate was not altered in complex media.
Acetate is formed from acetyl coenzyme A by the consecutive action of phosphotransacetylase and acetate kinase (Fig. 2). Work at Cetus Corp. demonstrated that mutational inactivation of phosphotransacetylase decreased acetate formation and increased the formation of human interleukin-2 in a high-cell-density fermentation (19). However, Diaz-Ricci et al. showed that inactivation of the acetate-forming enzymes caused lactate and pyruvate to accumulate (61). Expression of recombinant pyruvate decarboxylase and alcohol dehydrogenase reduced the pyruvate accumulation but resulted in significant ethanol formation. In later work, Bailey and coworkers showed that diverting glucose 6-phosphate to glycogen formation decreased acetate formation without significant pyruvate accumulation (57). In yet another interesting approach, Aristidou et al. overexpressed an acetolactate synthase gene to convert excess pyruvate into acetoin, a compound 50-fold less toxic than acetate (7). Less acetate accumulated, and growth in shake flasks was significantly improved. The growth advantage in a well-aerated fermentation, however, was less dramatic.
Although the measures described above were all successful in limiting the formation of acetate, they may not always be necessary. Often, high-cell-density fermentations become limited for glucose before acetate reaches inhibitory concentrations. Under glucose limitation, the culture may catabolize the accumulated acetate before induction of product synthesis. If the pH increase can be controlled or is not harmful, the cost of transient acetate accumulation may be minimal. Thus, alteration of the glycolytic pathway may not be necessary for efficient rDNA product formation.
Another class of modifications addresses the effectiveness of aerobic respiration in a large industrial fermentor. Uniform delivery of oxygen throughout a large bioreactor is difficult because of the high respiration rates of E. coli, a nonuniform mixing environment, and the low solubility of oxygen. In at least one case (J. R. Swartz, U.S. patent 5,342,763, August, 1995), spatial heterogeneity in the dissolved oxygen concentration caused the organism to rapidly switch from cytochome o oxidase to cytochrome d oxidase. (Cytochrome o oxidase has a lower affinity for oxygen but is more efficient in transferring protons than cytochrome d oxidase [35].) The switch caused the dissolved oxygen concentration in the vessel to drop rapidly to zero. Under certain circumstances, the dissolved oxygen level recovered, causing a switch back to the cytrochrome o oxidase and setting the stage for another precipitous drop in dissolved oxygen. This instability in environmental and metabolic control can disrupt the progress of large-scale cultures but can be prevented by using a host with a disabling mutation in either cytochrome oxidase.
In another modification to the respiration pathway, Khosla and Bailey cloned the gene for a hemoglobin protein expressed by an aerobic, pond-dwelling organism, Vitreoscilla sp., into E. coli (115). Expression of the protein improved both growth and protein production under oxygen-limited conditions (116). Even in large bioreactors with adequate bulk dissolved oxygen concentations, the cloned hemoglobin may protect the organism when it encounters regions that are oxygen depleted.
Improved Promoter Control.
The host organism can play a major role in contributing both to promoter control and to minimizing the effect of promoter induction. Probably the best known example is the use of a strain which overexpresses the lac repressor to control promoters employing the lac operator (30, 33). Similarly, the trp repressor can be overexpressed to avoid premature expression from the trp promoter (215). A more unusual approach was used by Bass and Swartz for control of the alkaline phosphatase promoter (Bass and Swartz, U.S. patent 5,304,472). The protein which senses extracellular phosphate concentration, PstS (PhoS), was mutated to decrease its affinity for phosphate. Thus, the promoter is induced at growth-permitting phosphate concentrations, and continuous phosphate feeding can be used without repressing the promoter. The result is higher rDNA product yields.
Increasing Product Stability.
Often, a serious challenge to rDNA product accumulation is degradation by E. coli proteases. One possible solution is to express a fusion protein that is more resistant to proteolysis (103). Another is to promote the incorporation of the target protein into insoluble inclusion bodies (46). Expression of protease inhibitors has also been evaluated (189). However, an increasingly important approach is to modify the host organism to reduce its proteolytic activity (145).
There are now several excellent reviews of E. coli proteases and their effects on rDNA proteins (14, 67, 87, 138). For intracellular accumulation of rDNA proteins, mutations in the lon (la) and clp proteases and in rpoH (htpR) have been most useful. The lon protease is the major ATP-stimulated intracellular protease (138, 139). The clp protease complex, also stimulated by ATP, contributes to the proteolysis of foreign proteins (113). Both lon and clp activities can be eliminated by introducing inactivating mutations, but removal of lon activity can cause cell mucoidy and UV sensitivity. Secondary mutations are often incorporated to avoid these phenotypes, as described by Gottesman (87).
The rpoH (htpR) gene encodes the σ 32 factor, which stimulates production of heat shock proteins (89, 157). Since a subset of these proteins are proteases (including the lon protease), mutations in htpR can reduce intracellular proteolysis, even in strains deficient in the lon protease (34). Since the htpR gene product is necessary for growth above 22°C (231), a variety of approaches have been used to decrease its activity during rDNA protein production without adversely affecting cell growth and product synthesis (87, 145).
rDNA proteins transported into the periplasm are threatened by a different set of proteases. These include HtrA (DegP), OmpT, protease III, and Prc (Tsp). Significant improvements in product accumulation have resulted from mutations which eliminate these proteases, either singly or in combination (13, 145, 200; G. Georgiou and F. Baneyx, U.S. patent 5,264,365, November, 1993; J. Beckwith and K. L. Strauch, International Patent Application WO 8805821, August, 1988). However, an htrA deletion causes temperature sensitivity. This can be suppressed (11), but increasing the number of protease lesions can lead to depressed growth rates, leaky outer membranes, and other undesirable phenotypes (145). The reasons for these limitations are not well understood.
Other Modifications.
Many other modifications have been or could be used: mutations in the fhuA (tonA) gene confer resistance to infection by T-odd bacteriophage (92), nonessential proteins which tend to copurify with the product protein can be eliminated, pathways which make nonnatural amino acids such as norleucine can be eliminated (27), and chaperonin or other helper proteins can be overexpressed (126).
A recent subset of E. coli rDNA technology allows the display of polypeptides on the surface of bacteriophage. Typically, a library of closely related genes for the activity of interest is fused to the gene encoding the surface protein of a filamentous bacteriophage virion. The virions which display polypeptides having the greatest affinity for an immobilized ligand can then be isolated. The isolated virions will also contain the desired DNA. When first introduced in 1985 by Smith (195), the technique was limited to smaller polypeptides. The affinity selection was also limited because more than a single fusion protein was displayed on each phage. These limitations were both addressed by Bass et al. (18), who used a "phagemid" vector containing two origins of replication, one for E. coli and the other for M13, a filamentous phage. Into this vector was inserted a gene encoding the fusion of the gene III coat protein and human growth hormone. Infection by a helper phage expressing the normal gene III protein and careful control of fusion protein expression allowed the production of phagemid particles which displayed approximately four normal coat proteins and at most only one molecule of the fusion protein per phage. Thus, on each recombinant phage virion, only one molecule of growth hormone was displayed, and this was found to be properly folded. The system was then used to isolate growth hormone mutants with increased affinity for immobilized growth hormone receptor.
The phage display technique has been used to screen for better protease substrates (136), enzyme inhibitors (172), and other receptor ligands (144). However, its most exciting and highly evolved application is the selection of antibody fragments with desired selectivity and affinity. This technique was first described by McCafferty et al. in 1990 (140) and has been developed rapidly into a powerful technology for screening DNA libraries isolated from both immunized and nonimmunized animals (216). A recent publication describes the development of a human antibody Fab fragment by using a rodent monoclonal antibody as a template and selecting first for a hybrid antibody with the desired properties and then for a fully human antibody (107).
There is nearly as much flexibility in developing the production process as in developing the engineered organism. In some cases, the protein itself can be modified, either for easier production or to have more desirable properties. The organism, process, and product options depend on many factors, such as the economic value of the product, the importance of having the natural molecule, the size and complexity of the protein, the propensity of the protein to be folded in vivo versus in vitro, the sensitivity of the protein to posttranslational modifications, the required homogeneity of the product, and other factors.
The following sections describe major process technologies as well as technology related to specific process steps.
The ideal production system would secrete and fold the mature protein so that it could be isolated from either the periplasm or the surrounding medium. However, the present technology accomplishes this for a limited number of proteins. Many proteins are readily exported to the periplasmic space but do not fold properly (218, 221). Others are difficult to secrete and may be more efficiently accumulated in the cytoplasm. The properties of a protein that limit its secretion and folding in E. coli are as yet poorly understood. Although size and disulfide bond complexity are important, results both with model systems (30, 175) and with potential products (220) suggest that other factors are important as well. These same results and a recent review (221) suggest that organism and process modifications can successfully overcome some limitations to protein secretion and folding. In other cases, it may be more beneficial to alter the protein’s primary amino acid sequence or to cotranslate it as a fusion protein.
Fusion Proteins.
When protein stability, secretion, or in vivo protein folding is problematic, it often has been productive to turn to the first rDNA production technology, the use of fusion proteins. Both somatostatin (103) and insulin (84) peptides were found to be excellent substrates for intracellular proteases, but both could be accumulated when attached to a larger protein, β-galactosidase. Although it was necessary to release the desired polypeptide by proteolytic cleavage, the product could be obtained with acceptable yields.
Fusion protein technology has progressed significantly since those early experiments. The fusion partner can confer many benefits in addition to resistance to degradation, most notably easier product isolation and purification either by antibody recognition (80, 97, 152) or by other interactions (8, 194). As described in reviews by Ford et al. (73) and by Nilsson et al. (159), fusion proteins have also been used to improve protein solubility and folding, to facilitate secretion into the periplasm and beyond, to anchor the protein either on the surface of a bacteriophage or the surface of the E. coli cell (74, 75; G. Georgiou and J. A. Francisco, International Patent Application WO 93/10214, May, 1993), and to produce proteins with new or improved therapeutic properties. Many patents have been granted on such innovations.
The principal limitation to the use of fusion proteins is the specific proteolytic cleavage required to liberate the product. Initially, cyanogen bromide was used to cleave the fusion protein at methionine residues (84, 103). However, if the desired product contains internal methionine residues, this approach is difficult at best. Other workers have used a variety of chemicals, proteases, and peptidases (36). One of the more interesting approaches uses a genetically engineered version of subtilisin (37, 38) to increase specificity by requiring the substrate to provide one of the amino acid residues of the enzyme’s catalytic triad.
Cytoplasmic Accumulation.
The first rDNA products were deposited in the cytoplasmic space. This continues to be a viable way to produce recombinant proteins. Although product can accumulate up to 50% of the cell’s protein, it is usually not soluble or properly folded (178). Because of its relatively low oxidation potential, the cytoplasm does not promote the proper folding of most proteins with disulfide bonds. Also, proteins accumulated in the cytoplasm often have an extra N-terminal methionine and may be aggressively proteolyzed.
However, there are now many examples of successful production using cytoplasmic accumulation. If the expressed protein rapidly aggregates to form inclusion bodies, proteolysis is reduced (46). Also, product deposited in inclusion bodies is easily separated from other cell components by differential centrifugation (72, 180) as an initial purification step. Alternatively, if the product is stable to proteolysis or if proteolysis can be controlled by the use of a protease-deficient host, soluble product can accumulate. Schein and Noteborn (179) and a variety of other investigators have shown that lower temperature often favors product solubility. Kopetzki et al. (120) also showed that slower expression rates resulted in less cytoplasmic aggregation, a result in agreement with the model expressed by Mitraki and King (149) that describes inclusion bodies as aggregates of protein folding intermediates. Cytoplasmic inclusion bodies can also be formed from properly folded protein (206), but this apparently is relatively rare.
The removal of the N-terminal methionine often is incomplete but sometimes can be enhanced by overexpression of the E. coli methionine amino peptidase (21). The efficency of the methionine removal depends on the side chain length of the second amino acid (95), with shorter chains allowing more efficient methionine removal. However, even with a short adjacent residue, rapid aggregation may inhibit methionine removal.
Periplasmic Accumulation or Export.
Early attempts to translocate rDNA proteins out of the cytoplasm and into the periplasm and beyond established feasibility (187, 188, 202, 203), but examples of commercial application have come slowly. Nutropin brand human growth hormone, approved in 1993, may have been the first. However, the secretion into the periplasmic space offers several potential advantages (41, 88, 99, 117, 202). Because the signal sequence is precisely removed, the proper N terminus is usually obtained. Deposition of the protein in the more highly oxidized periplasmic environment offers a greater chance for proper protein folding. The product, if soluble, may be more easily extracted, and its accumulation is less likely to affect cell metabolism. Finally, the periplasmic environment is more easily manipulated to further increase the degree of protein folding and solubility. Although the periplasm contains many proteases, hosts deficient in several of them are viable and are less likely to degrade secreted proteins (145).
The use of mammalian signal sequences to drive protein translocation can be effective (88, 203), but native E. coli signal sequences may be more efficient (41, 99). Commonly used signal sequences are the phoA, lamB, ompA, and STII leader sequences. For some proteins, the choice of signal sequence may be important, as suggested by Wong et al. (218; E. Wong and M. L. Bittner, U.S. patent 5,084,384, January, 1992) and by a recent report by Voss et al. (212). Pugsley has recently reviewed the secretion pathway in E. coli (167), and Izard and Kendall have reviewed the role of signal sequences (105). Although some proteins are difficult to secrete from E. coli, work by Rusch and Kendall suggests that increasing the hydrophobicity of the signal sequence may help (175).
There are now several examples of disulfide-containing proteins that fold properly when secreted into the periplasm. They include human growth hormone (88, 99), human alpha interferon (150, 212), epidermal growth factor (104), antibody fragments (25, 190), and bovine growth hormone (117). However, transport into the periplasm does not necessarily lead to proper folding. Aggregation of rDNA proteins occurs in the periplasm as well (30, 42, 218). The use of slower expression rates and the addition of nonmetabolizable sugars may discourage aggregation (30). But when product aggregation cannot be avoided, the aggregated protein may still be isolated and folded into its active conformation (42). The aggregated product often can be isolated by differential centrifugation. However, periplasmic aggregates may be more difficult to isolate in this manner because of their irregular shape (31) and possible interaction with other periplasmic components (211). In these cases, in situ solubilization and aqueous two-phase extraction may be better (94).
Various approaches have been used to release secreted polypeptide into the medium (118, 121). Although this potentially offers easier product isolation, some proteins may not be stable when exposed to the highly active gas-liquid interfaces of an intensely aerated microbial culture. A few proteins can leak into the medium without host or protein modifications. These include single-chain antibodies (25, 165), IGF-I (218), and E. coli β-lactamase (77). However, this happens infrequently and often destabilizes the outer membrane. The outer membrane can be intentionally destabilized by cell mutations or specific effectors to release secreted proteins during production. However, significant benefits from such efforts have not been reported, probably because they tend to negatively impact cell metabolism.
The most effective examples of product release have employed fusion proteins. When IGF-I is fused to a portion of staphylococcal protein A, the fusion protein is efficiently delivered to the surrounding medium (152). Specialized secretion systems such as the one for pullulanase (121) can also be used. Finally, L-form cells of Proteus mirabilis release secreted chymosin directly into the medium (118). Although these technologies may offer advantages for special cases, they have not come into general use.
Protein Folding In Vivo.
Poper folding of the expressed polypeptide is a critical part of rDNA protein production. Although a few proteins fold properly when they accumulate in the cytoplasm, and host cell mutations may allow disulfide bond formation (58), most naturally secreted proteins do not fold efficiently in the cytoplasm (178). Thus, most of the effort on in vivo protein folding has focused on the periplasm. Not only does the periplasm have the higher oxidation potential required for disulfide bond formation, but it also is more easily manipulated without compromising other metabolic functions. However, as pointed out by a recent review (221), the work in this area is relatively recent and is not highly evolved.
Because the mammalian proteins that we wish to secrete and fold are naturally folded in the endoplasmic reticulum (ER) of eukaryotic cells, the dominant theme has been to model the periplasmic space after the lumen of the ER. Fortunately, E. coli already provides several important properties, as shown in Table 3. A periplasmic prolyl isomerase has been found in E. coli which is similar to the one found in the ER (131). To date, however, overexpression of this enzyme has not enhanced folding of rDNA proteins (119, 221). At least two proteins with disulfide isomerase activity, DsbA (16, 111) and DsbC (148), are naturally present in the periplasm. The principal role of DsbA and DsbC appears to be the donation of disulfide bonds to nascent proteins (15, 186), and mutants deficient in DsbA do not efficiently fold either endogenous or recombinant proteins in the periplasm (119, 161). DsbA-mediated catalysis of disulfide rearrangement at neutral pH has been demonstrated (109), but DsbA is a weak disulfide isomerase compared with the mammalian enzyme. Nonetheless, Wunderlich and Glockshuber obtained improved folding of a secreted protease inhibitor when DsbA was overexpressed and reduced glutathione was added to the medium (222). DsbA overexpression did not improve folding for secreted antibody fragments (119), but it did improve the folding of secreted T-cell receptor fragments when combined with induction of the heat shock response at low temperature (220), suggesting a possible interaction with periplasmic chaperonin proteins, perhaps similar to the effect of the chaperonins in the eukaryotic ER. Although these results are encouraging, they underscore the complexity of the in vivo folding of proteins with disulfide bonds.
Table 3Protein export components |
Protein Folding In Vitro.
As mentioned earlier, in vitro protein folding has been an important contributor to the success of E. coli rDNA technology. Even though a large patent and nonpatent literature now exists, the power of this technology often is not fully appreciated. A recent review by Rudolph (174) concludes that "therefore, in vitro folding of any recombinant protein deposited in inclusion bodies will likely be successful." Occasionally, the inclusion body can be converted into properly folded protein in a single step (42, 199; L. A. Bentle, J. W. Mitchell, and S. B. Storrs, U.S. patent 4,652,630, March, 1981; C. Y. Chang and J. R. Swartz, U.S. patent 5,288,931, February, 1994). However, the process is usually more complicated.
Fortunately, the new rDNA industry was able to call upon a strong existing body of knowledge about protein folding (52) which propelled the early successes in rDNA protein production. Typically, the nascent rDNA protein is isolated in the form of inclusion bodies. It is then solubilized by using chaotropic agents such as urea, guanidine hydrochloride, and detergents in the presence of a reducing agent such as dithiothreitol or β-mercaptoethanol (56; A. J. S. Jones, K. C. Olson, and S. J. Shire, U.S. patent 4,512,922, April, 1985; K. C. Olson, U.S. patent 4,518,526, May, 1985; S. E. Builder and J. R. Ogez, U.S. patent 4,620,948, November, 1986). Alternatively, the sulfhydryl groups can be treated with sodium sulfite in a process called sulfitolysis to break intermolecular disulfide bonds and facilitate solubilization (39; R. B. Wetzel, U.S. patent 4,599,197, July, 1986; J. L. Bobbitt and J. V. Manetta, European Patent Office [EPO] 0 361 830, April, 1990).
The solution is then diluted or dialyzed to produce a suitable folding environment. The objective is to obtain proper protein folding with high yields at high protein concentrations in order to use smaller equipment (less capital cost) and to conserve reagents. However, at higher concentrations, denatured and partially folded polypeptides are more likely to aggregate. Two approaches have been taken to control aggregation. The first optimizes the folding environment to discourage aggregation. Detergents or agents such as polyethylene glycol (50) can be used to shield exposed hydrophobic regions which contribute to aggregation. Optimization of the ionic environment can also be useful (93). The second approach limits the concentration of partially folded species by slowly adding the denatured polypeptide at a rate which matches the rate of protein folding (R. Rudolph and S. Fischer, U.S. patent 4,933,434, June, 1990).
It is often important to control the redox potential of the solution to encourage disulfide bond formation and rearrangement (213). This can be effected by using redox buffers such as oxidized and reduced glutathione and conducting the folding reaction at high pH (174). However, these agents are expensive, and good folding results can often be obtained by using oxygen as the oxidant in the presence of small concentrations of Cu2+ (132).
Available space limits a more comprehensive coverage of this highly evolved technology. The reader is directed to the review by Rudolph (174) and to two recent books (49, 78). The use of the technology for the large-scale production of two cost-sensitive products, bovine growth hormone and human insulin, demonstrates the practicality of large-scale, low-cost in vitro protein folding.
After the product form has been chosen, the plasmid has been designed, and the host organism has been determined, there is a significant body of technology that addresses the performance of the production process. General guidelines for small-scale fermentations have been offered elsewhere (129). Thus, this section will focus on technologies developed for large-scale, commercial applications.
Transformation and Organism Storage.
The transformation of E. coli with plasmids has now become routine and is well described in the molecular cloning manual by Sambrook et al. (176). Although the CaCl2-based protocol is usually adequate, electroporation may be more effective in some cases. It may be important to transform a host which is restriction negative (hsdR) but modification positive. E. coli MM294 is a popular cloning strain for this reason. Additionally, it is often important to use a host background and a selection medium that inhibit expression of the cloned gene. For example, E. coli JM101 is often used as a cloning strain for expression systems using the lac operator region. JM101 contains a lacI q element to overexpress the lac repressor and avoid unintentional product expression from a multicopy plasmid.
A variety of methods can be used for organism storage. Lyophilization is best for permanent storage but is less convenient than freezing in liquid nitrogen or at –70°C (4). Although freezing in liquid nitrogen will preserve most bacteria for more than 30 years (158), freezing at –70°C is adequate for several years. The culture should be made approximately 15% in glycerol or 5 to 10% in dimethyl sulfoxide concentration before freezing. No special precautions are necessary to thaw for recovery.
Inoculum Preparation.
Little has been written about this part of the process, but it can have an important influence on the final outcome. Luria-Bertani broth (LB medium; 10 g of tryptone, 5 g of yeast extract, and 10 g of NaCl per liter) containing an antibiotic for plasmid selection is often used to prepare the initial inoculum. This medium is convenient to make and usually allows rapid recovery of frozen cultures. It supports growth to 2 to 4 units of optical density at 550 nm in baffled shake flasks. Inoculum culture volumes that are 0.5 to 5% of the next fermentation volume are normally used. In practice, a broad variety of inoculum media can be used, and it may be advantageous to customize the medium to better support growth, avoid pH changes, avoid product expression, and encourage a rapid adjustment to the next stage. pH drift and oxygen depletion deserve special attention. The degree of pH change during the flask culture can be influenced by the type of carbon and energy source used (for example, glucose versus amino acids) and by the addition of pH buffers. Oxygen transfer for shake flask cultures can also limit growth and cause unwanted metabolic changes. It is best to use baffled flasks with closures that allow oxygen transfer (209).
For large-scale fermentations, several inoculum stages may be necessary. The culture entering the final production vessel should consistently be at the same density and in the same metabolic state for process reproducibility. It is usually good practice to avoid severe metabolic challenges when the culture is transferred to the next stage of culture. For example, the inoculum should be grown at the same temperature as the next stage. If the inoculum is grown in a rich medium such as LB, it may be beneficial to add a small amount of yeast extract and protein hydrolysate to the next stage as well, even if growth on a defined medium is desired during product expression.
The Fermentation Process.
Environmental and metabolic influences can have major effects on both the yield and the quality of the product. These include obvious influences such as temperature, pH, and dissolved oxygen tension as well as others which are less obvious, such as the preinduction growth rate (53, 54) and the amount of yeast extract in the medium (43). Obviously, high product concentrations (grams per liter) and volumetric productivities (grams per liter per hour) lower fermentation costs by minimizing capital, raw material, and labor costs. However, because of the relative efficiency of E. coli rDNA expression, product isolation and purification costs can exceed fermentation costs by severalfold. Thus, to obtain an optimal overall process, the fermentation should be optimized for efficient and inexpensive downstream processing.
Proteins are easily modified during production, and altered forms of the product can be difficult to separate. Minimizing mistranslation and reactions such as protein oxidation, deamidation, and proteolysis can contribute significantly to lowering production costs, especially for human pharmaceuticals. Such optimization requires rapid and accurate assays. High-pressure liquid chromatography often provides a powerful tool to monitor both product concentration and quality in complex samples.
Much of the work done on large-scale process development has been done by industry and has not been published. However, academic laboratories have examined relevant issues, and several industrial publications and patents contribute to the literature. E. coli rDNA fermentation processes are usually designed with separate growth and expression phases. By avoiding product expression during the initial growth period, product toxicities are avoided, rDNA protein expression does not compete for metabolic resources, and the duration of product exposure to modification reactions can be minimized.
As discussed earlier, the chief challenges during the initial growth period are avoiding acetate accumulation and oxygen depletion. Yee and Blanch (226) have summarized methods used to obtain high-cell density E. coli cultures. Although oxygen depletion can be avoided by sparging with pure oxygen (81) or by slowing metabolic rates with reduced temperature (20), the preferred method is to limit the culture for its carbon and energy source, usually glucose (226). As the culture progresses to high density, glucose limitation has the additional advantage of limiting acetate formation. The use of other carbon and energy sources such as sucrose (81) or glycerol (P. M. Keith and W. Cain, U.S. patent 5,104,796, April, 1992) may help to limit acetate formation, but substrate limitation or pure oxygen sparging is still needed to avoid dissolved oxygen depletion.
A variety of media, both defined and complex, have been used for high-cell density rDNA protein production (226, 230). Fieschko and Ritch (71) described their methods for designing a defined medium to support high cell density. Other published recipes (129, 227) are similar, with the major difference depending on the mode of pH control. If pH is not controlled by NH4OH additions, more NH4 + is added to the initial medium. Growth is relatively insensitive to NH4 + concentrations below 170 mM (205). Ionic strength (106, 183) as well as the concentrations of specific ions (43) may be important, but the effects appear to vary with the production system. Dividing the required nutrients between the initial medium and medium feeds can be used to more precisely control the fermentation environment (71).
Complex nitrogen sources such as hydrolyzed proteins and yeast extract are most frequently used in complex media. Tsai et al. reported a 10-fold increase in intracellular human IGF-I accumulation by increasing yeast extract and tryptone additions (208). Since they induced product expression by increasing temperature to 42°C, and since Buell et al. saw a similar increase in IGF-I accumulation with lon and htpR mutations (34), it is reasonable to hypothesize that the complex nitrogen additions reduced product proteolysis. Chang et al. observed that the addition of yeast extract stimulated rDNA protein secretion (43).
The use of complex nitrogen components requires special precautions. For high-cell-density cultures, it is unlikely that sufficient amino acids will be added to supply all requirements. Since the rate of amino acid utilization can be quite high, amino acid depletion is correspondingly abrupt, particularly if complex nitrogen feeds are not used (90). These depletions cause periods of amino acid starvation which, at best, delay the culture and, more probably, cause metabolic changes such as protease induction (138).
Complex nitrogen feeds can play a role in improving product quality, for example, in avoiding norleucine substitution for methionine. Leucine can be fed to repress the pathway producing norleucine, methionine can be fed to saturate the methionyl-tRNA synthase, or leucine can be fed to a leucine auxotroph which is unable to make norleucine (27, 207; D. P. Brunner, G. C. Harbour, R. J. Kirschner, J. F. Pinner, and R. L. Garlick, WO 89/07651, August, 1989). Leucine feeding may have a separate effect as well (151), especially in light of the leucine-induced regulon in E. coli (137). Clearly, complex nitrogen sources can have significant effects on rDNA protein production. Unfortunately, with the exceptions of decreasing proteolysis and norleucine incorporation, the mechanisms for these effects are largely unknown.
The environmental conditions during product induction can have a significant effect on product formation. Kopetzki et al. (120) observed that the formation of active α-glucosidase was significantly improved at temperatures and pHs which were not optimal for growth. Interestingly, Curless et al. (53, 54) have found that the preinduction growth rate affects subsequent production. Unfortunately, optimal conditions for induction and product accumulation are not consistent and vary with the product, its disposition (secreted or cytoplasmic, soluble or inclusion body), and the expression system. Reports such as those of Wood and Peretti (219) and Bailey (10) have begun to describe the metabolic dependencies and influences of rDNA product expression. However, there is still much to learn. Particularly important are factors influencing proteolysis, protein secretion, and protein folding.
Isolation and Purification of the Product.
Optimal isolation and purification methods depend on the initial disposition of the product and the biochemical characteristics of both the product and the contaminants. In general, inclusion bodies are isolated by differential centrifugation after cell disruption. Soluble proteins may require cell disruption or, if secreted to the periplasmic space, may be extracted by osmotic shock. Chemical permeabilization of the outer membrane has also been used (155) to recover soluble periplasmic proteins, and Hart et al. (94) have used in situ solubilization and extraction for periplasmic inclusion bodies. Freeze-thaw cycles have also been reported to extract recombinant proteins (108).
Purification technology for recombinant proteins has received much attention, and representative technologies are described in a recent book (124). Purification processes generally begin with relatively low-resolution steps which reduce the liquid volume and remove the bulk of the contaminants. Such steps are precipitation, liquid-liquid extraction, and the use of general adsorption columns. Higher-resolution steps are then used to remove contaminants which derive from the host organism, the process steps, or from the product itself. These steps typically take advantage of a variety of specific properties of the product such as molecular size, hydrophobicity, specific affinities, and electrical charge and its dependence on pH. Although many authors suggest that removal of endotoxic lipopolysaccharides might be a limitation for E. coli-derived pharmaceuticals, this has not been the case. The steps required for removal of contaminating proteins usually are more than adequate for removal of host-derived lipopolysaccharides. However, lipopolysaccharide contamination from other organisms which contaminate purification columns or purification raw materials can be a problem, and endotoxin levels require careful attention independent of the producing organism.
The success of this technology is indicated by the characteristics of its products. Host proteins are often removed to the part-per-million levels. In fact, new analytical technology was required to allow measurements of this sensitivity. Modified product molecules such as oxidized or deamidated products are more difficult to remove but are also less likely to be problematic. These are generally removed to less than 1% of the desired product form. In fact, because E. coli-produced rDNA protein products are simpler and more easily analyzed than their glycosylated counterparts, they are usually significantly more homogeneous than proteins produced by mammalian cells.
Regulatory requirements play a large role in the use of rDNA organisms and especially in the production of human pharmaceuticals. Although the National Institutes of Health guidelines for the use of recombinant organisms have been relaxed for E. coli K-12, large-scale production generally requires that release of the live organism be minimized. The bioreactors are operated with containment filters for the exhaust gases, and the organism must be killed before containment is broken. This may be effected before, after, or coincident with product removal. Cell inactivation methods include treatment with high or low pH, addition of toxic agents, and heat inactivation. Since most of the treatments that kill bacteria also denature proteins, this step must be developed and tested carefully to avoid modifying the product or complicating its recovery.
Protein pharmaceuticals must be manufactured according to current Good Manufacturing Practice (cGMP) as defined by the Food and Drug Administration. In general, cGMP requires testing and control of all raw materials, validation of all equipment and procedures to ensure proper operation (including cleaning and sterilization), documentation of all procedures, repeated training of personnel, careful control of all steps in the process, and careful testing of process intermediates and final products. Compliance with cGMP requires an enormous effort. However, although precautions are sometimes taken to extremes, this type of system is important in protecting the quality of rDNA pharmaceuticals.
The input of Food and Drug Administration officials often helps determine acceptable process and product characteristics. The officials may specify direct requirements or may suggest safety or efficacy tests. It is usually necessary to demonstrate that the organism is genetically stable, that an adequate assay for contaminating cell proteins exists, that the production process is reproducible, and that the analytical procedures are adequate to determine the purity and the correct biochemical characteristics of the product. However, it is not yet accepted that analytical procedures can fully characterize relevant properties of an rDNA protein pharmaceutical, even one produced in E. coli. Thus, the production process must be carefully set and controlled. It is often necessary to repeat animal and human testing before process improvements can be implemented.
As the product moves through the three stages of testing in human subjects, the requirements become more stringent. For example, a change in the manufacturing process after the product has entered phase III trials will probably require additional clinical trials to demonstrate that the product is still equally safe and efficacious. Thus, it is important to define an acceptable and reproducible process as early as possible. Since it is always important to develop new products quickly, this often forces the process developer to fix upon a process which is reproducible but which may not be fully optimized.
It is always difficult to see into the future, but it can be useful to estimate what could take place. E. coli biotechnology has clearly established a formidable record of accomplishment, and many of the same applications will continue. In addition, recent progress in modeling and understanding metabolic pathways would suggest that E. coli rDNA may be used more for bioconversions and for the production of small biomolecules, particularly those with chiral centers. If methods that allow E. coli to function in more hostile environments could be found, application of the technology might be broader.
Even for the present applications, there is significant opportunity for innovation, for example, better plasmid selection without affecting cell metabolism, improved promoters which allow modulated expression and inexpensive induction without adversely affecting host cell metabolism, and more specific and less expensive methods for fusion protein cleavage.
It is also possible that more complex proteins will be made with E. coli. Even those that require posttranslational modifications such as glycosylation could be made if technology were developed for accurate in vitro modification. For example, synthetic polymers could be designed and attached at specific sites to produce biochemical characteristics the same as (or better than) those of the natural molecule. Modification methods for increasing molecular size by attaching polyethylene glycol polymers already exist (86, 114, 196). If such polymers could routinely be attached at specific sites, complex, modified proteins could be designed and produced as well-defined, homogeneous pharmaceuticals. Alternatively, in vitro glycosylation using immobilized glycosyltransferases might be feasible if the substrates and catalysts could be economically produced. However, so far there has not been sufficient economic motivation to implement these developments.
With knowledge comes power. Although we have considerable knowledge about E. coli physiology, we have little knowledge of cellular responses to rDNA protein production or of many factors which affect recombinant protein expression, secretion, and folding. Why does N-terminal codon selection have such a strong effect on translation? How do protease deletions affect cell metabolism, and can the effects be suppressed? What are the properties of the N terminus of a protein that limit its secretion? What are the biochemical properties of the periplasmic space that influence protein folding? What determines the cell’s response to rDNA protein production? Which proteases and chaperonins are induced and why? Why do complex nitrogen sources often provide substantial benefits? It is exciting to look forward to the time when the answers to these and many more questions will be known.
I thank the many people who contributed to this chapter, especially my editor, John Ingraham, for overall guidance, encouragement, and countless corrections. Roger Hart, Jeff Cleland, Brad Snedecor, and Norm Lin helped with pertinent references, and John Joly and John Jost provided many helpful suggestions.
References
1. Ahmed, S., and I. R. Booth. 1983. The effect of β-galactosides on the protonmotive force and growth of Escherichia coli. J. Gen. Microbiol. 129:2521–2529.
2. Aiba, S., T. Imanka, and H. Tsunekawa. 1980. Enhancement of tryptophan production by Escherichia coli as an application of genetic engineering. Biotechnol. Lett. 2:525–530.
3. Aiba, S., H. Tsunekawa, and T. Imanaka. 1982. New approach to tryptophan production by Escherichia coli: genetic manipulation of composite plasmids in vitro. Appl. Environ. Microbiol. 43:289–297.
4. Alexander, M., P. M. Daggett, R. Gherna, S. Jong, and F. Simione, Jr. 1980. Laboratory manual on preservation freezing and freeze-drying as applied to algae, bacteria, fungi, and protozoa, p. 1–57. In H. Hatt (ed.), American Type Culture Collection Methods. American Type Culture Collection, Rockville, Md.
5. Altherthum, F., and L. O. Ingram. 1989. Efficient ethanol production from glucose, lactose, and xylose by recombinant Escherichia coli. Appl. Environ. Microbiol. 55:1943–1948.
6. Anderson, A., and K. von Meyenburg. 1980. Are growth rates of Escherichia coli in batch cultures limited by respiration? J. Bacteriol. 144:114–123.
7. Aristidou, A. A., K. Y. San, and G. N. Bennett. 1994. Modification of central metabolic pathway in Escherichia coli to reduce acetate accumulation by heterologous expression of the Bacillus subtilis acetolactate synthase gene. Biotechnol. Bioeng. 44:944–951.
8. Arnold, F. H. 1991. Metal-affinity separations: a new dimension in protein processing. Bio/Technology 9:151–156.
9. Bailey, J. E. 1991. Toward a science of metabolic engineering. Science 252:1668–1675.
10. Bailey, J. E. 1993. Host-vector interactions in Escherichia coli. Adv. Biochem. Eng. Biotechnol. 48:29–52.
11. Baird, L., B. Lipinska, S. Raina, and C. Georgopoulos. 1991. Identification of the Escherichia coli sohB gene, a multicopy suppressor of the HtrA (DegP) null phenotype. J. Bacteriol. 173:5763–5770.
12. Balbas, P., and F. Bolivar. 1990. Design and construction of expression plasmid vectors in Escherichia coli. Methods Enzymol. 185:14–37.
13. Baneyx, F., and G. Georgiou. 1991. Construction and characterization of Escherichia coli strains deficient in multiple secreted proteases: protease III degrades high-molecular-weight substrates in vivo. J. Bacteriol. 173:2696–2703.
14. Baneyx, F., and G. Georgiou. 1992. Expression of proteolytically sensitive polypeptides in Escherichia coli, p. 69–108. In T. J. Ahern and M. C. Manning (ed.), Stability of Protein Pharmaceuticals, part A. Chemical and Physical Pathways of Protein Degradation. Plenum Press, New York.
15. Bardwell, J. C. A., J. O. Lee, G Jander, N. Martin, D. Belin, and J. Beckwith. 1993. A pathway for disulfide bond formation in vivo. Proc. Natl. Acad. Sci. USA 90:1038–1042.
16. Bardwell, J. C. A., K. McGovern, and J. Beckwith. 1991. Identification of a protein required for disulfide bond formation in vivo. Cell 67:581–589.
17. Barkley, M. D., and S. Bourgeois. 1978. Repressor recognition of operator and effectors, p. 177–220. In J. H. Miller and W. S. Reznikoff (ed.), The Operon. Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y.
18. Bass, S., R. Green, and J. A. Wells. 1990. Hormone phage: an enrichment method for variant proteins with altered binding properties. Proteins Struct. Funct. Genet. 8:309–314.
19. Bauer, K. A., A. Ben-Bassat, M. Dawson, V. T. de la Puente, and J. O. Neway. 1990. Improved expression of human interleukin-2 in high-cell-density fermentation cultures of Escherichia coli K-12 by a phosphotransacetylase mutant. Appl. Environ. Microbiol. 56:1296–1302.
20. Bauer, S., and M. D. White. 1976. Pilot scale exponential growth of Escherichia coli W to high cell concentration with temperature variation. Biotechnol. Bioeng. 18:839–846.
21. Ben-Bassat, A., K. Bauer, S. Y. Chang, K. Myambo, A. Boosman, and S. Chang. 1987. Processing of the initiation methionine from proteins: properties of the Escherichia coli methionine aminopeptidase and its gene structure. J. Bacteriol. 169:751–757.
22. Bentley, W. E., N. Mirjalili, D. C. Andersen, R. H. Davis, and D. Kompala. 1990. Plasmid-encoded protein: the principal factor in the metabolic burden associated with recombinant bacteria. Biotechnol. Bioeng. 35:668–681.
23. Betenbaugh, M. J., C. Beaty, and P. Dhurjati. 1989. Effects of plasmid amplification and recombinant gene expression on the growth kinetics of recombinant E. coli. Biotechnol. Bioeng. 33:1425–1436.
24. Better, M., C. P. Chang, R. R. Robinson, and A. H. Horwitz. 1988. Escherichia coli secretion of an active chimeric antibody fragment. Science 240:1041–1043.
25. Better, M., and A. H. Horwitz. 1993. In vivo expression of correctly folded antibody fragments from microorganisms, p. 203–217. In J. L. Cleland (ed.), Protein Folding: In Vivo and In Vitro. American Chemical Society, Washington, D.C.
26. Birnbaum, S., and J. E. Bailey. 1991. Plasmid presence changes the relative levels of many host cell proteins and ribosome components in recombinant Escherichia coli. Biotechnol. Bioeng. 37:736–745.
27. Bogosian, G., B. N. Violand, E. J. Dorward-King, W. E. Workman, P. E. Jung, and J. F. Kane. 1989. Biosynthesis and incorporation into protein of norleucine by Escherichia coli. J. Biol. Chem. 264:531–539.
28. Bolivar, F. 1978. Construction and characterization of new cloning vehicles. III. Derivatives of plasmid pBR322 carrying EcoR1 sites for selection of EcoR1 generated recombinant DNA molecules. Gene 4:121–136.
29. Bolivar, F., R. L. Rodriguez, P. J. Greene, M. O. Betlach, H. L. Heyneker, H. W. Boyer, I. H. Crosa, and S. Falkow. 1977. Construction and characterization of new cloning vehicles. II. A multiple cloning system. Gene 2:95–113.
30. Bowden, G. A., and G. Georgiou. 1990. Folding and aggregation of β-lactamase in the periplasmic space of Escherichia coli. J. Biol. Chem. 265:16760–16766.
31. Bowden, G. A., A. M. Paredes, and G. Georgiou. 1991. Structure and morphology of protein inclusion bodies in Escherichia coli. Bio/Technology 9:725–730.
32. Brau, B., and H. Sahm. 1986. Cloning and expression of the structural gene for pyruvate decarboxylase of Zymomonas mobilis in Escherichia coli. Arch. Microbiol. 144:296–301.
33. Brent, R., and M. Ptashne. 1981. Mechanism of action of the lexA gene product. Proc. Natl. Acad. Sci. USA 78:4204–4208.
34. Buell, G. M. F. Schulz, G. Selzer, A. Chollet, N. Rao Movva, D. Semon, S. Escanez, and E. Kawashima. 1985. Optimizing the expression in E. coli of a synthetic gene encoding somatomedin-C (IGF-I). Nucleic Acids Res. 13:1923–1938.
35. Calhoun, M. W., K. L. Oden, R. B. Gennis, M. J. Teixeira de Mattos, and O. M. Neijssel. 1993. Energetic efficiency of Escherichia coli: effects of mutations in components of the aerobic respiratory chain. J. Bacteriol. 175:3020–3025.
36. Carter, P. 1990. Site-specific proteolysis of fusion proteins, p. 181–193. In M. R. Ladisch, R. C. Willson, C. D. C. Painton, and S. E. Builder (ed.), Protein Purification: from Molecular Mechanisms to Large-Scale Processes. American Chemical Society, Washington, D.C.
37. Carter, P., L. Abrahmsen, and J. A. Wells. 1991. Probing the mechanism and improving the rate of substrate-assisted catalysis in subtilisin BPN'. Biochemistry 30:6142–6148.
38. Carter, P., B. Nilsson, J. P. Burnier, D. Burdick, and J. A. Wells. 1989. Engineering subtilisin BPN' for site-specific proteolysis. Proteins Struct. Funct. Genet. 6:240–248.
39. Chan, W. W. C. 1968. A method for the complete S sulfonation of cysteine residues in proteins. Biochemistry 7:4247–4253.
40. Chang, A. C. Y., and S. N. Cohen. 1974. Genome construction between bacterial species in vitro: replication and expression of Staphylococcus plasmid genes in Escherichia coli. Proc. Natl. Acad. Sci. USA 71:1030–1034.
41. Chang, C. N., M. Rey, B. Bochner, H. Heyneker, and G. Gray. 1987. High-level secretion of human growth hormone by Escherichia coli. Gene 55:189–196.
42. Chang, J. Y., and J. R. Swartz. 1993. Single-step solubilization and folding of IGF-1 aggregates from Escherichia coli, p. 178–188. In J. L. Cleland (ed.), Protein Folding: In Vivo and In Vitro. American Chemical Society, Washington, D.C.
43. Chang, J. Y. H., R. C. Pai, W. F. Bennett, R. G. Keck, and B. R. Bochner. 1986. Culture medium effects on periplasmic secretion of human growth hormone by Escherichia coli, p. 324–329. In L. Leive (ed.), Microbiology—1986. American Society for Microbiology, Washington, D.C.
44. Cheah, U. E., W. A. Weigand, and B. C. Stark. 1987. Effects of recombinant plasmid size on cellular processes in Escherichia coli. Plasmid 18:127–134.
45. Cheng, S. W. C., E. C. Lynch, K. R. Leason, D. L. Court, B. A. Shapiro, and D. I Freidman. 1991. Functional importance of sequence in the stem-loop of a transcription terminator. Science 254:1205–1207.
46. Cheng, Y. S. E., D. Y. Kwoh, T. J. Kwoh, B. C. Soltvedt, and D. Zipser. 1981. Stabilization of a degradable protein by its overexpression in Escherichia coli. Gene 14:121–130.
47. Chou, C. H., G. N. Bennett, and K. Y. San. 1994. Effect of modified glucose uptake using genetic engineering techniques on high-level recombinant protein production in Escherichia coli dense cultures. Biotechnol. Bioeng. 44:952–960.
48. Clackson, T., and J. A. Wells. 1995. A hot spot of binding energy in a hormone receptor interface. Science 267:383–386.
49. Cleland, J. L. (ed.). 1993. Protein Folding: In Vivo and In Vitro. American Chemical Society, Washington, D.C.
50. Cleland, J. L., S. E. Builder, J. R. Swartz, M. Winkler, J. Y. Chang, and D. I. C. Wang. 1992. Polyethylene glycol enhanced protein refolding. Bio/Technology 9:1013–1019.
51. Cohen, S. N., A. C. Y. Chang, H. W. Boyer, and R. B. Helling. 1973. Construction of biologically functional bacterial plasmids in vitro. Proc. Natl. Acad. Sci. USA 70:3240–3244.
52. Creighton, T. E. 1978. Experimental studies of protein folding and unfolding. Prog. Biophys. Mol. Biol. 33:231–297.
53. Curless, C., J. Pope, and L. Tsai. 1990. Effect of preinduction specific growth rate on recombinant alpha consensus interferon synthesis in Escherichia coli. Biotechnol. Prog. 6:149–152.
54. Curless, C. E., J. Pope, L. Loredo, and L. B. Tsai. 1994. Effect of preinduction specific growth rate on secretion of granulocyte macrophage colony stimulating factor by Escherichia coli. Biotechnol. Prog. 10:467–471.
55. Davis, B. D., S. M. Luger, and P. C. Tai. 1986. Role of ribosome degradation in the death of starved Escherichia coli cells. J. Bacteriol. 166:439–445.
56. De Bernardez-Clark, E., and G. Georgiou. 1991. Inclusion bodies and recovery of proteins from the aggregated state, p. 1–20. In G. Georgiou and E. De Bernardez-Clark (ed.), Protein Refolding. American Chemical Society, Washington D.C.
57. Dedhia, N. N., T. Hottinger, and J. E. Bailey. 1994. Overproduction of glycogen in Escherichia coli blocked in the acetate pathway improves cell growth. Biotechnol. Bioeng. 44:132–139.
58. Derman, A. I., W. A. Prinz, D. Belin, and J. Beckwith. 1993. Mutations that allow disulfide bond formation in the cytoplasm of Escherichia coli. Science 262:1744–1747.
59. . De Smit, M. H., and J. Van Duin. 1990. Secondary structure of the ribosome binding site determines translational efficiency: a quantitative analysis. Proc. Natl. Acad. Sci. USA 87:7668–7672.
60. Devlin, P., R. Drummond, P. Toy, D. Mark, K. Watt, and J. Devlin. 1988. Alteration of amino-terminal codons of human granulocyte-colony-stimulating factor increases expression levels and allows efficient processing by methionine aminopeptidase in E. coli. Gene 65:13–22.
61. Diaz-Ricci, J. C., L. Regan, and J. E. Bailey. 1991. Effect of alteration of the acetic acid synthesis pathway on the fermentation pattern of Escherichia coli. Biotechnol. Bioeng. 38:1318–1324.
62. Dosch, D. C., F. F. Salvacion, and W. Epstein. 1984. Tetracycline resistance element of pBR322 mediates potassium transport. J. Bacteriol. 160:1188–1190.
63. Dubé, S., J. W. Fisher, and J. S. Powell. 1988. Glycosylation at specific sites of erythropoietin is essential for biosynthesis, secretion, and biological function. J. Biol. Chem. 263:17516–17521.
64. Dykhuizen, D., and D. Hartl. 1978. Transport by the lactose permease of Escherichia coli as the basis of lactose killing. J. Bacteriol. 135:876–882.
65. Ellman, J., D. Mendel, S. Anthony-Cahill, C. J. Noren, and P. G. Schultz. 1991. Biosynthetic methods for introducing unnatural amino acids site-specifically into proteins. Methods Enzymol. 202:301–336.
66. El-Mansi, E. M. T., and W. H. Holms. 1989. Control of carbon flux to acetate excretion during growth of Escherichia coli in batch and continuous cultures. J. Gen. Microbiol. 135:2875–2883.
67. Enfors, S. O. 1992. Control of in vivo proteolysis in the production of recombinant proteins. Trends Biotechnol. 10:310–315.
68. Ensley, B. D. 1986. Stability of recombinant plasmids in industrial organisms. Crit. Rev. Biotechnol. 4:263–277.
69. Ensley, B. D., B. J. Ratzkin, T. D. Osslund, M. J. Simon, L. P. Wackett, and D. T. Gibson. 1983. Expression of naphthalene oxidation genes in Escherichia coli results in the biosynthesis of indigo. Science 222:167–169.
70. Ernst, J. F., and E. Kawashima. 1988. Variations in codon usage are not correlated with heterologous gene expression in Saccharomyces cerevisiae and Escherichia coli. J. Biotechnol. 7:1–10.
71. Fieschko, J., and T. Ritch. 1986. Production of human alpha consensus interferon in recombinant Escherichia coli. Chem. Eng. Commun. 45:229–240.
72. Fish, N. M., and M. Hoare. 1988. Recovery of protein inclusion bodies. Biochem. Soc. Trans. 16:102–104.
73. Ford, C. F., I. Suominen, and C. E. Glatz. 1991. Fusion tails for the recovery and purification of recombinant proteins. Protein Expr. Purif. 2:95–107.
74. Francisco, J. A., C. F. Earhart, and G. Georgiou. 1992. Transport and anchoring of β-lactamase to the external surface of Escherichia coli. Proc. Natl. Acad. Sci. USA 89:2713–2717.
75. Fuchs, P., F. Breitling, S. Dübel, T. Seehaus, and M. Little. 1991. Targeting recombinant antibodies to the surface of Escherichia coli: fusion to a peptidoglycan associated lipoprotein. Bio/Technology 9:1369–1372.
76. Georgiou, G. 1995. Expression of proteins in bacteria, p. 101–127. In J. L. Cleland and C. S. Craik (ed.), Principles and Practice of Protein Engineering. John Wiley & Sons, New York.
77. Georgiou, G., J. J. Chalmers, and M. L. Shuler. 1985. Continuous immobilized recombinant protein production from E. coli capable of selective protein excretion: a feasibility study. Biotechnol. Prog. 1:75–79.
78. Georgiou, G. and E. De Bernardez-Clark (ed.). 1991. Protein Refolding. American Chemical Society, Washington, D.C.
79. Gerdes, K. 1988. The parB (hok/sok) locus of plasmid R1: a general purpose plasmid stabilization system. Bio/Technology 6:1402–1405.
80. Germino, J., and D. Bastia. 1884. Rapid purification of a cloned gene product by genetic fusion and site-specific proteolysis. Proc. Natl. Acad. Sci. USA 81:4692–4696.
81. Gleister, I. E., and S. Bauer. 1981. Growth of E. coli W to high cell concentration by oxygen level linked control of carbon source concentration. Biotechnol. Bioeng. 23:1015–1021.
82. Glick, B. R., and J. J. Pasternak. 1994. Molecular Biotechnology: Principles and Applications of Recombinant DNA. American Society for Microbiology, Washington, D.C.
83. Goeddel, D. V., H. L. Heyneker, T. Hozumi, R. Arentzen, K. Itakura, D. G. Yansura, M. J. Ross, G. Miozzari, R. Crea, and P. H. Seeburg. 1979. Direct expression in Escherichia coli of a DNA sequence coding for human growth hormone. Nature (London) 281:544–548.
84. Goeddel, D. V., D. G. Kleid, F. Bolivar, H. L. Heyneker, D. G. Yansura, R. Crea, T. Hirose, A. Kraszewski, K. Itakura, and A. D. Riggs. 1979. Expression in Escherichia coli of chemically synthesized genes for human insulin. Proc. Natl. Acad. Sci. USA 76:106–110.
85. Goochee, C. F., M. J. Gramer, D. C. Anderson, J. B. Bahr, and J. R. Rasmussen. 1992. The oligosaccharides of glycoproteins: factors affecting their synthesis and their influence on glycoprotein properties, p. 199–240. In P. Todd, S. K. Sikdar, and M. Bier (ed.), Frontiers in Bioprocessing II. American Chemical Society, Washington, D.C.
86. Goodson, R. J., and N. V. Katre. 1990. Site-directed pegylation of recombinant interleukin-2 at its glycosylation site. Bio/Technology 8:343–346.
87. Gottesman, S. 1990. Minimizing proteolysis in Escherichia coli: genetic solutions. Methods Enzymol. 185:119–129.
88. Gray, G. L., J. S. Baldridge, K. S. McKeown, H. L. Heyneker, and C. N. Chang. 1985. Periplasmic production of correctly processed human growth hormone in Escherichia coli: natural and bacterial signal sequences are interchangeable. Gene 39:247–254.
89. Grossman, A. D., J. W. Erickson, and C. A. Gross. 1984. The htpR gene product of E. coli is a sigma factor for heat shock promoters. Cell 38:383–390.
90. Gschaedler, A., and J. Boudrant. 1994. Amino acid utilization during batch and continuous cultures of Escherichia coli on a semi-synthetic medium. J. Biotechnol. 37:235–251.
91. Hamilton, C. M., M. Aldea, B. K. Washburn, P. Babitzke, and S. R. Kushner. 1989. New method for generating deletions and gene replacements in Escherichia coli. J. Bacteriol. 171:4617–4622.
92. Hantke, K., and V. Braun. 1978. Functional interaction of the tonA/tonB receptor system in Escherichia coli. J. Bacteriol. 135:190–197.
93. Hart, R. A., D. M. Giltinan, P. M. Lester, D. H. Reifsnyder, J. R. Ogez, and S. E. Builder. 1994. Effect of environment on insulin-like growth factor I refolding selectivity. Biotechnol. Appl. Biochem. 20:217–232.
94. Hart, R. A., P. M. Lester, D. H. Reifsnyder, J. R. Ogez, and S. E. Builder. 1994. Large scale, in situ isolation of periplasmic IGF-I from E. coli. Bio/Technology 12:1113–1117.
95. Hirel, P. H., J. M. Schmitter, P. Dessen, G. Fayat, and S. Blanquet. 1989. Extent of N-terminal methionine excision from Escherichia coli proteins is governed by the side-chain length of the penultimate amino acid. Proc. Natl. Acad. Sci. USA 86:8247–8251.
96. Hoffman, S. J., D. L. Looker, J. M. Roehrich, P. E. Cozart, S. L. Durfee, J. L. Tedesco, and G. L. Stetler. 1990. Expression of fully functional tetrameric human hemoglobin in Escherichia coli. Proc. Natl. Acad. Sci. USA 87:8521–8525.
97. Hopp, T. P., K. S. Pricket, V. L. Price, R. T. Libby, C. J. March, D. P. Cerretti, D. L. Urdal, and P. J. Conlon. 1988. A short polypeptide marker sequence useful for recombinant protein identification and purification. Bio/Technology 6:1204–1210.
98. Hsiung, H. M., and W. C. MacKellar. 1987. Expression of bovine growth hormone derivative in Escherichia coli and the use of the derivatives to produce natural sequence growth hormone by cathepsin C cleavage. Methods Enzymol. 153:390–401.
99. Hsiung, H. M., N. G. Mayne, and G. W. Becker. 1986. High-level expression, efficient secretion and folding of human growth hormone in Escherichia coli. Bio/Technology 4:991–995.
100. Hui, A., and H. A. deBoer. 1987. Specialized ribosome system: preferential translation of a single mRNA species by a subpopulation of mutated ribosomes in Escherichia coli. Proc. Natl. Acad. Sci. USA 84:4762–4766.
101. Imanaka, T., H. Tsunekawa, and S. Aiba. 1980. Phenotypic stability of trp operon recombinant plasmids in Escherichia coli. J. Gen. Microbiol. 118:253–261.
102. Ingram, L. O., and T. Conway. 1988. Expression of different levels of ethanologenic enzymes from Zymomonas mobilis in recombinant strains of Escherichia coli. Appl. Environ. Microbiol. 54:397–404.
103. Itakura, K., H. Hirose, R. Crea, A. D. Riggs, H. L. Heyneker, F. Bolivar, and H. W. Boyer. 1977. Expression in Escherichia coli of a chemically synthesized gene for the hormone somatostatin. Science 198:1056–1063.
104. Ito, A., T. Katoh, H. Gomi, F. Kishimoto, H. Agui, S. Ogino, K. Yoda, M. Yamasaki, and G. Tamura. 1986. Expression and secretion of human epidermal growth factor in Escherichia coli. Agric. Biol. Chem. 50:1381–1388.
105. Izard, J. W., and D. A. Kendall. 1994. Signal peptides: exquisitely designed transport promoters. Mol. Microbiol. 13:765–773.
106. Jensen, E. B., and S. Carlsen. 1990. Production of recombinant human growth hormone in Escherichia coli: expression of different precursors and physiological effects of glucose, acetate, and salts. Biotechnol. Bioeng. 36:1–11.
107. Jespers, J. S., A. Roberts, S. M. Mahler, G. Winter, and H. R. Hoogenboom. 1994. Guiding the selection of human antibodies from phage display repertoires to a single epitope of an antigen. Bio/Technology 12:899–903.
108. Johnson, B. H., and M. H. Hecht. 1994. Recombinant proteins can be isolated from E. coli cells by repeated cycles of freezing and thawing. Bio/Technology 12:1357–1360.
109. Joly, J. C., and J. R. Swartz. 1994. Protein folding activities of Escherichia coli protein disulfide isomerase. Biochemistry 33:4231–4236.
110. Jung, G., P. Denefle, J. Becquart, and J. F. Mayaux. 1988. High-cell density fermentation studies of recombinant Escherichia coli strains expressing human interleukin-1β. Ann. Inst. Pasteur Microbiol. 139:129–146.
111. Kamitani, S., Y. Akiyama, and K. Ito. 1992. Identification and characterization of an Escherichia coli gene required for the formation of correctly folded alkaline phosphatase, a periplasmic enzyme. EMBO J. 11:57–62.
112. Kapralek, F., P. Jecmen, J. Sedlacek, M. Fabry, and S. Zadrazil. 1991. Fermentation conditions for high-level expression of the tac-promoter-controlled calf prochymosin cDNA in Escherichia coli HB101. Biotechnol. Bioeng. 37:71–79.
113. Katayama, Y., S. Gottesman, J. Pumphrey, S. Rudikoff, W. P. Clark, and M. R. Maurizi. 1988. The two-component, ATP-dependent Clp protease of Escherichia coli. J. Biol. Chem. 263:15226–15236.
114. Katre, N. V., M. J. Knauf, and W. J. Laird. 1987. Chemical modification of recombinant interleukin-2 by polyethylene glycol increases its potency in the murine Meth A sarcoma model. Proc. Natl. Acad. Sci. USA 84:1487–1491.
115. Khosla, C., and J. E. Bailey. 1988. Heterologous expression of a bacterial hemoglobin improves the growth properties of recombinant Escherichia coli. Nature (London) 331:633–635.
116. Khosla, C., J. E. Curtis, J. DeModena, U. Rinas, and J. E. Bailey. 1990. Expression of intracellular hemoglobin improves protein synthesis in oxygen-limited Escherichia coli. Bio/Technology 8:849–853.
117. Klein, B. K., S. R. Hill, C. S. Devine, E. Rowold, C. E. Smith, S. Galosy, and P. O. Olins. 1991. Secretion of active bovine somatotropin in Escherichia coli. Bio/Technology 9:869–872.
118. Klessen, C., K. H. Schmidt, J. Gumbert, H. H. Grosse, and H. Malke. 1989. Complete secretion of activable bovine prochyosin by genetically engineered L forms of Proteus mirabilis. Appl. Environ. Microbiol. 55:1009–1015.
119. Knappik, A. C. Krebber, and A. Plückthun. 1993. The effect of folding catalysts on the in vivo folding process of different antibody fragments expressed in Escherichia coli. Bio/Technology 11:77–83.
120. Kopetzki, E., G. Schumacher, and P. Buckel. 1989. Control of formation of active soluble or inactive insoluble baker’s yeast alpha-glucosidase PI in Escherichia coli by induction and growth conditions. Mol. Gen. Genet. 216:149–155.
121. Kornacker, M. G., and A. P. Pugsley. 1990. The normally periplasmic enzyme β-lactamase is specifically and efficiently translocated through the Escherichia coli outer membrane when it is fused to the cell-surface enzyme pullulanase. Mol. Microbiol. 4:1101–1109.
122. Kosinski, M. J., U. Rinas, and J. E. Bailey. 1992. Isopropyl-β-D-thiogalactopyranoside influences the metabolism of Escherichia coli. Appl. Microbiol. Biotechnol. 36:782–784.
123. Lacatena, R. M., and G. Cesareni. 1981. Base pairing of RNA I with its complementary sequence in the primer precursor inhibits ColE1 replication. Nature (London) 294:623–626.
124. Ladisch, M. R., R. C. Willson, C. D. C. Painton, and S. E. Builder. 1990. Protein Purification: from Molecular Mechanisms to Large-Scale Processes. American Chemical Society, Washington, D.C.
125. Larson, J., K. Gerdes, J. Light, and S. Molin. 1984. Low-copy-number plasmid-cloning vectors amplifiable by derepression of an inserted foreign promoter. Gene 28:45–54.
126. Lee, S. C., and P. O. Olins. 1992. Effect of overproduction of heat shock chaperones GroESL and DnaK on human procollagenase production in Escherichia coli. J. Biol. Chem. 267:2849–2852.
127. Leipold, R. J., and P. Dhurjati. 1993. Specialized ribosomes in Escherichia coli. Biotechnol. Prog. 9:443–449.
128. Liao, J. C., and J. Delgado. 1993. Advances in metabolic control analysis. Biotechnol. Prog. 9:221–233.
129. Lin, N. S., and J. R. Swartz. 1992. Production of heterologous proteins from recombinant DNA Escherichia coli in bench fermentors. Methods (Orlando) 4:159–168.
130. Lin-Chao, S., and S. N. Cohen. 1991. The rate of processing and degradation of antisense RNAI regulates the replication of ColE1-type plasmids in vivo. Cell 65:1233–1242.
131. Liu, J., and C. T. Walsh. 1990. Peptidyl-prolyl cis-trans-isomerase from Escherichia coli: a periplsmic homolog of cyclophilin that is not inhibited by cyclosporin A. Proc. Natl. Acad. Sci. USA 87:4028–4032.
132. Lu, H. S., C. L. Clogston, L. A. Merewether, L. O. Narhi, and T. C. Boone. 1993. Role of disulfide bonds in folding recombinant human granulocyte colony stimulating factor produced in Escherichia coli, p. 189–202. In J. L. Cleland (ed.), Protein Folding: In Vivo and In Vitro. American Chemical Society, Washington, D.C.
133. Luli, G. W., and W. R. Strohl. 1990. Comparison of growth, acetate production, and acetate inhibition of Escherichia coli strains in batch and fed-batch fermentations. Appl. Environ. Microbiol. 56:1004–1011.
134. Maniatis, T., M. Ptashne, K. Backman, D. Kleid, S. Flashman, A. Jeffrey, and R. Maurer. 1975. Recognition sequences of repressor and polymerase in the operators of bacteriophage lambda. Cell 5:109–113.
135. Martin, U., S. Fischer, U. Kohnert, U. Opitz, R. Rudolph, G. Sponer, A. Stern, and K. Strein. 1991. Thrombolysis with an Escherichia coli-produced recombinant plasminogen activator (BM 01.022) in the rabbit model of jugular vein thrombosis. Thromb. Haemostasis 65:560–564.
136. Matthews, D. J., and J. A. Wells. 1993. Substrate phage: selection of protease substrates by monovalent phage display. Science 260:1113–1117.
137. Matthews, G. 1994. The leucine-responsive regulatory protein, a global regulator of metabolism in Escherichia coli. Microbiol. Rev. 58:401–425.
138. Maurizi, M. R. 1992. Proteases and protein degradation in Escherichia coli. Experientia 48:178–201.
139. Maurizi, M. R., P. Trisler, and S. Gottesman. 1985. Insertional mutagenesis of the lon gene in Escherichia coli: lon is dispensable. J. Bacteriol. 164:1124–1135.
140. McCafferty, J., A. D. Griffiths, G. Winter, and D. J. Chiswell. 1990. Phage antibodies: filamentous phage displaying antibody variable domains. Nature (London) 348:552–554.
141. McDowell, J. C., J. W. Roberts, D. J. Jin, and C. Gross. 1994. Determination of intrinsic transcription termination efficiency by RNA polymerase elongation rate. Science 266:822–825.
142. McMurry, I., R. E. Petrucci, Jr., and S. B. Levy. 1980. Active efflux of tetracycline encoded by four genetically different tetracycline resistance determinants in Escherichia coli. Proc. Natl. Acad. Sci. USA 77:3974–3977.
143. Meacock, P. A., and S. N. Cohen. 1980. Partitioning of bacterial plasmids during cell division: a cis-acting locus that accomplishes stable plasmid inheritance. Cell 20:529–542.
144. Medynski, D. 1994. Phage display: all dressed up and ready to role. Bio/Technology 12:1134–1136.
145. Meerman, H. J., and G. Georgiou. 1994. Construction and characterization of a set of E. coli strains deficient in all known loci affecting the proteolytic stability of secreted recombinant proteins. Bio/Technology 12:1107–1110.
146. Merlin, T. L., G. E. Davis, W. L. Anderson, R. K. Moyzis, and J. K. Griffith. 1989. Aminoglycoside uptake increased by tet gene expression. Antimicrob. Agents Chemother. 33:1549–1552.
147. Meyer, H. P., C. Leist, and A. Fiechter. 1984. Acetate formation in continuous culture of Escherichia coli K12 D1 on defined and complex media. J. Biotechnol. 1:355–358.
148. Missiakas, D., C. Georgopoulos, and S. Raina. 1994. The Escherichia coli dsbC (xprA) gene encodes a periplasmic protein involved in disulfide bond formation. EMBO J. 13:2013–2020.
149. Mitraki, A., and J. King. 1989. Protein folding intermediates and inclusion body formation. Bio/Technology 7:690–697.
150. Miyake, T., T. Oka, T. Mishizawa, F. Misoka, T. Fuwa, K. Yoda, M. Yamasaki, and G. Tamura. 1985. Secretion of human interferon-alpha induced by using secretion vectors containing a promoter and signal sequence of alkaline phosphatase gene of Escherichia coli. J. Biochem. 97:1429–1436.
151. Mizutani, S. H. Mori, S. Shimizu, K. Sakaguchi, and T. Kobayashi. 1986. Effect of amino acid supplement on cell yield and gene product in Escherichia coli harboring plasmid. Biotechnol. Bioeng. 28:204–209.
152. Moks, T., L. Abrahmsen, E. Holmgren, M. Bilich, A. Olsson, M. Uhlen, G. Pohl, C. Sterky, H. Hultberg, S. Josephson, A. Holmgren, H. Jörnvall, and B. Nilsson. 1987. Expression of human insulin-like growth factor I in bacteria: use of optimized gene fusion vectors to facilitate protein purification. Biochemistry 26:5239–5244.
153. Morrow, J. F., S. N. Cohen, A. C. Y. Chang, H. W. Boyer, H. M. Goodman, and R. B. Helling. 1974. Replication and transcription of eukaryotic DNA in Escherichia coli. Proc. Natl. Acad. Sci. USA 71:1743–1747.
154. Moser, D. R., and J. L. Campbell. 1983. Characterization and complementation of pMB1 copy number mutant: effect of RNA I gene dosage on plasmid copy number and incompatibility. J. Bacteriol. 154:809–818.
155. Naglak, T. J., and H. Y. Wang. 1990. Recovery of a foreign protein from the periplasm of Escherichia coli by chemical permeabilization. Enzyme Microb. Technol. 12:603–611.
156. Neidhardt, F. C. 1987. Chemical composition of the cell, p. 3–6. In F. C. Neidhardt, J. L. Ingraham, K. B. Low, B. Magasanik, M. Schaechter, and H. E. Umbarger (ed.), Escherichia coli and Salmonella typhimurium: Cellular and Molecular Biology. American Society for Microbiology, Washington, D.C.
157. Neidhardt, F. C., R. A. VanBogelen, and E. T. Lau. 1983. Molecular cloning and expression of a gene that controls the high-temperature regulon of Escherichia coli. J. Bacteriol. 153:597–603.
158. Nierman, W. C., and T. Feldblyum. 1985. Cryopreservation of cultures that contain plasmids. Dev. Ind. Microbiol. 26:423–434.
159. Nilsson, B., G. Forsberg, T. Moks, M. Hartmannis, and M. Uhlen. 1992. Fusion proteins in biotechnology and structural biology. Curr. Opin. Struct. Biol. 2:569–575.
160. Noack, D., M. Roth, R. Geuther, G. Müller, K. Undisz, C. Hoffmeie, and S. Gaspar. 1981. Maintenance and genetic stability of vector plasmids pBR322 and pBR325 in Escherichia coli K12 strains grown in a chemostat. Mol. Gen. Genet. 184:121–124.
161. Ostermeier, M., and G. Georgiou. 1994. The folding of bovine pancreatic trypsin inhibitor in the Escherichia coli periplasm. J. Biol. Chem. 269:21072–21077.
162. Patnaik, R., and J. C. Liao. 1994. Engineering of Escherichia coli central metabolism for aromatic metabolite production with near theoretical yield. Appl. Environ. Microbiol. 60:3903–3908.
163. Phillips, T. A., R. A. VanBogelen, and F. C. Neidhardt. 1984. lon gene product of Escherichia coli is a heat-shock protein. J. Bacteriol. 159:283–287.
164. Pierce, J., and S. Gutteridge. 1985. Large-scale preparation of ribulosebisphosphate carboxylase from a recombinant system in Escherichia coli characterized by extreme plasmid instability. Appl. Environ. Microbiol. 49:1094–1100.
165. Plückthun, A., and A. Skerra. 1989. Expression of functional antibody Fv and Fab fragments in Escherichia coli. Methods Enzymol. 178:497–515.
166. Polisky, B. 1988. ColE1 replication control circuitry: sense from antisense. Cell 55:929–932.
167. Pugsley, A. P. 1993. The complete general secretory pathway in gram-negative bacteria. Microbiol. Rev. 57:50–108.
168. Reid, J. L., and A. Collmer. 1987. An nptI-scaB-sacR cartridge for constructing directed, unmarked mutations in gram-negative bacteria by marker exchange-eviction mutagenesis. Gene 57:239–246.
169. Reiling, H. E., H. Laurila, and A. Fiechter. 1985. Mass culture of Escherichia coli: medium developments for low and high censity cultivation of Escherichia coli B/r in minimal and complex media. J. Biotechnol. 2:191–206.
170. Remaut, E., H. Tsao, and W. Fiers. 1983. Improved plasmid vectors with a thermoinducible expression and temperature-regulated runaway replication. Gene 22:103–113.
171. Ringquist, S., S. Shinedling, D. Barrick, L. Green, J. Binkley, G. D. Stormo, and L. Gold. 1992. Translation initiation in Escherichia coli: sequences within the ribosome-binding site. Mol. Microbiol. 6:1219–1229.
172. Roberts, B. L., W. Markland, A. C. Ley, R. B. Kent, D. W. White, S. K. Guterman, and R. C. Ladner. 1992. Directed evolution of a protein: selection of potent neutrophil inhibitors displayed on M13 fusion phage. Proc. Natl. Acad. Sci. USA 89:2419–2433.
173. Rosteck, P. R., Jr., and C. L. Hershberger. 1983. Selective retention of recombinant plasmids coding for human insulin. Gene 25:29–38.
174. Rudolph, R. 1996. Successful protein folding on an industrial scale, p. 283–298. In J. L. Cleland and C. S. Craik (ed.), Principles and Practice of Protein Engineering. John Wiley & Sons, New York.
175. Rusch, S. L., and D. A. Kendall. 1994. Transport of an export-defective protein by a highly hydrophobic signal peptide. J. Biol. Chem. 269:1243–1248.
176. Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecular Cloning: a Laboratory Manual, 2nd ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y.
177. Schauder, B., and J. E. G. McCarthy. 1989. The role of bases upstream of the Shine-Dalgarno region and in the coding sequence in the control of gene expression in Escherichia coli: translation and stability of mRNA’s in vivo. Gene 78:59–72.
178. Schein, C. H. 1989. Production of soluble recombinant proteins in bacteria. Bio/Technology 7:1141–1149.
179. Schein, C. H., and M. H. M. Noteborn. 1988. Formation of soluble recombinant proteins in Escherichia coli is favored by lower growth temperature. Bio/Technology 6:291–294.
180. Schoner, R. G., L. F. Ellis, and B. E. Schoner. 1985. Isolation and purification of protein granules from Escherichia coli cells overproducing bovine growth hormone. Bio/Technology 3:151–154.
181. Seetharam, R., R. A. Heeren, E. Y. Wong, S. R. Braford, B. K. Klein, S. Aykent, C. E. Kotts, K. J. Mathis, B. F. Bishop, M. J. Jennings, C. E. Smith, and N. R. Siegel. 1988. Mistranslation in IGF-1 during over-expression of the protein in Escherichia coli using a synthetic gene containing low frequency codons. Biochem. Biophys. Res. Commun. 155:518–523.
182. Seo, J. H., and J. E. Bailey. 1985. Effects of recombinant plasmid content on growth properties and cloned gene product formation in Escherichia coli. Biotechnol. Bioeng. 27:1668–1674.
183. Seo, J. H., A. I. Löffler, and J. E. Bailey. 1988. A parametric study of cloned fusion protein expression in Escherichia coli. Biotechnol. Bioeng. 32:725–730.
184. Seow, H. F., C. R. Goh, L. Krishnan, and A. G. Porter. 1989. Bacterial expression, facile purification and properties of recombinant human lymphotoxin (tumor necrosis factor beta). Bio/Technology 7:363–368.
185. Shatzman, A. R. 1990. Gene expression using gram-negative bacteria. Curr. Opin. Biotechnol. 1:5–11.
186. Shevchik, V. E., G. Condemine, and J. Robert-Baudouy. 1994. Characterization of DsbC, a periplasmic protein of Erwinia chrysanthemi and Escherichia coli with disulfide isomerase activity. EMBO J. 13:2007–2012.
187. Silhavy, T. J., M. J. Casadaban, H. A. Shuman, and J. R. Beckwith. 1976. Conversion of β-galactosidase to a membrane-bound state by gene fusion. Proc. Natl. Acad. Sci. USA 73:3423–3427.
188. Silhavy, T. J., H. A. Shuman, J. Beckwith, and M. Schwartz. 1977. Use of gene fusions to study outer membrane protein localization in Escherichia coli. Proc. Natl. Acad. Sci. USA 74:5411–5415.
189. Simon, L. D., B. Randolph, N. Irwin, and G. Binkowski. 1983. Stabilization of proteins by a bacteriophage T4 gene cloned in Escherichia coli. Proc. Natl. Acad. Sci. USA 80:2059–2062.
190. Skerra, A., and A. Plückthun. 1988. Assembly of a functional immunoglobulin Fv fragment in Escherichia coli. Science 240:1038–1041.
191. Skogman, G., and J. Nilsson. 1984. Temperature-dependent retention of a tryptophan-operon-bearing plasmid in Escherichia coli. Gene 31:117–122.
192. Skogman, G., J. Nilsson, and P. Gustafsson. 1983. The use of a partition locus to increase stability of tryptophan-operon-bearing plasmids in Escherichia coli. Gene 23:105–115.
193. Slater, S. C., W. H. Voige, and D. E. Dennis. 1988. Cloning and expression in Escherichia coli of the Alcaligenes eutrophus H16 poly-β-hydroxybutyrate biosynthetic pathway. J. Bacteriol. 170:4431–4436.
194. Smith, D. B., and K. S. Johnson. 1988 One-step purification of polypeptides expressed in Escherichia coli as fusions with glutathione S-transferase. Gene 67:31–40.
195. Smith, G. P. 1985. Filamentous fusion phage: novel expression vectors that display cloned antigens on the virion surface. Science 228:1315–1317.
196. Smith, R. A. G., J. M. Dewdney, R. Fears, and G. Poste. 1993. Chemical derivatization of therapeutic proteins. Trends Biotechnol. 111:397–403.
197. Sorensen, M. A., C. G. Kurland, and S. Pedersen. 1989. Codon usage determines translation rate in Escherichia coli. J. Mol. Biol. 207:365–377.
198. St. John, A. C., and A. L. Goldberg. 1980. Effects of starvation for potassium and other inorganic ions on protein degradation and ribonucleic acid synthesis in Escherichia coli. J. Bacteriol. 143:1223–1233.
199. Storrs, S. B., and T. M. Przybycien. 1991. Commercial-scale refolding of recombinant methionyl bovine somatotropin, p. 197–205. In G. Georgiou and E. De Bernardez-Clark (ed.), Protein Refolding. American Chemical Society, Washington, D.C.
200. Strauch, K. L., and J. Beckwith. 1988. An Escherichia coli mutation preventing degradation of abnormal periplasmic proteins. Proc. Natl. Acad. Sci. USA 85:1576–1580.
201. Studier, F. W., A. H. Rosenberg, J. J. Dunn, and J. W. Dubendorff. 1990. Use of T7 polymerase to direct expression of cloned genes. Methods Enzymol. 185:60–89.
202. Talmadge, K., J. Kaufman, and W. Gilbert. 1980. Bacteria mature preproinsulin to proinsulin. Proc. Natl. Acad. Sci. USA 77:3988–3992.
203. Talmadge, K., S. Stahl, and W. Gilbert. 1980. Eukaryotic signal sequence transports insulin antigen in Escherichia coli. Proc. Natl. Acad. Sci. USA 77:3369–3373.
204. Terasawa, M., M. Fukushima, Y. Kurusu, and H. Yukawa. 1990. l-Tryptophan production by the application of high expressed tryptophanase in Escherichia coli. Proc. Biochem. Int. 25:172–175.
205. Thompson, B. G., M. Kole, and D. F. Gerson. 1985. Control of ammonium concentration in Escherichia coli. Biotechnol. Bioeng. 27:818–824.
206. Tokatlidis, K., P. Dhurjati, J. MIllet, P. Beguin, and J. P. Aubert. 1991. High activity of inclusion bodies formed in Escherichia coli overproducing Clostridium thermocellum endoglucanase D. FEBS Lett. 282:205–208.
207. Tsai, B. B. H. S. Lu, W. C. Kenney, C. C. Curless, M. L. Klein, P. H. Lai, D. M. Fenton, B. W. Altrock, and M. B. Mann. 1988. Control of misincorporation of de novo synthesized norleucine into recombinant interleukin-2 in E. coli. Biochem. Biophys. Res. Commun. 156:733–739.
208. Tsai, L. B., M. Mann, F. Morris, C. Rotgers, and D. Fenton. 1987. The effect of organic nitrogen and glucose on the production of recombinant human insulin-like growth factor in high cell density Escherichia coli fermentations. J. Ind. Microbiol. 2:181–187.
209. Tunac, J. B. 1989. High-aeration capacity shake-flask system. J. Ferment. Bioeng. 68:157–159.
210. Umbarger, H. E. 1987. Biosynthesis of the branched-chain amino acids, p. 352–367. In F. C. Neidhardt, J. L. Ingraham, K. B. Low, B. Magasanik, M. Schaechter, and H. E. Umbarger (ed.), Escherichia coli and Salmonella typhimurium: Cellular and Molecular Biology. American Society for Microbiology, Washington, D.C.
211. Valax, P., and G. Georgiou. 1993. Molecular characterization of β-lactamase inclusion bodies produced in Escherichia coli. 1. Composition. Biotechnol. Prog. 9:539–547.
212. Voss, T., E. Falkner, H. Ahorn, E. Krystek, I. Maurer-Fogy, G. Bodo, and R. Hauptmann. 1994. Periplasmic expression of human interferon-alpha2c in Escherichia coli results in a correctly folded molecule. Biochem. J. 298:719–725.
213. Walker, K. W., and H. F. Gilbert. 1994. Effect of redox environment on the in vitro and in vivo folding of RTEM-1β-lactamase and Escherichia coli alkaline phosphatase. J. Biol. Chem. 269:28487–28493.
214. Wanner, B. L. 1993. Gene regulation by phosphate in enteric bacteria. J. Cell. Biochem. 51:47–54.
215. Warne, S. R., C. M. Thomas, M. E. Nugent, and W. C. A. Tacon. 1986. Use of a modified Escherichia coli trpR gene to obtain tight regulation of high-copy-number expression vectors. Gene 46:103–112.
216. Winter, G., A. D. Griffiths, R. E. Hawkins, and H. R. Hoogenboom. 1994. Making antibodies by phage display technology. Annu. Rev. Immunol. 12:433–455.
217. Winter, R. B., K. M. Yen, and B. D. Ensley. 1989. Efficient degradation of trichloroethylene by a recombinant Escherichia coli. Bio/Technology 7:282–285.
218. Wong, E. Y., R. Seetharam, C. E. Kotts, R. A. Heeren, B. K. Klein, S. R. Braford, K. J. Mathis, B. F. Bishop, N. R. Siegel, C. E. Smith, and W. C. Tacon. 1988. Expression of secreted insulin-like growth factor-1 in Escherichia coli. Gene 68:193–203.
219. Wood, T. K., and S. W. Peretti. 1991. Effect of chemically-induced, cloned-gene expression on protein synthesis in E. coli. Biotechnol. Bioeng. 38:397–412.
220. Wülfing, C., and A. Plückthun. 1994. Correctly folded T-cell receptor fragments in the periplasm of Escherichia coli: influence of folding catalysts. J. Mol. Biol. 242:655–669.
221. Wülfing, C., and A. Plückthun. 1994. Protein folding in the periplasm of Escherichia coli. Mol. Microbiol. 12:685–692.
222. Wunderlich, M., and R. Glockshuber. 1993. In vivo control of redox potential during protein folding catalyzed by bacterial protein disulfide-isomerase (DsbA). J. Biol. Chem. 268:24547–24550.
223. Yamaguchi, A., N. Ono, T. Akasaka, T. Noumi, and T. Sawai. 1990. Metal-tetracycline/H+ antiporter of Escherichia coli encoded by a transposon, Tn10. J. Biol. Chem. 265:15525–15530.
224. Yamaguchi, A., T. Udagawa, and T. Sawai. 1990. Transport of divalent cations with tetracycline as mediated by the transposon Tn10-encoded tetracycline resistance protein. J. Biol. Chem. 265:4809–4813.
225. Yansura, D. G., and D. J. Henner. 1990. Use of Escherichia coli trp promoter for direct expression of proteins. Methods Enzymol. 185:54–60.
226. Yee, L., and H. W. Blanch. 1992. Recombinant protein expression in high cell density fed-batch cultures of Escherichia coli. Bio/Technology 10:1550–1556.
227. Yee, L., and H. W. Blanch. 1993. Recombinant trypsin production in high cell density fed-batch cultures in Escherichia coli. Biotechnol. Bioeng. 41:781–790.
228. Yukawa, H., Y. Kurusu, M. Shimazu, M. Terasawa, A. Ohta, and I. Shibuya. 1985. Stabilization by the mini-F fragment of a pBR322 derivative bearing the tryptophan operon in Escherichia coli. Agric. Biol. Chem. 49:3619–3622.
229. Yukawa, H., Y. Kurusu, M. Shimazu, H. Yamagata, and M. Terasawa. 1988. Stabilization of an E. coli plasmid by a mini-F fragment of DNA. J. Ind. Microbiol. 2:323–328.
230. Zabriskie, D. W., and E. J. Arcuri. 1986. Factors influencing productivity of fermentations employing recombinant microorganisms. Enzyme Microb. Technol. 8:706–717.
231. Zhou, Y. N., N. Kusulawa, J. W. Erickson, C. A. Gross, and T. Yura. 1988. Isolation and characterization of Escherichia coli mutants that lack the heat shock sigma factor (σ32). J. Bacteriol. 170:3640–3649.
Chapter108
