`Vol. 76, No. 1, pp. 106-110, January 1979
`Biochemistry
`
`Expression in Escherichia coli of chemically synthesized genes
`for human insulin
`
`(plasmid construction/lac operon/fused proteins/radioimmunoassay/peptide purification)
`
`DAVID V. GOEDDEL", DENNIS G. KLEID*, FRANCISCO BoLIvAR*, HERBERT L. HEYNEKER*,
`DANIEL G. YANsURA*, ROBERTO CaEA**, TADAAKI Hmosei, ADAM KRAszEwsK1*,
`KEIICHI I'rAKUnA*, AND ARTHUR D. Riccsii
`
`‘Division of Molecular Biology. Cenentech, Inc., 460 Point San Bruno Boulevard, South San Francisco, California 94080; and *Division of Biology, City of Hope
`National Medical Center, Duarte, California 91010
`
`Communicated by Ernest Beutler, October 3, 1978
`
`Synthetic genes for human insulin A and B
`ABSTRACT
`chains were cloned separately in plasmid pBR322. The cloned
`synthetic genes were then fused to an Escherichia coll’ fl- a-
`lactosidase gene to provide efficient transcri tion and trans a-
`tion and a stable recursor protein. The insu in peptides were
`cleaved from B-ga actosidase, detected by radioimmunoassay
`and purified. Comglete purification of the A chain and partial
`puri ication of the
`chain were achieved. These products were
`mixed, reduced, and reoxidized. The presence of insulin was
`detected by radioimmunoassay.
`
`Recently improved methods of DNA chemical synthesis,
`combined with recombinant DNA technology, permit the de-
`sign and relatively rapid synthesis of modest-sized genes that
`can be incorporated into prokaryotic cells for gene expression.
`The feasibility of this general approach was first demonstrated
`by the synthesis, and expression in Escherichia coli, of a gene
`for the mammalian peptide somatostatin (1).
`Following the precursor protein approach used for soma-
`tostatin,(1), the experimental design for this work was such that
`the insulin peptide chains would be made in vivo as short tails
`joined by a methionine to the end of 43-galactosidase. After
`synthesis, the insulin chains, which contain no methionine, can
`be cleaved off efficiently by treatment with cyanogen bromide.
`We deliberately chose to construct two separate bacterial
`strains, one for each of the two peptide chains of insulin: the
`21-amino-acid A chain and the 30—amino-acid B chain. In na-
`tive insulin, the two chains are held together by two disulfide
`bonds, and methods have been available for years for joining
`the chains correctly, in ultra, by air oxidation (2). The efficiency
`of correct joining has been variable and often low. However,
`by using S-sulfonated derivatives and an excess of A chain,
`50-80% correct joining has been obtained (3).
`The synthetic plan and chemical synthesis of the DNA
`fragments coding for the A and B chains of human insulin were
`described in a previous paper (4) and were summarized in Fig.
`1 of that paper. In this communication, we describe the as-
`sembly and cloning of the genes for the A and B chains, their
`insertion into the carboxy terminus of the E. coil 5-galactosidase
`structural gene, the expression and purification of the separate
`A and B chains, and their joining to form native human insu-
`lin.
`
`MATERIALS AND METHODS
`
`Bacterial Strains. E. coli K-12 strain 294 (endA, thi", hsr‘,
`hsmfi) (5) was provided by K. Backman. E. coli K-12 strain
`D1210, a lac+ (iQo+z"'y+) derivative of HB101, was con-
`structed by ]. Betz and I. Sadler and obtained from J. Sadler.
`
`The publication costs of this article were defrayed in part by page
`charge payment. This article must therefore be hereb marked “ad-
`oerttsement" in accordance with 18 U. S. C. §l734 so ely to indicate
`this fact.
`
`Enzymes and DNA Preparations. T4 DNA ligase and T4
`polynucleotide kinase were purified as described (6). Restriction
`endonuclease Eco R1 was purified by the procedure of Greene
`et al. (7); HindIII was purified by a method developed by D.
`Goeddel (unpublished). Restriction endonuclease BamHI was
`purchased from Bethesda Research (Rockville, MD); E. colt
`alkaline phosphatase was purchased from Worthington.
`Plasmids, including pBR322 (8), were isolated by a published
`procedure (9) with some modifications. The chemical synthesis
`of the deoxyoligonucleotides (figure 1 of ref. 4) has been de-
`scribed (4). )\plac5 DNA was isolated as described (10).
`The following reaction buffers were used: kinase buffer, 60
`mM Tris-HCI, pH 8/ 15 mM 2-mercaptoethanol/ 10 mM MgCl2;
`ligase buffer, 20 mM Tris-HCI, pH 7.5/10 mM dithiothreitol/ 10
`mM MgCl2; BamHI buffer, 20 mM Tris-HCl, pH 7.5/7 mM
`MgCl2/2 mM 2-mercaptoethanol; EcoRI—HindIII buffer,
`BamHI buffer containing 50 mM NaCl; and phosphatase
`buffer, 50 mM Tris-HCI, pH 8/10 mM MgCl2.
`Assembly of Insulin Genes. The assembly of the right (BB)
`half of the B-chain gene (see figure 1 of ref. 4) will be described
`in detail. Oligonucleotides B2-B9 were phosphorylated indi-
`vidually. Fifty microcuries of [7-32P]ATP (#2000 Ci/mmol,
`New England Nuclear) was evaporated to dryness in a 1.5-ml
`polypropylene tube, then incubated with the oligonucleotide
`(10 pg) and 8 units of T4 polynucleotide kinase in 60 pl of kinase
`buffer. After 20 min at 87°C, 10 nmol of ATP and 10 units of
`T4 kinase were added and the reaction was continued for an
`additional hour. The kinase was inactivated by heating at 90°C
`for 5 min.
`'
`Phosphorylated fragments B2, B3, B6, and B7 (2.5 pg each)
`were combined with 2.5 pg of 5’-OH fragment B1 and
`for 2 hr against 1 liter of ligase buffer. ATP was added to a
`concentration of 0.2 mM, the reaction mixture (60 pl) was
`cooled to 12°C, and T4 ligase (50 units) was added. A separate
`ligation reaction involving phosphorylated fragments B4, B5,
`B8, and B9 and unphosphorylated B10 was performed identi-
`cally. After 12 hr at 12°C, the two ligation reaction mixtures
`were combined, additional T4 ligase (40 units) was added, and
`the mixture was incubated at 12°C for ,4 hr. The mixture was
`extracted with phenol/chloroform and precipitated with eth-
`anol, and the DNA fragments were purified by electrophoresis
`on a 10% acrylamide gel (11). The most slowly migrating band
`was sliced from the gel and the DNA was extracted (11).
`A similar procedure, with the following exceptions, was used
`to assemble the left (BH) half of the B-chain gene. All eight
`
`Abbreviations: BB, left half of insulin B gene; BH, right half of insulin
`B gene; A(SSO3‘), S-sulfonated derivatives of the insulin A-chain
`peptide; B(SSO3'), S-sulfonated derivatives of the insulin B-chain
`peptide.
`1' To whom reprint requests should be addressed.
`
`106
`
`Page 1
`
`SANOFI V. GENENTECH
`
`IPR2015-01624
`
`EXHIBIT 2011
`
`
`
`Biochemistry: Goeddel et al.
`
`Proc. Natl. Acad. Sci. USA 76 (1979)
`
`107
`
`oligonucleotides (H1-H8 in figure 1 of ref. 4, 30 pg each) were
`phosphorylated. Therefore, after complete ligation and before
`purification by gel electrophoresis, the reaction -mixture was
`treated with 400 units of Ec0RI and 400 units of HindIII for
`2 hr at 37°C. The BH band migrating at 46 base pairs was
`eluted from a 10% acrylamide gel.
`_
`The procedure used to construct the A-chain gene was also
`similar to that described for the BB fragment. The only major
`difference was that, after assembly, the 5’ ends of the complete
`A gene were phosphorylated.
`-
`.
`Construction and Characterization of Iac—Insulin Hybrid
`Plasmids. The BB fragment was cloned as follows: 1 pg of
`pBR322 was treated with 5 units of BamHI in BamHI buffer
`for 1 hr at 37°C. After addition of NaCl to 50 mM, Hindlll (5
`units) was added and the reaction was continued for 1 hr. The
`enzymes were inactivated by heating at 70°C for 10 min. The
`prepared pBR322 was ligated to the BB fragment for 3 hr at
`12°C in 25 pl of ligase buffer (containing 0.16 mM ATP) by
`using 20 units of T4 ligase. Half of the resulting DNA was used
`to transform E. coli 294 by a published procedure (12). The BH
`fragment and the A-chain gene were cloned similarly, with the
`appropriate restriction endonucleases to cut pBR322.
`Construction of the plasmids for expression of the synthetic
`insulin genes is described in the legend of Fig. 1. The separate
`chains in insulin are biologically inactive (2) and were synthe-
`sized attached to large precursor proteins. Therefore, the con-
`tainment level of P2-EK1, recommended by the National In-
`stitutes of Health guideline, was used.
`DNA Sequences. The method of Maxam and Gilbert (11)
`was used to determine DNA sequences. Sequence data are not
`included, but will be provided upon request.
`Preparation of Insulin Reagents. Porcine and bovine insulin
`were purchased from Sigma. The S -sulfonated derivatives
`(SSO3") of their A and B chains were prepared and purified as
`described (13). 355-Labeled A(SSO3") and B(SSO3') were
`prepared similarly except that 5 mCi of sodium [35S]sulfite (69
`mCi/mol, New England Nuclear) was substituted for unlabeled
`
`sodium sulfite. After separation of the chains, the specific ac-
`tivity was 92,000 cpm/pg and 32,000 cpm/pg, respectively,
`for A and B chains. The radioimmunoassay for the insulin
`chains is described in the legend of Fig. 3.
`Purification of B Chain of Human Insulin. E. coli
`D1210/plB1 was grown to late logarithmic phase in 7 liters of
`LB medium (10) containing 20 mg of ampicillin per liter. Iso-
`propyl-B-D-thiogalactoside was added to a final concentration
`of 2 mM, and the cells were grown for one more doubling. Wet
`cell paste (24 g) was suspended in 30 ml of BB buffer (10) and
`cells were lysed by one passage through a French press at 4000
`lb/inch” (27.6_MPa). The cell debris was pelleted by centrifu-
`gation at 15,000 rpm for 30 min. The pellet was dissolved in 40
`ml of 6 M guanidinium chloride/ 1% 2-mercaptoethanol and
`centrifuged at 30,000 rpm for 1 hr. The supernatant was di-
`alyzed overnight against 20 liters of H20. The precipitate,
`containing about 1 g of protein, was dissolved in 25 ml of 70%
`formic acid. Cyanogen bromide (1.3 g) was added and the
`mixture was allowed to react overnight at room temperature.
`Formic acid and cyanogen bromide was removed by rotary
`evaporation and the residue was dissolved in 50 ml of 8 M
`guanidinium chloride. S -Sulfonated derivates of the peptide
`mixture were prepared by adding 1 g of sodium tetrathionate
`and 2 g of sodium sulfite, adjusting the pH to 9 with NH4OH,
`and stirring the mixtures at room temperature for 2'4 hr. The
`pH was then adjusted to 5 with acetic acid and the mixture was
`dialyzed twice against 3 liters of H20. The resulting white
`precipitate (~06 g of protein) was pelleted by centrifuging at
`10,000 rpm for 10 min.
`Purification of A Chain of Human Insulin. E. coli 294/
`pIA1 was grown to A 550 of 2 in 5 liters of LB medium con-
`taining 20 mg of ampicillin per liter. This strain is constitutive
`for B-galactosidase and so was not induced. The 15 g (wet
`weight) of cells obtained were processed by the same procedure
`used for the B chain up through the preparation of the S-sul-
`fonated derivatives. After the pH was adjusted to 5 and the
`mixture was dialyzed against H20 to an ionic strength of about
`
`Ecofil
`B 1?-—>
`Hindlll
`pa 1°
`
`BB4-pBR322->
`
`BH
`Ecofil
`aH1-——> o—v
`Hindlll
`
`D
`
`
`
`FIG. 1.
`Construction of lac—-insulin plasmids. pBB101 (2 pg) (pBR322 containing the BB sequence) was treated with EcoRI and Hindlll (20
`units each), and the large fragment was purified on a 10% acrylamide gel. pBH1 (8 pg) (pBR322 containing BH sequence) also was treated with
`EcoRI and HindIII, and the small fragment was purified on a 10% acrylamide gel. These two fragments were ligated to 2 pg of EcoRI-digested
`Aplac5 in 30 pl of ligase buffer with 20 units of ligase. This mixture was used to transform E. coli 294. The configuration of restriction site ends
`(V represents Hindlll; 0 represents EcoRI) was such that only correct joining of the two halves of the B gene would lead to viable plasmids.
`To screen for the presence of the lac fragment, we plated the transformed culture on glucose minimal plates (10) containing 40 pg of 5-bromo-
`4-chloro-3-indoyl-)3-galactoside (X-gal) and 20 pg of ampicillin per ml. Plasmids were prepared from B-galactosidase constitutive (blue) colonies.
`Because the Xplac5 lac operon fragment contains an asymmetrical HindIII site (14), the orientation of that fragment in the resulting plasmids
`can be determined. Plasmid samples of 1 pg were digested with Hindlll and sized on 0.7% agarose gels. Plasmids (15-pg samples) having the
`desired orientation of the lac fragment were then treated with EcoRI, HindIII, and BamHI, and sized on a 10% acrylamide gel to verify the
`presence of both the BH and BB fragments. The diagram of pIB1 (7.1 megadaltons) is not drawn to scale. To construct the lac—insulin A plasmid
`(pIA1, not shown), we ligated 1 pg of EcoRI-treated pA11 (pBR322 containing the A gene) and 3 pg of EcoRI-treated )\plac5 for 4 hr at 4°C.
`Transformants of E. coli 294 were selected for resistance to ampicillin on X-gal plates. Orientation of the lac fragment was determined by digesting
`plasmids purified from the blue colonies with HindIII and BamHI.
`
`Page 2
`
`
`
`108
`
`Biochemistry: Goeddel et al.
`
`Proc. Natl. Acad. Sci. USA 76 (1979)
`
`ABCDEF
`
`
`
`FIG. 2. Sodium dodecyl sulfate/polyacrylamide gel electropho-
`resis of extracts of strain 294/pIA1. Samples were heated in 0’FarrelI
`sample buffer and electrophoresed in a sodium dodecyl sulfate/10%
`gel as described (15). Lanes A, B, and C: total cells; 20, 10, and 5 ul,
`respectively. Lanes D, E, and F: insoluble cell debris; 20, 10, and 5 pl,
`respectively.
`
`0.01 M, the mixture was centrifuged and the supernatant was
`used for further purification (see Results and Discussion).
`
`RESULTS AND DISCUSSION
`
`Assembly and Cloning of B-Chain Gene and A-Chain
`Gene. The gene for the B chain of insulin was designed to have
`an EcoRI restriction site on the left end, a Hindlll site in the
`middle, and a BamHI site at the right end. This was done so that
`both halves, the left EcoRI—HmdIII half (BH) and the right
`HindIII—BamHI half (BB), could be separately cloned in the
`convenient cloning vehicle pBR322 (8) and, after their se-
`quences had been verified, joined to give the complete B ‘gene
`(Fig. l). The BB half was assembled by ligation from 10 oligo-
`deoxyribonucleotides, labeled Bl—B10 in figure 1 of ref. 4,
`made by phosphotriester chemical synthesis. B1 and B10 were
`not phosphorylated, thereby eliminating unwanted polymer-
`ization of these fragments through their cohesive ends (HindIII
`and Burn HI). After purification by preparative acrylamide gel
`electrophoresis and elution of the largest DNA band, the BB
`fragment was inserted into plasmid pBR322 that had been
`cleaved with HindIII and BamHI. About 50% of the ampicil-
`lin-resistant colonies derived from the DNA were sensitive to
`tetracyline, indicating that a nonplasmid HmdIII-BamHI
`fragment had been inserted. The sequences of the small Hin-
`dIII—BamHI fragments from four of these colonies (pBBl01
`to pBB104) were determined (11) and were correct as de-
`signed.
`' The BH fragment was prepared in a similar manner and
`inserted into pBR322 that had been cleaved with EcoRI and
`HindIII restriction endonucleases. Plasmids from three am-
`picillin-resistant, tetracycline-sensitive transformants (pBHl
`to pBH3) were analyzed. The small EcoRI—HindIII fragments
`had the expected nucleotide sequence.
`V
`The A—chain gene was assembled in three parts. The left four,
`middle four, and right four oligonucleotides (see figure 1 of ref.
`4) were ligated separately, then mixed and ligated (oligonu—
`cleotides Al and A12 were unphosphorylated). The assembled
`.A-chain gene was phosphorylated, purified by gel electro-
`phoresis, and cloned in pBR322 at the EcoRI—BamHI sites. The
`
`Page 3
`
`o 0.6
`
`8/8
`
`0
`
`2
`
`4
`
`8
`6
`1/dilution X 10“
`
`10
`
`40
`
`FIG. 3. Reconstitution radioimmunoassay for insulin chains. The
`S -sulfonated A sample was mixed with the S-sulfonated B sample
`in a 1.5-ml conical polypropylene tube and dried under reduced
`pressure. The dried proteins were suspended in 25 pl of 10 mM sodium
`acetate (pH 4.5). Two microliters of 10% (vol/vol) 2-mercaptoethanol
`was added and the mixture was heated (~95°C) for 10 min. The
`mercaptoethanol was removed by ethyl acetate extraction. Two mi-
`croliters of 0.1 M glycine buffer (pH 10.6) was added, and the pH was
`adjusted, if necessary, to 9.6—10.6. The open tube was placed in a
`dessicator over moist hya.mine hydroxide at room temperature for 26
`hr. A diluted aliquot of the reaction mixture was assayed for insulin
`radioimmune activity by use of a commercially available radioim-
`rnunoassay kit (Phadebus Insulin Test, Pharmacia). Dilution and
`assay were done according to the instructions supplied. The ordinate
`B/Bo is the cpm in the pellet of the experimental sample divided by
`the cpm in the pellet obtained with buffer only. 0, 40 pg of porcine
`A(SSO3‘) in 10 mM NI-I41-[CO3 (pH 9) was mixed with 10 ug of bovine
`B(SS03") in the same buffer; 0, porcine A(SS03’) only; El, bovine
`B(SS03‘) only; A, 100 #8 bf porcine A(SSO3') and 93 ug of E. coli
`B-chain fraction F-10 (cleaved by CNBr and S-sulfonated, insoluble
`at pH 5); X, fraction F-10 only; A, 100 ug of porcine A(SS03‘), 93 pg
`of fraction F-10, and 3 ug of bovine B(SSO3‘).
`
`EcoRI-BamHI fragments from two ampicillin-resistant, tet-
`racycline-sensitive clones (pA10 and pA1l) contained the de-
`sired A-gene sequence.
`Construction of Plasmids for Expression of A and B Insulin
`Genes. Fig. 1 illustrates the construction of the lac—insulin B
`plasmid (pIBl). Plasmids pBH1 and pBBl01 were digested with
`EcoRI and Hindlll endonucleases. The small BH fragment of
`pBH1 and the large fragment of pBBl01 (containing the BB
`fragment and most of pBR322) were purified by gel electro-
`phoresis, mixed, and ligated in the presence of EcoRI-cleaved
`Aplac5. The 4.4-megadalton EcaRI fragment of Aplac5 con-
`tains the lac control region and the majority of the fl-galacto-
`sidase structural gene (1, 14). The configuration of the restric-
`tion sites ensures correct joining of BH to BB. The lac Ecolil
`fragment can be inserted in two orientations; thus, only half of
`the clones obtained after transformation should have the desired
`orientation. The orientation of 10 ampicillin-resistant, fi-ga-
`lactosidase-constitutive clones were checked by restriction
`analysis (see legend of Fig. 1). Five of these colonies contained
`the entire B-gene sequence and the correct reading frame from
`the B-galactosidase gene into the B-chain gene. One, pIBl, was
`chosen for subsequent experiments.
`In a similar experiment, the 4.4-megadalton lac fragment
`from )\plac5 was introduced into the pA11 plasmid at the EcoRI
`site to give pIAl. pIA1 is identical to pIBl except that the A-
`gene fragment is substituted for the B-gene fragment. DNA
`sequence analysis showed that the correct A- and B-chain gene
`sequences were retained in pIAl and pIBl, respectively.
`
`
`
`Biochemistry: Goeddel et al.
`
`Proc. Natl. Acad. Sci. USA 76 (1979)
`
`109
`
`28
`
`24
`
`20
`
`n
`E3 16
`><
`.'-E12
`
`8
`4
`
`o
`
`o
`
`A
`
`BovineB
`J,
`
`1.
`‘)2
`3
`
`5
`
`E
`4 .5
`
`3 ‘<3
`
`2 E
`1
`-,9,’
`E
`
`4
`
`8
`
`12
`
`16
`
`20
`
` 32
`
`Porciine A
`
`24
`
`“o
`X
`.516
`
`8
`
`8 T?x
`7 2
`6 E
`
`6
`4 ‘QG:C
`3
`
`E
`2 :5
`.‘2cc
`..a
`
`E ,
`0
`
`4
`
`8
`
`_/3
`
`.
`24
`
`28
`
`‘\_..._.
`1 6
`20
`1 2
`Time, min
`FIG. 5. Reversed-phase high-performance liquid chromatogra-
`phy. (A) A sample (500 ug), purified by DEAE-cellulose chromatog-
`raphy (B chain, Fig. 4), was subjected to high-performance liquid
`chromatography at room temperature on an SP 3500 liquid chro-
`matograph (Spectra-Physics) equipped with a LiChrosorb RP-8 0.3
`X 25 cm column (Merck EM). The eluting buffer was 50 mM NH4OAc
`with an acetonitrile gradient of 24-60%. Fractions were collected and
`dried, and samples were assayed for B-chain radioimmune activity
`by our reconstitution assay. (B) An A-chain sample (500 ug total
`protein), partially purified by aminoethyl-cellulose chromatography,
`was subjected to high-performance liquid chromatography with an
`acetonitrile gradient of 15-60%. Fractions were collected and samples
`were assayed for A-chain radioimmune activity. Solid line, A230; A,
`radioimmune activity.
`
`used a radioimmunoassay based on the reconstitution of com-
`plete insulin from the separate chains. The insulin reconstitution
`procedure of Katsoyannis et al. (3), adapted to a 27-ul assay
`volume, provided a very suitable assay. Easily detectable insulin
`radioimmune activity was obtained after S -sulfonated deriv-
`atives of theinsulin chains were mixed and reconstituted by the
`procedure described in the le end of Fig. 3. The separate S-
`sulfonated chains of insulin 0 not react significantl , after
`reduction and oxidation, with the anti-insulin anti
`y used.
`Our reconstitution assay, though not extremely sensitive (limits
`of detection about 1 pg), was specific and suitable for following
`insulin chain radioimmune activity during purification.
`To use the reconstitution assay, we partially purified the
`(3-galactosidase—A or B chain hybrid protein, cleaved it with
`cyanogen bromide, formed S -sulfonated derivatives, and
`partially purified the peptides as described in Materials and
`Methods. This procedure was based on our earlier experience
`with the purification of somatostatin from E. coli (unpublished
`data) and the known properties of the insulin chains.
`
`2
`
`'1‘C-
`3>
`Q.
`E3U,
`.5111
`.2+-u:11
`:1»c:3
`
`X1
`
`D:
`Bovine[”Slinsulin,cpmX10"
`
`
`
`E E.
`
`9.'0:1:
`
`20
`
`40
`
`60
`
`80
`
`100
`
`120
`
`140
`
`FIG. 4. DEAE-cellulose chromatography of an extract of strain
`D1210/pIB1. The fraction that was insoluble at pH 5 (580 mg of
`protein) was dissolved in 20 ml of 10 mM NH4l-{CO3 and adjusted to
`pH 9. 35S-Labeled bovine B(SSO3") (135,000 cpm; 4.4 pg) was added
`and the sample was applied to a 2 X 60 cm column of Whatman DE52.
`Elution was with a 1-liter gradient of 0.01—2.0 M NH4HCO3 (pH 9).
`Fractions of 4 ml were collected. 0, A230; 0, 100-141 aliquots were used
`to measure radioactivity of 35S-labeled bovine B(SSO3‘); A, 100-ul
`aliquots were assayed for B-chain radioimmune activity by being
`mixed with 100 pg of porcine A(SS03') and using the reconstitution
`assay (Fig. 3).
`
`Expression. The strains that contain the insulin genes cor-
`rectly attached to [3-galactosidase (D1210/pIBl and 294/pIA1)
`both produce large quantities of a protein the size of 5-galac-
`tosidase (Fig. 2). Approximately 20% of the total cellular protein
`was this 13-galactosidase—insulin A or B chain hybrid. The hybrid
`proteins are insoluble and were found in the first low-speed
`pellet in which they constitute ~50% of the protein (Fig. 2).
`To detect the expression of the insulin A and B chains, we
`
`Table 1. Amino acid composition of E. coli insulin A chain
`Amino
`Residues er
`e tide
`acid
`E. coli A(SSO3')
`Porcine A(SSO3‘)
`His
`0.08
`0.08
`Lys
`0.00
`0.00
`Trp
`0.00
`0.00
`Arg
`0.00
`0.00
`Phe
`0.00
`0.00
`Asx
`2.38
`2.50
`Th!‘
`0.24
`0.28
`Ser
`0.14
`0.23
`H-Ser
`0.02
`0.00
`Glx
`3.97
`4.58
`Pro
`0.00
`0.09
`Gly
`1.40
`1.48
`Ala
`0.20
`0.11
`Cys
`0.55
`0.00
`Val
`1.15
`1.06
`Met
`0.62
`0.43
`Ile
`1.99
`1.48
`Leu
`2.33
`2.35
`Tyr
`1.89
`2.30
`
`Predicted
`0
`0
`0
`0
`0
`2
`1
`2
`0
`4
`0
`1
`0
`0
`1
`0
`2
`2
`2
`
`Approximately 25 ug of porcine A(SSO3‘) (which is identical in
`sequence to human A) and 25 ug of E. coli A(SSO3‘) purified twice
`by high-performance liquid chromatography were hydrolyzed and
`analyzed in parallel. The SS03" derivatives of cysteine were destroyed
`during hydrolysis and do not register as amino acids with the program
`used. Serine and threonine were also partially destroyed.
`
`Page 4
`
`
`
`110
`
`Biochemistry: Goeddel et al.
`
`Proc. Natl. Acad. Sci. USA 76 (1979)
`
`is high (approximately 10 mgfrom 24 g wet weight of cells) and
`consistent with the amount of insoluble B-galactosidase protein
`obtained (at least 105 molecules per cell). This estimated yield
`is 10 times higher than that reported for somatostatin (1).
`The evidence that we have obtained correct expression from
`chemically synthesized genes for human insulin can be sum-
`marized as follows. (i) Radioimmune activity has been detected
`for both chains. (ii) The DNA sequences obtained after cloning
`and plasmid construction have been directly verified to be
`correct as designed. Because radioimmune activity is obtained,
`translation must be in phase. Therefore, the genetic code dic-
`tates that peptides with the sequences of human insulin are
`being produced. (iii) The E. C0ll products, after cyanogen
`bromide cleavage, behave as insulin chains in three different
`chromatographic systems that separate on different principles
`(gel filtration, ion exchange, and reversed-phase high-perfor-
`mance liquid chromatography). (to) The A chain produced by
`E. colt‘ has been purified on a small scale by high-performance
`liquid chromatography and has the correct amino acid com-
`position (Table 1).
`Table 2 illustrates that insulin radioimmune activity can be
`obtained entirely from E. coli products. Easily detectable ra-
`dioimmune insulin activity is produced when purified E. coli
`A chain is mixed and reconstituted with partially purified
`(~10% pure) E. coli B chain.
`
`We gratefully acknowledge the excellent technical assistance of
`Louise Shively, Rochelle Sailor, Frances Fields, and Mark Backer. We
`give special thanks to Herbert W. Boyer for his encouragement and
`scientific consultation and to Robert A. Swanson for making this work
`possible. Work at the City of Hope was supported by contracts from
`Genentech, Inc.
`
`1.
`
`Itakura, K., Hirose, T., Crea, R., Riggs, A. D., Heyneker, H. L.,
`Bolivar, F. & Boyer, H. w. (1977) Science 196, 1056-1063.
`2. Humbel, R. E., Bosshard, H. R. & Zahn, H. (1972) in Handbook
`of Physiology, eds. Steiner, D. F. & Freinkel, N. (Williams &
`Wilkins, Baltimore, MD), Vol. 1, pp. 111-132.
`8. Katsoyannis, P. G., Trakatellis, A. C., Johnson, S., Zalut, C. 81
`Schwartz, G. (1967) Biochemistry 6, 2642-2655.
`4. Crea, R., Hirose, T., Kraszewski, A. & Itakura, K. (1978) Proc.
`Natl. Acad. Sci. USA 75, 5765-5769.
`5. Backman, K., Ptashne, M. 61 Gilbert, W. (1976) Proc. Natl. Acad.
`Sci. USA 73, 4174-4178.
`6. Panet, A., van de Sande, I. H., Loewen, P. C., Khorana, H. G.,
`Raae, A. I., Lillenhaug, I. R. & Kleppe, K. (1973) Biochemistry
`12, 5045-5049.
`7. Greene, P. J., Betlach, M., Bolivar, F., Heyneker, H. L., Tait, R.
`& Boyer, H. W. (1978) Nucleic Acids Res. 12, 2373—2sso.
`8. Bolivar, F., Rodriguez, R. L., Greene, P. ]., Betlach, M. C.,
`Heyneker, H. L., Boyer, H. W., Crosa, J. H. & Falkow, S. (1977)
`Gene 2, 95-119.
`9. Clewell, D. B. (1972) Bacteriol. 110, 667-676.
`10. Miller, I. H. (1972) Experiments in Molecular Genetics (Cold
`Spring Harbor Laboratory, Cold Spring Harbor, NY).
`11. Maxam, A. M. 81 Gilbert, W. (1977) Proc. Natl. Acad. Sci. USA
`74, 560-564.
`12. Hershfield, V., Boyer, H. W., Yanofsky, C., Lovett, M. A. &
`Helinski, D. B. (1974) Proc. Natl. Acad. Sci. USA 71, 3455-
`3459.
`13. Katsoyannis, P. G., Tometsko, A., Zalut, C., Johnson, S. & Trak-
`atellis, A. C. (1967) Biochemistry 6, 2635-2642.
`Polisky, 13., Bishop, R. J. & Gelfand, D. H. (1976) Proc. Natl. Acad.
`Sci. USA 73, 3900-3904.
`15. O'Farrell, P. H. (1975) J. Biol. Chem. 250, 4007-4021.
`
`14.
`
`Table 2. Reconstitution of radioimmune human insulin
`
`“A" sample
`E. coli 58-HPLC’‘'
`——
`Porcine A3
`E. coli 58-HPLC
`E. coli 58-HPLC
`
`“B” sample
`..
`E. coli DE1177
`E. coli DE117
`Bovine B5
`E. coli DE117
`
`Radioimmune active
`insulin, ng
`<0.5
`<0.5
`74
`45
`20
`
`Our standard reconstitutionassay procedure was used (Fig. 3). The
`results are given as ng of radioimmune active insulin produced per
`20 pl of the reaction mixture. HPLC, high-performance liquid chro-
`matography.
`* Five hundred microliters of fraction 58 from an aminoethyl-cellulose
`column was chromatographed on an RP-8 column and the “A” peak
`was collected. As estimated from the peak height, the sample con-
`tained approximately 25 pg of protein.
`1 Ten microliters of DEAE-cellulose fraction 117 (Fig. 4), concentrated
`to 1.6 mg of total protein per ml, was used as the “B” sample.
`3 S-Sulfonated porcine A (70 pg).
`§ S -Sulfonated bovine B (10 pg).
`
`Insulin B-chain radioimmune activity was detected, first
`among the S -sulfonated cyanogen bromide peptides insoluble
`at pH 5 (fraction F-10, Fig. 3). The activity was enriched fur-
`ther by chromatography on DEAE-cellulose (Fsiig. 4). The B-
`chain radioimmune activity coeluted with S—[ Slsulfonated
`bovine B chain.
`A portion of the material purified by DEAE-cellulose chro-
`matography was analyzed by high-performance liquid chro-
`matography on a reversed-phase RP-8 column (Fig. 5A ). This
`column separates primarily on the basis of hydrophobic inter-
`actions. The insulin B-chain radioimmune activity eluted at a
`position very close to that of bovine B chain. Good purification
`was obtained by high-perfomiance li uid chromatography, but
`the breadth of the peak indicated t at the chromatographic
`fraction was not pure.
`‘
`_ Another sample (1 mg total protein) of the material purified
`by DEAE-cellulose chromatography was subjected to gel fil-
`tration on Sephadex G-75 in 50% acetic acid, a system that
`completely resolves A chain from B chain. The B-chain ra-
`dioimmune activity eluted at the same position as S-sulfonated
`bovine B chain, indicating similar sizes (data not shown).
`Insulin A-chain radioimmune activity was detected first in
`the total mixture of cyanogen bromide peptide fragments ob-
`tained from the partially purified )3-galactosidase-A chain
`hybrid. The activity was enriched by pH 5 precipitation and
`aminoethyl-cellulose chromatography and purified on a mi-
`crogram scale by high-performance liquid chromatography
`(Fig. 5B). The insulin.A-chain radioimmune activity eluted
`from the column at a position indistinguishable from that of
`porcine S-sulfonated A chain. Porcine A chain is identical to
`human A chain (2).
`-
`When an excess of porcine A(SSO3‘) (40 pg) was mixed, re-
`duced, and oxidized with bovine B(SSO3’) (10 pg), we usually
`obtained 10-15% correct joining to yield radioimmune active
`insulin. Reconstitution in impure mixtures was lower, as ex-
`pected. Because of this strong and variable competitive inhi-
`bition by other peptides, the amount of insulin chains in the
`extracts can best be estimated by adding to the extract a known
`amount of the chain to be assayed. This type of experiment
`(illustrated in Fig. 3) indicates that the yield of insulin chains
`
`Page 5