Volume 12 Number 9 1984
`
`Nucleic Acids Research
`
`Assembly of functional antibodies from lmmunoglobuHn heavy and light chains SYDthesised in
`E. coli
`
`Michael A.Boss, John H.Kenten, Clive R.Wood and J.Spencer Emtage
`
`Celltech Ltd, 244- 250 Bath Road, Slough SLI 4DY, Berks, UK
`
`Received 5 March 1984; Revised and Accepted 16 April1984
`
`ABSTRAcr
`Genes for a murine ~ heavy chain and a ~ light chain immunoglobulin
`have been inserted into bacterial expression plasmids containing the
`Escherichia coli trp promoter and ribosome binding site.
`Induction of
`transcription from the trp promoter results in accumulation of both light
`and heavy chain polypeptides in appropriate host strains.
`Both proteins
`were found as insoluble products.
`Following extraction and purification of
`the immunoglobulin containing fractions, antigen binding activity was
`recovered.
`The activity demonstrates essentially the same properties as the
`antibody from the hybridoma from which the genes were cloned.
`
`INTRODUcriON
`
`Immunoglobulin genes and their products represent one of the most
`
`Immunoglobulin
`extensively studied families of eukaryotic macromolecules.
`polypeptides are secreted proteins and are synthesised with an amino-terminal
`
`signal peptide which is cleaved to yield the mature protein.
`
`The expression
`
`of immunoglobulin genes in E. coli forms the initial stage in
`
`the production
`
`of antibodies produced via recombinant DNA techniques.
`
`Such antibodies
`
`would have many uses.
`
`For example, detailed studies on antigen-antibody
`interactions following alterations of the antigen combining site by site
`
`directed mutagenesis could be carried out, or the Fe regions of the molecules
`
`could be altered for specific uses such as binding to matrices for
`
`immunopurification.
`
`Thus, it is surprising that with the many studies on
`
`expression of eukaryotic genes in E. coli (1), little has been done on
`
`So far immunoglobulin genes have been expressed in
`immunoglobulin genes.
`modified forms at low levels in E. coli, usually as incomplete amino-terminal
`fusion proteins (2,3,4).
`In one case, a trpE-IgE fusion has been expressed
`at 10% total E. coli protein (5).
`
`Here we describe the bacterial expression of a murine ~ heavy chain and a
`murine ~ light chain immunoglobulin eDNA.
`The Ig ~ and ~ genes used in
`these studies are from eDNA clones isolated from the hybridomas B1-8 and S43
`
`@ IRL Press Limited, Oxford, England.
`
`3791
`
`BEQ 1033
`Page 1
`
`

`
`Nucleic Acids Research
`
`(6), respectively.
`
`These hybridomas were raised to the hapten 4-hydroxy-3-
`
`nitrophenyl acetyl (NP) and produce IgM antibodies which are termed
`heteroclitic, that is binding a related hapten e.g. 4-hydroxy-5-iodo-3
`nitrophenyl acetyl (NIP) more strongly than NP (7).
`The~ gene was cloned
`from the hybridoma line Bl-8 (8) and the A gene was cloned from the related
`hybridoma S43 (9). However, the sequence of the A from 543 varied by only
`two amino acids from Bl-8 A sequence, assuming that Bl-8 A has germ-line
`sequence (9).
`Both changes were conservative and outside of the
`complementarity determining regions.
`So in effect, the antibody from Bl-8
`
`can be used to represent the parental monoclonal antibody.
`The two polypeptides were synthesised in E. coli as native proteins
`lacking eukaryotic signal sequences and presumably possessing amino(cid:173)
`terminal methionine residues. High levels of expression were achieved
`using the E. coli K12 strain E103S or E. coli B but only low levels of
`expression occurred in HB101 (10).
`Following solubilisation of ~ and A
`polypeptides expressed in the same cell or different cells, the protein
`products were purified and antigen binding activity recovered.
`This
`activity demonstrates essentially the same properties as those found for an
`NP binding IgM hybridoma antibody.
`
`MATERIALS AND METHODS
`Chemicals and Cloning Procedures
`Materials were purchased as follows: restriction enzymes (Bethesda
`Research Laboratories and New England Biolabs), T4 DNA polymerase (P-L
`Biochemicals), DNase I (Sigma), radioisotopes (Amersham), rabbit anti-mouse
`
`IgM (Bionetics), rabbit anti-IgM (Tago), rabbit anti-A (Miles), MOPC 104E an
`IgM (~A1) myeloma protein (Bionetics), calf intestinal alkaline phosphatase
`and 51 nuclease (Boehringer Mannheim).
`Unless otherwise stated cloning
`procedures were as described (10).
`Oligodeoxyribonucleotides were synthesised by the phosphotriester
`procedures (11) and were designed to have the sequences; 5'(cid:173)
`GATCAATGCAGGCTGTTGTG-3'
`(R45), and 5'-ATTCCTGAGTCACAACAGCC-3'
`Bacterial strains and Plasmids
`Plasmids were transformed into E. coli strain HB101, DH1, E. coli B (10)
`and E. coli K12 strain E103S (Dr. Lee Simon, Waksman Institute of
`Microbiology, Piscataway, New Jersey 08854-0759, personal communication),
`and grown in L-broth containing 0.1g carbenicillin per litre.
`Plasmids
`pAB~-11 (8) and pABAl-15 (9) were a gift from Drs. A. Bothwell and D.
`
`(R44).
`
`3792
`
`BEQ 1033
`Page 2
`
`

`
`Nucleic Acids Research
`
`Baltimore.
`
`B1-8 proteins were gifts from Drs. M. Neuberger and T.Imanishi-
`
`Kari.
`Pulse Chase Analysis
`For pulse chase analysis inductions were set up as described above,
`except that the medium used consisted of: proline (0.3g/L), leucine (0.1g/L),
`Difco methionine assay medium (5g/L), glucose (60mg/L), thiamine (10mg/L),
`cac12 (22mg/L), Mgso4 (0.25g/L)m and carbenicillin (0.1g/L). During
`exponential growth cells were pulse labelled with 30 ~Ci/ml L-[ 35s)
`methionine for 2 minutes, after which unlabelled methione (100 ~g/ml) was
`added and the incubation continued for the times indicated.
`Other Methods
`Procedures used for bacterial lysis and fractionation were as described
`(12), as were procedures for inductions and protein assays (13).
`Protein Purification
`For further purification of the ~ light chain, the cell debris were
`dissolved in 10mM Tris-HCl pH8.0, 25% formamide, 7M urea, 1mM EDTA and 2mM
`dithiothreitol. This material was loaded onto a DEAE Sephacel column
`(Pharmacia) (1 x 25cm at a flow rate of 5ml/hr) which had been equilibrated
`in 9M urea, 10mM Tris-HCl pH8.0, 1mM EDTA and 2mM DTT.
`The DEAE Sephacel
`column was developed using a 0-150mM NaCl gradient in loading buffer. The
`eluted peak of ~ light chain immunoreactivity, corresponding to the major
`peak of protein was diluted to a final concentration of 2.25M urea, 10mM
`Tris-HCl pHB.O, 1mM EDTA, 2mM DTT and loaded onto an octyl-Sepharose column
`(Pharmacia) (2.5 x 10cm). Material was eluted by use of a urea gradient of
`2.25-9M urea.
`The peak material was pooled, dialysed into ammonium
`bicarbonate and lyophilised.
`The ~ heavy chain was purified from 9M urea solubilised pellets by
`anion exchange chromatography and chromatofocussing (Pharmacia).
`Reconstitution of Activity
`Production of functional antibodies from E. coli expressing both heavy
`and light chains was achieved by lysing the cells and clarifying the
`The insoluble material was washed, followed
`supernatant by centrifugation.
`by sonication (3 times for 3 minutes), and finally dissolved in 9M urea, 50mM
`This extract was
`glycine-NaOH pH10 .8, 1mM EDTA, and 20mM 2-mercaptoethanol.
`dialysed for 40 hours against 3 changes of 20 vols. of 100mM KC1, 50mM
`glycine-NaOH pH10.8, 5% glycerol, 0.05mM EDTA, 0.5mM reduced glutathione
`The dialysate was cleared by centrifugation
`and 0.1mM oxidised glutathione.
`at 30,000g for 15 minutes and loaded directly onto DEAE Sephacel, followed by
`
`3793
`
`BEQ 1033
`Page 3
`
`

`
`Nucleic Acids Research
`
`development with a 0-0.5M KC1 linear gradient in 10mM Tris-HC1, 0.5mM EDTA,
`
`pH8.0.
`
`The purified Ig ~and A were treated as above, except that no anion
`
`exchange chromatography was carried out.
`
`The preparation was finally
`
`dialysed into phosphate buffered saline, 5% glycerol, 0.01% sodium azide and
`
`0.5mM EDTA pH7.4.
`
`RESULTS
`
`Construction of Plasmids for Expression of A Light Chain
`
`We chose to express the A gene in E. coli by direct expression of the
`
`gene lacking the eukaryotic signal peptide but containing a methionine
`
`initiator residue at the amino-terminus (met lambda) .
`
`The approach used for
`
`bacterial synthesis of met-A was to reconstruct the gene in vitro from
`
`restriction fragments of a eDNA clone and to utilise synthetic DNA fragments
`
`for insertion into the bacterial plasmid pCT54 (12) (figure 1).
`
`As a
`
`source of light chain we used the plasmid pABA1-15 which contains a full(cid:173)
`length A1 light chain eDNA cloned into the Psti site of pBR322 (9).
`have previously outlined the construction of plasmids for the expression of A
`
`We
`
`light chain (14).
`
`A plasmid was isolated (designated pCT54 19-1) and shown to have the
`
`anticipated sequence except that there was an alteration at the fifth codon
`
`from GTG to ATG, changing ti;e amino acid at this point from valine to
`
`methionine (figure 1). Valine is an invariant residue at this position in
`
`mouse A chains. Methionine, however, is the residue most frequently found
`
`in mouse K chains at this position (15).
`
`As most E. coli mRNAs have 6-11 nucleotides between the Shine-Dalgarno
`
`(SD) sequence and the AUG (16) the distance in pCT54 19-1 was reduced by
`
`modification at the Cla1 site. Altering the distance between the SD
`
`sequence and the ATG has been demonstrated to alter the expression of a
`
`number of genes (13,14,17-20) presumably by placing the SD and ATG sequences
`
`in the optimal configuration for formation of the initiation complex. pCT54
`
`19-1 was cut with Cla1 and incubated with S1 nuclease.
`nuclease was adjusted so that some DNA molecules would lose 1-2 extra base
`This DNA on religation with
`
`pairs as a result of 'nibbling' by the enzyme.
`T4 DNA ligase and transformation into E. coli strain HB101 gave rise to a
`
`The amount of S1
`
`number of plasmids which had lost the Cla1 site.
`
`The nucleotide sequence
`
`across the modified region of two of these plasmids was determined (figure 1).
`
`pNP4 and pNP3 were shorter than pCT54 19-1 by 5 and 4 nucleotides
`
`3794
`
`BEQ 1033
`Page 4
`
`

`
`Nucleic Acids Research
`
`\JM' T T
`ID
`
`op
`
`•
`1 -... _
`
`1
`....
`1
`
`1
`
`MliliGTATCGATTGATCA&Ili
`MlillGTA
`TTGATCAAJli
`MIIAGT
`TTGATCA6Di.
`
`Figure 1. Construction of plasmids for the direct synthesis of A light
`chain in E. coli.
`Plasmid pATA1-15 contains the A gene inserted into the Hindiii site of
`pAT153.
`(A) 5' Hinfi - Sac! fragment 1 was isolated by polyacrylamide gel
`electrophoresis.
`The 3' fragment 2 of the gene was isolated as a Sac! -
`Hindiii fragment.
`pCT54 was cut with Bel! + Hindiii and the A gene
`fragments together with oliqodeoxyribonucleotides R45 and R44 ligated to
`yield plasmid pCT54 19-1. (B) Digestion of pCT54 19-1 with Clai and S1
`nuclease produced plasmids pNP3 and pNP4, with reduced SD-ATG distances,
`E, EcoRI; H, Hinfi; H3, Hindiii.
`
`Secondary
`respectively, giving SD-ATG distances of 9 and 10 nucleotides.
`structure analysis, as described (13), revealed no hairpin loop sequestering
`the SD or initiation codon into double-stranded regions of the mRNA of
`either pCT54 19-1 or the S1 derivatives.
`SUch base pairing interactions
`have been shown drastically to reduce translational efficiency of a number of
`genes, notably those for phageA ~ (17), fibroblast and leukocyte
`
`3795
`
`BEQ 1033
`Page 5
`
`

`
`Nucleic Acids Research
`
`-
`
`- 2
`
` 3 4 5 6
`
`8
`
`9 10
`
`Figure 2. Accumulation and distribution of A protein from E. coli E103S.
`E103S cells containing pNP3 were grown under inducing conditions and
`samples taken throughout the induction cycle.
`Following electrophoresis,
`gels were stained with Coomassie blue (lanes 1-7) or subjected to analysis
`by Western blotting (lanes 8-10) using rabbit anti-A serum and iodinated
`protein A (2~Ci/ml).
`Lane 1, purified MOPC104E myeloma protein standard
`indicating the position of authentic A protein; lane 2, E103S containing
`pNP3 after growth to stationary phase in L-broth; lanes 3-5, samples from
`E103S containing pNP3 taken at increasing absorbance during induction; lanes
`6 and 7, respectively, soluble and insoluble fractions from pNP3 containing
`E103S;
`lane 8, unfractionated extract;
`lanes 9 and 10, respectively,
`soluble and insoluble fractions.
`
`interferons (18), SV40 small-t antigen gene (19) and a murine~ heavy
`chain (13).
`Expression of A Protein
`HB101 cells containing pNP3 were found to express A light chain as
`determined by immunoprecipitation (14) and Western blot analyses.
`No such
`protein was evident in extracts from pCT54 19-1, cells containing pNP4
`which had a SD-ATG distance one base pair shorter than that of pNP3
`expressed only a very low level of A light chain (14).
`In the absence of specific immunoprecipitation a novel protein band was
`not visible from extracts of HB101 containing pNP3 nor was the A protein
`
`3796
`
`BEQ 1033
`Page 6
`
`

`
`Nucleic Acids Research
`
`found to accumulate (data not shown).
`
`However, there was a dramatic
`In this strain A
`difference when pNP3 was induced in the K12 strain E103S.
`protein was found to accumulate during induction until the cells reached
`stationary phase (figure 2, lanes 3-5) to a level of about 150 times that
`found in HB101 as determined by an ELISA (enzyme linked immunosorbent assay).
`These cells were found to contain inclusion bodies which appeared refractile
`under light microscopy, a phenomenon characteristic of high level expression
`of foreign proteins (21).
`An estimate of the percentage of total E. coli
`protein represented by recombinant A protein was obtained by separating the
`proteins by gel electrophoresis, staining them with Coomassie blue and
`This method
`scanning the stained gel with a Joyce-Loebl chromoscan 3.
`showed that A was the major protein present (figure 2, lane 5) and
`The A protein had a half-life of
`represented 13% of total E. coli protein.
`20 minutes in HB101 (data not shown) but accumulated to very high levels in
`E103S, suggesting that the lambda protein was much more stable in the latter
`strain. After cell lysis and centrifugation of HB101 or E103S containing
`pNP3, A light chain was detected in the insoluble (figure 2, lanes 7 and 10)
`but not in the soluble fractions (figure 2, lanes 6), as determined by
`The identity of the major Coomassie blue stained
`Coomassie blue staining.
`band as A protein was confirmed by western blot analysis (figure 2, lanes
`The presence of such immunoreactive bands was specific to pNP3
`8-10).
`containing cells. When extracts from cells containing pCT70, a prochymosin
`expressing plasmid (12), were subjected to the same analysis, no bands were
`
`This more sensitive technique showed that a
`detected (data not shown).
`small amount of the A protein was in the soluble fraction (figure 2, lane 9).
`The presence of a number of distinct immunoreactive proteins all smaller
`than full-length A protein were also detected.
`These may result from
`proteolytic degradation of A protein, from premature termination of
`transcription or from internal initiation of translation.
`
`Expression of ~ Protein
`The construction of plasmids pNP11 and pNP14 for the expression of
`
`full-length mature~ protein under the control of the ~promoter has been
`described (13).
`E. coli B cells containing the~ expression plasmid pNP11
`were grown under inducing conditions and soluble and insoluble extracts
`prepared, and analysed by SDS-PAGE.
`A novel band was seen after staining
`the gel with Coomassie blue in the lane containing proteins from the
`insoluble fraction (figure 3, lane 2).
`This band was not seen in the
`negative control lane which contained proteins from the same fraction from
`
`3797
`
`BEQ 1033
`Page 7
`
`

`
`Nucleic Acids Research
`
`-, __
`jJ,... - ---
`
`-
`
`•
`
`1
`
`2
`
`3
`
`4 5
`
`6 7 8 ·
`
`Figure 3. Expression and distribution of ~protein from E. coli B.
`E. coli B cells containing pNP14 were grown under inducing conditions.
`Following electrophoresis, gels were stained with Coomassie blue (lanes 1-3)
`or subjected to analysis by western blotting (lanes 4-8) using rabbit anti(cid:173)
`IgM serum.
`Lane 1, molecular weight markers (94, 67, 43, 30 and 20Kd);
`lanes 2 and 4, respectively, pNP14 containing insoluble fraction; lanes 3 and
`5, pCT70 containing insoluble fraction; lanes 6 and 7, respectively, pNP14 and
`pCT70 containing soluble fractions; lane 8, unfractionated sample from cells
`containing pNP14.
`
`The novel band was found to
`
`cells harbouring pCT70 (figure 3, lane 3).
`migrate to a position corresponding to a protein of a molecular weight
`within less than 5% of the actual molecular weight of non-glycosylated ~ of
`A duplicate set of lanes were transferred to nitrocellulose, and
`62.5Kd.
`Western blotted. Alignment of the stained gel and the blot autoradiogram
`confirms that this novel band is antigenically related to IgM (figure 3,
`No band was found in extracts from cells containing pCT70
`lanes 4 and 8).
`Only a low amount of ~ was found in the soluble
`(figure 3, lanes 5 and 7).
`fraction (figure 3, lane 6).
`A greatly increased level of expression of ~ was found in E. coli B
`
`PUlse chase analysis demonstrated that in E. coli B,
`compared to HB101.
`a similar level of ~ protein was detected after a 60 minute chase (figure
`4, lane 3) as was seen after the initial labelling period (figure 4, lane
`1) .
`In HB101, however, very little~ protein could be seen after a 10
`
`3798
`
`BEQ 1033
`Page 8
`
`

`
`Nucleic Acids Research
`
`1 2 3 4 5 6 7 8
`
`-
`
`94-
`
`67-
`mu~
`
`-
`
`43-
`
`30-
`
`Figure 4. Pulse chase autoradiogram of pNP14 in E. coli Band HB101.
`E. coli B and HB101 cells harbouring pNP14 were grown under inducing
`conditions and pulsed with L-[35s]-methionine for 2 mins. after which
`unlabelled L-methionine was added to 100 ~g/ml (zero time). Cells were
`harvested at varying times, and samples analysed by SDS-PAGE following
`immunoprecipitation.
`E. coli B: lane 1, zero time; lane 2, 30 min; lane 3,
`60 min.
`Lane 4 is a zero time sample for pCT70 in HB101.
`HB101: lane 5,
`zero time; lane 6, 5 min; lane 7, 10 min; lane 8, 30 min.
`
`minute chase (figure 4, lane 7), and none after 30 minutes (figure 4, lane
`
`8) , compared to the amount detected after the initial labelling period
`
`(figure 4, lane 5).
`
`Induced E. coli B cells harbouring pNP14 when
`
`examined by phase contrast microscopy were found to contain inclusion bodies.
`
`Purification of Recombinant
`The presence of A light chain in the insoluble fraction was a useful
`purification step since it both concentrated the protein and separated it
`
`from the bulk of E. coli soluble proteins.
`
`For further purification the E. coli insoluble material was solubilised
`in a Tris-HCl buffer containing 25% formamide and 7M urea and loaded onto a
`
`DEAE Sephacel column which had been preequilibrated in the same buffer with
`
`9M urea, but without formamide.
`
`0-150mM NaCl gradient in the urea buffer.
`
`The bound material was eluted using a
`The A protein was the major
`The A protein
`eluted peak as determined by gel electrophoresis and ELISA.
`was diluted to a final concentration of 2.25 urea and loaded onto an octyl-
`
`3799
`
`BEQ 1033
`Page 9
`
`

`
`Nucleic Acids Research
`
`Sepharose column.
`
`The A protein bound and was eluted by using a 2.25-9M
`
`Following this step, only a single band of Coomassie blue
`urea gradient.
`stainable material corresponding to recombinant A protein was visualised by
`
`SDS-PAGE (data not shown).
`Expression of ~ and A Polypeptides in the same cell
`Each of the Ig ~ and A genes in expression plasmids were transformed
`into the same E. coli cell to direct the synthesis of both Ig ~ and A
`polypeptides.
`In order to overcome plasmid incompatibility and provide a
`
`second antibiotic resistance marker, the~ promoter and A sequences were
`excised from pNP3 on a Hindiii-Bam HI fragment and inserted into the
`Hindiii-Bam HI fragment of pACYC184 (10) to create pACYCA.
`HB101 cells
`This weak growth
`containing this plasmid were found to grow very poorly.
`was thought to result from read through of RNA polymerase into the origin of
`replication, and inhibition of growth was virtually eliminated, by inserting
`the bacteriophage T7 early transcriptional terminator (22) at the Hindiii
`
`The resultant plasmid pACA T7-1 has a chloramphenicol
`site of pACYCA.
`resistance gene and an origin compatible with the pBR322-derived origin
`on pNP14, the Ig ~ expressing plasmid.
`Transformation of both plasmids
`into the same E. coli B cell was achieved in two steps, firstly pNP14 was
`introduced, and then pACA T7-1, in two separate transformations to give
`
`ampicillin and chloramphenicol resistant clones.
`E. coli B cells derived from double-transformant clones showed the
`presence of inclusion bodies and two novel polypeptide bands on stained gels
`of the insoluble fraction after lysis.
`These two bands correlated both
`with immunological activity by Western blotting for Ig ~ and A and their
`expected molecular weights of 63,500 and 25,000 daltons respectively (data
`
`Stability of the plasmids was also investigated and it was
`not shown) o
`found that after 36 hours in shake flasks only 5% of the E. coli contained
`both antibiotic resistance markers, although 35% were carbenicillin
`resistance and 74% were chloramphenicol resistant.
`This illustrates the
`selective pressure against both plasmids together.
`Reconstitution of Antigen Binding
`It was of great interest to determine whether the concomitant
`In
`expression of ~ and A would lead to the formation of functional IgM.
`order to determine this extracts were made from E. coli containing both
`Ig ~ and A polypeptides and these tested for antigen binding.
`we used a
`two-site sandwich ELISA which detects ~ chain binding to haptenalated
`This assay demonstrates
`bovine serum albumin (NIP-caproate-BSA).
`
`3800
`
`BEQ 1033
`Page 10
`
`

`
`Cl-2
`
`10
`
`Nucleic Acids Research
`
`Figure 5. Purification of reconstituted antibodies.
`Fractions from DEAE sephacel anion exchange chromatography of E. coli B
`expressed Ig ~ and A. Analysis shows level of Ig ~ expressed as Bl-8 IgM
`equivalents; NIP-cap-BSA and BSA binding activities from ELISA's; total
`protein determined by A280nm readings (1.5 A280~mg/ml).
`
`The extracts were prepared as soluble
`sensitivity to 60 pg of Bl-8 IgM.
`The insoluble material was solubilized in the
`and insoluble material.
`same buffer used in lysis but containing 8M urea followed by its dilution
`
`However, no antigen binding activity was detected.
`for assay.
`In order to obtain activity for the Ig ~and A, extracts were made of
`the insoluble fraction and these dialysed into buffer conditions in which
`
`The results from
`disulphide interchange will occur at a higher frequency.
`assays of material processed in this way indicated that some activity was
`The level of activity obtained in this way was too low to do
`obtained.
`any detailed studies on, so the resultant dialysate was purified by anion
`This process resulted in the isolation
`exchange chromatography (figure 5).
`of significant NIP-cap-BSA binding activity over that of background binding
`to BSA (figure 5).
`The assay of the fractions for the level of Ig ~,
`expressed as Bl-8 IgM equivalents demonstrated two peaks of activity. This
`was not found to correlate with full length Ig ~ by western blotting (data
`The first peak observed may represent a fragment of Ig ~.
`not shown).
`The separation of NIP-cap-BSA binding activity from the majority of full
`length Ig ~ and protein indicates that the hapten binding activity is
`contained within a particular molecular species formed at low efficiency.
`The processing of insoluble material obtained from Ig ~ expression in
`E. coli produced a similar IgM protein profile but without NIP-cap-BSA
`This demonstrates that the activity recovered was a
`binding activity.
`property of the combined immunoglobulin expression, not of some E. coli
`factor, or of the Ig ~heavy chain alone.
`
`3801
`
`BEQ 1033
`Page 11
`
`

`
`Nucleic Acids Research
`
`0·3
`
`0·2
`
`0·1
`
`A540nm
`in ELISA
`
`•t
`
`Figure 6. Specific hapten binding of reconstituted antibodies.
`NIP-cap-BSA binding activity from fraction 26 (A), purified Ig >t and A
`( 0) and B1-8 ( 0). Binding in the presence of the 301-!M NIP-cap. ( •,•,•
`respectively).
`
`Further studies of the characteristics of the hapten (NIP-cap-BSA)
`
`The reduction of hapten binding activity on
`binding were carried out.
`dilution of renatured >t and A expressed and purified together was less than
`that found for either B1-8 IgM or purified >t and A expressed separately
`
`The dilution curves for B1-8 antibody and separately expressed
`(figure 6).
`>1 and A were virtually identical.
`the binding activity in both undiluted and diluted samples. Using B1-8
`antibody as a standard for both IgM and hapten binding, the specific
`4
`activity of the assembled antibody was calculated to be 1.4 x 10 gm/gm of
`
`Free hapten was found to inhibit most of
`
`IgM equivalents.
`
`This value demonstrates the inefficient recovery of
`
`activity, but possibly represents an underestimate of the specific activity
`
`due to an overestimate of full-length Ig >t in these fractions, as described
`
`above.
`Heteroclitic Nature of Recombined Antibody
`
`Detailed specificity of binding to NIP-cap-BSA was investigated by
`comparing the assembled antibodies with B1-8 IgM in the presence of free
`
`NIP-cap and NP-cap (figure 7).
`
`Both B1-8 IgM and the assembled antibodies
`
`showed that higherNP-cap than NIP-cap concentrations were required to
`
`inhibit NIP-cap-BSA binding.
`
`3802
`
`BEQ 1033
`Page 12
`
`

`
`100
`
`.......
`
`.......
`'
`
`•
`'
`' ' '
`'
`'
`'
`'
`
`"'
`'
`'
`'·.........
`
`.......
`
`Nucleic Acids Research
`
`........
`
`• ,.7
`
`tO ..
`- - (M )
`
`Figure 7. Heteroclitic binding of reconstituted antibody.
`Binding of antibodies to NIP-cap-BSA; B1-8 IgM ( •), fraction 26 ( •),
`purified Ig 1.1. and A ( •), in the presence of free NIP-cap (---~ or NP-cap (-).
`
`The heteroclitic nature is demonstrated by the molar ratio of NIP to
`NP at 50% inhibition.
`The concentrations of NIP and NP at 50% inhibition
`(ISO) were found to be similar for both B1-8 and the assembled antibodies
`(Table 1).
`Also the specificity ratios (NP ISO/NIP ISO) were similar
`(Table 1).
`The specificity ratios for the 1.1. and A expressed together were somewhat
`lower than those found for B1-8 IgM or separately expressed 1.1. and A which
`were both very similar (Table 1). This is due to the identical
`concentrations of NP but the greater concentrations of NIP required to
`inhibit binding of 1.1. and A expressed together compared with B1-8 IgM.
`Although the specificity ratios for B1-8 IgM and separately expressed 1.1. and
`A are very similar, lower concentrations of NIP and NP are required for
`inhibition of separately expressed 1.1. and A compared to B1-8 IgM.
`
`Table 1. Hapten concentration at 50% inhibition (ISO) of binding of
`antibodies to NIP-cap-BSA solid phase.
`
`NIP ISO 1.1.M
`
`NP ISO 1.1.M
`
`0.13 (SD, 0.05)
`B1-8 IgM
`0.34 (SD, 0.09)
`Fraction 26
`0.11 (SD, 0.02)
`Fraction 27 and 28
`Purified 1.1. and A
`0.04
`SD = Standard Deviation.
`
`3.7 (SD, 2.9)
`1.9 (SD, 0.4)
`1.1 (SD, 0.3)
`0.84
`
`NP ISO
`NIP ISO
`
`29
`6
`10
`22
`
`3803
`
`BEQ 1033
`Page 13
`
`

`
`Nucleic Acids Research
`
`DISCUSSION
`Low level expression of A and ~ polypeptides was demonstrated in
`
`A greater level of expression of A was found in strains
`E. coli HB101.
`E103S and E. coli B (J. Schoemaker, personal communication) and represented
`
`A higher ~ concentration was also found in
`13% total E. coli protein.
`E. coli B than HB101 cells containing pNP14 and was equivalent to 1% total
`E. coli protein.
`These differences in steady-state levels of the ~ and A
`proteins produced is most likely to result from different levels of protein
`
`stability. A protein had a half life of 20 minutes in HB101 but was stable
`and accumulated in E103s.
`Similarly the pulse chase data shows that ~ is
`more stable in E. coli B than HB101.
`This increased stability in E. coli B
`may be explained by the formation of inclusions, perhaps compartmentalizing ~
`away from proteases.
`E, coli B is known to be deficient in lon protease
`(23,24), so that the absence of this protease either acting upon~. or being
`responsible for activating other proteases that act on ~. may be the major
`factor resulting in accumulation in E. coli B.
`
`Both ~ and A proteins were expressed on compatible plasmids in the same
`cells. This expression of two different polypeptides necessary for the
`formation of a higher eukaryotic multi-subunit protein in E. coli represents
`the first of its kind.
`Thus it was of great interest to see if this
`concomitant expression of Ig ~ and A would lead to the formation of complete
`and functional IgM.
`No functional antibody was found following
`solubilisation of insoluble material.
`This indicated that the Ig ~ and A
`were not covalently interacting in the insoluble material in a specific way
`and that the inclusions represented non-active protein.
`The lack of
`activity correlated with insignificant amounts of Ig ~ or A in the soluble
`fraction,
`However, only a low percentage of cells were found to contain
`both plasmids after 36 hours induction.
`Thus the failure to find active
`(and presumably soluble) antibodies in vivo might simply be a reflection of
`the possibility that both polypeptides were not expressed at high levels in
`the same cells, and only a low percentage of cells contained both plasmids
`anyway.
`Functional antibody activity, as defined by antigen binding, was
`obtained following dialysis of extracts in conditions in which disulphide
`interchange occurs.
`Significant antigen binding activity was recovered,
`which was abolished by competition with free hapten in a binding assay.
`Serial dilution of samples revealed a similar gradient in the reduction of
`binding compared to the monoclonal B1-8.
`The bacterially synthesised
`
`3804
`
`BEQ 1033
`Page 14
`
`

`
`Nucleic Acids Research
`
`antibody also demonstrated higher affinity binding to the related hapten
`
`NIP in a manner similar to that of the B1-8 antibody.
`
`Recovery of
`
`functional activity was not dependent on coexpression of the ~ and A
`polypeptide chains since results from purified ~ and A expressed from
`
`different cells showed similar results.
`
`Functional activity was, however,
`
`dependent on association between heavy and light chain polypeptides, since
`
`folding the ~ heavy chain in the absence of A light chain produced no
`
`antigen binding molecules.
`
`Although the overall efficiency of production of functional antibodies
`
`was low, those molecules which did assemble, demonstrated activity in a way
`
`very similar to the hybridoma synthesised molecules.
`
`It is unclear why
`
`the efficiency of assembly was low, but was possibly due to the large
`
`The
`number of disulphide bridges needed to assemble an IgM molecule.
`heavy and light chain polypeptides were completely unfolded before assembly
`
`and so the intra-domain disulphide bridges as well as the inter-chain
`disulphide bridges need to be made correctly. Classically, when
`
`immunoglobulin heavy and light chains are separated and reassembled, only
`
`the inter-chain disulphide bridges are reduced and these molecules refold
`
`with high efficiency (25,26).
`
`The conditions used in our experiments for
`
`the formation of disulphide bridges were probably adequate, since the
`
`bacterially synthesised A light chain when folded under such conditions led
`
`to the appearance of a discrete lower molecular weight species, as expected
`for accurately folded and oxidised A light chain, which comigrated with
`
`oxidised A protein from B1-8 (data not shown).
`
`These results indicate that non-glycosylated immunoglobulin molecules
`
`produced in E. coli can be used for studies concerning antigen-binding.
`
`It is now possible to investigate the effects on antibody-antigen
`
`interactions following mutation of specific amino-acids of the antibody by
`site-directed mutagenesis and expression of the polypeptides in E. coli.
`
`ACKNOWLEDGEMENTS
`
`We thank Drs. A. Bothwell and D. Baltimore for giving us plasmids
`pABA1-15 and pAB~-11; Drs. M. Neuberger and T. Imanishi-Kari for gifts of
`B1-8 protein; and many colleagues at Celltech who have given us advice
`during the course of this work.
`
`REFERENCES
`1. Harris, T.J.R. (1983) in Genetic Engineering, Williamson, R. Ed.,
`Vol. 4, pp.127-183, Academic Press, London.
`
`3805
`
`BEQ 1033
`Page 15
`
`

`
`Nucleic Acids Research
`
`(1978) Eur. J. Immunol. 8,
`
`2. Amster, 0., Salomon, D., Zemel, 0., Zamier, A., Zeelon, E.P., Zantor,
`F. and Schechter, I. (1980) Nucl. Acids Res. 8, 2055-2065.
`3. Kemp, D.J. and Cowman, A.F. (1981) Proc. Natl. Acad. Sci. USA 78,
`4520-4524.
`4. Kurokaw