ADONIS
Immunology 1991 72 471-480
00192805910008OU
Blot-sequencing of antibodies: application to analysis of V gene usage among anti-steroid monoclonal antibodies M. J. TAUSSIG, P. R. SYMINGTON, M. WOODS, M. J. SIMS, D. BEALE, A. S. HUMPHREYS, A. J. NORTHROP,* P. J. BARKER,* N. S. HUSKISSON,* J. COLEYt J. PIRONt & M. GANIt Department of Immunology and *Microchemical Facility, AFRC Institute of Animal Physiology and Genetics Research, Babraham, Cambridge and tUnilever Research Laboratories, Colworth, Bedford
Acceptedfor publication 28 December 1990
SUMMARY Automated gas-phase protein sequencing has been used to characterize variable regions of antibody heavy and light chains separated by SDS-polyacrylamide gel electrophoresis (PAGE) and electroblotted onto Immobilon polyvinylidene difluoride membranes ('blot-sequencing'). Starting from 100 yg of antibody, 20 or more residues of N-terminal VH and VL sequences can regularly be obtained, which is often sufficient to assign the V region to a known family or subgroup. We have applied the blot-sequencing method to analysis of VH and VL usage among a panel of monoclonal anti-steroid antibodies, namely anti-progesterone, anti-pregnanediol, anti-estrone and antitestosterone. The results demonstrate restricted, repetitive usage of VL subgroups and VH families related to anti-steroid specificities. VL regions of the VK1 group were particularly associated with anti-progesterone, VK21 with anti-estrone, and VK8 and VK9 with anti-pregnanediol. VH regions of anti-progesterone antibodies were all derived from the VHVGAM3.8 family; anti-estrone and antipregnanediol antibodies were derived from the VH7183 and VH36-60 families. The latter two families appear to characterize antibodies raised against steroids conjugated to proteins via a sugar bridge. Differences in VH/VL combination were associated with diversity of antibody specificity. In order to extend the sequence data obtained by this technique and confirm family assignments, we have shown that internal V-region sequences can be obtained by limited chemical cleavage of whole antibody with cyanogen bromide, followed by separation of individual fragments by SDS-PAGE and blotsequencing.
Different methods are available for identification of VH and VL segments in expressed antibodies. They include slot or Northern blotting of mRNA with appropriate family- or subgroup-specific DNA probes; nucleotide sequencing of mRNA, cDNA or genomic clones; and amino acid sequencing of isolated heavy and light chains. We have adopted a method of microprotein sequencing which, in a relatively short time (48-72 h from protein to VH and VL sequences), can provide sufficient N-terminal sequence data to assign the V regions to particular families or subgroups. Monoclonal antibody heavy and light chains are separated by SDS-PAGE, followed by electroblotting onto a polyvinylidene difluoride (PVDF, 'Immobilon') membrane, which withstands the reagents required for subsequent amino-terminal protein sequencing;'°0" bands cut from the membrane are analysed directly by automated gas phase sequencing ('blot-sequencing')." This method is particularly useful for V region assignment and comparison, assessment of antibody homogeneity, identification of individual antibodies and sequence confirmation. In this report, we describe the use of N-terminal blot-sequencing to characterize the VH and VL regions of monoclonal antibodies in a panel of anti-steroid
INTRODUCTION Antibody diversity is the result of several levels of genetic variation: combinations of VL and JL segments for the light chain and VH, D and JH for the heavy chain; junctional and N region variation; combinations of complete VH and VL domains; and somatic mutation of rearranged V regions during response maturation." 2 There is an extensive germline V segment repertoire for light and heavy chains. The K light chain repertoire in the mouse consists of about 100-300 VK gene segments which can be divided into 24 subgroups based on sequence relatedness.3 For mouse heavy chains, 11 VH families have been identified, ranging in size from two segments (VHX24) to over 50 (VHJ558).4 Restricted VH/VL gene usage occurs in some responses, especially where the antigen is a haptenF-7 while in others, V genes of several families are used. Recently there has been much interest in selective VH gene usage in autoimmunity' and the idiotypic network.9 Correspondence: Dr M. J. Taussig, AFRC Institute of Animal Physiology and Genetics Research, Babraham, Cambridge CB2 4AT, U.K.
471
M. J. Taussig et al.
472
antibodies of diverse specificities (progesterone hemisuccinate, testosterone hemisuccinate, estrone glucuronide and pregnanediol glucuronide). We have also extended the technique to determination of internal V region sequences after limited chemical cleavage with cyanogen bromide (CNBr) followed by separation of polypeptide fragments by SDS-PAGE and blotsequencing. MATERIALS AND METHODS
Steroids and steroid-protein conjugates Steroids were obtained from Sigma, Poole, Dorset, U.K. The conjugates used in immunization were: progesterone coupled at the I a-position via a succinyl ester bridge to bovine serum albumin (BSA) (progesterone-i La-hemisuccinyl-BSA or progesterone-BSA); testosterone coupled similarly at the 1 7fposition; and estrone and pregnanediol coupled via a glucuronide bridge at the 3-position, also to BSA (estrone-3-glucuronide-BSA, pregnanediol-3-glucuronide-BSA). Progesterone was conjugated to BSA and ovalbumin (OVA) as previously described.'2 The 1 a-hemisuccinyl derivative of progesterone (progesterone hemisuccinate) was prepared by reaction between the 11a-hydroxy steroid and succinic anhydride, and progesterone-lII c-succinyl protein conjugates were prepared by mixed anhydride reaction.'3"14 The molar coupling ratios were 10: 1 (progesterone: BSA) and 17: 1 (progesterone:OVA). Testosterone was conjugated similarly at the i7fl-position to BSA and OVA using testosterone 1 7#-hemisuccinate. Estrone and pregnanediol were conjugated to BSA via a glucuronide residue at the 3-position, using an adaptation of the mixed anhydride method.'3"15"16 Steroids were similarly conjugated to alkaline phosphatase. To steroid 3-glucuronides or progesterone I ol-hemisuccinate dissolved in dimethyl formamide (I mg/ml) were added 2 molar equivalents of tri-n-butylamine (5 pl/ml in dimethyl formamide) and the mixture cooled to 100 for 10 min. One molar equivalent of isobutyl chloroformate (5 pl/ml in dimethyl formamide) was added and the mixture left for 35 min at 100. Three hundred and fifty microlitres were then transferred to a solution of alkaline phosphatase (0-5 mg) in 1 ml 0-5% sodium bicarbonate and left for 2 hr at 4°. Unreacted steroid was removed by dialysis against 0-1 M phosphate buffer, pH 7-4, containing Norit A charcoal. After dialysis the conjugate was added to 5 ml 0-05 M Tris buffer, pH 8, containing 5% OVA, 0-01 M MgCl2 and 0-05% NaN3.
Monoclonal antibodies Mouse monoclonal anti-steroid antibodies (mAb) were prepared by repeated immunization of BALB/c mice with steroidBSA conjugates, followed by fusion of their spleen cells with the non-secreting mouse myeloma line NSO or hybridoma line SP2/ 0.1' Hybridomas secreting anti-progesterone and anti-testosterone antibodies (DB3, 10/8, 11/32, 11/64, AN3/40) were detected by radioimmunoassay.'2 The remaining antibodies to estrone, pregnanediol and progesterone were screened by enzyme-linked immunoassay (below). Hybridomas were cloned by limiting dilution and grown as ascites or in serum-free medium in vitro. Anti-progesterone IgG antibodies were purified by precipitation at 18% Na2SO4 followed by ion exchange chromatography on DE-32 cellulose; other anti-steroids were purified by chromatography on protein A-Sepharose.'8 Purity
of Ig was confirmed by immunoelectrophoresis, isoelectric focusing and SDS-PAGE (below). All antibodies were characterized as IgG, with the exception of anti-progesterone 1/64, an IgM.
Antibody assay and determination of specificity Radioimmunoassay for anti-steroid detection. This was as described elsewhere,'2 using progesterone-OVA or testosteroneOVA bound to wells of microtitre plates and developed with 25Ilabelled sheep anti-mouse Ig (Amersham International, Amersham, Bucks, U.K.). Enzyme-linked immunoassay for anti-steroid detection. Polystyrene 96-well microtitre plates were coated overnight at 4° with rabbit anti-mouse Ig (5 ug/ml in 0-1 M sodium bicarbonate, pH 9 8), followed by washing in phosphate-buffered saline (PBS) containing 0-15% Tween 20 and 0-01% sodium azide (PBSTA). One-hundred microlitres of each clone supernatant were added per well, incubated for 1 hr at 370 and the wells washed again. One-hundred microlitres of a steroid-alkaline phosphatase conjugate were added and incubated for I hr at 37°. The plates were washed again, 200 PI substrate solution (I mg/ ml paranitrophenyl phosphate in I M diethanolamine, pH 9-8, containing I mM MgCl2) added per well and the OD measured at 405 nm after 30 min incubation. Specificity. Cross-reactivity of antibodies with related steroids was determined in a competition assay.'9 Aliquots (200 PI) ofantibody suitably diluted in PBSTA were incubated for I hr at room temperature with nylon pegs presensitized with rabbit anti-mouse Ig.20 After washing in PBSTA, the pegs were placed in microtitre wells containing 100 PI steroid-alkaline phosphatase conjugate and 100 P1 of a dilution of steroid, and incubated for 1 hr. After rewashing, the pegs were incubated with 200 Ml substrate solution for I hr and the optical densities read at 405 nm after removing the pegs. Calibration graphs with homologous steroid were constructed by plotting log steroid concentration against percentage inhibition of binding of steroid-enzyme conjugate to antibody. Cross-reaction with other steroids is defined as 100X/ Y%, where X is the mass of homologous steroid and Y is the mass of heterologous steroid required to produce 50% inhibition.
Cyanogen bromide digestion of antibodies2' Monoclonal antibodies at a concentration of 3-4 mg/ml were dialysed against 0-1 M ammonium bicarbonate, divided into 2mg aliquots and lyophilized. An individual 2-mg aliquot was made up to 400 P1 in 70% formic acid and a single crystal of CNBr added; the reaction vial was flushed with nitrogen, sealed and left overnight at room temperature in the dark. The reaction mixture was then diluted to 4 ml with water and divided into four 500 pg portions which were lyophilized separately; each was then resuspended in a few drops of water and relyophilized. Individual 500 pg portions were taken up in SDS sample buffer and run on SDS-PAGE (17% gel) for analysis and blot sequencing as below.
Blot-sequencing
SDS-polyacrylamide gel electrophoresis (SDS-PAGE). Polyacrylamide slab gels, 10-17% according to the proteins to be separated, were run in Tris/glycine/SDS buffer (pH 8 3) using the discontinuous system of Laemmli.22 All samples were prepared fully reduced in sample buffer, the final mixture
473
V gene usage Table 1. N-terminal sequences of anti-steroid light chains
VL group
Specificity
Progesteronehemisuccinate
Testosterone hemisuccinate
Estroneglucuronide
Pregnanediol glucuronide
mAb
1
11/32 10/8 DB3 11/64 2528:1
D D D D D
AN3/40
D I V L T Q S P A S L A V S L G Q R A T
2412:19
2433:16 2278:7 2323:18 2633:20 2923:22
D D D D D D
2380:18 2555:6 2384:5
D I V M S Q S P S ? L A V S P G E ? V T D I VM S Q S P S S L A V S V G E ? T T D I Q M T Q S S S ? L S I ? L G G K V T
10
Potter
Kabat
VVMTQTP L S L P V S L G D Q A S I S
VKI
MT Q T P L S L P V S L G D Q A S I S
VK1 VK1 VK1 VK1
KII KII KII
V V D V
V I I I I I
V V V V
V V V V V V
20
M T Q I P L S LPVNLGD Q A S I S MT Q T P L S L S V G L G D Q A S ? S
M T Q T P L S L
M L L L L L
T T T T
Q T P L S L P V S L G D Q A S I S
Q Q Q T Q T Q
S S S S S
consisting of 25 p1 stacking gel buffer (0-25 M Tris-HCl, pH 6-8), 20 p1 10% SDS, 10 p1 glycerol, 5 yl 2-mercaptoethanol and 50 pI protein solution; samples were boiled for 2 min before applying to the gel. Dimensions of the resolving gel were 13 cm x 12 cm, with stacking gel of 13 cm x 4 cm; gel thickness was 2 mm and tracks were 1 cm wide. Electrophoresis was performed until the bromophenol blue dye front had migrated a distance of 10-11 cm into the resolving gel. Gels to be used for electroblotting were divided into two parts. Tracks on one side of the gel were used for markers and for small aliquots of the protein samples, the purpose being to confirm the success of the run and to analyse the samples; after electrophoresis these tracks were detached and stained with Coomassie blue. Tracks on the remainder of the gel were used for electroblotting (below); after blotting they were also stained with Coomassie blue to estimate the amount of individual protein bands remaining in the gel. Electroblotting. Blotting onto Immobilon PVDF membrane (Millipore Corp., Bedford, MA) was carried out largely according to the manufacturer's instructions." Immobilon is a hydrophobic, mechanically strong membrane with a high protein adsorption capacity.'0 Before use, Immobilon strips were cut to size (slightly larger than the gel), treated for a few seconds in methanol and washed in water for 5 min before equilibrating in transfer buffer consisting of 50 mm sodium borate, pH 8-0, 20% methanol and 0-02% mercaptoethanol.23 Blotting took place in an LKB 2005 Transphor Electroblotting Unit (LKB, Bromma, Sweden), for 1 hr without cooling at the maximum voltage setting, which generated about 0.5 amps, then for a further hour with water cooling. After blotting, the membranes were stained with 0-05% w/v Coomassie blue for 5 min and destained for a similar time; staining and destaining solutions contained a minimum of 50% methanol. Stained bands of interest were excised for gas-phase sequencing and stored in Eppendorf tubes; there was no apparent deterioration of the protein on the dried Immobilon when stored at 40 for several weeks.
P P P P P
A A A A T
S S S S S
L L L L L
A A ? A A
V V V V V
S ? S S S
L L L L
G Q ? A T I G Q ? A T I G Q G
KII KII
VK21
KIII
VKl
KII KIII KIII KIII KIII KIII
VK21
VK21 VK21 VK21 VK21 VK8
VK8 VK9
KI KI KV
Automated N-terminal protein sequencing. Sequencing was carried out using an Applied Biosystems (Warrington, Cheshire, U.K.) 470A gas-phase protein sequencer equipped with a 120A on-line phenylthiohydantoin (PTH) analyser. Protein bands cut from Immobilon blots were placed directly into the sequencer reaction cartridge and held in place with TFA-etched glass fibre filter discs. RESULTS N-terminal light and heavy chain sequences of anti-steroid antibodies Samples of 100 pg of anti-steroid monoclonal antibodies were reduced with 2-mercaptoethanol and run on 12-5% SDSPAGE. Separated heavy and light chains electroblotted onto Immobilon were stained with Coomassie blue. Stained bands were cut from the blot and subjected to automated gas-phase sequencing. The HPLC profiles (not shown) of PTH amino acids from electroblotted samples were characterized by exceptionally low backgrounds and high repetitive yields. In general, 15-20 cycles of sequence determination were carried out. Light chains. N-terminal light chain sequences of 14 antisteroid monoclonal antibodies are shown in Table 1. For the anti-progesterone group (five antibodies), the sequences are virtually identical and all derive from the VK1 group in the Potter classification3 or KII in the Kabat classification.24 Sequencing of mRNA25 has confirmed the probable common origin of these light chains from the VKJ05 germline gene. No exceptions have been observed in seven sequences, including others sequenced from mRNA (M. J. Sims and M. J. Taussig, unpublished data). Among the monoclonal anti-estrone antibodies, five out of six also have virtually identical N-terminal light chain sequences, but in this case derived from the VK21 group (Potter) or Kabat group KIII. The one exception is antibody 2412:19,
474
M. J. Taussig et al. Table 2. N-terminal sequences of anti-steroid heavy chains
Specificity
1
mAb
10
20
Progesterone hemisuccinate
DB3* 2528:1
QI QL V QS G P E L K K P G E T V K I S VQ S G P E L K K P G E S V K I S
Estroneglucuronide
2412:19 2433:16 2323:18 2278:7 2633:20 2923:22
EV EM EV EV EV EV
2380:18 2384:5
E V K L V E S G G G L V K P G G S L K I S DV Q L Q E S G P G L V K P S Q S L
Pregnanediol glucuronide
QL I L P L I L I L Q L
V V V V V Q
E S G G D L V K P G G S L ? L S G G G L V K P G G S L K L S
A E E E E
S S S S
G G G G
G G L V K P G Q S L G G L V K P G G S L ? L P P G L A K P S Q T
VH family VGAM3.8 VGAM3.8
7183 7183 7183 7183 t 36-60 7183 36-60
* By mRNA sequencing.
t Not assigned. Table 3. Representative N-terminal germline sequences of mouse VH families
Family
Gene
1
10
20
References
VHJ558
H30 H18
Q V QLQQ S G A E L V K T G A S V K M S E V QL QQS G P E L V K P G A S V K I S
26, 27
VHQ52N
PJ14
Q V QL K E S G P G L V A P S QS L S I T
28
VH36-60
1210.7 SB322
E V QL QE S G P S L V K P S QT L S L T D V QL QE S G P S L V K P S QS L S L T
29-31
VHX24 VH7183
441
E V K L L E S G G G L V QP G G S L K L S
81X E4.15 37.1 50.1
E V Q E V M E V K E V K
32, 33 32, 34
VHJ606
14B 22.1
E V K L D E T G G G L V QP G R P M K L S E V K L E E S G G G L V QP G G S M K L S
35
VHS107
VI
E V K L V E S G G G L V QP G G S L R L S E V K L V E S G G G L V QP G A S L R L S
36, 37
V3
V31
QV T L K V S G P G I L QP S Q T L G L A
38
QI Q L V Q S GPELKKP G E T V K I S
25; M. J. Sims and
VH3609 VHVGAM3.8
L L L L
V V V V
E E E E
S S S S
G G G G
G G G G
G G G G
L L L L
V QP V K P V K P V QP
R G G G
E G G G
S S S S
L L L L
K K K K
L L L L
S S S S
M. J. Taussig
(unpublished data) MRL DNA4
VCP
DNA4
EVQLVET G G G L V Q P K G S S K L S EVQLLET G G G L V Q P G G S R G L S
which uses a VK1 gene identical in the first 23 residues to that of anti-progesterone antibodies. The single anti-testosterone antibody sequenced also uses a VK21 light chain. In contrast, neither VK1 nor VK21 light chains occur in the three antipregnanediol antibodies, two of which (2380:18, 2553:20) express VK8 (KI) sequences and the third (2384:5) a VK9 (KV) light chain. Heavy chains. The N-terminal sequences of heavy chains of 10 anti-steroid antibodies are shown in Table 2; for comparison, repesentative N-terminal germline sequences of 11 murine VH families are shown in Table 3. Most of the heavy chains of the
39 40
anti-progesterone antibodies were N-terminally blocked; however, one heavy chain, 2528: 1, could be sequenced by this method as it started at residue 5. (It is not known whether proteolytic cleavage of 2528: 1 occurred during preparation of the sample or whether the molecule as secreted lacks the normal first four residues of VH.) The sequence of the first 21 N-terminal residues of 2528: 1 corresponded to the germline of the VHVGAM3.8 family, with one substitution (serine replacing threonine at position 17). Elsewhere25'4' we have shown by mRNA sequencing that five out of five other anti-progesterone monoclonal antibodies, produced by three different mice, are
V gene coded by a VGAM3.8 VH segment and all start with an Nterminal glutamine, a residue which frequently causes Nterminal blocking by cyclisation to pyroglutamic acid. Among the anti-estrone monoclonals, four out of six heavy chains derive from the VH7183 family. VH genes of the VH7183, VHS107, VHX24 and VHJ606 families are similar but not identical over the first 20 N-terminal residues (Table 3); the antiestrone antibodies 2412:19, 2433:16, 2323:18 and 2278:7 possess a combination of residues characteristic of VH7183, namely valine at position 5, lysine at 13 and lysine at 19 (for 2433:16). In contrast, the VH of anti-estrone 2923:22 is a member of the VH36-60 family. Only 10 residues were determined for 2633:20, from which it was uncertain whether this antibody should be placed in VH7183/VHS107 (valine expected at position 5) or VH36-60 (proline at 9). Some unusual substitutions were seen at position 3 among the anti-estrones, notably isoleucine in three antibodies (2433:16, 2278:7 and 2633:20) and proline in 2323:18. At this position, isoleucine occurs only once in 634 antibodies reported in Kabat et al.24 and in no germline genes so far reported, but is seen three times in six anti-estrones (three mAb from two different mice); similarly, proline at position 3 has only two other reported occurrences out of 634 antibodies.24 Such changes can be attributed to somatic mutation, though the repeated occurrence of isoleucine-3 suggests the possibility of a
475
usage
Table 4. VH and VL combinations used in anti-steroid monoclonal antibodies
VH family
Antibody
Specificity
VL group
Progesterone hemisuccinate
2528:1 DB3 10/8 11/32 11/64
VGAM3.8 VGAM3.8* VGAM3.8* VGAM3.8* VGAM3.8*
VKl VK1 VK1 VK1 VK1
Testosterone hemisuccinate
AN3/40
ND
VK21
Estrone glucuronide
2412: 19 2433:16 2278:7 2323:18 2633:20 2923:22
7183 7183 7183 7183 t 36-60
VK1 VK21 VK21 VK21 VK21
2380:18 2555:6 2384:5
7183 ND 36-60
VK8 VK8 VK9
Pregnanediol glucuronide
VK21
* By mRNA or cDNA sequencing. t Not definitely assigned.
common germline origin.
Of the two anti-pregnanediol antibodies sequenced, one (2380:18) clearly belongs to the VH7183 family, possibly deriving from the germline gene 37.1, while the other (2384: 5) is coded in the VH36-60 family and is identical over the region sequenced to myeloma protein MOPC 315. The VH36-60 genes used by anti-estrone 2923:22 and anti-pregnanediol 2384:5 are probably different (glutamic acid versus aspartic acid at residue 1; threonine versus serine at 17). The VH/VL combinations used in these antibodies are summarized in Table 4.
6945-
36 29 24
-
-
9
-
7
20-
Internal sequences of VH and VL by cyanogen bromide cleavage The N-terminal residues of VH genes are frequently blocked by cyclisation of glutamine to pyroglutamic acid and are not available for the automated Edman degradation. As noted above, this is the case for all the anti-progesterone antibodies with the exception of 2528: 1 (Table 2). Two approaches are available to negotiate this problem. One is to deblock the Nterminus with pyroglutamate aminopeptidase,42 a procedure which has not been successful with anti-progesterones in our hands. The other is to subject the antibody to partial degradation with chemical reagents which produce proteolytic cleavage at relatively infrequent amino acid residues, a familiar example being specific cleavage at methionine with cyanogen bromide.43 We have employed this reagent together with the blot-sequencing technique to confirm that anti-progesterones use VHVGAM3.8 genes. In this VH family, a germline-encoded methionine at residue 48 immediately precedes the CDR2, the sequence of which (GWINTYTGEPTY) is unique to the family (M. J. Sims and M. J. Taussig, unpublished data).25 Five-hundred microgram samples of anti-progesterone antibodies DB3 and 10/8 were treated with CNBr; the resulting fragments of heavy and light chains were separated by 17% SDS-PAGE and blotted onto Immobilon (Fig. 1). Ten distinct
10
1
6 5 g=33535
=
37E~m3 7
KD557a71 6
4
3f n
5
142
..
14 3
_~2 _~ 1
I
Markers
I-
DBL
digest
10/8 digest
Figure 1. Position on SDS gel and Immobilon blot of fragments of the anti-progesterone antibodies DB3 and 10/8 after CNBr degradation. The bands are drawn from the blot and numbered; bands which were shown by sequencing to contain V region fragments are indicated with single diagonals (VL fragment) or cross-hatching (VH fragment). Molecular weights (MW x 10-3) ofmarkers are shown. See also Table 5.
bands were obtained from DB3 after CNBr cleavage and 12 from 10/8, the absence of an intact heavy chain band indicating that digestion was complete in each case. Individual bands were cut out of the blot and sequenced for 6-10 cycles in order to identify the fragments. The results of this analysis are given in Table 5 and the deduced location of methionine residues in each molecule is shown in Fig. 2. All the bands yielded interpretable sequences
M. J. Taussig et al.
476
Table 5. Identification, by blot-sequencing, of fragments obtained after CNBr degradation of anti-progesterone antibodies
Antibody DB3
10/8
Band
MW*
1
5000-6000
2
10,000
3
16,000
4 5 6 7 8
17,000 19,000 20,000 23,000 30,000
1 2
3000 4000
3 4 5 6 7 8 9 10 11
10,000 13,000 16,000 19,000 21,000 23,000 24,000 29,000 32,000
N-terminal sequence of fragment DVVMTQ DTDGS GWINIY ITDFFP VTLG(C)L YKVSN(R) ND YKVSN(R) DVVMTQ VTLG(C)L VTLG(C)L
GWINTY ATYF(C) DTDGS SSTLT ITDF ND ND
DVVMTQ TQTPL TQTPL VTLG(C) DVVMTQ ATYF(C) VTLG(C)
Location in molecule and residuest VL, 1-48 CH3, 427-477 VH, 49-137 CH3, 392-477 CH1, 138-328 VL, 49-175 VL, 49-214 VL, 1-175
CH1-CH3, 138-381 CHI-CH3, 138-426 VH, 49-87
VH-CHl, 88-137 CH3, 427-477 CL, 176-214 CH3, 392-477
VL-CL, 1-175 VL-CL, 5-175 VL-CL, 5-214 CHI-CH3, 138-381 VL-CL, 1-214 VH-CH3, 88-381 VH-CH3, 138-477
* Apparent MW from gel markers. t Positions of methionines in constant region taken from sequence for MOPC 21 (mouse IgG I myeloma), numbering according to Kabat et al.24 Residues in parentheses not identified by sequencer, but assumed from known sequences to be cysteine (C) or arginine (R) as appropriate. ND, not done.
from which the fragments could be identified and assigned to variable or constant regions; the sizes of each identified fragment were in most cases consistent with the VH mRNA sequences of these antibodies25 and the positions of methionines in the y I and K constant regions24. For DB3, sequences of bands 1, 3, 5 and 6 were identified as light chain fragments (all from VL), while bands 1, 2, 3, 7 and 8 contained heavy chain fragments (VH and CH); the presence of mixed sequences in 1, 2 and 3 did not present a problem of interpretation. A sequence in band 2 was identified as the CDR2 of the VHVGAM3.8 family (GWINIY, residues 49-53), the isoleucine at position 52a being a divergence from the germline. This fragment was the only internal VH sequence of DB3 revealed by CNBr cleavage. Similarly, for antibody 10/8, bands 2, 6, 7 and 9 contained light chain fragments (all VL except band 2) and bands, 1, 2, 3, 8, 10 and I I were heavy chain derived (VH and CH); band I contained the characteristic VHVGAM3.8 CDR2 sequence, GWINTY. These results demonstrate that it is possible, by fragmentation and blot sequencing, to confirm the presence of the VHVGAM3.8 segment in antibodies which are N-terminally blocked.
A second 10/8 VH fragment was obtained with the sequence ATYFC (bands 2 and 10), indicating the presence ofmethionine at position 87 in the heavy chain of 10/8; this fragment was not found in DB3. This difference is encoded in the germline of the VHVGAM3.8 family, where three germline genes have threonine-87 and two have methionine-87 (M. J. Sims, U. Krawinkel and M. J. Taussig, manuscript in preparation). Thus, the presence of the appropriate CNBr fragments in the 10/8 digest and their absence from DB3 is evidence that DB3 and 10/8 derive from different VHVGAM3.8 germline genes. For the light chain, one internal VL sequence was obtained from DB3, namely the CDR2 (YKVSNR), which occurred in bands 3 and 5. The presence of this VL fragment, identified as starting at residue 49, was unexpected, as residue 48 had originally been identified as isoleucine rather than methionine by mRNA sequencing25 and is also isoleucine in the putative germline gene Vc105.44 Re-examination of the mRNA sequencing gel showed it to be ambiguous at the third base of codon 48 and that the original reading had probably been erroneous. No corresponding fragment was found in 10/8 light chain, which is otherwise homologous to DB3.
Specificity profiles of anti-steroid antibodies The specificity of anti-estrone and anti-pregnanediol antibodies was studied in detail by competition between steroid-enzyme conjugate and a panel of free steroids (Tables 6 and 7); a similar study of anti-progesterone antibodies has been published previously.45 Among the anti-estrone monoclonals (Table 6), all have greater reactivity with the immunizing ligand, estrone-3-glucuronide, than with estrone itself or estrone-3-sulphate, indicating that the sugar linker group contributes significantly to binding affinity. 2412: 19, the only one to use the VH7183/VK1 combination, is the most specific antibody, cross-reacting significantly only with 17fl-estradiol-3-glucuronide; in particular, it did not bind detectably to estrone-3-sulphate, androsterone or androsterone-glucuronide. In contrast, antibodies in which VH7183 is combined with VK21 all show greater binding reactivity than 2412: 19 against estrone, estrone-3-sulphate, androsterone and androsterone glucuronide and less against 17,B-estradiol-3glucuronide; this may indicate an influence of light chain on specificity. The patterns of cross-reaction of 2433:16 and 2278:7 are virtually identical; these antibodies, which both use the VH7183/VK21 combination, arose in the same fusion and are likely to be clonally related. 2323:18, also VH7183/VK21, shows a similar pattern but greater cross-reaction with androsterone and its derivatives. 2633:20, for which the heavy chain family could not be definitely assigned, has a particularly high cross-reaction for androsterone and androstenedione. 2923: 22, in which a 36-60 VH segment is combined with VK21, has a more restricted specificity than the VH7183/VK21 group; it crossreacts with estrone, estrone-3-sulphate and 17,B-estradiol to about the same extent as the other antibodies, but has lower reactivity against androsterone-glucuronide and several other ligands. In the anti-pregnanediol antibodies (Table 7), binding was also highest with the immunizing ligand (pregnanediol-3glucuronide), about 10 times greater than with pregnanediol itself. Antibodies 2380:18 and 2384:5, coded by different VH and VL families, diverge considerably in detailed specificity. Antibody 2384:5 (VH36-60/VK9) is highly pregnanediol speci-
477
V gene usage DB3 LIGHT
Met: 4
175
48
1
1
,
5
Fragments_
6
.
100
2-1
26 0-06
244 > 100
35
44-5
Immunology 1991 72 471-480
00192805910008OU
Blot-sequencing of antibodies: application to analysis of V gene usage among anti-steroid monoclonal antibodies M. J. TAUSSIG, P. R. SYMINGTON, M. WOODS, M. J. SIMS, D. BEALE, A. S. HUMPHREYS, A. J. NORTHROP,* P. J. BARKER,* N. S. HUSKISSON,* J. COLEYt J. PIRONt & M. GANIt Department of Immunology and *Microchemical Facility, AFRC Institute of Animal Physiology and Genetics Research, Babraham, Cambridge and tUnilever Research Laboratories, Colworth, Bedford
Acceptedfor publication 28 December 1990
SUMMARY Automated gas-phase protein sequencing has been used to characterize variable regions of antibody heavy and light chains separated by SDS-polyacrylamide gel electrophoresis (PAGE) and electroblotted onto Immobilon polyvinylidene difluoride membranes ('blot-sequencing'). Starting from 100 yg of antibody, 20 or more residues of N-terminal VH and VL sequences can regularly be obtained, which is often sufficient to assign the V region to a known family or subgroup. We have applied the blot-sequencing method to analysis of VH and VL usage among a panel of monoclonal anti-steroid antibodies, namely anti-progesterone, anti-pregnanediol, anti-estrone and antitestosterone. The results demonstrate restricted, repetitive usage of VL subgroups and VH families related to anti-steroid specificities. VL regions of the VK1 group were particularly associated with anti-progesterone, VK21 with anti-estrone, and VK8 and VK9 with anti-pregnanediol. VH regions of anti-progesterone antibodies were all derived from the VHVGAM3.8 family; anti-estrone and antipregnanediol antibodies were derived from the VH7183 and VH36-60 families. The latter two families appear to characterize antibodies raised against steroids conjugated to proteins via a sugar bridge. Differences in VH/VL combination were associated with diversity of antibody specificity. In order to extend the sequence data obtained by this technique and confirm family assignments, we have shown that internal V-region sequences can be obtained by limited chemical cleavage of whole antibody with cyanogen bromide, followed by separation of individual fragments by SDS-PAGE and blotsequencing.
Different methods are available for identification of VH and VL segments in expressed antibodies. They include slot or Northern blotting of mRNA with appropriate family- or subgroup-specific DNA probes; nucleotide sequencing of mRNA, cDNA or genomic clones; and amino acid sequencing of isolated heavy and light chains. We have adopted a method of microprotein sequencing which, in a relatively short time (48-72 h from protein to VH and VL sequences), can provide sufficient N-terminal sequence data to assign the V regions to particular families or subgroups. Monoclonal antibody heavy and light chains are separated by SDS-PAGE, followed by electroblotting onto a polyvinylidene difluoride (PVDF, 'Immobilon') membrane, which withstands the reagents required for subsequent amino-terminal protein sequencing;'°0" bands cut from the membrane are analysed directly by automated gas phase sequencing ('blot-sequencing')." This method is particularly useful for V region assignment and comparison, assessment of antibody homogeneity, identification of individual antibodies and sequence confirmation. In this report, we describe the use of N-terminal blot-sequencing to characterize the VH and VL regions of monoclonal antibodies in a panel of anti-steroid
INTRODUCTION Antibody diversity is the result of several levels of genetic variation: combinations of VL and JL segments for the light chain and VH, D and JH for the heavy chain; junctional and N region variation; combinations of complete VH and VL domains; and somatic mutation of rearranged V regions during response maturation." 2 There is an extensive germline V segment repertoire for light and heavy chains. The K light chain repertoire in the mouse consists of about 100-300 VK gene segments which can be divided into 24 subgroups based on sequence relatedness.3 For mouse heavy chains, 11 VH families have been identified, ranging in size from two segments (VHX24) to over 50 (VHJ558).4 Restricted VH/VL gene usage occurs in some responses, especially where the antigen is a haptenF-7 while in others, V genes of several families are used. Recently there has been much interest in selective VH gene usage in autoimmunity' and the idiotypic network.9 Correspondence: Dr M. J. Taussig, AFRC Institute of Animal Physiology and Genetics Research, Babraham, Cambridge CB2 4AT, U.K.
471
M. J. Taussig et al.
472
antibodies of diverse specificities (progesterone hemisuccinate, testosterone hemisuccinate, estrone glucuronide and pregnanediol glucuronide). We have also extended the technique to determination of internal V region sequences after limited chemical cleavage with cyanogen bromide (CNBr) followed by separation of polypeptide fragments by SDS-PAGE and blotsequencing. MATERIALS AND METHODS
Steroids and steroid-protein conjugates Steroids were obtained from Sigma, Poole, Dorset, U.K. The conjugates used in immunization were: progesterone coupled at the I a-position via a succinyl ester bridge to bovine serum albumin (BSA) (progesterone-i La-hemisuccinyl-BSA or progesterone-BSA); testosterone coupled similarly at the 1 7fposition; and estrone and pregnanediol coupled via a glucuronide bridge at the 3-position, also to BSA (estrone-3-glucuronide-BSA, pregnanediol-3-glucuronide-BSA). Progesterone was conjugated to BSA and ovalbumin (OVA) as previously described.'2 The 1 a-hemisuccinyl derivative of progesterone (progesterone hemisuccinate) was prepared by reaction between the 11a-hydroxy steroid and succinic anhydride, and progesterone-lII c-succinyl protein conjugates were prepared by mixed anhydride reaction.'3"14 The molar coupling ratios were 10: 1 (progesterone: BSA) and 17: 1 (progesterone:OVA). Testosterone was conjugated similarly at the i7fl-position to BSA and OVA using testosterone 1 7#-hemisuccinate. Estrone and pregnanediol were conjugated to BSA via a glucuronide residue at the 3-position, using an adaptation of the mixed anhydride method.'3"15"16 Steroids were similarly conjugated to alkaline phosphatase. To steroid 3-glucuronides or progesterone I ol-hemisuccinate dissolved in dimethyl formamide (I mg/ml) were added 2 molar equivalents of tri-n-butylamine (5 pl/ml in dimethyl formamide) and the mixture cooled to 100 for 10 min. One molar equivalent of isobutyl chloroformate (5 pl/ml in dimethyl formamide) was added and the mixture left for 35 min at 100. Three hundred and fifty microlitres were then transferred to a solution of alkaline phosphatase (0-5 mg) in 1 ml 0-5% sodium bicarbonate and left for 2 hr at 4°. Unreacted steroid was removed by dialysis against 0-1 M phosphate buffer, pH 7-4, containing Norit A charcoal. After dialysis the conjugate was added to 5 ml 0-05 M Tris buffer, pH 8, containing 5% OVA, 0-01 M MgCl2 and 0-05% NaN3.
Monoclonal antibodies Mouse monoclonal anti-steroid antibodies (mAb) were prepared by repeated immunization of BALB/c mice with steroidBSA conjugates, followed by fusion of their spleen cells with the non-secreting mouse myeloma line NSO or hybridoma line SP2/ 0.1' Hybridomas secreting anti-progesterone and anti-testosterone antibodies (DB3, 10/8, 11/32, 11/64, AN3/40) were detected by radioimmunoassay.'2 The remaining antibodies to estrone, pregnanediol and progesterone were screened by enzyme-linked immunoassay (below). Hybridomas were cloned by limiting dilution and grown as ascites or in serum-free medium in vitro. Anti-progesterone IgG antibodies were purified by precipitation at 18% Na2SO4 followed by ion exchange chromatography on DE-32 cellulose; other anti-steroids were purified by chromatography on protein A-Sepharose.'8 Purity
of Ig was confirmed by immunoelectrophoresis, isoelectric focusing and SDS-PAGE (below). All antibodies were characterized as IgG, with the exception of anti-progesterone 1/64, an IgM.
Antibody assay and determination of specificity Radioimmunoassay for anti-steroid detection. This was as described elsewhere,'2 using progesterone-OVA or testosteroneOVA bound to wells of microtitre plates and developed with 25Ilabelled sheep anti-mouse Ig (Amersham International, Amersham, Bucks, U.K.). Enzyme-linked immunoassay for anti-steroid detection. Polystyrene 96-well microtitre plates were coated overnight at 4° with rabbit anti-mouse Ig (5 ug/ml in 0-1 M sodium bicarbonate, pH 9 8), followed by washing in phosphate-buffered saline (PBS) containing 0-15% Tween 20 and 0-01% sodium azide (PBSTA). One-hundred microlitres of each clone supernatant were added per well, incubated for 1 hr at 370 and the wells washed again. One-hundred microlitres of a steroid-alkaline phosphatase conjugate were added and incubated for I hr at 37°. The plates were washed again, 200 PI substrate solution (I mg/ ml paranitrophenyl phosphate in I M diethanolamine, pH 9-8, containing I mM MgCl2) added per well and the OD measured at 405 nm after 30 min incubation. Specificity. Cross-reactivity of antibodies with related steroids was determined in a competition assay.'9 Aliquots (200 PI) ofantibody suitably diluted in PBSTA were incubated for I hr at room temperature with nylon pegs presensitized with rabbit anti-mouse Ig.20 After washing in PBSTA, the pegs were placed in microtitre wells containing 100 PI steroid-alkaline phosphatase conjugate and 100 P1 of a dilution of steroid, and incubated for 1 hr. After rewashing, the pegs were incubated with 200 Ml substrate solution for I hr and the optical densities read at 405 nm after removing the pegs. Calibration graphs with homologous steroid were constructed by plotting log steroid concentration against percentage inhibition of binding of steroid-enzyme conjugate to antibody. Cross-reaction with other steroids is defined as 100X/ Y%, where X is the mass of homologous steroid and Y is the mass of heterologous steroid required to produce 50% inhibition.
Cyanogen bromide digestion of antibodies2' Monoclonal antibodies at a concentration of 3-4 mg/ml were dialysed against 0-1 M ammonium bicarbonate, divided into 2mg aliquots and lyophilized. An individual 2-mg aliquot was made up to 400 P1 in 70% formic acid and a single crystal of CNBr added; the reaction vial was flushed with nitrogen, sealed and left overnight at room temperature in the dark. The reaction mixture was then diluted to 4 ml with water and divided into four 500 pg portions which were lyophilized separately; each was then resuspended in a few drops of water and relyophilized. Individual 500 pg portions were taken up in SDS sample buffer and run on SDS-PAGE (17% gel) for analysis and blot sequencing as below.
Blot-sequencing
SDS-polyacrylamide gel electrophoresis (SDS-PAGE). Polyacrylamide slab gels, 10-17% according to the proteins to be separated, were run in Tris/glycine/SDS buffer (pH 8 3) using the discontinuous system of Laemmli.22 All samples were prepared fully reduced in sample buffer, the final mixture
473
V gene usage Table 1. N-terminal sequences of anti-steroid light chains
VL group
Specificity
Progesteronehemisuccinate
Testosterone hemisuccinate
Estroneglucuronide
Pregnanediol glucuronide
mAb
1
11/32 10/8 DB3 11/64 2528:1
D D D D D
AN3/40
D I V L T Q S P A S L A V S L G Q R A T
2412:19
2433:16 2278:7 2323:18 2633:20 2923:22
D D D D D D
2380:18 2555:6 2384:5
D I V M S Q S P S ? L A V S P G E ? V T D I VM S Q S P S S L A V S V G E ? T T D I Q M T Q S S S ? L S I ? L G G K V T
10
Potter
Kabat
VVMTQTP L S L P V S L G D Q A S I S
VKI
MT Q T P L S L P V S L G D Q A S I S
VK1 VK1 VK1 VK1
KII KII KII
V V D V
V I I I I I
V V V V
V V V V V V
20
M T Q I P L S LPVNLGD Q A S I S MT Q T P L S L S V G L G D Q A S ? S
M T Q T P L S L
M L L L L L
T T T T
Q T P L S L P V S L G D Q A S I S
Q Q Q T Q T Q
S S S S S
consisting of 25 p1 stacking gel buffer (0-25 M Tris-HCl, pH 6-8), 20 p1 10% SDS, 10 p1 glycerol, 5 yl 2-mercaptoethanol and 50 pI protein solution; samples were boiled for 2 min before applying to the gel. Dimensions of the resolving gel were 13 cm x 12 cm, with stacking gel of 13 cm x 4 cm; gel thickness was 2 mm and tracks were 1 cm wide. Electrophoresis was performed until the bromophenol blue dye front had migrated a distance of 10-11 cm into the resolving gel. Gels to be used for electroblotting were divided into two parts. Tracks on one side of the gel were used for markers and for small aliquots of the protein samples, the purpose being to confirm the success of the run and to analyse the samples; after electrophoresis these tracks were detached and stained with Coomassie blue. Tracks on the remainder of the gel were used for electroblotting (below); after blotting they were also stained with Coomassie blue to estimate the amount of individual protein bands remaining in the gel. Electroblotting. Blotting onto Immobilon PVDF membrane (Millipore Corp., Bedford, MA) was carried out largely according to the manufacturer's instructions." Immobilon is a hydrophobic, mechanically strong membrane with a high protein adsorption capacity.'0 Before use, Immobilon strips were cut to size (slightly larger than the gel), treated for a few seconds in methanol and washed in water for 5 min before equilibrating in transfer buffer consisting of 50 mm sodium borate, pH 8-0, 20% methanol and 0-02% mercaptoethanol.23 Blotting took place in an LKB 2005 Transphor Electroblotting Unit (LKB, Bromma, Sweden), for 1 hr without cooling at the maximum voltage setting, which generated about 0.5 amps, then for a further hour with water cooling. After blotting, the membranes were stained with 0-05% w/v Coomassie blue for 5 min and destained for a similar time; staining and destaining solutions contained a minimum of 50% methanol. Stained bands of interest were excised for gas-phase sequencing and stored in Eppendorf tubes; there was no apparent deterioration of the protein on the dried Immobilon when stored at 40 for several weeks.
P P P P P
A A A A T
S S S S S
L L L L L
A A ? A A
V V V V V
S ? S S S
L L L L
G Q ? A T I G Q ? A T I G Q G
KII KII
VK21
KIII
VKl
KII KIII KIII KIII KIII KIII
VK21
VK21 VK21 VK21 VK21 VK8
VK8 VK9
KI KI KV
Automated N-terminal protein sequencing. Sequencing was carried out using an Applied Biosystems (Warrington, Cheshire, U.K.) 470A gas-phase protein sequencer equipped with a 120A on-line phenylthiohydantoin (PTH) analyser. Protein bands cut from Immobilon blots were placed directly into the sequencer reaction cartridge and held in place with TFA-etched glass fibre filter discs. RESULTS N-terminal light and heavy chain sequences of anti-steroid antibodies Samples of 100 pg of anti-steroid monoclonal antibodies were reduced with 2-mercaptoethanol and run on 12-5% SDSPAGE. Separated heavy and light chains electroblotted onto Immobilon were stained with Coomassie blue. Stained bands were cut from the blot and subjected to automated gas-phase sequencing. The HPLC profiles (not shown) of PTH amino acids from electroblotted samples were characterized by exceptionally low backgrounds and high repetitive yields. In general, 15-20 cycles of sequence determination were carried out. Light chains. N-terminal light chain sequences of 14 antisteroid monoclonal antibodies are shown in Table 1. For the anti-progesterone group (five antibodies), the sequences are virtually identical and all derive from the VK1 group in the Potter classification3 or KII in the Kabat classification.24 Sequencing of mRNA25 has confirmed the probable common origin of these light chains from the VKJ05 germline gene. No exceptions have been observed in seven sequences, including others sequenced from mRNA (M. J. Sims and M. J. Taussig, unpublished data). Among the monoclonal anti-estrone antibodies, five out of six also have virtually identical N-terminal light chain sequences, but in this case derived from the VK21 group (Potter) or Kabat group KIII. The one exception is antibody 2412:19,
474
M. J. Taussig et al. Table 2. N-terminal sequences of anti-steroid heavy chains
Specificity
1
mAb
10
20
Progesterone hemisuccinate
DB3* 2528:1
QI QL V QS G P E L K K P G E T V K I S VQ S G P E L K K P G E S V K I S
Estroneglucuronide
2412:19 2433:16 2323:18 2278:7 2633:20 2923:22
EV EM EV EV EV EV
2380:18 2384:5
E V K L V E S G G G L V K P G G S L K I S DV Q L Q E S G P G L V K P S Q S L
Pregnanediol glucuronide
QL I L P L I L I L Q L
V V V V V Q
E S G G D L V K P G G S L ? L S G G G L V K P G G S L K L S
A E E E E
S S S S
G G G G
G G L V K P G Q S L G G L V K P G G S L ? L P P G L A K P S Q T
VH family VGAM3.8 VGAM3.8
7183 7183 7183 7183 t 36-60 7183 36-60
* By mRNA sequencing.
t Not assigned. Table 3. Representative N-terminal germline sequences of mouse VH families
Family
Gene
1
10
20
References
VHJ558
H30 H18
Q V QLQQ S G A E L V K T G A S V K M S E V QL QQS G P E L V K P G A S V K I S
26, 27
VHQ52N
PJ14
Q V QL K E S G P G L V A P S QS L S I T
28
VH36-60
1210.7 SB322
E V QL QE S G P S L V K P S QT L S L T D V QL QE S G P S L V K P S QS L S L T
29-31
VHX24 VH7183
441
E V K L L E S G G G L V QP G G S L K L S
81X E4.15 37.1 50.1
E V Q E V M E V K E V K
32, 33 32, 34
VHJ606
14B 22.1
E V K L D E T G G G L V QP G R P M K L S E V K L E E S G G G L V QP G G S M K L S
35
VHS107
VI
E V K L V E S G G G L V QP G G S L R L S E V K L V E S G G G L V QP G A S L R L S
36, 37
V3
V31
QV T L K V S G P G I L QP S Q T L G L A
38
QI Q L V Q S GPELKKP G E T V K I S
25; M. J. Sims and
VH3609 VHVGAM3.8
L L L L
V V V V
E E E E
S S S S
G G G G
G G G G
G G G G
L L L L
V QP V K P V K P V QP
R G G G
E G G G
S S S S
L L L L
K K K K
L L L L
S S S S
M. J. Taussig
(unpublished data) MRL DNA4
VCP
DNA4
EVQLVET G G G L V Q P K G S S K L S EVQLLET G G G L V Q P G G S R G L S
which uses a VK1 gene identical in the first 23 residues to that of anti-progesterone antibodies. The single anti-testosterone antibody sequenced also uses a VK21 light chain. In contrast, neither VK1 nor VK21 light chains occur in the three antipregnanediol antibodies, two of which (2380:18, 2553:20) express VK8 (KI) sequences and the third (2384:5) a VK9 (KV) light chain. Heavy chains. The N-terminal sequences of heavy chains of 10 anti-steroid antibodies are shown in Table 2; for comparison, repesentative N-terminal germline sequences of 11 murine VH families are shown in Table 3. Most of the heavy chains of the
39 40
anti-progesterone antibodies were N-terminally blocked; however, one heavy chain, 2528: 1, could be sequenced by this method as it started at residue 5. (It is not known whether proteolytic cleavage of 2528: 1 occurred during preparation of the sample or whether the molecule as secreted lacks the normal first four residues of VH.) The sequence of the first 21 N-terminal residues of 2528: 1 corresponded to the germline of the VHVGAM3.8 family, with one substitution (serine replacing threonine at position 17). Elsewhere25'4' we have shown by mRNA sequencing that five out of five other anti-progesterone monoclonal antibodies, produced by three different mice, are
V gene coded by a VGAM3.8 VH segment and all start with an Nterminal glutamine, a residue which frequently causes Nterminal blocking by cyclisation to pyroglutamic acid. Among the anti-estrone monoclonals, four out of six heavy chains derive from the VH7183 family. VH genes of the VH7183, VHS107, VHX24 and VHJ606 families are similar but not identical over the first 20 N-terminal residues (Table 3); the antiestrone antibodies 2412:19, 2433:16, 2323:18 and 2278:7 possess a combination of residues characteristic of VH7183, namely valine at position 5, lysine at 13 and lysine at 19 (for 2433:16). In contrast, the VH of anti-estrone 2923:22 is a member of the VH36-60 family. Only 10 residues were determined for 2633:20, from which it was uncertain whether this antibody should be placed in VH7183/VHS107 (valine expected at position 5) or VH36-60 (proline at 9). Some unusual substitutions were seen at position 3 among the anti-estrones, notably isoleucine in three antibodies (2433:16, 2278:7 and 2633:20) and proline in 2323:18. At this position, isoleucine occurs only once in 634 antibodies reported in Kabat et al.24 and in no germline genes so far reported, but is seen three times in six anti-estrones (three mAb from two different mice); similarly, proline at position 3 has only two other reported occurrences out of 634 antibodies.24 Such changes can be attributed to somatic mutation, though the repeated occurrence of isoleucine-3 suggests the possibility of a
475
usage
Table 4. VH and VL combinations used in anti-steroid monoclonal antibodies
VH family
Antibody
Specificity
VL group
Progesterone hemisuccinate
2528:1 DB3 10/8 11/32 11/64
VGAM3.8 VGAM3.8* VGAM3.8* VGAM3.8* VGAM3.8*
VKl VK1 VK1 VK1 VK1
Testosterone hemisuccinate
AN3/40
ND
VK21
Estrone glucuronide
2412: 19 2433:16 2278:7 2323:18 2633:20 2923:22
7183 7183 7183 7183 t 36-60
VK1 VK21 VK21 VK21 VK21
2380:18 2555:6 2384:5
7183 ND 36-60
VK8 VK8 VK9
Pregnanediol glucuronide
VK21
* By mRNA or cDNA sequencing. t Not definitely assigned.
common germline origin.
Of the two anti-pregnanediol antibodies sequenced, one (2380:18) clearly belongs to the VH7183 family, possibly deriving from the germline gene 37.1, while the other (2384: 5) is coded in the VH36-60 family and is identical over the region sequenced to myeloma protein MOPC 315. The VH36-60 genes used by anti-estrone 2923:22 and anti-pregnanediol 2384:5 are probably different (glutamic acid versus aspartic acid at residue 1; threonine versus serine at 17). The VH/VL combinations used in these antibodies are summarized in Table 4.
6945-
36 29 24
-
-
9
-
7
20-
Internal sequences of VH and VL by cyanogen bromide cleavage The N-terminal residues of VH genes are frequently blocked by cyclisation of glutamine to pyroglutamic acid and are not available for the automated Edman degradation. As noted above, this is the case for all the anti-progesterone antibodies with the exception of 2528: 1 (Table 2). Two approaches are available to negotiate this problem. One is to deblock the Nterminus with pyroglutamate aminopeptidase,42 a procedure which has not been successful with anti-progesterones in our hands. The other is to subject the antibody to partial degradation with chemical reagents which produce proteolytic cleavage at relatively infrequent amino acid residues, a familiar example being specific cleavage at methionine with cyanogen bromide.43 We have employed this reagent together with the blot-sequencing technique to confirm that anti-progesterones use VHVGAM3.8 genes. In this VH family, a germline-encoded methionine at residue 48 immediately precedes the CDR2, the sequence of which (GWINTYTGEPTY) is unique to the family (M. J. Sims and M. J. Taussig, unpublished data).25 Five-hundred microgram samples of anti-progesterone antibodies DB3 and 10/8 were treated with CNBr; the resulting fragments of heavy and light chains were separated by 17% SDS-PAGE and blotted onto Immobilon (Fig. 1). Ten distinct
10
1
6 5 g=33535
=
37E~m3 7
KD557a71 6
4
3f n
5
142
..
14 3
_~2 _~ 1
I
Markers
I-
DBL
digest
10/8 digest
Figure 1. Position on SDS gel and Immobilon blot of fragments of the anti-progesterone antibodies DB3 and 10/8 after CNBr degradation. The bands are drawn from the blot and numbered; bands which were shown by sequencing to contain V region fragments are indicated with single diagonals (VL fragment) or cross-hatching (VH fragment). Molecular weights (MW x 10-3) ofmarkers are shown. See also Table 5.
bands were obtained from DB3 after CNBr cleavage and 12 from 10/8, the absence of an intact heavy chain band indicating that digestion was complete in each case. Individual bands were cut out of the blot and sequenced for 6-10 cycles in order to identify the fragments. The results of this analysis are given in Table 5 and the deduced location of methionine residues in each molecule is shown in Fig. 2. All the bands yielded interpretable sequences
M. J. Taussig et al.
476
Table 5. Identification, by blot-sequencing, of fragments obtained after CNBr degradation of anti-progesterone antibodies
Antibody DB3
10/8
Band
MW*
1
5000-6000
2
10,000
3
16,000
4 5 6 7 8
17,000 19,000 20,000 23,000 30,000
1 2
3000 4000
3 4 5 6 7 8 9 10 11
10,000 13,000 16,000 19,000 21,000 23,000 24,000 29,000 32,000
N-terminal sequence of fragment DVVMTQ DTDGS GWINIY ITDFFP VTLG(C)L YKVSN(R) ND YKVSN(R) DVVMTQ VTLG(C)L VTLG(C)L
GWINTY ATYF(C) DTDGS SSTLT ITDF ND ND
DVVMTQ TQTPL TQTPL VTLG(C) DVVMTQ ATYF(C) VTLG(C)
Location in molecule and residuest VL, 1-48 CH3, 427-477 VH, 49-137 CH3, 392-477 CH1, 138-328 VL, 49-175 VL, 49-214 VL, 1-175
CH1-CH3, 138-381 CHI-CH3, 138-426 VH, 49-87
VH-CHl, 88-137 CH3, 427-477 CL, 176-214 CH3, 392-477
VL-CL, 1-175 VL-CL, 5-175 VL-CL, 5-214 CHI-CH3, 138-381 VL-CL, 1-214 VH-CH3, 88-381 VH-CH3, 138-477
* Apparent MW from gel markers. t Positions of methionines in constant region taken from sequence for MOPC 21 (mouse IgG I myeloma), numbering according to Kabat et al.24 Residues in parentheses not identified by sequencer, but assumed from known sequences to be cysteine (C) or arginine (R) as appropriate. ND, not done.
from which the fragments could be identified and assigned to variable or constant regions; the sizes of each identified fragment were in most cases consistent with the VH mRNA sequences of these antibodies25 and the positions of methionines in the y I and K constant regions24. For DB3, sequences of bands 1, 3, 5 and 6 were identified as light chain fragments (all from VL), while bands 1, 2, 3, 7 and 8 contained heavy chain fragments (VH and CH); the presence of mixed sequences in 1, 2 and 3 did not present a problem of interpretation. A sequence in band 2 was identified as the CDR2 of the VHVGAM3.8 family (GWINIY, residues 49-53), the isoleucine at position 52a being a divergence from the germline. This fragment was the only internal VH sequence of DB3 revealed by CNBr cleavage. Similarly, for antibody 10/8, bands 2, 6, 7 and 9 contained light chain fragments (all VL except band 2) and bands, 1, 2, 3, 8, 10 and I I were heavy chain derived (VH and CH); band I contained the characteristic VHVGAM3.8 CDR2 sequence, GWINTY. These results demonstrate that it is possible, by fragmentation and blot sequencing, to confirm the presence of the VHVGAM3.8 segment in antibodies which are N-terminally blocked.
A second 10/8 VH fragment was obtained with the sequence ATYFC (bands 2 and 10), indicating the presence ofmethionine at position 87 in the heavy chain of 10/8; this fragment was not found in DB3. This difference is encoded in the germline of the VHVGAM3.8 family, where three germline genes have threonine-87 and two have methionine-87 (M. J. Sims, U. Krawinkel and M. J. Taussig, manuscript in preparation). Thus, the presence of the appropriate CNBr fragments in the 10/8 digest and their absence from DB3 is evidence that DB3 and 10/8 derive from different VHVGAM3.8 germline genes. For the light chain, one internal VL sequence was obtained from DB3, namely the CDR2 (YKVSNR), which occurred in bands 3 and 5. The presence of this VL fragment, identified as starting at residue 49, was unexpected, as residue 48 had originally been identified as isoleucine rather than methionine by mRNA sequencing25 and is also isoleucine in the putative germline gene Vc105.44 Re-examination of the mRNA sequencing gel showed it to be ambiguous at the third base of codon 48 and that the original reading had probably been erroneous. No corresponding fragment was found in 10/8 light chain, which is otherwise homologous to DB3.
Specificity profiles of anti-steroid antibodies The specificity of anti-estrone and anti-pregnanediol antibodies was studied in detail by competition between steroid-enzyme conjugate and a panel of free steroids (Tables 6 and 7); a similar study of anti-progesterone antibodies has been published previously.45 Among the anti-estrone monoclonals (Table 6), all have greater reactivity with the immunizing ligand, estrone-3-glucuronide, than with estrone itself or estrone-3-sulphate, indicating that the sugar linker group contributes significantly to binding affinity. 2412: 19, the only one to use the VH7183/VK1 combination, is the most specific antibody, cross-reacting significantly only with 17fl-estradiol-3-glucuronide; in particular, it did not bind detectably to estrone-3-sulphate, androsterone or androsterone-glucuronide. In contrast, antibodies in which VH7183 is combined with VK21 all show greater binding reactivity than 2412: 19 against estrone, estrone-3-sulphate, androsterone and androsterone glucuronide and less against 17,B-estradiol-3glucuronide; this may indicate an influence of light chain on specificity. The patterns of cross-reaction of 2433:16 and 2278:7 are virtually identical; these antibodies, which both use the VH7183/VK21 combination, arose in the same fusion and are likely to be clonally related. 2323:18, also VH7183/VK21, shows a similar pattern but greater cross-reaction with androsterone and its derivatives. 2633:20, for which the heavy chain family could not be definitely assigned, has a particularly high cross-reaction for androsterone and androstenedione. 2923: 22, in which a 36-60 VH segment is combined with VK21, has a more restricted specificity than the VH7183/VK21 group; it crossreacts with estrone, estrone-3-sulphate and 17,B-estradiol to about the same extent as the other antibodies, but has lower reactivity against androsterone-glucuronide and several other ligands. In the anti-pregnanediol antibodies (Table 7), binding was also highest with the immunizing ligand (pregnanediol-3glucuronide), about 10 times greater than with pregnanediol itself. Antibodies 2380:18 and 2384:5, coded by different VH and VL families, diverge considerably in detailed specificity. Antibody 2384:5 (VH36-60/VK9) is highly pregnanediol speci-
477
V gene usage DB3 LIGHT
Met: 4
175
48
1
1
,
5
Fragments_
6
.
100
2-1
26 0-06
244 > 100
35
44-5
