F protein sequence
Eur. J. Immunol. 1991. 21: 1235-1240
J. Paul Schofieldov, R. K.VijayakumarAn and David B. G. OliveiraAo MRC Molecular Genetics Unit. and Department of Medicine, University of Cambridge School of Clinical MedicineA, Addenbrooke’s Hospital, Cambridge
1235
Sequences of the mouse F protein alleles and identification of a T cell epitope F protein is found predominantly in the liver and is of unknown function. The protein has been of interest to immunologists in the areas of self tolerance and the immunogenetics of the anti-F protein response. In the mouse there are two alleles (F1 and F2), and although mice are completely tolerant to the self form of the protein, mice of responder strains make a good antibody response to immunization with the non-self form. This response cross-reacts with the self form, implying firstly, that autoreactive B cells are present and that tolerance is therefore maintained at theTcell level, and secondly, that the difference between the two allelic products defines a Tcell epitope. Primers based on the published sequence for rat F protein were used in the polymerase chain reaction to amplify the cDNA for the two mouse alleles. Subsequent sequencing shows a high degree of sequence identity between the rat and mouse cDNA.The two mouse cDNA are identical apart from a single A to G base change which predicts an asparagine (F1 protein) to aspartate (F2protein) amino acid residue change. Using allele-specific oligonucleotide probes we confirmed that this base change has the same strain distribution as the previously determined F protein type. Isoelectric focusing shows that F1 protein migrates in a more basic position than €2protein, as predicted by the asparagine to aspartate change. Finally, a synthetic peptide from the allovariable site of F2 protein will successfully restimulateTcells in vifro from an Fl type mouse primed in vivo with whole F2 protein, whereas the corresponding peptide from F1 protein will not. This is evidence that, as predicted, the allovariable site does indeed define aTcell epitope. Peptides covering the rest of the F2 protein molecule were not stimulatory.
1 Introduction
respect to self-tolerance include APC [6] and the suppressor cell system [7].
F protein, first described in 1968111, is found predominantly in the liver, although smaller amounts are detectable in all organs studied in both mouse [2] and man [3]. It is also evolutionarily conserved in that it has been found in all vertebrates so far examined [4] as well as in an invertebrate (a starfish, species unknown; unpublished observations). Despite this indirect evidence for the importance of F protein, its function remains unknown. The protein has been of interest to immunologists in two main areas. Firstly, it has been used as a probe to determine the extent to which different compartments of the immune system are rendered tolerant to self [5]. Briefly, there are two alleles in mice, F1 and F2,and mouse strains are completely tolerant to the self form of the protein. Mice of responder strains (see below), however, make a good antibody response when immunized with the non-self form of the protein. This antibody response cross-reacts with the self form of the protein, demonstrating that autoreactive B cells are present and that tolerance is maintained by the T cell. Other compartments that have been studied with
[I 91481
The second main area of interest is the immunogenetics of the anti-Fprotein response in the mouse. There are believed to be both immune response and immune suppression gene effects operating to control this response [8,9], and in particular the MHC class I1 A molecule of the k haplotype is required in order to mount an immune response to either allelic form (allomorph) of the protein. A prediction from the differing T and B cell reactivities towards F protein outlined above is that the two F protein allomorphs must obviously be different, and that this difference will define, either directly or indirectly, a Tcell epitope. Progress in this area would clearly be greatly aided by a knowledge of the primary sequences of the two mouse F protein alleles. Following the publication of the cDNA sequence encoding the rat Fprotein [lo], and given the known conservation of the protein, we attempted the reverse transcriptase polymerase chain reaction (RT-PCR) amplification of mouse cDNA using primers derived from the rat sequence. Peptides where then synthesized based on the mouse cDNA sequences to test the prediction concerning the Tcell epitope.
MRC Training Fellow. Supported by a grant from Smith Kline Beecham. 0 Lister Institute Research Fellow.
2 Materials and methods
Correspondence: David B. G. Oliveira, Department of Medicine,
2.1 Animals and cDNA preparation
Addenbrooke’s Hospital, Hills Road, Cambridge CB22, GB Abbreviation: PCR Polymerase chain reaction 0 VCH Verlagsgesellschaft mbH, D-6940 Weinheim, 1991
Mice of F1 (CBA) and F2 type (BALB/c) were obtained from Olac (Bicester, GB). Liver total RNA was isolated 0014-2980/91/0505-1235$3.50+.25/0
1236
J. P. Schofield, R. KVijayakumar and D. B. G. Oliveira
[ 111 and first-strand cDNA reverse transcribed [12] using oligo(dT) primer, without prior purification of poly(A)+ mRNA. 2.2 PCR and DNA sequencing About 250-500 ng of first strand cDNA was input into a 50 p1 PCR. Amplification primers were derived from the published rat F protein cDNA sequence and modified at their 5’ ends by incorporating a restriction endonuclease cleavage recognition site to facilitate “sticky-end’’ cloning:
Eur. J. Immunol. 1991. 21: 1235-1240 F1 protein gene 5‘-AAGACTGCAACCACATTGT-3‘
F2 protein gene 5’-AAGACTGCGACCACATTGT-3’ The probes were end-labeled with [Y-~~P]ATP (> 3000 Ci/ mmol = 111TBq/mmol, Amersham) using T4 polynucleotide kinase [15] and purified on 0.5 ml DE52 columns (Whatman, Maidstone, GB). Hybridization was at 50°C for 4-8 h, followed by 2 x 15-min washes at 60°C in 6 x SSC.The filters were then exposed for 4-24 h (depending on bound radioactivity).
2.4 2-D gel electrophoresis Forward primer 5 ‘-ACTGTCGACTACTGGGACA AAGGACCAAAG-3’ Sal I Reverse primer 5’-ACTCTGCAGACACCGTGGTCTCCAGGTCA-3’ Pst I The reaction mixture was overlaid with 50 p1 light mineral oil (Sigma, Poole, GB) and amplified for 30 cycles on a Techne PHC-1 programmable thermocycler according to published protocols [13, 141.The PCR temperature profile was: 95 “C 30 s denaturation, 58°C 30 s annealing, and 72°C 1 min extension. A 5-pl aliquot from the PCR was checked for purity and yield by agarose gel electrophoresis, and the remainder was gel-purified and extracted onto glassmilk (GenecleanTM, Bio-101, La Jolla, CA). The purified PCR product was digested with SalI and PstI restriction enzymes, phenol extracted, and ethanol precipitated [15]. Ligation into prepared cloning vector pBluescript-KSf (Stratagene, Cambridge, GB) was performed overnight at 15“C, followed by transformation into E. cofi NM522 frozen competent cells [16].White colonies were screened for insert by PCR [ 171, and Qiagen (Diagen, Dusseldorf, FRG) purified DNA was sequenced on both strands by dideoxy-chain termination [18] with oligo-walking [19].
2.3 Allele-specific oligonucleotide dot-blotting The following PCR primers were used to amplify a 159-bp fragment encompassing the allovariable site: Forward primer 5’-CACCTGGTGAAGCATGG-3‘ Reverse primer 5’-CTGCAGCACAGCGAACT-3’ cDNA was prepared from the livers of a variety of mouse strains of known Fprotein type [20]. cDNA used to determine the sequences in Sect. 2.2 was also included to verify probe specificity. PCR conditions were as described above. The PCR products were dot-blotted onto two duplicate filters (Hybond N, Amersham Int., Amersham, GB) via a Hybri-Dot manifold (BRL, Gaithersburg, MD). Following prehybridization, one of the filters was hybridized with an oligonucleotide probe complementary to the allovariable site of the F1 protein gene, and the other with a similar probe for the F2 protein gene.The probe sequences were:
Both F1 and F2 proteins were affinity purified from mouse liver extracts (CBA and BALB/c, respectively) using an anti-F protein mAb affinity column as previously described [21]. 2-D gel electrophoresis was based on published protocols [22]. The first IEF dimension incorporated 6% of ampholyte 5-8 and 1.2% of ampholyte 3.5-10 (LKB, Cambridge, GB). Running buffers were 5% 1,2-diaminoethane (cathode) and 5% orthophosphoric acid (anode). The second dimension was run in a discontinuous buffer system [23] with 12% acrylamide in the resolving gel. Proteins were visualized by staining with Coomassie blue.
2.5 Peptide synthesis A series of peptides (15-21 residues long, overlapping by 6 residues) encompassing the entire F protein sequence (additional peptide included in series to provide the alternative at the allovariable site) was synthesized as described by Houghton et al. [24]. Briefly, standard Boc (t-butoxycarbonyl) amino acid resin (170 mg) was sealed in polypropylene mesh bags for simultaneous multiple peptide synthesis. Standard de-protecting, neutralization, coupling and wash protocols were employed [25,26]. To prevent acylation and racemization of histidine residues during coupling DNP-protected histidine was used. DNP was removed at the end of synthesis by treatment with 1% (v/v) thiophenol in dimethyl formamide.The peptides were cleaved simultaneously from the resin using a conventional HF procedure [25]. Cleaved peptides were extracted from the bags in 5%-30% acetic acid, lyophilized and dissolved in either PBS or DMSO at 5 mg/ml, according to their solubililty. DMSO-soluble peptides were used in proliferation assays at a final DMSO concentration of < 1%. Purity of peptides was assessed by reverse phase HPLC using a Novapak C18 column (Waters Chromoatgraphy, Watford, GB) running in a gradient from water to 50% acetonitrile (in 0.02% triethylamine, buffered to pH 6.8 with acetic acid). Peptides with a purity > 70% were used in experiments.
2.6 T cell proliferation assay Mice of F1 type (CBA) were primed with 100 pg F2 protein emulsified with an equal volume of CFA (Difco, Detroit, MI); the total volume (200 pl) was divided equally between both rear footpads and both sides of the base of the tail.
Eur. J. Immunol. 1991. 21: 1235-1240 Figure 1. Rat Fand mouse F1 protein cDNA and predicted amino acid sequences (rat data from [lo]).The underlined bases represent the PCR primers. Shown in bold is the site of the difference between the two mouse F protein alleles (see Fig. 2).
After 7-10 days draining LN were removed and a singlecell suspension prepared. Cells (2 x 105-5 x 105) were cultured in 200 p1 of Iscove's modified Dulbecco's medium (Gibco, Paisley, Scotland) with the addition of 10% FCS in a 5% COz atmosphere at 37 "C. Additions of proteins and peptides were at the start of the culture period which was for a total of 96 h. [3H]dThd (1 pCi) was added to each well for the last 8 h of culture. Cells were harvested using a Titertek cell harvester and incorporated [3H]dThd measured in a Packard Tri-Carb liquid scintillation analyzer. All cultures were in triplicate and the results are expressed as the mean k SEM.
3 Results 3.1 Mouse F protein allele cDNA sequence A comparison between rat Fand mouse F1 protein cDNA, together with the predicted primary amino acid sequences, is shown in Fig. 1. There is 92% identity between rat and mouse Fprotein cDNA, and 97% identity between the predicted amino acid sequences. The mouse F protein sequence has an apparent deletion of the first base of the termination codon (not shown in Fig. 1). This type of deletion has been noted before (unpublished observations) and is believed to represent a PCR artefact. The mouse F protein cDNA are identical apart from a single nucleotide substitution which predicts an asparagine residue in F1 protein and an aspartate residue in F2 protein (Fig. 2). The predicted molecular mass is 43 kDa for the mouse F proteins. Apart from the expected match with rat F protein, no other significant matches were found in the EMBL or Genbank databases.
3.2 Allele-specific oligonucleotide dot-blotting Primers based on the sequences of the mouse Fprotein alleles were used to successfully amplify a 159-bp region centered on the single base difference found between the two alleles. The resultant dot blot is shown in Fig. 3, demonstrating concordance between the known F type of the various strains and the nucleotide difference. 3.3 2D gel electrophoresis The nucleotide difference between the F1 and F2 protein genes predicts that F2 protein (with an aspartate residue) should be more negatively charged than F1 protein (with an asparagine residue). Fig. 4 shows that although a number of spots are visible for each form of Fprotein, F1 protein migrates in a more basic position than F2 protein.
F protein sequence
1237
J. P. Schofield, R. K.Vijayakumar and D. B. G. Oliveira
1238
Eur. J. Immunol. 1991.21: 1235-1240
I
A-A
2l-mrp.ptidefmmF2
W-W
21-mwp.pthfmmFl
0
0
n 1
i J
1
U
-
F1
F2
10 29 I ml
12
I 6
3
1.5
paptlde concmtmtionpq/ ml)
Figure 5. Proliferation assay of Tcells primed to F2 protein (see Sect. 2.6 for details). ( 0 ) Represents background proliferation (note that the scale starts at 1 x 104cpm). Points are mean of IIIIIIIIIIII IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII triplicates f SEM. ; l-r
V E D C N H I V Q K A R E R G A K I V a l G l ~ iI lse V a l G l n L y s A l a A r g G l u I l e Muse F1 Muse F2
V a l c l ~ ~I l i e sV a l G l n L y s A l a A r g G l I l e V E D C D H I V Q K A R E R G A K I
Figure2. Fragment of the sequencing gel showing the only identified difference between mouse F1 and F2 protein cDNA, and the resulting predicted amino acid sequence.
3.4 T cell proliferation assay
A synthetic peptide from the F2 protein molecule, centered on the allovariable site, could successfully stimulate Tcells primed against the whole F2 protein molecule; these cells did not proliferate to whole F1 protein or to a corresponding synthetic peptide centered on the allovariable site (Fig. 5). A series of peptides covering the entire F2 protein sequence was also tested; no other stimulatory peptides were detected despite using each peptide over a concentration range of 1-20 pg/ml (data not shown).
4 Discussion Figure 3. Allele-specific oligonucleotide dot blot of amplified fragments of F protein cDNA centered on the allovariable site. Fragments were amplified from the cDNA used to obtain the original sequences (labeled F1 and F2), and from previously typed strains of mice [20]: F1 type CBA, DBAR; F2 type: BALBk,
C57BL/6. OVA
-
+ F1
F2
F1 and F2 Figure 4 . 2-D electrophoresis of immunopurified F1 protein (top panel), F2 protein (middle panel), and a mixture of the two (bottom panel: same amounts of protein as in the above panels). IEFdimension horizontal, SDS-PAGE dimension vertical. Marker protein on the left is OVA (45 kDa).
The principal results of this work are (a) the amplification and sequencing of the cDNA encoding the two mouse F proteins; (b) evidence that the identified sequence difference between the two cDNA does indeed constitute the allelic difference between the two F protein genes and (c) evidence that the predicted difference in primary amino acid sequence defines a T cell epitope. Our use of the PCR for the amplification of the mouse F protein cDNA introduces the possibility of amplification artefacts in the resultant sequences. Taq polymerase has an error rate of approximately nucleotide/cycle [27], and any such error will of course be duplicated in all subsequent amplification steps. Although only one clone was sequenced for each F protein cDNA, the existence of two alleles allows a cross-check against such artefacts. The fact that the independent primary sequences were identical, apart from a single base (discussed below), provides strong confirmation that the sequences are correct. The 21 5' bases of the sequences represent the PCR primer and are obviously the same for all three cDNA. However, there is clearly a high degree of sequence similarity between rat and mouse F protein cDNA. Most of the differences are third base position changes which make no difference to the predicted amino acid sequence, and many of the predicted differences are conserved substitutions. As with the rat F protein cDNA the intitiation codon for the two mouse
F protein sequence
Eur. J. Immunol. 1991. 21: 1235-1240
cDNA has not been located, but the close agreement between predicted and measured molecular weights suggest that there is very little missing 5' coding sequence. It was important to obtain independent corroboration of the single base difference found between the two mouse Fprotein cDNA: it was possible that this represented a PCR artefact, and that the true allelic difference lay in the short missing 5' coding sequence. The first piece of confirming evidence is provided by the dot blot data. The specific hybridization to six independently amplified cDNA fragments found with probes differing by only one base argues strongly against a PCR artefact. The observed correlation between the base difference and known F protein type also provides evidence that this nucleotide difference is the source of allelic variation in the F protein system, as opposed to an irrelevant mutation peculiar to the initial strains used for cDNA amplification. The identified nucleotide difference between the two mouse F protein cDNA does not exclude other allelic differences present at the 5' or 3' ends (non-coding region in the latter case), either in the region of the PCR primers or beyond. The two further pieces of confirmatory evidence, IEF and T cell epitope analysis, are therefore particularly useful, as they also demonstrate that known differences between the F protein alleles at the protein level can be accounted for by the single amino acid substitution. Previous work on IEF of F proteins revealed a rather complex pattern, with mul$ple bands detectable on immunoblotting of focused crude liver extracts [20]. However, it was found that a number of bands migrating in the basic PI region were present in F1 type liver extracts, but absent in €2type liver extracts. We also found that the picture is complicated by the presence of a number of spots, but confirmed that F1 protein does indeed migrate in a more basic position than F2 protein. This is the difference expected from the predicted amino acid substitution. The differently charged species found for each of the allomorphs may be due to differential phosphorylation or possibly in vitro de-amidation of asparagine residues, but further work is required to confirm or refute these speculations. The explanation for the much more complex picture found with crude extracts is also unclear. We used affinity-purified proteins for our focusing, whereas the previous study used crude liver extracts run under nonreducing conditions. We have also found that Western blotting (with anti-F protein antisera) of SDS-polyacrylamide gels of crude liver extracts run under nonreducing conditions reveals a number of high molecular weight bands (unpublished data). It is therefore possible that F protein may associate with other proteins (or itself) under some conditions, explaining these more complex patterns. This property may also explain the anomalously high molecular weights reported for F protein (typically about twice the weight of the monomeric protein) in some earlier studies [28,29]. Although it is clear that Tcells are capable of distinguishing between the two allelic forms of Fprotein [5] it did not follow that any amino acid difference between the two would necessarily define aTcell epitope directly.There are examples in other systems where variations at one point in a protein structure can influence the presentation of epitopes
1239
at another point relatively distant in the primary sequence, perhaps due to differences in processing [30]. Other mechanisms for this effect have also been proposed, such as inhibitory interactions between the portion of a peptide lying outside the MHC molecule and the portion actually in the MHC molecule groove [31]. It was therefore not inconceivable that the immunodominant T cell epitope(s) would lie in a portion of the molecule common to the two F protein allomorphs, with the allelic difference determining whether this would be presented successfully. However, there was a strongapriori argument suggesting that this was unlikely, namely that both allomorphs of F protein can be presented successfully (as judged by their immunogenicity) and both exhibit the same very restricted immune response gene dependency, that is a requirement for the MHC class I1 Ak molecule [9].The simplest explanation for this is that the same portion of each allomorph is involved in each case, accounting for the identical MHC molecule binding properties, with the difference at the allovariable site allowing self-nonself discrimination. Our data indeed provide evidence for the straight forward situation in which the allovariable site directly defines a T cell epitope. This epitope does not bear any obvious features in common with other epitopes known to be presented by the Ak molecule [32], and is distinct from the candidate epitope predicted on this basis from the rat Fprotein gene cDNA sequence [lo]. The argument outlined above concerning the ability of both forms of F protein to be successfully presented implies that nonresponsiveness to the self form of the protein in responder strains must be due to creation of the relevant hole in theTcell repertoire during the establishment of self tolerance. This is not particularly surprising, but the nature of nonresponsiveness in nonresponder (non-Ak-bearing) strains is less clear. One possibility is that other class I1 MHC molecules are.unable to bind the allovariable peptide. Another possibility is that such molecules, although capable of binding this peptide, do so in such a way that the resultant MHC-antigen complex appears the same toTcells irrespective of the allomorph of F protein involved. Alternatively, the Tcell repertoire produced in the presence of non-Ak molecules may lack the ability to discriminate between the bound epitopes of the two allomorphs. Further work is required to distinguish between these possibilities, and to determine the features of the epitope that are required for successful interaction with the Ak molecule and the T cell receptor. Finally, we are no closer to determining the function of F protein. The lack of sequence similarity with any other known cDNA or protein sequence is disappointing. It is possible that further clues may arise from studies on the human F protein. We have recently screened a human genomic DNA cosmid library using a randomly primed 1 kb mouse F1 probe, and are in the process of characterizing a strongly positive clone. However, it is perhaps more likely that we will continue to increase our understanding of the immunology of F protein while remaining ignorant of its function. We are indebted to Karen Worfreys for expert technical assistance.
Received December 20, 1990.
1240
J. P. Schofield, R. K.Vijayakumar and D. B. G. Oliveira
5 References 1 Fravi, G. and Lindenmann, J., Nature 1968. 218: 141. 2 Griffiths, J. A. and Oliveira, D. B. G., Scand. J. lmmunol. 1988. 27: 357. 3 Grewal, G. K. and Oliveira, D. B. G., Clin. Chim. Acta 1990. 191: 93. 4 Oliveira, D. B. G. and Vindlacheruvu, S., Comp. Biochem. Physiol. ( B ) 1987. 87: 87. 5 Mitchison, N. A., Clin. lmmunol. Newsl. 1985. 6: 12. 6 Winchester, G., Sunshine, G. H., Nardi, N. and Mitchison, N. A., lmmunogenetics 1984. 19: 487. 7 Lukic, M. L. and Mitchison, N. A., Eur. J. Immunol. 1984.14: 766. 8 Oliveira, D. B. G., Blackwell, N..Virchis, A. E. and Axelrod, R. A., Immunogenetics 1985. 22: 169. 9 Oliveira, D. B. G. and Nardi, N. B.. Immunogenetics 1987.26: 359. 10 Gerswhin, M. E., Coppel, R. L., Bearer, E., Peterson, M. G., Sturgess, A. and Mackay, I. R., J. lmmunol. 1987. 139: 3828. 11 Chirgwin, J. M., Przybyla, A. E., MacDonald, R. J. and Rutter, W. J., Biochemistry 1979. 18: 5294. 12 Gubler, U. and Hoffman, B. J., Gene 1983. 25: 263. 13 Saiki, R. K., Gelfand, D. H., Stoffel, S., Scharf, S. J., Higuchi, R., Horn, G. T. and Erlich, H. A., Science 1988. 239: 487. 14 Rappolee, D. A., Mark, D., Banda, M. J. and Werb, Z., Science 1988. 241: 708. 15 Manitis,T., Fritsch, E. and Sambrook J., Molecular cloning: a laboratory manual, 2nd Edit., Cold Spring Harbor Laboratory Press, New York 1990. 16 Hanahan, D., J. Mol. Biol. 1983. 166: 557. 17 Schofield, J. €?, Vaudin, M., Kettle, S. and Jones, D. S. C., Nucleic Acids Res. 1989. 17: 9498. 18 Sanger, F., Nicklens, S. and Coulson, A. R., Proc. Natl. Acad. Sci. USA 1977. 74: 5463. 19 Bankier, A.T. and Barrell, B. G., in Howe, C. J. and Ward, E. S. (Eds.), Nucleic acids sequencing, a practial approach, IRL Press, Oxford 1990.
Eur. J. Immunol. 1991. 21: 1235-1240 20 Hayakawa, J. and Nikaido, H., lmmunogenetics 1987. 26: 366. 21 Wedderburn, L., Lukic, M. L., Edwards, S., Kahan, M. C., Nardi, N. and Mitchison, N. A., Mol. Immunol. 1984. 21: 979. 22 OFarrell, F! H., J. Biol. Chem. 1975. 250: 4007. 23 Laemmli, U. K., Nature 1970. 227: 680. 24 Houghton, R. A., Proc. Natl. Acad. Sci. USA 1985. 82: 5131. 25 Houghton, R. A., Chang, W. C. and Li, C. H., lnt. J. Pept. Protein Res. 1980. 16: 311. 26 Houghton, R. A., Ostrech, J. M. and Klipstein, F. A., Eur. J. Biochem. 1984. 145: 157. 27 Tindall, K. R. and Kunkel, T. A., Biochemistry 1988. 27: 6008. 28 Mihas, A. A., Hirschowitz, B. I. and Saccomani, G., J. lmmunol. 1976. 116: 1228. 29 Mori,Y., Iesato, K., Ueda, S.,Wakashin,Y., Wakashin, W. and Okuda. K., Clin. Chim. Acta 1981. 112: 125. 30 Shastri, N., Miller, A. and Sercarz, E. E., J. Immunol. 1986. 136: 371. 31 Vacchio, M. S., Berzofsky, J. A., Krzych, U., Smith, J. A., Hodes, R. J. and Finnegan, A., J. lmmunol. 1989. 143: 2814. 32 Buus, S . , Sette, A., Colon, S. M., Miles, C. and Grey, H. M., Science 1987. 235: 1353. Noteaddedinproof: We haverecentlylearnt that ageneencodinga protein homologous to rat and mouse F protein has been cloned from the ciliated protozoan Tetrahymena therrnophila (K. Kristiansen, personal communication; submitted for publication). Similarities at the 3' end of the mouse and Tetrahymena genes strongly suggest that the apparent deletion of the first base of the termination codon of the mouse genes (Sect. 3.1) is real, and that the coding region for the mouse genes extends into the 3' PCR primer. Received March 23, 1991.
Eur. J. Immunol. 1991. 21: 1235-1240
J. Paul Schofieldov, R. K.VijayakumarAn and David B. G. OliveiraAo MRC Molecular Genetics Unit. and Department of Medicine, University of Cambridge School of Clinical MedicineA, Addenbrooke’s Hospital, Cambridge
1235
Sequences of the mouse F protein alleles and identification of a T cell epitope F protein is found predominantly in the liver and is of unknown function. The protein has been of interest to immunologists in the areas of self tolerance and the immunogenetics of the anti-F protein response. In the mouse there are two alleles (F1 and F2), and although mice are completely tolerant to the self form of the protein, mice of responder strains make a good antibody response to immunization with the non-self form. This response cross-reacts with the self form, implying firstly, that autoreactive B cells are present and that tolerance is therefore maintained at theTcell level, and secondly, that the difference between the two allelic products defines a Tcell epitope. Primers based on the published sequence for rat F protein were used in the polymerase chain reaction to amplify the cDNA for the two mouse alleles. Subsequent sequencing shows a high degree of sequence identity between the rat and mouse cDNA.The two mouse cDNA are identical apart from a single A to G base change which predicts an asparagine (F1 protein) to aspartate (F2protein) amino acid residue change. Using allele-specific oligonucleotide probes we confirmed that this base change has the same strain distribution as the previously determined F protein type. Isoelectric focusing shows that F1 protein migrates in a more basic position than €2protein, as predicted by the asparagine to aspartate change. Finally, a synthetic peptide from the allovariable site of F2 protein will successfully restimulateTcells in vifro from an Fl type mouse primed in vivo with whole F2 protein, whereas the corresponding peptide from F1 protein will not. This is evidence that, as predicted, the allovariable site does indeed define aTcell epitope. Peptides covering the rest of the F2 protein molecule were not stimulatory.
1 Introduction
respect to self-tolerance include APC [6] and the suppressor cell system [7].
F protein, first described in 1968111, is found predominantly in the liver, although smaller amounts are detectable in all organs studied in both mouse [2] and man [3]. It is also evolutionarily conserved in that it has been found in all vertebrates so far examined [4] as well as in an invertebrate (a starfish, species unknown; unpublished observations). Despite this indirect evidence for the importance of F protein, its function remains unknown. The protein has been of interest to immunologists in two main areas. Firstly, it has been used as a probe to determine the extent to which different compartments of the immune system are rendered tolerant to self [5]. Briefly, there are two alleles in mice, F1 and F2,and mouse strains are completely tolerant to the self form of the protein. Mice of responder strains (see below), however, make a good antibody response when immunized with the non-self form of the protein. This antibody response cross-reacts with the self form of the protein, demonstrating that autoreactive B cells are present and that tolerance is maintained by the T cell. Other compartments that have been studied with
[I 91481
The second main area of interest is the immunogenetics of the anti-Fprotein response in the mouse. There are believed to be both immune response and immune suppression gene effects operating to control this response [8,9], and in particular the MHC class I1 A molecule of the k haplotype is required in order to mount an immune response to either allelic form (allomorph) of the protein. A prediction from the differing T and B cell reactivities towards F protein outlined above is that the two F protein allomorphs must obviously be different, and that this difference will define, either directly or indirectly, a Tcell epitope. Progress in this area would clearly be greatly aided by a knowledge of the primary sequences of the two mouse F protein alleles. Following the publication of the cDNA sequence encoding the rat Fprotein [lo], and given the known conservation of the protein, we attempted the reverse transcriptase polymerase chain reaction (RT-PCR) amplification of mouse cDNA using primers derived from the rat sequence. Peptides where then synthesized based on the mouse cDNA sequences to test the prediction concerning the Tcell epitope.
MRC Training Fellow. Supported by a grant from Smith Kline Beecham. 0 Lister Institute Research Fellow.
2 Materials and methods
Correspondence: David B. G. Oliveira, Department of Medicine,
2.1 Animals and cDNA preparation
Addenbrooke’s Hospital, Hills Road, Cambridge CB22, GB Abbreviation: PCR Polymerase chain reaction 0 VCH Verlagsgesellschaft mbH, D-6940 Weinheim, 1991
Mice of F1 (CBA) and F2 type (BALB/c) were obtained from Olac (Bicester, GB). Liver total RNA was isolated 0014-2980/91/0505-1235$3.50+.25/0
1236
J. P. Schofield, R. KVijayakumar and D. B. G. Oliveira
[ 111 and first-strand cDNA reverse transcribed [12] using oligo(dT) primer, without prior purification of poly(A)+ mRNA. 2.2 PCR and DNA sequencing About 250-500 ng of first strand cDNA was input into a 50 p1 PCR. Amplification primers were derived from the published rat F protein cDNA sequence and modified at their 5’ ends by incorporating a restriction endonuclease cleavage recognition site to facilitate “sticky-end’’ cloning:
Eur. J. Immunol. 1991. 21: 1235-1240 F1 protein gene 5‘-AAGACTGCAACCACATTGT-3‘
F2 protein gene 5’-AAGACTGCGACCACATTGT-3’ The probes were end-labeled with [Y-~~P]ATP (> 3000 Ci/ mmol = 111TBq/mmol, Amersham) using T4 polynucleotide kinase [15] and purified on 0.5 ml DE52 columns (Whatman, Maidstone, GB). Hybridization was at 50°C for 4-8 h, followed by 2 x 15-min washes at 60°C in 6 x SSC.The filters were then exposed for 4-24 h (depending on bound radioactivity).
2.4 2-D gel electrophoresis Forward primer 5 ‘-ACTGTCGACTACTGGGACA AAGGACCAAAG-3’ Sal I Reverse primer 5’-ACTCTGCAGACACCGTGGTCTCCAGGTCA-3’ Pst I The reaction mixture was overlaid with 50 p1 light mineral oil (Sigma, Poole, GB) and amplified for 30 cycles on a Techne PHC-1 programmable thermocycler according to published protocols [13, 141.The PCR temperature profile was: 95 “C 30 s denaturation, 58°C 30 s annealing, and 72°C 1 min extension. A 5-pl aliquot from the PCR was checked for purity and yield by agarose gel electrophoresis, and the remainder was gel-purified and extracted onto glassmilk (GenecleanTM, Bio-101, La Jolla, CA). The purified PCR product was digested with SalI and PstI restriction enzymes, phenol extracted, and ethanol precipitated [15]. Ligation into prepared cloning vector pBluescript-KSf (Stratagene, Cambridge, GB) was performed overnight at 15“C, followed by transformation into E. cofi NM522 frozen competent cells [16].White colonies were screened for insert by PCR [ 171, and Qiagen (Diagen, Dusseldorf, FRG) purified DNA was sequenced on both strands by dideoxy-chain termination [18] with oligo-walking [19].
2.3 Allele-specific oligonucleotide dot-blotting The following PCR primers were used to amplify a 159-bp fragment encompassing the allovariable site: Forward primer 5’-CACCTGGTGAAGCATGG-3‘ Reverse primer 5’-CTGCAGCACAGCGAACT-3’ cDNA was prepared from the livers of a variety of mouse strains of known Fprotein type [20]. cDNA used to determine the sequences in Sect. 2.2 was also included to verify probe specificity. PCR conditions were as described above. The PCR products were dot-blotted onto two duplicate filters (Hybond N, Amersham Int., Amersham, GB) via a Hybri-Dot manifold (BRL, Gaithersburg, MD). Following prehybridization, one of the filters was hybridized with an oligonucleotide probe complementary to the allovariable site of the F1 protein gene, and the other with a similar probe for the F2 protein gene.The probe sequences were:
Both F1 and F2 proteins were affinity purified from mouse liver extracts (CBA and BALB/c, respectively) using an anti-F protein mAb affinity column as previously described [21]. 2-D gel electrophoresis was based on published protocols [22]. The first IEF dimension incorporated 6% of ampholyte 5-8 and 1.2% of ampholyte 3.5-10 (LKB, Cambridge, GB). Running buffers were 5% 1,2-diaminoethane (cathode) and 5% orthophosphoric acid (anode). The second dimension was run in a discontinuous buffer system [23] with 12% acrylamide in the resolving gel. Proteins were visualized by staining with Coomassie blue.
2.5 Peptide synthesis A series of peptides (15-21 residues long, overlapping by 6 residues) encompassing the entire F protein sequence (additional peptide included in series to provide the alternative at the allovariable site) was synthesized as described by Houghton et al. [24]. Briefly, standard Boc (t-butoxycarbonyl) amino acid resin (170 mg) was sealed in polypropylene mesh bags for simultaneous multiple peptide synthesis. Standard de-protecting, neutralization, coupling and wash protocols were employed [25,26]. To prevent acylation and racemization of histidine residues during coupling DNP-protected histidine was used. DNP was removed at the end of synthesis by treatment with 1% (v/v) thiophenol in dimethyl formamide.The peptides were cleaved simultaneously from the resin using a conventional HF procedure [25]. Cleaved peptides were extracted from the bags in 5%-30% acetic acid, lyophilized and dissolved in either PBS or DMSO at 5 mg/ml, according to their solubililty. DMSO-soluble peptides were used in proliferation assays at a final DMSO concentration of < 1%. Purity of peptides was assessed by reverse phase HPLC using a Novapak C18 column (Waters Chromoatgraphy, Watford, GB) running in a gradient from water to 50% acetonitrile (in 0.02% triethylamine, buffered to pH 6.8 with acetic acid). Peptides with a purity > 70% were used in experiments.
2.6 T cell proliferation assay Mice of F1 type (CBA) were primed with 100 pg F2 protein emulsified with an equal volume of CFA (Difco, Detroit, MI); the total volume (200 pl) was divided equally between both rear footpads and both sides of the base of the tail.
Eur. J. Immunol. 1991. 21: 1235-1240 Figure 1. Rat Fand mouse F1 protein cDNA and predicted amino acid sequences (rat data from [lo]).The underlined bases represent the PCR primers. Shown in bold is the site of the difference between the two mouse F protein alleles (see Fig. 2).
After 7-10 days draining LN were removed and a singlecell suspension prepared. Cells (2 x 105-5 x 105) were cultured in 200 p1 of Iscove's modified Dulbecco's medium (Gibco, Paisley, Scotland) with the addition of 10% FCS in a 5% COz atmosphere at 37 "C. Additions of proteins and peptides were at the start of the culture period which was for a total of 96 h. [3H]dThd (1 pCi) was added to each well for the last 8 h of culture. Cells were harvested using a Titertek cell harvester and incorporated [3H]dThd measured in a Packard Tri-Carb liquid scintillation analyzer. All cultures were in triplicate and the results are expressed as the mean k SEM.
3 Results 3.1 Mouse F protein allele cDNA sequence A comparison between rat Fand mouse F1 protein cDNA, together with the predicted primary amino acid sequences, is shown in Fig. 1. There is 92% identity between rat and mouse Fprotein cDNA, and 97% identity between the predicted amino acid sequences. The mouse F protein sequence has an apparent deletion of the first base of the termination codon (not shown in Fig. 1). This type of deletion has been noted before (unpublished observations) and is believed to represent a PCR artefact. The mouse F protein cDNA are identical apart from a single nucleotide substitution which predicts an asparagine residue in F1 protein and an aspartate residue in F2 protein (Fig. 2). The predicted molecular mass is 43 kDa for the mouse F proteins. Apart from the expected match with rat F protein, no other significant matches were found in the EMBL or Genbank databases.
3.2 Allele-specific oligonucleotide dot-blotting Primers based on the sequences of the mouse Fprotein alleles were used to successfully amplify a 159-bp region centered on the single base difference found between the two alleles. The resultant dot blot is shown in Fig. 3, demonstrating concordance between the known F type of the various strains and the nucleotide difference. 3.3 2D gel electrophoresis The nucleotide difference between the F1 and F2 protein genes predicts that F2 protein (with an aspartate residue) should be more negatively charged than F1 protein (with an asparagine residue). Fig. 4 shows that although a number of spots are visible for each form of Fprotein, F1 protein migrates in a more basic position than F2 protein.
F protein sequence
1237
J. P. Schofield, R. K.Vijayakumar and D. B. G. Oliveira
1238
Eur. J. Immunol. 1991.21: 1235-1240
I
A-A
2l-mrp.ptidefmmF2
W-W
21-mwp.pthfmmFl
0
0
n 1
i J
1
U
-
F1
F2
10 29 I ml
12
I 6
3
1.5
paptlde concmtmtionpq/ ml)
Figure 5. Proliferation assay of Tcells primed to F2 protein (see Sect. 2.6 for details). ( 0 ) Represents background proliferation (note that the scale starts at 1 x 104cpm). Points are mean of IIIIIIIIIIII IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII triplicates f SEM. ; l-r
V E D C N H I V Q K A R E R G A K I V a l G l ~ iI lse V a l G l n L y s A l a A r g G l u I l e Muse F1 Muse F2
V a l c l ~ ~I l i e sV a l G l n L y s A l a A r g G l I l e V E D C D H I V Q K A R E R G A K I
Figure2. Fragment of the sequencing gel showing the only identified difference between mouse F1 and F2 protein cDNA, and the resulting predicted amino acid sequence.
3.4 T cell proliferation assay
A synthetic peptide from the F2 protein molecule, centered on the allovariable site, could successfully stimulate Tcells primed against the whole F2 protein molecule; these cells did not proliferate to whole F1 protein or to a corresponding synthetic peptide centered on the allovariable site (Fig. 5). A series of peptides covering the entire F2 protein sequence was also tested; no other stimulatory peptides were detected despite using each peptide over a concentration range of 1-20 pg/ml (data not shown).
4 Discussion Figure 3. Allele-specific oligonucleotide dot blot of amplified fragments of F protein cDNA centered on the allovariable site. Fragments were amplified from the cDNA used to obtain the original sequences (labeled F1 and F2), and from previously typed strains of mice [20]: F1 type CBA, DBAR; F2 type: BALBk,
C57BL/6. OVA
-
+ F1
F2
F1 and F2 Figure 4 . 2-D electrophoresis of immunopurified F1 protein (top panel), F2 protein (middle panel), and a mixture of the two (bottom panel: same amounts of protein as in the above panels). IEFdimension horizontal, SDS-PAGE dimension vertical. Marker protein on the left is OVA (45 kDa).
The principal results of this work are (a) the amplification and sequencing of the cDNA encoding the two mouse F proteins; (b) evidence that the identified sequence difference between the two cDNA does indeed constitute the allelic difference between the two F protein genes and (c) evidence that the predicted difference in primary amino acid sequence defines a T cell epitope. Our use of the PCR for the amplification of the mouse F protein cDNA introduces the possibility of amplification artefacts in the resultant sequences. Taq polymerase has an error rate of approximately nucleotide/cycle [27], and any such error will of course be duplicated in all subsequent amplification steps. Although only one clone was sequenced for each F protein cDNA, the existence of two alleles allows a cross-check against such artefacts. The fact that the independent primary sequences were identical, apart from a single base (discussed below), provides strong confirmation that the sequences are correct. The 21 5' bases of the sequences represent the PCR primer and are obviously the same for all three cDNA. However, there is clearly a high degree of sequence similarity between rat and mouse F protein cDNA. Most of the differences are third base position changes which make no difference to the predicted amino acid sequence, and many of the predicted differences are conserved substitutions. As with the rat F protein cDNA the intitiation codon for the two mouse
F protein sequence
Eur. J. Immunol. 1991. 21: 1235-1240
cDNA has not been located, but the close agreement between predicted and measured molecular weights suggest that there is very little missing 5' coding sequence. It was important to obtain independent corroboration of the single base difference found between the two mouse Fprotein cDNA: it was possible that this represented a PCR artefact, and that the true allelic difference lay in the short missing 5' coding sequence. The first piece of confirming evidence is provided by the dot blot data. The specific hybridization to six independently amplified cDNA fragments found with probes differing by only one base argues strongly against a PCR artefact. The observed correlation between the base difference and known F protein type also provides evidence that this nucleotide difference is the source of allelic variation in the F protein system, as opposed to an irrelevant mutation peculiar to the initial strains used for cDNA amplification. The identified nucleotide difference between the two mouse F protein cDNA does not exclude other allelic differences present at the 5' or 3' ends (non-coding region in the latter case), either in the region of the PCR primers or beyond. The two further pieces of confirmatory evidence, IEF and T cell epitope analysis, are therefore particularly useful, as they also demonstrate that known differences between the F protein alleles at the protein level can be accounted for by the single amino acid substitution. Previous work on IEF of F proteins revealed a rather complex pattern, with mul$ple bands detectable on immunoblotting of focused crude liver extracts [20]. However, it was found that a number of bands migrating in the basic PI region were present in F1 type liver extracts, but absent in €2type liver extracts. We also found that the picture is complicated by the presence of a number of spots, but confirmed that F1 protein does indeed migrate in a more basic position than F2 protein. This is the difference expected from the predicted amino acid substitution. The differently charged species found for each of the allomorphs may be due to differential phosphorylation or possibly in vitro de-amidation of asparagine residues, but further work is required to confirm or refute these speculations. The explanation for the much more complex picture found with crude extracts is also unclear. We used affinity-purified proteins for our focusing, whereas the previous study used crude liver extracts run under nonreducing conditions. We have also found that Western blotting (with anti-F protein antisera) of SDS-polyacrylamide gels of crude liver extracts run under nonreducing conditions reveals a number of high molecular weight bands (unpublished data). It is therefore possible that F protein may associate with other proteins (or itself) under some conditions, explaining these more complex patterns. This property may also explain the anomalously high molecular weights reported for F protein (typically about twice the weight of the monomeric protein) in some earlier studies [28,29]. Although it is clear that Tcells are capable of distinguishing between the two allelic forms of Fprotein [5] it did not follow that any amino acid difference between the two would necessarily define aTcell epitope directly.There are examples in other systems where variations at one point in a protein structure can influence the presentation of epitopes
1239
at another point relatively distant in the primary sequence, perhaps due to differences in processing [30]. Other mechanisms for this effect have also been proposed, such as inhibitory interactions between the portion of a peptide lying outside the MHC molecule and the portion actually in the MHC molecule groove [31]. It was therefore not inconceivable that the immunodominant T cell epitope(s) would lie in a portion of the molecule common to the two F protein allomorphs, with the allelic difference determining whether this would be presented successfully. However, there was a strongapriori argument suggesting that this was unlikely, namely that both allomorphs of F protein can be presented successfully (as judged by their immunogenicity) and both exhibit the same very restricted immune response gene dependency, that is a requirement for the MHC class I1 Ak molecule [9].The simplest explanation for this is that the same portion of each allomorph is involved in each case, accounting for the identical MHC molecule binding properties, with the difference at the allovariable site allowing self-nonself discrimination. Our data indeed provide evidence for the straight forward situation in which the allovariable site directly defines a T cell epitope. This epitope does not bear any obvious features in common with other epitopes known to be presented by the Ak molecule [32], and is distinct from the candidate epitope predicted on this basis from the rat Fprotein gene cDNA sequence [lo]. The argument outlined above concerning the ability of both forms of F protein to be successfully presented implies that nonresponsiveness to the self form of the protein in responder strains must be due to creation of the relevant hole in theTcell repertoire during the establishment of self tolerance. This is not particularly surprising, but the nature of nonresponsiveness in nonresponder (non-Ak-bearing) strains is less clear. One possibility is that other class I1 MHC molecules are.unable to bind the allovariable peptide. Another possibility is that such molecules, although capable of binding this peptide, do so in such a way that the resultant MHC-antigen complex appears the same toTcells irrespective of the allomorph of F protein involved. Alternatively, the Tcell repertoire produced in the presence of non-Ak molecules may lack the ability to discriminate between the bound epitopes of the two allomorphs. Further work is required to distinguish between these possibilities, and to determine the features of the epitope that are required for successful interaction with the Ak molecule and the T cell receptor. Finally, we are no closer to determining the function of F protein. The lack of sequence similarity with any other known cDNA or protein sequence is disappointing. It is possible that further clues may arise from studies on the human F protein. We have recently screened a human genomic DNA cosmid library using a randomly primed 1 kb mouse F1 probe, and are in the process of characterizing a strongly positive clone. However, it is perhaps more likely that we will continue to increase our understanding of the immunology of F protein while remaining ignorant of its function. We are indebted to Karen Worfreys for expert technical assistance.
Received December 20, 1990.
1240
J. P. Schofield, R. K.Vijayakumar and D. B. G. Oliveira
5 References 1 Fravi, G. and Lindenmann, J., Nature 1968. 218: 141. 2 Griffiths, J. A. and Oliveira, D. B. G., Scand. J. lmmunol. 1988. 27: 357. 3 Grewal, G. K. and Oliveira, D. B. G., Clin. Chim. Acta 1990. 191: 93. 4 Oliveira, D. B. G. and Vindlacheruvu, S., Comp. Biochem. Physiol. ( B ) 1987. 87: 87. 5 Mitchison, N. A., Clin. lmmunol. Newsl. 1985. 6: 12. 6 Winchester, G., Sunshine, G. H., Nardi, N. and Mitchison, N. A., lmmunogenetics 1984. 19: 487. 7 Lukic, M. L. and Mitchison, N. A., Eur. J. Immunol. 1984.14: 766. 8 Oliveira, D. B. G., Blackwell, N..Virchis, A. E. and Axelrod, R. A., Immunogenetics 1985. 22: 169. 9 Oliveira, D. B. G. and Nardi, N. B.. Immunogenetics 1987.26: 359. 10 Gerswhin, M. E., Coppel, R. L., Bearer, E., Peterson, M. G., Sturgess, A. and Mackay, I. R., J. lmmunol. 1987. 139: 3828. 11 Chirgwin, J. M., Przybyla, A. E., MacDonald, R. J. and Rutter, W. J., Biochemistry 1979. 18: 5294. 12 Gubler, U. and Hoffman, B. J., Gene 1983. 25: 263. 13 Saiki, R. K., Gelfand, D. H., Stoffel, S., Scharf, S. J., Higuchi, R., Horn, G. T. and Erlich, H. A., Science 1988. 239: 487. 14 Rappolee, D. A., Mark, D., Banda, M. J. and Werb, Z., Science 1988. 241: 708. 15 Manitis,T., Fritsch, E. and Sambrook J., Molecular cloning: a laboratory manual, 2nd Edit., Cold Spring Harbor Laboratory Press, New York 1990. 16 Hanahan, D., J. Mol. Biol. 1983. 166: 557. 17 Schofield, J. €?, Vaudin, M., Kettle, S. and Jones, D. S. C., Nucleic Acids Res. 1989. 17: 9498. 18 Sanger, F., Nicklens, S. and Coulson, A. R., Proc. Natl. Acad. Sci. USA 1977. 74: 5463. 19 Bankier, A.T. and Barrell, B. G., in Howe, C. J. and Ward, E. S. (Eds.), Nucleic acids sequencing, a practial approach, IRL Press, Oxford 1990.
Eur. J. Immunol. 1991. 21: 1235-1240 20 Hayakawa, J. and Nikaido, H., lmmunogenetics 1987. 26: 366. 21 Wedderburn, L., Lukic, M. L., Edwards, S., Kahan, M. C., Nardi, N. and Mitchison, N. A., Mol. Immunol. 1984. 21: 979. 22 OFarrell, F! H., J. Biol. Chem. 1975. 250: 4007. 23 Laemmli, U. K., Nature 1970. 227: 680. 24 Houghton, R. A., Proc. Natl. Acad. Sci. USA 1985. 82: 5131. 25 Houghton, R. A., Chang, W. C. and Li, C. H., lnt. J. Pept. Protein Res. 1980. 16: 311. 26 Houghton, R. A., Ostrech, J. M. and Klipstein, F. A., Eur. J. Biochem. 1984. 145: 157. 27 Tindall, K. R. and Kunkel, T. A., Biochemistry 1988. 27: 6008. 28 Mihas, A. A., Hirschowitz, B. I. and Saccomani, G., J. lmmunol. 1976. 116: 1228. 29 Mori,Y., Iesato, K., Ueda, S.,Wakashin,Y., Wakashin, W. and Okuda. K., Clin. Chim. Acta 1981. 112: 125. 30 Shastri, N., Miller, A. and Sercarz, E. E., J. Immunol. 1986. 136: 371. 31 Vacchio, M. S., Berzofsky, J. A., Krzych, U., Smith, J. A., Hodes, R. J. and Finnegan, A., J. lmmunol. 1989. 143: 2814. 32 Buus, S . , Sette, A., Colon, S. M., Miles, C. and Grey, H. M., Science 1987. 235: 1353. Noteaddedinproof: We haverecentlylearnt that ageneencodinga protein homologous to rat and mouse F protein has been cloned from the ciliated protozoan Tetrahymena therrnophila (K. Kristiansen, personal communication; submitted for publication). Similarities at the 3' end of the mouse and Tetrahymena genes strongly suggest that the apparent deletion of the first base of the termination codon of the mouse genes (Sect. 3.1) is real, and that the coding region for the mouse genes extends into the 3' PCR primer. Received March 23, 1991.
