1229
Analysis of Interstrain Variation in Cytomegalovirus Glycoprotein B Sequences Encoding Neutralization-Related Epitopes Sunwen Chou and Karen M. Dennison
Medical and Research Services, VA Medical Center, and Division of Infectious Diseases, Oregon Health Sciences University, Portland
Host immunity to human cytomegalovirus (CMV) infection is directed at numerousviral proteinsand glycoproteins [1]. The specific viral targets are being characterized in increasing detail through genetic analysis and use of murine and humanmonoclonal antibodies. CMVenvelope glycoproteins are of special interest because monoclonal antibodies withvirus-neutralizing activityrecognize the major envelope glycoprotein gp130/55 (gB) [2-4] and other envelope glycoproteins [5, 6]. gB is formed as a glycosylated precursor of 130-160 kDa [7-9] that is cleaved by cellular enzymes at a specific site to form gp55 (encoded by the 3'-region of the gB gene), which is incorporated into the viral envelope [10, 11]. Epitopesrecognized by neutralizing murinemonoclonal antibodies andhuman antiserahave beenmapped to theaminoterminal part of gp55 [3, 4, 10]. The role of gB in protectiveimmunity needs further study regarding the typeof response(e.g., neutralization, cytotoxic T cells, antibody-dependent cytotoxicity) and the viral epitopes involved. The latter include the important issue of whether strainvariation affects immune recognition. Although restriction enzyme analyses have showngenetic variation in every region of the CMV genome [12], including gB [13], there is less information on the functional consequences of such variation. Published sequence data of the laboratory strainsAD169 and Towne indicatethere are someaminoacid sequence differences between thesestrains[10]. An unresolved question is the extentofvariation ofgB among clinical strains.
Received 15 November 1990; revised 28 January 1991. Financial support: Department of Veterans Affairs research funds. Reprints orcorrespondence: Dr. Sunwen Chou, VA Medical Center 111F, P.O. Box 1034, Portland, OR 972CJ7. The Journal of Infectious Diseases 1991;163:1229-1234 © 1991 by The University of Chicago. All rightsreserved.
0022-1899/91/6306-0009$01.00
To explore thisissuefurther, recentclinical isolates ofCMV were examined by polymerase chain reaction (PCR) amplification of the gp55 region of gB and sequence analysis of the amplified products. By comparing homologous sequences of gp55 among strains, an effortwas made to discover the effect of genetic variation on peptide encoding in the vicinity of neutralization-related epitopes.
Materials and Methods Strains. Twelve CMV strainswereselectedfor sequencing from clinical isolatesshed by transplant recipientsat our medicalcenter overthepast5 years. Thesestrainswereisolatedfromurine or blood specimens ofepidemiologically unrelated individuals andwereshown to be genetically distinct by restriction enzyme digest analysis of variousregionsof the viral genome [13, 14]. Virus was culturedin human fibroblasts and DNA was extracted for analysis within 12 passages as previously described [14]. To avoid possible artifacts arising from additional passagein cell culture, strains were generally not plaque-purified. Clinical isolates shownby restrictionenzymeanalysis to contain mixtures of strainswereexcluded. However, two of the strains used in this study (C325 and C327)were plaquepurifiedfrom a mixtureof the strains shed in the same urine specimen [15]. Two other strains(C326and C330) wereshedat different times by the same individual. Oligonucleotide primers and DNA amplification. Target DNA sequences for use as sequencing templateswere amplified fromextracted viral DNA or its previous amplification products. Primers for amplification and sequencing were 17- to 21-base oligonucleotides that were selected from the published gB sequence of CMV strain AD169, supplemented by information on variant sequences accumulated over the course of this study. Oligonucleotides were synthesized using standardsolid-phase cyanoethyl phosphoramidite chemistryon an automatedsynthesizer and were partly purifiedby size-exclusion chromatography. Primersthat provided successful sequencingtemplates and reactionsfroma varietyof strainsare listed in table 1. Manycombinations of forward and reverseprimers were
Downloaded from http://jid.oxfordjournals.org/ at University of Aberdeen on February 15, 2015
Nucleotide sequences ofa part ofthe envelope glycoprotein B(gB) gene ofhumancytomegalovirus(CMV), encoding epitopes recognized byvirus-neutralizing monoclonal antibodies, were determined for 12distinct clinical strainsofCMV afteramplification ofsuitable templates using the polymerase chainreaction. Sequence analysis of this region (codons 384-717) revealed that the clinical strainsand previously sequenced laboratory strains Towne and AD169 belong to one offourvariantgroups, eachwitha characteristic nucleotide and peptide sequence. Peptide homology was >99% forstrainswithin a group, and varied from 91% to 98% forstrainsin different groups. Variation was most frequent between codons 448and 480. ThegBgroup ofa CMV strain could bedetermined by restriction analysis ofa small target sequence amplified from viral genomic DNA, and an additional 28clinical strains were grouped in thismanner. Theexistence ofa limited number ofvariants ofgBamong clinical strainsfacilitates analysis ofbiologic function and crossreactivity of immune responses.
1230
Chou & Dennison
JID 1991;163 (June)
Table 1. Amplification and sequencing primers in cytomegalovirus gB sequences.
Forward (coding) strand
Reverse (noncoding) strand)
Name
Sequence
Name
Sequence
gB3 gBl71* gB178 gBI043 gB1170 gB1246 gB1319 gB1438 gB1615 gB1795 gB1987
GCGACGAACATGGAATCC CTGGTGCCTGGTAGTCTG CTGGTAGTCTGCGTTAAC TCTGGGAAGCCTCGGAACG CTGCGTACGTGATGAGGC GGAAACGTGTCCGTCTT TGGAACTGGAACGTTTGGC TACGCCCAGCTGCAGTT CGTTTCATGGGTGATGT GAAATCCTGTTGGGCAA CTGGAACTTTACTCGCA
gB1529 gBI604 gBI724 gB1868 gB2058 gB2534* gB2857 gBZ
CGTTGATCCACACACCAGGC GAAACGCGCGGCAATCGG GAGTAGCAGCGTCCTGGCGA GAGTTCCCGGCGATGAAGAT GAATTCGCGCATGATCTC TAAGCCTGCTCGTTGGTGTA CGGAGTCTTTCAAGTGTC GGTTCAGACGTTCTCTTC
used, but most of the templates were 2.4- to 2.6-kb long and thus spannedmostof gB.Conditionsfor PCR amplification were as previously described [13]. PeR products wereusedforsequencing without subcloning. Sequencing strategy. Preliminary experiments indicated that single-stranded sequencing templates werenecessary for reasonable productivity. Two approaches for generating such templates were used. One wasto do a "single-sided" amplification from previously amplified material, using one primer annealingto a site internalto the originalamplification primers. This wasaccomplished withTaq polymeraseduring 30 cyclesof denaturation (94°C for 1 min), annealing(55"Cfor 2 min), andextension(nOC for2 min). The reaction mixturewas then ethanol precipitated, dried, and dissolved in 10 ILl of sequenase buffer(see below) containing 10ng of an appropriate sequencing primer derived from the complementary strand. Primer annealing wasconductedin a heat block preheatedto 65°C and cooledto 37°C over 30 min. This approachwasoften satisfactory but wasnot optimal in providing readablesequences. Wethereforeadopted a second approach, modified froma published technique [16]. PCR amplification for templateproductionwasdone as usual using forward and reverse primers, except that one of the primers was biotinylated at its 5' end through use of an amino-linker phosphoramidite(Aminolink 2, AppliedBiosystems, Foster City, CA), and a biotin N-hydroxysuccinimide ester (Enzotin; Enzo Biochem, New York). After PCR, the reaction mixture was agitated for 45 min at 37°C with streptavidin-coated magneticbeads (Dynabeads M280; Dynal, Oslo) to bind the biotinylated DNA strands generated during PCR. The complementary strand was washed off the solidphase with alkali as described [16]. After a neutralizing rinse (1 M NaCl, 50 mM TRIS, pH 7.4), the magnetic beads were resuspended in 20 ILl of the same buffer, containing 20 ng!ILl of the sequencing primer, and agitated at 37"C for 30 min to anneal. Excess primer waswashed away, and a prealiquotedmixtureof buffer, nucleotides, and polymerasewas added to the beads to initiatethe sequencing reactions. Sequencing reactions weredonebythe dideoxy chain-termination techniqueusing mutated T7 polymerase (Sequenase version2; US Biochem, Cleveland), according to the instructions furnished by the vendorand as published [17]. P5S]deoxyadenosine triphosphate was used as the label. Completed sequencing reactions were analyzed
on O.4-mm thick 6% polyacrylamide gels. Sequences from 50 to 300 bases downstream from each primer were best resolvedon sequencing gelsand weretheprimarysourceof data. Information from furtherdownstream wasusedprimarilyto confirma readingderived froma subsequent primer.Variant or indeterminate sequences noted on individual experiments wereclarified by sequencing both strands andrepeated analyses. Compressed bandsobscuring thetruesequence werenotespecially frequent intheregion sequenced butwereresolved by resequencing usingdeoxyinosine triphosphate in placeof deoxyguanosine triphosphate. Comparisons and alignments. By use of multiple sequencing primers spaced 100-300bp apart,thegBcoding sequence frombases 1150-2151 (codons 384-717, Towne strain numbering) was determined for each of 12 clinical strains. This part of gB includes the cleavage siteofgp55(codon461),the neutralization-related epitopes recognized bymurinemonoclonal antibodies (codons 461-680),and extends into the transmembrane region (codon 715). Sequence alignments were made with strains ADl69 and Towne (reference sequences) and among the clinical strains. Restriction enzyme analysisfor strain grouping. The 293-to 296bp target(sizevariesby strain)amplified byprimersgB1319/gB1604 (table 1) fromextracted viral DNAwasdigested with restrictionenzymeHinfl and withRsaI. Results ofthisanalysis wereusedto identify the sequence variant to which a given strain belonged. Apart fromstrainsselected for sequencing, an additional 28 randomCMV isolatescollectedin the last 6 years from our localorgan transplant recipients were studied.
Results
Readability ofsequences derivedfrom peR-amplifiedtemplates. Direct sequencing of PCRproducts, without subcloning, gave clearandconsistent data(figure 1), despiterepeated PCR amplification of target sequences from the same CMV DNAextractand the use of multiple forward and reverse sequencing primers. As data weregenerated from the project, additional primersweresynthesized as needed to avoid priming within regions of sequence variability. In our experience,
Downloaded from http://jid.oxfordjournals.org/ at University of Aberdeen on February 15, 2015
NOTE. All primer sequences are shown in 5' to 3' direction and are based on published sequences for strain ADl69 [2], except that one base change was made in gB13l9 to reduce mismatch with variant strains. * 5'-biotinylated primers used to make sequencing templates.
Sequence Analysis of CMV gp55 Variation
JID 1991;163 (June)
.= ------ ...-ACGT
........---It::
~
..::-=:: .....-
Analysis of Interstrain Variation in Cytomegalovirus Glycoprotein B Sequences Encoding Neutralization-Related Epitopes Sunwen Chou and Karen M. Dennison
Medical and Research Services, VA Medical Center, and Division of Infectious Diseases, Oregon Health Sciences University, Portland
Host immunity to human cytomegalovirus (CMV) infection is directed at numerousviral proteinsand glycoproteins [1]. The specific viral targets are being characterized in increasing detail through genetic analysis and use of murine and humanmonoclonal antibodies. CMVenvelope glycoproteins are of special interest because monoclonal antibodies withvirus-neutralizing activityrecognize the major envelope glycoprotein gp130/55 (gB) [2-4] and other envelope glycoproteins [5, 6]. gB is formed as a glycosylated precursor of 130-160 kDa [7-9] that is cleaved by cellular enzymes at a specific site to form gp55 (encoded by the 3'-region of the gB gene), which is incorporated into the viral envelope [10, 11]. Epitopesrecognized by neutralizing murinemonoclonal antibodies andhuman antiserahave beenmapped to theaminoterminal part of gp55 [3, 4, 10]. The role of gB in protectiveimmunity needs further study regarding the typeof response(e.g., neutralization, cytotoxic T cells, antibody-dependent cytotoxicity) and the viral epitopes involved. The latter include the important issue of whether strainvariation affects immune recognition. Although restriction enzyme analyses have showngenetic variation in every region of the CMV genome [12], including gB [13], there is less information on the functional consequences of such variation. Published sequence data of the laboratory strainsAD169 and Towne indicatethere are someaminoacid sequence differences between thesestrains[10]. An unresolved question is the extentofvariation ofgB among clinical strains.
Received 15 November 1990; revised 28 January 1991. Financial support: Department of Veterans Affairs research funds. Reprints orcorrespondence: Dr. Sunwen Chou, VA Medical Center 111F, P.O. Box 1034, Portland, OR 972CJ7. The Journal of Infectious Diseases 1991;163:1229-1234 © 1991 by The University of Chicago. All rightsreserved.
0022-1899/91/6306-0009$01.00
To explore thisissuefurther, recentclinical isolates ofCMV were examined by polymerase chain reaction (PCR) amplification of the gp55 region of gB and sequence analysis of the amplified products. By comparing homologous sequences of gp55 among strains, an effortwas made to discover the effect of genetic variation on peptide encoding in the vicinity of neutralization-related epitopes.
Materials and Methods Strains. Twelve CMV strainswereselectedfor sequencing from clinical isolatesshed by transplant recipientsat our medicalcenter overthepast5 years. Thesestrainswereisolatedfromurine or blood specimens ofepidemiologically unrelated individuals andwereshown to be genetically distinct by restriction enzyme digest analysis of variousregionsof the viral genome [13, 14]. Virus was culturedin human fibroblasts and DNA was extracted for analysis within 12 passages as previously described [14]. To avoid possible artifacts arising from additional passagein cell culture, strains were generally not plaque-purified. Clinical isolates shownby restrictionenzymeanalysis to contain mixtures of strainswereexcluded. However, two of the strains used in this study (C325 and C327)were plaquepurifiedfrom a mixtureof the strains shed in the same urine specimen [15]. Two other strains(C326and C330) wereshedat different times by the same individual. Oligonucleotide primers and DNA amplification. Target DNA sequences for use as sequencing templateswere amplified fromextracted viral DNA or its previous amplification products. Primers for amplification and sequencing were 17- to 21-base oligonucleotides that were selected from the published gB sequence of CMV strain AD169, supplemented by information on variant sequences accumulated over the course of this study. Oligonucleotides were synthesized using standardsolid-phase cyanoethyl phosphoramidite chemistryon an automatedsynthesizer and were partly purifiedby size-exclusion chromatography. Primersthat provided successful sequencingtemplates and reactionsfroma varietyof strainsare listed in table 1. Manycombinations of forward and reverseprimers were
Downloaded from http://jid.oxfordjournals.org/ at University of Aberdeen on February 15, 2015
Nucleotide sequences ofa part ofthe envelope glycoprotein B(gB) gene ofhumancytomegalovirus(CMV), encoding epitopes recognized byvirus-neutralizing monoclonal antibodies, were determined for 12distinct clinical strainsofCMV afteramplification ofsuitable templates using the polymerase chainreaction. Sequence analysis of this region (codons 384-717) revealed that the clinical strainsand previously sequenced laboratory strains Towne and AD169 belong to one offourvariantgroups, eachwitha characteristic nucleotide and peptide sequence. Peptide homology was >99% forstrainswithin a group, and varied from 91% to 98% forstrainsin different groups. Variation was most frequent between codons 448and 480. ThegBgroup ofa CMV strain could bedetermined by restriction analysis ofa small target sequence amplified from viral genomic DNA, and an additional 28clinical strains were grouped in thismanner. Theexistence ofa limited number ofvariants ofgBamong clinical strainsfacilitates analysis ofbiologic function and crossreactivity of immune responses.
1230
Chou & Dennison
JID 1991;163 (June)
Table 1. Amplification and sequencing primers in cytomegalovirus gB sequences.
Forward (coding) strand
Reverse (noncoding) strand)
Name
Sequence
Name
Sequence
gB3 gBl71* gB178 gBI043 gB1170 gB1246 gB1319 gB1438 gB1615 gB1795 gB1987
GCGACGAACATGGAATCC CTGGTGCCTGGTAGTCTG CTGGTAGTCTGCGTTAAC TCTGGGAAGCCTCGGAACG CTGCGTACGTGATGAGGC GGAAACGTGTCCGTCTT TGGAACTGGAACGTTTGGC TACGCCCAGCTGCAGTT CGTTTCATGGGTGATGT GAAATCCTGTTGGGCAA CTGGAACTTTACTCGCA
gB1529 gBI604 gBI724 gB1868 gB2058 gB2534* gB2857 gBZ
CGTTGATCCACACACCAGGC GAAACGCGCGGCAATCGG GAGTAGCAGCGTCCTGGCGA GAGTTCCCGGCGATGAAGAT GAATTCGCGCATGATCTC TAAGCCTGCTCGTTGGTGTA CGGAGTCTTTCAAGTGTC GGTTCAGACGTTCTCTTC
used, but most of the templates were 2.4- to 2.6-kb long and thus spannedmostof gB.Conditionsfor PCR amplification were as previously described [13]. PeR products wereusedforsequencing without subcloning. Sequencing strategy. Preliminary experiments indicated that single-stranded sequencing templates werenecessary for reasonable productivity. Two approaches for generating such templates were used. One wasto do a "single-sided" amplification from previously amplified material, using one primer annealingto a site internalto the originalamplification primers. This wasaccomplished withTaq polymeraseduring 30 cyclesof denaturation (94°C for 1 min), annealing(55"Cfor 2 min), andextension(nOC for2 min). The reaction mixturewas then ethanol precipitated, dried, and dissolved in 10 ILl of sequenase buffer(see below) containing 10ng of an appropriate sequencing primer derived from the complementary strand. Primer annealing wasconductedin a heat block preheatedto 65°C and cooledto 37°C over 30 min. This approachwasoften satisfactory but wasnot optimal in providing readablesequences. Wethereforeadopted a second approach, modified froma published technique [16]. PCR amplification for templateproductionwasdone as usual using forward and reverse primers, except that one of the primers was biotinylated at its 5' end through use of an amino-linker phosphoramidite(Aminolink 2, AppliedBiosystems, Foster City, CA), and a biotin N-hydroxysuccinimide ester (Enzotin; Enzo Biochem, New York). After PCR, the reaction mixture was agitated for 45 min at 37°C with streptavidin-coated magneticbeads (Dynabeads M280; Dynal, Oslo) to bind the biotinylated DNA strands generated during PCR. The complementary strand was washed off the solidphase with alkali as described [16]. After a neutralizing rinse (1 M NaCl, 50 mM TRIS, pH 7.4), the magnetic beads were resuspended in 20 ILl of the same buffer, containing 20 ng!ILl of the sequencing primer, and agitated at 37"C for 30 min to anneal. Excess primer waswashed away, and a prealiquotedmixtureof buffer, nucleotides, and polymerasewas added to the beads to initiatethe sequencing reactions. Sequencing reactions weredonebythe dideoxy chain-termination techniqueusing mutated T7 polymerase (Sequenase version2; US Biochem, Cleveland), according to the instructions furnished by the vendorand as published [17]. P5S]deoxyadenosine triphosphate was used as the label. Completed sequencing reactions were analyzed
on O.4-mm thick 6% polyacrylamide gels. Sequences from 50 to 300 bases downstream from each primer were best resolvedon sequencing gelsand weretheprimarysourceof data. Information from furtherdownstream wasusedprimarilyto confirma readingderived froma subsequent primer.Variant or indeterminate sequences noted on individual experiments wereclarified by sequencing both strands andrepeated analyses. Compressed bandsobscuring thetruesequence werenotespecially frequent intheregion sequenced butwereresolved by resequencing usingdeoxyinosine triphosphate in placeof deoxyguanosine triphosphate. Comparisons and alignments. By use of multiple sequencing primers spaced 100-300bp apart,thegBcoding sequence frombases 1150-2151 (codons 384-717, Towne strain numbering) was determined for each of 12 clinical strains. This part of gB includes the cleavage siteofgp55(codon461),the neutralization-related epitopes recognized bymurinemonoclonal antibodies (codons 461-680),and extends into the transmembrane region (codon 715). Sequence alignments were made with strains ADl69 and Towne (reference sequences) and among the clinical strains. Restriction enzyme analysisfor strain grouping. The 293-to 296bp target(sizevariesby strain)amplified byprimersgB1319/gB1604 (table 1) fromextracted viral DNAwasdigested with restrictionenzymeHinfl and withRsaI. Results ofthisanalysis wereusedto identify the sequence variant to which a given strain belonged. Apart fromstrainsselected for sequencing, an additional 28 randomCMV isolatescollectedin the last 6 years from our localorgan transplant recipients were studied.
Results
Readability ofsequences derivedfrom peR-amplifiedtemplates. Direct sequencing of PCRproducts, without subcloning, gave clearandconsistent data(figure 1), despiterepeated PCR amplification of target sequences from the same CMV DNAextractand the use of multiple forward and reverse sequencing primers. As data weregenerated from the project, additional primersweresynthesized as needed to avoid priming within regions of sequence variability. In our experience,
Downloaded from http://jid.oxfordjournals.org/ at University of Aberdeen on February 15, 2015
NOTE. All primer sequences are shown in 5' to 3' direction and are based on published sequences for strain ADl69 [2], except that one base change was made in gB13l9 to reduce mismatch with variant strains. * 5'-biotinylated primers used to make sequencing templates.
Sequence Analysis of CMV gp55 Variation
JID 1991;163 (June)
.= ------ ...-ACGT
........---It::
~
..::-=:: .....-
