JOURNAL OF CLINICAL MICROBIOLOGY, Mar. 2008, p. 882–886 Vol. 46, No.

3
0095-1137/08/$08.00⫹0 doi:10.1128/JCM.01079-07
Copyright © 2008, American Society for Microbiology. All Rights Reserved.

Cytomegalovirus Strain Diversity in Seropositive Women䌤

Zdenek Novak,1* Shannon A. Ross,1 Raj Kumar Patro,2 Sunil Kumar Pati,2 Rekha A. Kumbla,1
Sallie Brice,1 and Suresh B. Boppana1
Department of Pediatrics, The University of Alabama at Birmingham, Birmingham, Alabama,1 and All India Institute of
Medical Sciences, New Delhi, India2
Received 26 May 2007/Returned for modification 10 August 2007/Accepted 12 January 2008

Infection and reinfection with multiple cytomegalovirus (CMV) strains have been shown to occur in immu-
nocompromised individuals, sexually transmitted disease clinic attendees, and children attending day care
centers. To characterize the CMV diversity in healthy seropositive individuals, 16 CMV PCR-positive speci-
mens from 113 seropositive women were analyzed for glycoprotein gN and gB genotypes by cloning, followed
by nucleotide sequencing of the plasmid DNA and/or restriction fragment length polymorphism (RFLP). The
results showed that most (93.7%) of the PCR-positive specimens contained multiple gN and/or gB genomic
variants, suggesting that the majority of women were infected with more than one virus strain. The results also
showed that the RFLP technique might not be sufficiently sensitive to detect all of the genomic variants present
in a sample.

Cytomegalovirus (CMV) species are important opportunistic number of samples containing mixtures of two gB genotypes
agents in infection of immunocompromised individuals and a (3, 7). However, a recent study using hybridization with type-
frequent cause of congenital infection. Infection with multiple specific probes (10) showed mixtures of all genotypes. To de-
strains of CMV has been shown to occur frequently in immu- termine the CMV strain diversity in healthy seropositive
nocompromised individuals and sexually transmitted disease women, the presence of multiple gN and gB genomic variants
(STD) clinic attendees (8, 10). In addition, reinfection with in urine or peripheral blood was examined by two different
different CMV strains was documented to occur in children methods, RFLP and cloning followed by nucleotide sequence
attending day care centers (2). More recently, CMV reinfec- analysis of the plasmid DNA.
tions were demonstrated in seropositive women, and such re- (This research was presented in part at the 43rd Annual
infections can result in intrauterine transmission and damaging Meeting of the Infectious Diseases Society of America, San
fetal infection (5). Francisco, CA, 7 October 2005, abstract 924.)
Extensive genetic polymorphisms in envelope glycoproteins
of CMV, including glycoprotein B (gpUL55), glycoprotein O
MATERIALS AND METHODS
(gpUL74), and glycoprotein N (gpUL73), have been demon-
Specimens and subjects. The specimens studied consisted of 306 urine and 248
strated among clinical CMV isolates. Major envelope glycopro-
peripheral blood samples from 113 healthy, CMV-seropositive women who were
tein B (gB) of CMV has been demonstrated to elicit a strong tested for the presence of CMV immunoglobulin G antibodies in the postpartum
neutralizing antibody response (6), and on the basis of restric- period between February 2000 and June 2004. The women in the study were
tion fragment length polymorphism (RFLP) analysis of clinical derived from a predominantly urban, low-income, African American population.
samples, four unique genomic variants, gB types 1 to 4, have Informed consent was obtained from all study participants, and the study was
conducted in accordance with the guidelines of the Institutional Review Board
been identified (9). Glycoprotein N has been shown to be for Human Use of the University of Alabama at Birmingham.
highly polymorphic at the amino-terminal region, and most Detection of CMV DNA. DNA was extracted from urine and peripheral blood
clinical CMV isolates have been shown to cluster into four specimens with a commercial spin column kit (Qiagen Inc., Chatsworth, CA).
distinct genomic variants, gN-1, gN-2, gN-3a, gN-3b, gN-4a, The samples were initially tested for the presence of CMV DNA by PCR with
primers from the antigen domain 1 region of the gene encoding glycoprotein B
gN-4b, and gN-4c (11). Recent studies have shown that a
as described previously (4). The antigen domain 1 region of gB has been shown
significant proportion of the virus-neutralizing response was to be highly conserved among clinic isolates of CMV (9). The PCR-positive
also directed against the gM/gN complex (17). No linkage samples were further analyzed to determine gN and gB genotypes.
between gN genotypes and gB genotypes has been observed Characterization of gN genomic variants. (i) Nucleotide sequence analysis
(11). following cloning of the PCR-amplified gN products. The samples that were
CMV PCR positive were further subjected to PCR to amplify the gN region with
Published studies using RFLP analyses to determine the gN primers gN-Fw (5⬘ GGC GGT GGT GTG ATG GAG TG) and gN-Rev (5⬘ AAT
genotypes have identified only a single gN type in a given AGC CTT TGG TGG TGG TTG C). After an initial 2-min denaturation at
sample (11, 13). Studies of glycoprotein B based on RFLP 96°C, the samples underwent seven cycles of denaturation at 96°C for 30 s,
analyses showed the presence of a single genotype or a limited annealing at 65°C for 30 s, and extension at 72°C for 40 s and the annealing
temperature was decreased by 1°C each cycle. The samples were further sub-
jected to 28 cycles with an annealing temperature of 58°C and a final extension
step at 72°C for 7 min. The PCR products were directly cloned into TOPO TA
* Corresponding author. Mailing address: University of Alabama at cloning vector pCR 2.1 (Invitrogen Inc., Carlsbad, CA). The colonies were
Birmingham, Children’s Hospital, CHB 150, 1600 6th Ave. S., Bir- screened for the presence of the gN gene insert by direct PCR amplification and
mingham, AL 35233. Phone: (205) 996-7896. Fax: (205) 996-7150. then grown to an appropriate volume of culture medium. Initially, the plasmid
E-mail: [email protected]. DNA from five individual colonies was sequenced with the gN-Fw primer at the
䌤
Published ahead of print on 23 January 2008. University of Alabama at Birmingham sequencing core facility. If two or more

882
VOL. 46, 2008 CYTOMEGALOVIRUS STRAIN DIVERSITY 883

FIG. 1. Alignment of gN (UL73) nucleotide sequences with only part of the variable region shown. Dots indicate identity, and dashes indicate
deletions. Strains are grouped by comparing the sequences of the recombinants with the prototypic gN genotypes previously described (11). The
sequences are listed with the unique study subject identifiers. Prototypical laboratory-adapted strains are in parentheses. The nucleotide sequences
were aligned with that of AD169 (gN1 prototype) with the Vector NTI Advance software package V.10 (Invitrogen, Carlsbad, CA) and displayed
as a printable output by the BOXSHADE server (http://www.ch.embnet.org/software/BOX_form.html).

variants were found, no more colonies were screened. However, in the event that RESULTS
the first five colonies contained a single genotype, five additional colonies were
sequenced. The nucleotide sequences were compared to the published gN sub- Of the 554 samples examined for the presence of CMV
type sequences of the four major gN genotypes (GenBank accession numbers
DNA, 16 (2.9%) were CMV PCR positive (9/306 urine samples
AF309971, AF309976, AF309980, AF390773, AF309987, AF309997, and
AF310004). A limited number of colonies from each recombinant were se-
and 7/248 blood samples). The positive samples were collected
quenced in both directions to confirm the sequence diversity and the genotype from 16 different study subjects. To determine the sensitivity of
assignment (Fig. 1). the PCR assay, the positive samples were subjected to a real-
(ii) RFLP analysis. The PCR products amplified with the gN-Fw and gN-Rev time PCR assay to estimate the amount of CMV DNA (4).
primers described above were digested in four separate reactions with the re- This analysis showed that the sensitivity of the PCR assay was
striction endonucleases SalI, SacI, BsaXI, and MmeI. The resulting restriction
300 copies/ml (data not shown).
fragments were resolved by agarose gel electrophoresis, and the band patterns
were analyzed in accordance with the restriction sites contained in different gN The gpUL73 (gN) and gpUL55 (gB) diversity was examined
subtypes (12). in all 16 samples by RFLP and cloning of the gN gene and by
Characterization of gB genotypes by nested PCR followed by cloning. The cloning of the gB gene. Of the 16 samples, 15 (93%) were
DNA samples were initially subjected to PCR to amplify the target gB region found to have more than one gN genomic variant (Table 1).
located in the variable region between bp 1138 and 1638 with gB primers
The only sample that was found to contain a single gN geno-
gB1138 (5⬘ CAA GAR GTG AAC ATG TCC GA) and gB1638 (5⬘ GTC ACG
CAG CTG GCC AG). The PCR products were diluted 1:10 and subjected to
type on cloning and nucleotide sequence analysis was also
nested PCR with primers gB1319 (5⬘ TGG AAC TGG AAC GTT TGG C) found to have a single gN genomic variant on RFLP examina-
and gB1604 (5⬘ GAA ACG CGC GGC AAT CGG), yielding a 285-bp product tion (Table 2). Although the RFLP examination revealed that
(3). The PCR products were gel purified and cloned into the pCR 2.1 TOPO four additional samples contained a single gN genotype, nu-
TA cloning vector (Invitrogen Inc., Carlsbad, CA). The colonies were cleotide sequence analysis documented the presence of multi-
screened by direct PCR, and those containing the insert were grown in culture
medium. Plasmid DNA from five individual colonies from each recombinant
ple gN genomic variants in the samples (Table 1).
was sequenced with the M13 forward primer at the University of Alabama at Of the 16 samples analyzed for gB genotypes, 69% (11/16)
Birmingham sequencing core facility, and if two or more variants were found, were found to contain more than one gB genotype; eight sam-
no more colonies were screened. However, in the event that the first five ples had two genotypes, and three samples had three genotypes
colonies contained a single genotype, five additional colonies were se- (Table 2). Among the 16 samples that underwent gN and gB
quenced. The nucleotide sequences were compared to the published gB
genotype analysis by cloning, 15 (93.7%) samples contained
genotype sequences (GenBank accession numbers M60928, M60930,
M609931, and M609933). As described above for gN genotype characteriza- multiple genotypes and only one sample had a single gN and
tion, a limited number of colonies were sequenced in both directions to gB genotype. The sample with only one gN genotype was from
confirm the genotype assignment (Fig. 2). a blood specimen. Of the samples with a single gB genotype,
884 NOVAK ET AL. J. CLIN. MICROBIOL.

FIG. 2. Alignment of representative gB (UL55) sequences with the prototypes (only part of the variable region of gB is shown). Dots indicate
identity, and dashes indicate deletions. Strains are grouped by comparing the sequences of the recombinants with the prototypic gB genotypes
previously described (9). The sequences are listed with the unique study subject identifiers. The nucleotide sequences were aligned with the gB1
prototype with the Vector NTI Advance software package V.10 (Invitrogen, Carlsbad, CA) and displayed as a printable output by the BOXSHADE
server (http://www.ch.embnet.org/software/BOX_form.html).

two (40%) were from blood and three (60%) were from urine ence of different and multiple gN and/or gB genomic variants.
specimens. Similar findings were reported in a recent study in our labora-
The most common glycoprotein N genotype in our study tory in which multiple gN genotypes were detected in genital
group was type 3 (41%), followed by type 1 (35%). Glycopro- tract specimens from women who attended an STD clinic (15).
tein B genotype 2 (43%) was the most frequent, followed by In an earlier report, we documented the occurrence of CMV
type 4 (33%) (Fig. 3). reinfection between pregnancies in seropositive women, and
such reinfection could lead to intrauterine transmission and
DISCUSSION symptomatic congenital CMV infection (5). Several other stud-
ies have also documented CMV infection with multiple virus
The findings of the present study clearly document the pres- strains in a variety of population groups, including children
ence of multiple CMV strains in the majority of the CMV attending day care centers, human immunodeficiency virus-
PCR-positive urine and peripheral blood specimens from the infected individuals, allograft recipients, and infants with con-
healthy seropositive women studied. The existence of multiple genital CMV infection (1, 2, 10, 14). Together, the findings of
virus strains in the specimens was demonstrated by the pres- these studies of different population groups suggest that infec-
tions with multiple CMV strains occur frequently.
It is possible that the observed strain diversity in our study
TABLE 1. Comparison of gN genotyping results by RFLP and the population of healthy, seropositive women is due to an in-
cloning techniques for each of the 16 PCR-positive samples creased exposure to CMV, resulting in CMV reinfections.
Observed genotype(s) However, we could not determine the timing of reinfection(s)
Sample in our study. Previous studies have reported reinfection with
RFLP Cloning
multiple strains of CMV in immunocompetent hosts. Chandler
0-013 4 1, 4 et al. reported that four of eight women attending an STD
1-025 1, 3 1, 3
1-028 1, 3 1, 3
clinic were found to be reinfected with a new strain of CMV
1-030 1, 3 1, 3
1-037 1, 2 1, 2
1-039 1, 3 1, 3 TABLE 2. Results of gN and gB genotyping of CMV in the urine
1-047 1 1, 3 and blood of CMV-seropositive women
2-006 1, 3 1, 3
2-027 1, 3 1, 3 % of samples with indicated no. of
2-037 2, 3, 4 2, 3, 4 Glycoprotein Technique genotypes (no.a of samples):
2-038 1, 3 1, 3 1 2 3
2-043 3 3
2-046 1 1, 3 gN RFLP 31.2 (5) 56.3 (9) 12.5 (2)
2-049 1, 3 1, 3, 4 gN Cloning 6.3 (1) 75 (12) 18.7 (3)
3-049 1, 3, 4 3, 4 gB Cloning 31.3 (5) 50 (8) 18.7 (3)
4-005 4 2, 3, 4 a
The total number of samples tested was 16.
VOL. 46, 2008 CYTOMEGALOVIRUS STRAIN DIVERSITY 885

pattern, and therefore the amount of DNA in the samples

tested is critical. Multiple gN genomic variants present in a
sample may not be easily distinguished because the intensity of
the fragments on gel electrophoresis depends on the amount
and ratio of the type-specific DNA. Furthermore, because of
overlapping and sometimes minute differences in band size, it
may be difficult to properly identify all of the bands necessary
for precise genotype identification, even on high-concentration
agarose gel electrophoresis. The restriction sites could be af-
fected by a single nucleotide change, which in turn makes it
difficult to recognize correct genotypes. However, such a
change in the restriction sites is unlikely to affect the genotype
assignment because the grouping of genotypes is based on
multiple nucleotide changes. The cloning and sequence anal-
ysis method is more accurate in determining the number of gN
or gB genotypes because it is not limited by the relative amount
of type-specific DNA contained in the sample. However, there
are limitations to the use of cloning and sequence analysis in
the detection of multiple strains in a sample. These include the
sensitivity and specificity of the PCR assay used to amplify
FIG. 3. Relative frequencies of CMV gN and gB genotypes in urine CMV DNA in the samples and the number of bacterial colo-
and blood samples from CMV-seropositive women.
nies that can be screened.
Like glycoprotein N, genotyping of glycoprotein B by RFLP
has only shown one or two types present in a clinical sample. In
(8). Bale et al. examined serial samples from 37 children at- a study based on samples from liver transplant patients, 87%
tending day care centers and found that 19% had evidence of (53/61) of the samples contained only one gB genotype (3).
infection with more than one CMV strain (2). Although expo- The remaining 13% of the samples contained two genotypes.
sure to CMV through sexual activity or contact with young This could be due to the fact that the use of the enzymes HinfI
children could explain the strain diversity observed in our and RsaI produced indistinguishable restriction patterns in
study, the association between increased CMV exposure and certain combinations of three and four different gB genotypes.
CMV strain diversity has not been documented. Additionally, the above-described limitations of the ability of
Several previous studies that examined the genetic diversity RFLP to detect multiple gN genotypes are also relevant to the
of CMV glycoprotein gN only reported the presence of a single detection of gB genotypes. However, mixtures of three or four
gN genotype in the specimens tested. Pignatelli et al. examined different gB genotypes per sample were detected in a study
223 viral isolates from a variety of patient populations, includ- using non-RFLP-based techniques. A recent study by Co-
ing allograft recipients, AIDS patients, congenitally infected aquette et al. used PCR amplification followed by hybridization
children, and children with postnatally acquired CMV infec- with a single-stranded DNA probe specific for each CMV gB
tion from four different geographic regions. The virus isolates genotype detected by DNA enzyme immunoassay. In that
examined in that study were obtained from urine, saliva, blood, study, of the 97 CMV isolates collected from 92 immunocom-
amniotic fluid, and biopsy specimens (12). Although all four promised patients (transplant recipients and lymphoma and
gN genotypes were present among the specimens tested, only leukemia patients), 74% (72/97) contained a single gB geno-
one gN type was detected in a given specimen. In contrast, we type, 22% (21/97) contained two genotypes, 3% (3/97) con-
could document the presence of multiple gN genomic variants tained three genotypes, and 1% (1/97) contained all four geno-
in the majority of CMV PCR-positive specimens in our study. types (10).
In the present study, we used two different techniques, RFLP Similar to the results of previous studies, we did not observe
and cloning of the amplified gN product, followed by nucleo- a linkage between the gN and gB genotypes (11). However, the
tide sequence analysis, to determine the gN genotypes, number of samples examined was small and we found only one
whereas the studies by Pignatelli et al. and others only used sample that contained a single gN genotype and five samples
RFLP (16). It is likely that the addition of the cloning method containing a single variant of gB. Only one study sample con-
may have improved our ability to distinguish different genomic tained a single genotype of both the gN and gB glycoproteins.
variants present in a specimen. Additional factors, including Therefore, our study clearly did not have a sufficient sample
the populations studied and the geographic origin of the spec- size to examine the possible linkage between the gN and gB
imens, may account for the differences between the present genotypes. In addition, our study was based on a predomi-
study and previously published reports. Furthermore, Pig- nantly African American, low-income population and there-
natelli et al. studied virus isolates that had been propagated in fore the findings of our study may not be representative of the
cell culture rather than DNA extracted from the original spec- general population.
imen. The propagation of virus isolates in cell culture could Using RFLP and cloning followed by nucleotide sequence
have selected for a single virus strain. analysis, we could demonstrate infection with multiple CMV
Identification of gN and gB genotypes by RFLP is based on strains in most healthy seropositive women who had CMV
the ability to clearly visualize the restriction fragment band DNA in their urine and/or peripheral blood. These findings
886 NOVAK ET AL. J. CLIN. MICROBIOL.

need to be confirmed in a study with a larger number of 8. Chandler, S. H., H. H. Handsfield, and J. K. McDougall. 1987. Isolation of
multiple strains of cytomegalovirus from women attending a clinic for sex-
samples and more diverse population groups. ually transmitted diseases. J. Infect. Dis. 155:655–660.
9. Chou, S. 1992. Comparative analysis of sequence variation in gp116 and gp55
ACKNOWLEDGMENTS components of glycoprotein B of human cytomegalovirus. Virology 188:388–
390.
Financial support for this study came from the National Institute on 10. Coaquette, A., A. Bourgeois, C. Dirand, A. Varin, W. Chen, and G. Herbein.
Deafness and Other Communication Disorders (R01 DC04163) and 2004. Mixed cytomegalovirus glycoprotein B genotypes in immunocompro-
mised patients. Clin. Infect. Dis. 39:155–161.
the General Clinical Research Center (M01 R00032).
11. Pignatelli, S., P. Dal Monte, and M. P. Landini. 2001. gpUL73 (gN) genomic
We have no commercial conflicts of interest. variants of human cytomegalovirus isolates are clustered into four distinct
genotypes. J. Gen. Virol. 82:2777–2784.
REFERENCES 12. Pignatelli, S., P. Dal Monte, G. Rossini, S. Chou, T. Gojobori, K. Hanada,
1. Arav-Boger, R., R. E. Willoughby, R. F. Pass, J. C. Zong, W. J. Jang, D. J. J. Guo, W. Rawlinson, W. Britt, M. Mach, and M. P. Landini. 2003.
Alcendor, and G. S. Hayward. 2002. Polymorphisms of the cytomegalovirus Human cytomegalovirus glycoprotein N (gpUL73-gN) genomic variants:
(CMV)-encoded tumor necrosis factor-alpha and beta-chemokine receptors identification of a novel subgroup, geographical distribution and evidence of
in congenital CMV disease. J. Infect. Dis. 186:1057–1064. positive selective pressure. J. Gen. Virol. 84:647–655.
2. Bale, J. F., Jr., S. J. Petheram, I. E. Souza, and J. R. Murph. 1996. Cyto- 13. Pignatelli, S., P. Dal Monte, G. Rossini, T. Lazzarotto, M. R. Gatto, and
megalovirus reinfection in young children. J. Pediatr. 128:347–352. M. P. Landini. 2003. Intrauterine cytomegalovirus infection and glycoprotein
3. Binder, T., W. Siegert, A. Kruse, H. Oettle, F. Wilborn, R. Peng, H. Timm, N (gN) genotypes. J. Clin. Virol. 28:38–43.
P. Neuhaus, and C. A. Schmidt. 1999. Identification of human cytomegalo- 14. Rasmussen, L., C. Hong, D. Zipeto, S. Morris, D. Sherman, S. Chou, R.
virus variants by analysis of single strand conformation polymorphism and Miner, W. L. Drew, R. Wolitz, A. Dowling, A. Warford, and T. C. Merigan.
DNA sequencing of the envelope glycoprotein B gene region-distribution 1997. Cytomegalovirus gB genotype distribution differs in human immuno-
frequency in liver transplant recipients. J. Virol. Methods 78:153–162. deficiency virus-infected patients and immunocompromised allograft recip-
4. Boppana, S. B., K. B. Fowler, R. F. Pass, L. B. Rivera, R. D. Bradford, F. D. ients. J. Infect. Dis. 175:179–184.
Lakeman, and W. J. Britt. 2005. Congenital cytomegalovirus infection: as- 15. Ross, S. A., Z. Novak, G. Ashrith, L. B. Rivera, W. J. Britt, S. Hedges, J. R.
sociation between virus burden in infancy and hearing loss. J. Pediatr. 146: Schwebke, and A. S. Boppana. 2005. Association between genital tract
817–823. cytomegalovirus infection and bacterial vaginosis. J. Infect. Dis. 192:1727–
5. Boppana, S. B., L. B. Rivera, K. B. Fowler, M. Mach, and W. J. Britt. 2001. 1730.
Intrauterine transmission of cytomegalovirus to infants of women with pre- 16. Rossini, G., S. Pignatelli, P. Dal Monte, D. Camozzi, T. Lazzarotto, L.
conceptional immunity. N. Engl. J. Med. 344:1366–1371. Gabrielli, M. R. Gatto, and M. P. Landini. 2005. Monitoring for human
6. Britt, W. J., L. Vugler, and E. B. Stephens. 1988. Induction of complement- cytomegalovirus infection in solid organ transplant recipients through anti-
dependent and -independent neutralizing antibodies by recombinant-derived genemia and glycoprotein N (gN) variants: evidence of correlation and
human cytomegalovirus gp55-116 (gB). J. Virol. 62:3309–3318. potential prognostic value of gN genotypes. Microbes Infect. 7:890–896.
7. Carraro, E., and C. F. Granato. 2003. Single human cytomegalovirus gB 17. Shimamura, M., M. Mach, and W. J. Britt. 2006. Human cytomegalovirus
genotype shed in multiple sites at the time of diagnosis in renal transplant infection elicits a glycoprotein M (gM)/gN-specific virus-neutralizing anti-
recipients. J. Med. Virol. 70:240–243. body response. J. Virol. 80:4591–4600.