ABSTRACT
Sequence analysis of the UL144 gene of human cytomegalovirus (CMV) was used to investigate the epidemiology of CMV infections in preterm infants. Nosocomial transmission of CMV from congenitally infected infant to preterm twins was excluded based on distinct molecular profiles of CMV strains. Indistinguishable molecular profiles between strains from the mother and the infant indicated postnatal acquisition of CMV through breastfeeding.
Analysis of polymorphic regions in the human cytomegalovirus (CMV) genome can be used to study the epidemiology of CMV infections (1, 4, 8, 11). The commonly used methods include analysis of restriction fragment length polymorphism of the entire CMV genome and PCR- or PCR-restriction fragment length polymorphism-based stepwise analysis of polymorphic regions (1, 3). Here we describe the application of a molecular typing approach based on sequence analysis of the UL144 gene to examine possible epidemiological relationships between CMV infections in preterm infants.
Case 1 was a suspected nosocomial transmission of CMV within the newborn nursery. Female twins were born at 30 weeks of gestation and weighed 1,450 g (subject DM [here and throughout, two-letter patient designations are composed of randomly chosen letters]) and 1,650 g (subject DV). The Apgar scores were 9 and 10 at 1 and 5 min, respectively, in these patients. At 12 weeks cerebral calcifications were identified in DM, while the cranial ultrasonography was normal in DV. CMV was detected in urine by virus culture and PCR in both infants. A congenital infection was excluded by a negative PCR on Guthrie cards. The mother was CMV immunoglobulin G positive, but immunoglobulin M and A negative. The infants were breastfed for 12 weeks after birth. Up to 20 months of age CMV remained detectable in the urine of both infants but the hearing and the neurological development of both infants were normal. The twins had been hospitalized in the same newborn nursery as a male infant (subject KC), born at 36 weeks of gestation (2,150 g), who was diagnosed with a congenital CMV infection. The viral load in urine on day 5 was 3 × 106 copies/ml. Cranial ultrasonography revealed cerebral calcifications (lenticulostriate vasculopathy) and septation of the occipital horns. Antiviral therapy reduced the viral load in the urine to undetectable levels. The outcome at 2 years was poor, with severe learning disabilities and epilepsy.
Case 2 was a suspected acquisition of CMV through breastfeeding. A female infant (subject PC) was born at 26 weeks of gestation and weighed 800 g. The Apgar scores were 5 and 9 at 1 and 5 min, respectively. She developed a mild respiratory distress and because of apnea she was treated with continuous positive airway pressure for several weeks. Cranial ultrasonography was normal until 7 weeks after birth, when she developed respiratory insufficiency due to pneumonitis. CMV was isolated from bronchial secretions. CMV DNA was detected in high quantities in urine (6.8 × 105 copies/ml) and plasma (6.7 × 104 copies/ml). Cranial ultrasonography showed echogenicity (lenticulostriate vasculopathy) in the basal ganglia and germinal matrix, which increased over the next weeks. Cerebrospinal fluid analysis was normal and CMV PCR was negative. Congenital infection was excluded by a negative CMV PCR on a Guthrie card. CMV-specific immunoglobulin G and A were detected in the blood of the mother (PCm) and CMV DNA was detected in the breast milk (5 × 103 copies/ml) at the time of diagnosis. CMV was detected in the urine and blood of the infant up to 9 and 6 months after birth, respectively, and in the breast milk of the mother up to 6 months. Hearing and neurological development were normal at the age of 18 months.
DNA isolation from clinical specimens and CMV detection by real-time PCR were performed as described earlier (9). For UL144 genotyping a 399-bp fragment of the UL144 coding sequence, including the hypervariable 5′ part (nucleotides 6 to 405), was amplified by nested PCR. For the first-round PCR (728-bp fragment) the previously described primer set UL144-B (6) was slightly modified: 5′-AACCGCGGAGAGGATGATAC-3′ (forward) and 5′-ACTCAGACACGGTTCCGTAA-3′ (reverse). For the nested PCR, previously described primer set UL144-A was used (6). For clinical isolates that failed to be amplified by the UL144-A primer set an alternative UL144-N nested primer set was designed, 5′-GTTCGGCCCCATGAGTTATT-3′ (forward) and 5′-GTGTGACTTCATCGTACCGT-3′ (reverse).
In addition, sequence analysis of glycoprotein B (gB) was performed to verify the results of UL144 genotyping. A 757-bp fragment containing the gB cleavage site was amplified (nucleotides 1072 to 1829) using the previously described forward primer (6) and the reverse primer 5′-CATTCCTCAGTGCGGTGGTT-3′. If the first-round PCR yielded no product, a gB-N nested primer set was used, 5′-CTGCCAAAATGACTGCAACT-3′ (forward) and 5′-AGACATCACCCATGAAACGC-3′ (reverse), yielding a 531-bp product.
The PCR products were sequenced and phylogenetic analysis was conducted using ClustalW (http://www.ebi.ac.uk/clustalw/ ). The phylogenetic tree was constructed by using the neighbor-joining method and was based on the N-terminal portion of the UL144 sequence (nucleotides 70 to 342).
The typing approach was first validated on isolates from an infant (KP) and the mother (KPm) from a known case of a congenital CMV infection, which were predicted to be indistinguishable. As expected, the UL144 and gB molecular profiles of the strains from the infant and the mother were identical (Table 1).
In case 1, the UL144 sequences of isolates from the twins (DM and DV) were all identical, and clustered in UL144 genogroup 2 (Table 1, Fig. 1), according to the classification by Lurain et al. (6). The sequences of three serial isolates from the infant with congenital CMV (KC) were all identical and also clustered in UL144 genogroup 2. However, these sequences differed markedly (by more than 11 nucleotides) from those of the twins (Fig. 1). In addition, the twins both excreted gB genotype 1 strains, according to the classification by Chou and Dennison (4), while the congenitally infected infant (KC) excreted a gB genotype 3 strain. These results confirmed infection with different strains and thus excluded the nosocomial transmission of CMV from KC to the twins during their concurrent hospitalization.
The mode of transmission of CMV in preterm twins is not known, but acquisition of CMV through breast milk from their seropositive mother is most likely. The mixture of gB genotypes in DV suggests the presence of an additional unrelated strain. It could be speculated that both strains were present in the mother but were differentially transmitted to each of the twins.
In case 2, six longitudinal HCMV isolates from the infant (PC) had identical sequences clustering in UL144 genogroup 3. Two isolates from the breast milk of the mother (PCm) showed total UL144 sequence identity with the isolates from the infant. In addition, the gB genotypes of the isolates from the infant and the mother were identical, clustering in gB genotype 3 (Fig. 1, Table 1). These results clearly demonstrated transmission of CMV through breast milk.
Characterization of CMV for epidemiological purposes relies on the presence of polymorphic regions within the genome. Regions showing hypervariability among unrelated strains but stability during virus passage, transmission and long-term excretion are potentially good candidates for analysis of interstrain variability. Here we used sequence analysis of the 5′ half of the UL144 gene, which displays significant strain-specific sequence variability (6). The UL144 sequences from strains that were clearly epidemiologically linked (twins and congenital infections) were identical, whereas they clearly differed in the absence of an epidemiological relationship.
In view of the fact that unrelated strains can infrequently have identical UL144 sequence (2), another polymorphic region should be sequenced before concluding that strains are indistinguishable. Likewise, detection of unrelated strains (although easier to demonstrate) should also be verified the same way. Confirmatory sequencing of the gB gene was performed in our study to further support the results of UL144 typing.
Despite the small number of cases studied, our results demonstrated the potential usefulness of sequence-based molecular typing for investigation of the epidemiology of CMV infections in the clinical setting. With the ongoing improvement and wider availability of genomic sequencing, DNA sequence analysis of polymorphic regions should become the method of choice for straightforward and precise molecular characterization of CMV.
Of interest is our observation of a postnatal development of intracerebral calcifications in two out of three infants. The development of long-term sequelae following postnatal CMV infection in preterm infants is a phenomenon which has not been sufficiently investigated until now (5, 10). There is some evidence that early postnatal acquisition of CMV might increase the risk for neurological sequelae and abnormal neuromotor function at 3 years of age (7). Despite the apparent neurological abnormalities found in our report, no adverse effects were observed up to 18 months of age. It will be interesting to follow these children till school age to evaluate the effect of postnatal CMV infection on their neurodevelopmental outcome.
Phylogenetic analysis of the UL144 sequences of the CMV strains from preterm infants described in this study. Results are displayed as an unrooted tree generated by the neighbor-joining method. The Toledo strain sequence was used as the reference. Selected bootstrap values are based on 100 repetitions. Bold numbers indicate the UL144 genotype.
CMV strains: summary of CMV genotypes in preterm infantsa
FOOTNOTES
- Received 22 April 2005.
- Returned for modification 22 July 2005.
- Accepted 24 October 2005.
- Copyright © 2006 American Society for Microbiology
