Distribution of the CMV glycoprotein gH/gL/gO and gH/gL/pUL128/pUL130/pUL131A complex variants and associated clinical manifestations in infants infected congenitally or postnatally

Human cytomegalovirus (CMV) is a major cause of morbidity in fetuses following intrauterine infection. The glycoprotein (g) envelope trimeric gH/gL/gO and pentameric gH/gL/pUL128/pUL130/pUL131A complexes are required for CMV entry into fibroblasts and endothelial/epithelial cells, respectively, and both are targets for neutralizing antibodies. The role of sequence variability among viral strains in the outcome of congenital CMV infection is controversial. Variation in the CMV UL75 gene encoding glycoprotein H (gH), the UL115 (gL), the UL74 (gO), and the UL128 locus (UL128L) encoding three structural proteins (pUL128, pUL130, and pUL131A) was determined in 82 newborns with congenital CMV infection and 113 infants with postnatal or unproven congenital CMV infection. Genotyping was performed by sequencing analysis of PCR‐amplified fragments and the PCR-restriction fragment length polymorphism (RFLP) method, and the viral load was measured by quantitative real‐time PCR. The obtained results demonstrated that (1) different CMV variants and mixed CMV infections can be detected in newborns infected congenitally; (2) the gH1 genotype, UL130 variant 6, and UL131A variant 1 were associated with some signs/symptoms within cohort of pediatric patients, mainly consisting of infants with symptomatic CMV infection. The results revealed that pUL130, pUL131A, and gH polymorphisms seemed to be associated with the outcome of CMV infection in infants.

all CMV-infected infants. These studies confirm the findings that various CMV variants can be vertically transmitted. The results support the hypothesis that variability in pentamer genes is an important factor that affects clinical sequelae following CMV infection.

Study population and clinical outcome. One hundred and ninety-five pediatric patients with CMV
DNAemia were enrolled in the study. The patients were selected based on clinical diagnosis and diagnostic markers. Congenital CMV infection was confirmed in 82 newborns (positive CMV DNA in urine ≤ 21 day of life), and the remaining 113 children were classified as having postnatal or unproven congenital CMV infection (pCMV). Among infants in pCMV group, negative CMV DNA in urine ≤ 21 day of life was found in 83% cases. The average age at which the newborns with cCMV infection were examined was 9.6 days (median age 8.0 days; range 1-21 days), while in the group of infants with pCMV infection, it was 3.4 months (median age 3.0 months; range 1-11.5 months). The demographic and clinical characteristics of congenitally or postnatally infected children are summarized in Table 1. Almost all examined patients demonstrated cytomegaly symptoms, and this high frequency of symptomatic infection was due to selection bias. The most prevalent symptoms in CMV-infected children from both groups were neurological dysfunction and hematological disorders (especially anemia and thrombocytopenia). In addition, newborns with cCMV infection usually demonstrated CNS damage (73%), including microcephaly and abnormal brain ultrasound findings (cystic lesions, intracranial calcification, and ventriculomegaly). In contrast to postnatal infection, symptoms such as CNS damage (P < 0.0001), neurological dysfunction (P = 0.003), intrauterine growth restriction (P < 0.0001), unilateral or bilateral hearing loss (P < 0.0001), ocular defects (P = 0.015), thrombocytopenia (P = 0.0001), and petechiae (P < 0.0001) occurred more commonly after cCMV infection. Some complications, such as pneumonia and hepatitis were more common in infants in pCMV group (P = 0.005 and P = 0.022, respectively).
Prevalence of CMV variants. The trimer gH/gL/gO gene products were amplified for all examined pediatric patients ( Table 2). Genotyping of gH was performed by analysis of nested PCR (nPCR)-amplified fragments and the two genomic variants were identified by different amplicon size lengths. The applied PCR-restriction fragment length polymorphism (PCR-RFLP) method allowed the detection of major gL or gO genotypes (Fig. 1). Genotypes determined by RFLP were sequenced, and no discrepancies were found. In contrast, the UL128 locus gene products were amplified and sequenced successfully for 79/82 (96.3%) newborns with cCMV infection and 101/113 (89.4%) infants in pCMV group (Table 3). The distribution of the gH genotypes were similar in both patient groups (P > 0.05). To explore the data, one hundred and five infants with CMV infection, including 28 cases of cCMV and 77 cases of pCMV, were examined for the presence of UL75 genotypes as described in our previous study 17 . No significant advantages of the gH1 genotype in congenital cases and gH2 in postnatal cases were observed. The gH1 genotype was detected in 54/82 (65.9%), while the gH2 in 46/82 (56.1%) newborns with cCMV infection ( Table 2). Among children with pCMV infection, the gH1 genotype was found in 63/113 (55.8%), while the gH2 genotype was found in 74/113 (65.5%) cases. The nucleotide sequence analysis confirmed the gH genomic variants previously identified by a nPCR analysis. Mixed infections with both gH1 and gH2 genotypes were detected in 18/82 (22.0%) patients with cCMV infection and in 24/113 (21.2%) cases with pCMV infection.
The UL115 gene encoding gL had a low proportion of nucleotide and amino acid variability in clinical isolates (5-9% and 1.4-2.5%, respectively), while low sequence conservation for UL74 (gO) was observed (sequence variability: 20-50% and 19.4-26.4%, respectively). All obtained amino acid sequences are illustrated in Fig. 2A-C. The gL3 genotype was prevalent in congenital infections, whereas this genomic variant was less common in postnatal infection (P < 0.0001). The most prevalent gO genotype in both groups, newborns with cCMV infection and infants with pCMV infection, was gO1 (approximately 91%). The gO4 genotype was detected more frequently in www.nature.com/scientificreports www.nature.com/scientificreports/ congenital compared to postnatal infection (P = 0.018). Congenital infection with mixed gL and gO genotypes was detected in 56.1% and 31.7% cases, respectively. Mixed pCMV infections were detected in 30.1% (gL) and 11.5% (gO) of infants. The children exhibited mixed infection with two gO genotypes, especially gO1 and gO4 or gO3, and with two or three distinct CMV gL genotypes. Among examined infants, mixed infections were identified mostly with viral gL1-gL3 (29 cases), gL2-gL3 (22 cases), gL3-gL4 (14 cases), and gL1-gL3-gL4 (5 cases) genomic variants. Analysis of the gH/gL/gO genotypes revealed that multiple CMV strains were detected in congenitally infected patients. Unexpectedly, considering all trimer gH/gL/gO genotypes, mixed infections were more commonly detected in congenital than in postnatal infections (75.6% vs. 48.7%; P = 0.0002).
Several viral variants were identified in newborns with cCMV infection, indicating that all variants of the virus could be passed from mother to child. The UL128 variant 3 was identified most commonly in both pediatric patient groups (43.0% and 44.6%), and no statistically significant association with the UL128 variant distribution among children with congenital or postnatal CMV infection was observed ( Table 3). The UL130 variant 6 was prevalent in newborns with cCMV and infants in pCMV group (46.8% and 43.6%, respectively). The UL130 variant 8 was identified more commonly in cCMV than pCMV patients (P = 0.027). Analysis of the UL131A region in clinical isolates revealed that the variant 1 was prevalent in both patient groups, and it occurred more frequently in congenital than postnatal infections (79.7% vs. 58.4%; P = 0.002). In contrast, the UL131A variant 4 was found in 20.3% newborns with cCMV and in 34.7% infants in pCMV group (P = 0.045). Some mixing of pUL128, pUL130, and pUL131A gene variants was observed, but the frequency was ≤ 10% among isolates.
CMV gene variability is associated with an increased risk of specific disease outcomes. In the patients carrying the gH1 genotype, hearing loss was diagnosed with higher incidence compared to those with the gH2 genotype (P = 0.023; Table 5). As was previously found in a smaller group of cases 17 , infants carrying the gH2 genotype were diagnosed with deafness at a lower incidence compared to those with the gH1 genotype (P = 0.004). In addition, gH2-infected children exhibited an increased risk of developing purpuric and petechial www.nature.com/scientificreports www.nature.com/scientificreports/ rashes. The gH2 genotype was associated with a diminished risk of hearing loss and psychomotor retardation in infants in adjusted and unadjusted models (Table 5).
Infection with gH1 genotype was also associated with a three-fold increased risk of neurological dysfunction and microcephaly (P < 0.05). Infection with the gH2 genotype was associated with a decreased risk of hearing loss and ventriculomegaly (P = 0.007 and P = 0.024, respectively), though this genotype was observed only in a small number of the symptomatic infants (data not shown).
The UL130 variant 6 was detected in 62.5% of newborns with IUGR (P = 0.013) and was associated with at least a three-fold increased risk of this symptom ( Table 5). The risk of ocular defects was more than three-fold increased following congenital infection with this variant (P = 0.024). Infection with the UL131A variant 1 was associated with a four-fold increased risk of IUGR, neurological dysfunction, and hepatitis (P = 0.001, P = 0.025, and P = 0.021, respectively) and with at least a two-fold increased risk of CNS damage (P = 0.005). In contrast, gL2-infected patients exhibited a decreased risk of developing hematological disorders (P = 0.025 for neonates with cCMV infection; P = 0.018 for all pediatric patients).

Discussion
Several studies have analyzed sequences from CMV clinical isolates and revealed that genomic variability caused additional infectivity or immunomodulation functions [55][56][57][58][59][60] . The increasing evidence suggests a correlation of clinical importance of CMV genetic diversity with pathogenesis [15][16][17][18]20,61 . To assess whether the variability in the CMV genes encoding trimeric and pentameric complexes are associated with congenital transmission, we compared the frequencies of variants in children with and without confirmed congenital infection. Our study focused on the variation in the all CMV genes encoding envelope protein complexes, gH/gL/gO and gH/gL/ www.nature.com/scientificreports www.nature.com/scientificreports/ pUL128-131A, and the distribution of genotypes in pediatric patients. We have determined the sequence variation in the UL128-UL131A genes, named these specific viral variants and identified variants transmitted during congenital and postnatal infection. Our study provides the first demonstration that the CMV variants of the pentamer complex seems to be assigned to a clinical outcome in children, including, gH1, UL130 variant 6, and UL131A variant 1. In addition, the gL3 and gO4 genotypes of CMV might be an important virological marker of congenital infection in children.
Detailed sequence analysis showed that UL128, UL130, and UL131A genes were highly conserved in clinical isolates, suggesting an important role of these genes in viral pathogenesis 55,[61][62][63] . Most amino acid substitutions were limited to the N-terminal region of the UL128 and UL131A genes, although they can also be found in other parts of the UL130 gene. The UL128 and UL130 proteins share conserved cysteine residues that are characteristic of CC-and CXC-chemokines, respectively. Our findings showed 96-99%, 84-97%, and 97-99% sequence identity at the aa level in the UL128, UL130, and UL131A proteins, respectively, in comparison to the protein sequences obtained from the publishing database NCBI. The overall identity of these CMV genes was similar and ranged between 92% and 96% in isolates from Italian patients 55 61 . The conserved cysteine residues in exons 2 and 3 of UL128 were detected in all variants 61 . These conserved cysteines in CMV UL128 protein are especially important for CC chemokine motifs. Vogel et al. found no evidence for a connection between UL144/UL128/UL130/UL131A genotypes and the incidence of CMV retinitis in AIDS patients 61 . We have observed that none of the detected UL128 variants were significantly linked with symptomatic CMV infection. Our results revealed that the UL130 variant 6 and the UL131A variant 1 were associated with an increased risk of specific symptoms in children. No significant difference was detected between the CMV variants determined on the basis of the UL128L and the signs of infection in Chinese infants 63 . The UL130 variant 6 showed a few amino acid substitutions such as P40L, L60I, and S78L that were undetected in common CMV variants. It was found that a 2-bp insertion (TT) led to a frameshift mutation in the UL130 (Towne strain) and a 1-bp insertion (A) in UL131A (AD-169 strain) genes leading to a frameshift mutation have impaired the immunogenicity of viral strains 58,59 . The mutations include nucleotide substitutions that introduce in-frame translational termination codons or splicing are predicted to ablate gene functions, insertions, and deletions that affect one or more genes. These mutations impact the growth properties and tropism of the virus. Moreover, these changes have likely impaired the immunogenicity, particularly with respect to induction of epithelial cell and endothelial cell neutralizing antibodies 57,60 .
The findings of Maidji et al. support a potential role of endothelial cells in CMV transmission from the uterus to endovascular cytotrophoblasts 64 . The syncytiotrophoblast constitutes a barrier to vertical transmission, and first-trimester chorionic villi are largely resistant to CMV infection, whereas cytotrophoblasts and other villous cells are susceptible 65 . It is suggested that virions transmitted to cytotrophoblasts could then spread the infection to the placenta and to fetal blood vessels in the villus core. Considering the structure of the human placenta as an active barrier against infection, the pentamer gH/gL/pUL128-131A complex is of utmost importance in vertical CMV transmission 53 . The pUL128/pUL130/pUL131A, when assembled with the gH/gL heterodimer to form the pentameric complex, is necessary for entry into endothelial and epithelial cells, as well as for CMV transmission to leukocytes. The pentamer complex mediates virus entry into epithelial and endothelial cells by endocytosis and fusion. The entry process into fibroblasts is independent of the proteins of the pentamer complex, and the UL128L genes may be lost during long-term cultivation in human fibroblasts. Thus, this genetic locus is supposed to play an important role in cell tropism. Recent studies of the virus tropism concentrated on the UL128L of wild-type CMV strains. The pUL128/pUL130/pUL131A complex is also the target of the most potent neutralizing antibodies following natural infection 31,41,52 . The neutralization potency of CMV infection in epithelial cells by monoclonal antibodies against the UL128L complex and the gH/gL complex was 1000-fold and 10-fold higher, respectively, than that by CMV-specific hyperimmune globulin 66 . The pentamer gH/gL/pUL128-131A complex is of interest in the CMV vaccine field. gH/gL dimer may be complexed either with pUL128L forming the pentamer complex, or gO, forming the gH/ gL/gO complex. Greater amounts of pentamer gH/gL/pUL128/pUL130/pUL131A complex may result in there being less gH/gL available for the formation of trimer gH/gL/gO complex. Two viral factors, UL148 and US16, were identified to impact the composition of gH/gL complexes in CMV strains 67,68 . Moreover, the expression levels of UL128-131A and gO seem to influence the abundance of gH/gL/pUL128-131A and gH/gL/gO complexes   69 . There are at least eight genotypes of gO that differ by 10 to 30% of amino acids. Used set of primers was adequate for amplifying a region of the UL74 gene that was digested for identification of the five major gO genotypes, with two of them divided in five sub-genotypes. However, no relationship was observed between gO genotype and the outcome of CMV infection in infants. The distribution of the gO genotypes in clinical isolates from different disease settings confirmed that no correlation exists between gO type and CMV disease 23 . The present results performed on larger groups of pediatric patients confirmed our earlier findings that infection with the gH2 genotype diminishes the risk of hearing loss, whereas detection of the gH1 genotype is associated with hearing loss 17 . In addition, an association between the gH1 genotype and the incidence of neurological dysfunction was found. The heterogeneity between gH1 and gH2 genotypes is characterized by the deletion of a proline at position 36 and the substitution of lysine for histidine at position 37 in the AD-169 strain.
Most of congenital infections occur during nonprimary maternal infection, and it has been estimated that approximately three-quarters of congenital CMV infections occur in the setting of recurrent maternal infection during pregnancy 5 . A previously acquired maternal CMV infection does not provide complete protection against infection of the fetus, but it reduces this risk. It was found that antibodies targeting the CMV pentameric complex are able to neutralize genetically diverse clinical isolates of CMV 66 . Moreover, a pentameric complex of proteins elicited far more neutralizing antibodies than gB 41 . Antibodies to anti-UL128L complex are increased in the groups of non-transmitting mothers 54 . The increase in antibodies after CMV infection in pregnant women is associated with a decreased risk of congenital infection, suggesting that anti-UL128L complex antibodies play a critical role in protection against transmission to the fetus 54,70 . However, it is unknown if antibodies specific for pentamer complex can block infection of trophoblasts and restrict cCMV transmission. It was observed that these antibodies inhibit infection of term cytotrophoblasts, while no inhibition in the first-trimester trophoblast progenitor cells was found 53,71 .
This study has a number of strengths, but also some limitations. This is the first study to focus on the genes encoding the gH/gL/pUL128-131A complex of CMV in infants infected congenitally or postnatally. The main strength of the present study is the clinical evaluation of children with a CMV infection and important clinical implications. We have found significant associations of CMV variants with specific clinical symptoms. It should be noted that our results are not representative of the entire infant population and that the detected CMV variant distribution is only valid for symptomatic patients. Because the sample population of patients with specific symptoms was small, the significance of these results must be handled with caution. Further studies using larger patient groups from different geographical regions are needed to confirm our findings. Moreover, investigations of the role of intra-host recombination and functions of variants of key viral genes are necessary to elucidate the role of CMV diversity in pathogenesis. It should be underlined that we have no universal CMV screening for newborns  www.nature.com/scientificreports www.nature.com/scientificreports/ in our country. As pCMV infection was diagnosed in children above three weeks of age, it is possible that, apart from the postnatally infected majority, there are also congenitally infected infants (with no clinical findings at birth) in this group.
In summary, this report demonstrates that a relationship between the CMV variant in trimeric and pentameric complexes, including gH, pUL130 and pUL131A, and clinical outcome exists. This study suggests that the CMV variant may be one of the virological markers of congenital CMV disease. Quantification of CMV DNA. The CMV DNA copy numbers in DNA isolates were determined using a 7900HT Fast Real-Time PCR system (Applied Biosystems, Foster City, CA, USA) as previously described 72,73 . DNA was amplified using primers specific for the CMV gB (UL55) gene. The sensitivity of the assay was determined to be approximately 2 × 10 2 copies/mL. A negative control without template DNA was included in every amplification run.

Patients.
Genotyping of three genes encoding the gH/gL/gO complex of CMV. The UL75 gene (gH) was amplified by nPCR by using two sets of primers, as described elsewhere 17 . The genomic variants gH1 and gH2 were identified by different amplicon size lengths (240 bp and 237 bp, respectively). The identification of UL74 (gL) and UL115 (gO) was performed by nPCR-RFLP with the primers 5′-TAACGGGCGCTTGTTTACGT-3′ and 5′-CAGCAAAACGACCAGAATCAG-3′ (UL74) and 5′-GACGCACGGCGCGGTTGGTACG-3′ and 5′-CGTGCCGCAGACTTGATGTGCCG-3′ (UL115) in the first round and primer sets designed in our Laboratory or described previously by Stanton et al. 74 . The sequences of all primers used in PCR assays are listed in Table 6. Each run consisted of an initial denaturation at 95 °C for 1 min following by 45 Table 7. Briefly, a digestion with HpaII was used to characterize the five major gO genotypes, while double digestion with Eco24I and CseI was used to identify the four gL genotypes. The nPCR products and the digested CMV UL128 locus genotyping. To successfully amplify the genes encoding the CMV UL128L, nPCR (UL130 and UL131A) and hemi-nested PCR (UL128) assays were used. The sequences of all primers used in UL128L genotyping are listed in Table 6. The reaction mixture for UL128 and UL131A gene amplifications consisted of 0.5 μg (blood) or 0.2 μg (urine) template DNA (5 μl) and 45 μl reaction solution, as described above, while for the UL130 gene, it was as follows: 5 μl template DNA, 10 μl 5 × KAPA HiFi Fidelity Buffer, 1.5 μl 10 mM KAPA dNTP, 0.15 μl gene-specific primers, 1 μl KAPA HiFi polymerase (1 U/μl, KAPA Biosystems, Boston, MA, USA), and 32.2 μl nuclease-free water. The UL128 and UL131A reactions were performed at 95 °C for 5 min, followed by 30 cycles of 95 °C for 1 min, annealing at 55 °C for UL128 and 57 °C for UL131A for 1 min, and 72 °C for 1 min, followed by a final extension at 72 °C for 10 min. The PCR was performed in a Veriti ® 96 Well Thermal Cycler (Applied Biosystems). The PCR parameters for UL130 were as follows: 95 °C for 3 min and 35 cycles each at 98 °C for 20 s, at 59 °C for 15 s, and at 72 °C for 15 s with a final extension at 72 °C for 5 min. Positive and negative controls were included in the PCR assays, as described above. The amplicons were verified by the QIAxcel capillary   Table 7. Restriction enzymes and length of the restriction fragments. bp, base pairs.