Maternal CD4⁺ T cells protect against severe congenital cytomegalovirus disease in a novel nonhuman primate model of placental cytomegalovirus transmission

Significance

Congenital cytomegalovirus (CMV) is the leading infectious cause of childhood hearing loss and brain damage worldwide. Yet, despite its high prevalence and ranking as a top priority for vaccine development, the immune correlates of protection that could guide vaccine development remain undefined. Using a novel nonhuman primate model of congenital CMV transmission, we demonstrate a critical role for maternal CD4⁺ T cells in the induction of protective maternal immune responses that prevent fetal demise. In addition to establishing placental CMV transmission for the first time (to our knowledge) in nonhuman primates, this study reveals an association between delayed maternal virus-specific neutralizing antibody responses and severe fetal outcome, providing insight into the mechanism by which maternal CD4⁺ T cells impact congenital CMV disease.

Abstract

Elucidation of maternal immune correlates of protection against congenital cytomegalovirus (CMV) is necessary to inform future vaccine design. Here, we present a novel rhesus macaque model of placental rhesus CMV (rhCMV) transmission and use it to dissect determinants of protection against congenital transmission following primary maternal rhCMV infection. In this model, asymptomatic intrauterine infection was observed following i.v. rhCMV inoculation during the early second trimester in two of three rhCMV-seronegative pregnant females. In contrast, fetal loss or infant CMV-associated sequelae occurred in four rhCMV-seronegative pregnant macaques that were CD4⁺ T-cell depleted at the time of inoculation. Animals that received the CD4⁺ T-cell–depleting antibody also exhibited higher plasma and amniotic fluid viral loads, dampened virus-specific CD8⁺ T-cell responses, and delayed production of autologous neutralizing antibodies compared with immunocompetent monkeys. Thus, maternal CD4⁺ T-cell immunity during primary rhCMV infection is important for controlling maternal viremia and inducing protective immune responses that prevent severe CMV-associated fetal disease.

Human congenital cytomegalovirus (CMV) infection occurs in 0.7% of all pregnancies (1) and is a major cause of permanent sensorineural and neurocognitive disabilities in infants worldwide. The rate of congenital CMV transmission is as high as 50% among women who acquire primary CMV infection during pregnancy, compared with less than 2% in women with chronic infection (2). Furthermore, congenital CMV transmission following primary maternal infection causes the most severe fetal outcomes including microcephaly, intracranial cyst formation, seizures, and intrauterine growth restriction (3).

Although evident, the protective immune correlates of preconceptual immunity have not been determined. Currently, the guinea pig animal model is used to study protective immune responses against congenital CMV infection, as it is the only other species known to be susceptible to placental transmission of its species-specific CMV (4). Despite having limited sequence homology to human CMV (HCMV) (5), guinea pig CMV (GPCMV) crosses the placental barrier at a similar rate following acute maternal infection and establishes fetal infection with comparable CMV-associated sequelae (6). Several vaccine strategies including live-attenuated virus (7, 8), passive immunization of antibodies specific for glycoprotein B (gB) and the gH/gL complex (9, 10), and recombinant gB subunit immunization (11) have proven to be effective at reducing the rate of congenital GPCMV infection and preventing fetal demise. In human clinical trials, gB immunization was only 50% effective in reducing postpartum maternal virus acquisition (12) and passive immunization with CMV hyperimmune globulin of women with primary CMV infection within 6 wk of presumed acquisition did not achieve a significant reduction in the rate of fetal infection compared with placebo controls (13). These findings address the need for a relevant large-animal model with a more closely related CMV genome and wider availability of tools to assess the maternal immune system.

Nonhuman primate models are widely used in the preclinical evaluation of vaccine candidates, as they are anatomically, physiologically, and immunologically similar to humans. Furthermore, their widespread use has led to the development of an extensive set of immunological tools that allow rigorous probing and characterization of vaccine-elicited immune responses. Rhesus macaques, a common nonhuman primate animal model, have previously been used to study CMV pathogenesis in the setting of primary and secondary infection (14–16). Rhesus CMV (rhCMV), which has greater sequence and structural homology to HCMV than GPCMV (17–19), results in asymptomatic infection and establishment of a persistent and lifelong infection, similar to that in humans.. Importantly, previous studies have shown that rhCMV inoculated intraperitoneally or intracranially into the developing fetus can induce neurological defects similar to those observed in congenitally infected human infants (20, 21). However, because of the high rate of rhCMV seroprevalence among animals of reproductive age (22) and repeated exposure to rhCMV within breeding colonies where rhCMV is endemic (23), there have been no previous reports of fetal or neonatal macaques exhibiting sequelae consistent with congenital rhCMV infection.

Here, to our knowledge, we report the first nonhuman primate model of congenital CMV transmission using rhCMV-seronegative rhesus monkeys. In this model, CD4⁺ T-cell–depleted or immunocompetent rhCMV-seronegative monkeys were inoculated with a defined mixture of rhCMV strains in the early second trimester of pregnancy. Following detection of placental rhCMV transmission, fetuses were observed for signs of rhCMV-associated sequelae, and the maternal immune responses between mothers with severely affected fetuses and asymptomatic or noninfected infants were evaluated to identify potential immune correlates of protection against congenital CMV disease in nonhuman primates.

Results

Rhesus Macaque Model of Congenital rhCMV Transmission.

Two groups of rhCMV-seronegative (rhCMVsn) pregnant rhesus macaques (n = 3 each) were i.v. inoculated with a high-titer mixture of rhCMV strains, including the fibroblast-passaged 180.92 (24) and limited fibroblast-passaged, epithelial cell-tropic UCD52 and UCD59 (25) viruses in a 2:1:1 ratio during week 8 of a 24-wk gestation (Fig. S1). To maximize the likelihood of successful in utero rhCMV transmission, the first group of rhCMVsn dams (group 1) was treated i.v. with a recombinant rhesus anti-CD4⁺ T-cell–depleting antibody (CD4R1) 1 wk before rhCMV infection, as reduced CD4⁺ T-cell counts are associated with increased susceptibility to CMV infection in humans and monkeys (26–28). The second group of rhCMVsn dams (group 2) remained immunocompetent. To control for the effect of maternal CD4⁺ T-cell depletion alone on fetal outcome, three rhCMV-seropositive (rhCMVsp) pregnant females (group 3) were treated with the CD4⁺ T-cell–depleting antibody at the same gestational time point as group 1 but were not challenged with rhCMV.

Fig. S1.

Timeline for administration of CD4⁺ T-cell–depleting antibody and rhCMV inoculation of pregnant rhesus macaques. A recombinant rhesus macaque CD4⁺ T-cell–depleting antibody was administered i.v. at 50 mg/kg to four rhCMV-seronegative (rhCMVsn) (group 1) and three rhCMV-seropositive (group 3) rhesus monkeys during late first trimester of pregnancy at week 7 of gestation. At week 8 of gestation, three animals in group 1, and three additional rhCMVsn, non-CD4⁺ T-cell–depleted pregnant females (group 2) were inoculated i.v. with a mixture of rhCMV strains 180.92, UCD52, and UCD59. The fourth animal in group 1 was inoculated with rhCMV 180.92 only.

The inoculum chosen for this study included multiple rhCMV virus stocks with differential cell tropism to enhance the probability of transplacental transmission. However, a fourth rhCMVsn animal (274-98) included in group 1 was inoculated with rhCMV 180.92 only, which itself is a mixture of rhCMV variants that either encode a full-length genome or a partially deleted ULb′ region, resulting in elimination of several immune-modulatory proteins but retaining an intact UL128-UL131 region encoding surface proteins important for in vivo replication (29). The dosage of the inoculum [1–2 × 10⁶ 50% tissue culture infective dose (TCID₅₀)/strain], consistent with previous rhCMV challenge experiments (2 × 10⁵ to 2 × 10⁶ TCID₅₀) (29, 30), was administered i.v. based on studies conducted by Barry and colleagues (25) that have shown variable kinetics and magnitude of viremia following s.c. inoculation. Intravenous inoculation was expected to achieve localization of rhCMV progeny at the maternal–fetal interface with consistent kinetics. In addition, it was important to establish a proof-of-principle that rhCMV was indeed capable of transplacental transmission. Subsequent studies will refine routes of maternal inoculation to better recapitulate maternal HCMV exposure and subsequent fetal infection.

Efficiency of the maternal CD4⁺ T-cell antibody depletion was determined by flow cytometry of peripheral blood CD4⁺ T cells (Fig. S2). Within 1 wk of CD4⁺ T-cell–depleting antibody treatment, we observed a complete loss of circulating CD4⁺ T cells in group 1 and group 3 dams (Fig. S3A). Peripheral blood CD4⁺ T cells were detected again after 1–2 wks but persisted at low levels throughout the duration of pregnancy. In contrast, the number of CD4⁺ T cells in group 2 animals remained unchanged during acute and chronic rhCMV infection. As expected, there was a reciprocal increase in CD8⁺ T cells in group 1 and 3 macaques (Fig. S3B), and an expansion in peripheral effector memory CD4⁺ and CD8⁺ T cells in all rhCMV-infected animals following acute infection (Fig. S3 C and D).

Fig. S2.

FLOW cytometry gating of CD4⁺ and CD8⁺ T-cell populations. (A) A flow cytometry gating strategy was designed to positively select for CD4⁺ and CD8⁺ T cells from maternal whole blood. Sequential selection of lymphocytes; FSC and SSC singlets; CD45⁺ leukocytes; CD3⁺ T cells; and CD4⁺ versus CD8⁺ T cells from an rhCMV-seronegative, CD4⁺ T-cell–depleted group 1 monkey before rhCMV infection is shown as a reference. Central memory (CM), effector memory (EM), and naïve (N) CD4⁺ and CD8⁺ T-cell subsets were classified as CD28⁺/CD95⁺, CD28⁻CD95⁺, and CD28⁺/CD95⁻, respectively. Transitional memory (TM) T cells were characterized as CD28⁺/CCR7⁻. (B) Representative CD4⁺ and CD8⁺ T-cell plots from all three groups of rhesus monkeys before (week 7) and after (week 8) CD4⁺ T-cell depletion of group 1 and group 3 females.

Fig. S3.

CD4⁺ and CD8⁺ T-cell kinetics in CD4⁺ T-cell–depleted and immunocompetent pregnant monkeys during acute and chronic rhCMV infection. Flow cytometry analysis of CD4⁺ (A) and CD8⁺ (B) T-cell kinetics in whole blood of rhCMV-seronegative and rhCMV-seropositive pregnant females. Cell frequency, measured as the percentage of CD3⁺ T cells, and absolute CD4⁺ T-cell numbers are shown. The frequency of central memory (CM), effector memory (EM), and naïve CD4⁺ (C) and CD8⁺ (D) T cells are reported as measured by flow cytometry.

Quantitative real-time PCR (QPCR) was used to assess maternal plasma viremia in rhCMVsn CD4⁺ T-cell–depleted and immunocompetent females following rhCMV inoculation. In all dams, peak viremia occurred between 3 and 5 wks postinfection, reaching a mean peak titer of 2.87 × 10⁶ DNA copies per mL (range: 1.76 × 10⁵ to 7.5 × 10⁶) in group 1 females (Fig. 1 A–D). This was approximately one-half log higher than that of group 2 females (mean, 6.85 × 10⁵ DNA copies per mL; range, 6.79 × 10⁴ to 9.31 × 10⁵) (Fig. 1 E–G).

Fig. 1.

Congenital rhCMV transmission and pregnancy outcome in rhCMV-seronegative, CD4⁺ T-cell–depleted and nondepleted females following peak maternal viremia. (A–G) rhCMV copy number in maternal plasma and AF was assessed by QPCR in rhCMV-seronegative, CD4⁺ T-cell–depleted group 1 and immunocompetent group 2 females. Data are shown as the mean ± SD from three or more replicates. Identification numbers for the pregnant females are displayed above each graph. Crosses signify maternal euthanasia. (H) Percent survival of fetuses following maternal rhCMV inoculation or CD4⁺ T-cell depletion.

Diagnosis and Outcome of Congenital rhCMV Infection.

Detection of CMV DNA in amniotic fluid (AF) by PCR is the gold standard for diagnosing congenital CMV transmission in humans. Using amniocentesis and AF QPCR, we observed a 100% (four of four) transmission rate in CD4⁺ T-cell–depleted group 1 animals (Fig. 1 A–D) and a 66% (two of three) transmission rate in immunocompetent group 2 animals (Fig. 1 E–G), defined by having at least two of six AF PCR replicate wells positive for detectable rhCMV DNA following inoculation. The mean virus load in the AF of group 1 dams ranged from 24 to 580 rhCMV DNA copies per mL and appeared as early as 1 or 2 wks postinfection, yet appeared slightly later (2 or 4 wks postinfection) and was lower magnitude (16–71 copies per mL) in group 2 dams, potentially attributable in part to the higher viral loads in the plasma of dams lacking peripheral CD4⁺ T cells. In contrast, animals in group 3 chronically infected with rhCMV that were depleted of CD4⁺ T cells, had little to no detectable plasma viremia during pregnancy, and AF rhCMV DNA was only detected in a single PCR replicate from animal 234-07 at week 9 of gestation. This rare and inconsistent virus detection was not considered to be confirmatory of congenital rhCMV transmission. Importantly, three of the four group 1 females experienced spontaneous abortion at week 3 postinfection, whereas all group 2 dams carried fetuses to term (Fig. 1H). This high rate of fetal loss was not caused by CD4⁺ T-cell depletion alone, as the control group 3 females all delivered full-term infants. Comparison of placental tissue from CD4⁺ T-cell–depleted females that underwent spontaneous abortion to that of immunocompetent animals with asymptomatic infants showed no evidence of pathology that would indicate poor placental health as a cause of fetal loss (Fig. 2A).

Fig. 2.

rhCMV detection in placental tissue and potential CMV-associated sequelae in congenitally rhCMV-infected live-born infants. (A) Hematoxylin–eosin staining of placenta tissue from all CD4⁺ T-cell–depleted group 1 females that experienced fetal loss (369-09, 174-97, and 274-98) and from 273-98, an immunocompetent group 2 dam with an asymptomatic, congenitally infected infant. DS, decidual stroma; FV, fetal villi. (B) Immunohistochemical staining of rhCMV-IE1 protein in placental tissue from two of the group 1 females. (C) Fetal sonogram of the live-born infant from group 1 revealed a liver calcification at week 20 of gestation, indicated by the white arrow. (D) Absolute neutrophil counts from the infant(s) born to group 1, group 2, and group 3 dams are displayed as red, blue, and black lines, respectively. Dashed lines signify infants with intrauterine rhCMV infection. The lower limit of normal neutrophil counts for rhesus infants, aged 7–30 d, is marked by the black dotted line.

Of note, euthanasia was required for two of the four group 1 dams around the time of peak maternal viremia (Fig. 1 C and D). Necropsy of dam 174-97 demonstrated a massive small bowel intussusception but lacked evidence of rhCMV enteritis or other obvious causal condition, making it unclear whether loss of the fetus, which occurred days before maternal death, was a result of poor maternal health or congenital rhCMV infection. Necropsy of dam 274-98 revealed widespread rhCMV dissemination, viral myocarditis, and congestive heart failure. Although the timing of fetal loss in this instance also preceded maternal death, we cannot exclude the possibility that the spontaneous abortion was a consequence of poor maternal health related to rhCMV dissemination. Still, the fetal loss, combined with detectable rhCMV in AF, suggest an important role for maternal CD4⁺ T cells in protection against both severe maternal disease and fetal outcome during primary maternal rhCMV infection.

To investigate the extent of intrauterine rhCMV infection, we performed immunohistochemical staining for the rhCMV immediate early protein 1 (IE1) on available placental and aborted fetal tissue of group 1 dams (Fig. 2B). rhCMV-IE1 was visible in the decidual stroma and villous trophoblasts in two of the three placentas. Additionally, sonography of the rhCMV-exposed fetuses was conducted in utero to identify early signs of congenital rhCMV infection characteristic of infected human fetuses, including microcephaly, intracranial or intrahepatic calcifications, and intrauterine growth restriction. Fetal sonography of the single live-born infant from group 1 (175-13) revealed an intrahepatic calcification at week 20 of gestation (Fig. 2C). No additional signs of rhCMV infection were present in this or other rhCMV-exposed fetuses (Fig. S4 A and B).

Fig. S4.

Growth assessment and hematologic findings of rhCMV-exposed fetuses and live-born infants. Fetuses were monitored regularly for signs of intrauterine growth restriction via ultrasound. Growth as measured by (A) biparietal diameter (BPD) and (B) femur length (FL) were recorded throughout gestation for infants born to CD4⁺ T-cell–depleted group 1 (red) and inmmunocompetent group 2 (blue) females. The black line represents the predicted BPD and FL for normal rhesus macaques throughout the late first to third trimester as previously reported (49, 50). Complete blood count parameters measured within the first week of life for all live-born infants are reported (C).

Another method used to diagnose congenital CMV infection in human infants is the detection of CMV DNA in urine or saliva within the first weeks of life (31). We performed QPCR on available infant urine and saliva samples collected between 1 and 3 wks of life, and also on aborted fetal tissue and placenta from group 1 females (Table S1). rhCMV DNA was present in the placenta of all three spontaneous abortions and in all aborted fetal tissue. The three live-born infants born to rhCMVsn females with rhCMV DNA in the AF also had detectable rhCMV DNA in saliva between weeks 1 and 3 of life. Interestingly, the saliva viral load in the single live-born infant of the group 1 dams (175-13) was ∼1.5 orders of magnitude higher than other asymptomatic, rhCMV-infected infants, suggesting a higher virus burden in this infant. rhCMV was not detected in the urine or saliva of infants 267-13 and 19-14, born to the rhCMVsp group 3 dams, one of whom had detectable rhCMV DNA in the AF. However, saliva from another infant in this group (42-14) had low-level virus detectable in the first week of life, in addition to 1 of 12 PCR-positive replicates in the urine at week 3 of life and very low rhCMV DNA copies in the placenta tissue in 4 of 12 PCR replicates. These results suggest the possibility that congenital rhCMV infection can occur in naturally seropositive animals in the setting of CD4⁺ T-cell depletion, an important finding for the assessment of maternal immune responses and congenital transmission during virus reactivation or reinfection.

Table S1.

Detection of rhCMV DNA confirms intrauterine rhCMV transmission

Placenta/aborted fetuses						Live-born Infants
							Urine		Saliva		Breast milk
Group (no.)	Mother	Sample type	*Mean (±SD)	†Freq	Infant	Infant week of life	Mean (±SD)	Freq	Mean (±SD)	Freq	Mean (±SD)	Freq
(1)	145-97	NA			175/13	1	‡–	0/12	790 (±589)	11/12	30,370 (+5,193)	12/12
rhCMVsn						3	–	0/12	1,057 (±460)	12/12
CD4−	369-09	Placenta	24,228 (±920)	12/12	NA
		Cord	2,699 (±268)	12/12
		Liver	709 (±78)	12/12
	174-97	Placenta	2,477 (±199)	12/12	NA
		Cord	3,920 (±1,156)	12/18
		Liver	1,849 (±364)	12/12
		Spleen	880 (±127)	12/18
		Lung	1,114 (±458)	12/12
		Jejunum	1,220 (±119)	12/12
		Heart	2,497 (±206)	12/12
		Occipital cortex	595 (±459)	12/18
		Frontal cortex	696 (±865)	12/12
	274-98	Placenta	179 (±105)	10/12	NA
(2)	273-98	Placenta	1 (±2)	1/12	27-14	2	32 (±77)	2/12	34 (±81)	2/12	28,048 (+6,143)	12/12
rhCMVsn						3	–	0/12	67 (±140)	3/12	5,416 (+2,906)	12/12
CD4+
	223-98	NA			270-13	1	–	0/12	–	0/12
						3	NA		36 (±85)	2/12
	251-05	Placenta	–	0/12	58-14	1	–	0/12	–	0/12
						2	–	0/12	–	0/12
						3	–	0/12	–	0/12
(3)	234-07	NA			267-13	2	–	0/12	–	0/12
rhCMVsp
CD4−	309-09	Placenta	1 (±1)	4/12	42-14	1	–	0/12	28 (±67)	2/12
						3	22 (±55)	1/12	–	0/12
	222-02	§NA			19-14	2	–	0/12	–	0/12

^*Mean (±SD), mean CMV copy number per microgram of DNA input.

^†Freq, number of PCR positives/number of PCR replicates.

^‡–, not detected.

^§NA, not available.

Following birth, rhesus infants were also assessed for hematologic abnormalities, such as thrombocytopenia and neutropenia, which are characteristic of congenital CMV infection in humans (32). Infant 175-13, with the high saliva virus load and identified liver lesion, also displayed neutrophil counts below the normal level for juvenile monkeys at the time of birth and throughout 28 wks of life (Fig. 2D). One uninfected infant (58-14) born to a group 2 female displayed low levels of all hematopoietic cell types, yet no other major hematologic abnormalities were observed in the rhCMV-exposed infants at birth (Fig. S4C).

Identification of the Congenitally Transmitted/Founder Virus.

Recent deep-sequence analysis revealed that, despite high CMV intrahost genetic variability in adults, human congenital CMV infection appears to result from a limited number of transmitted/founder variants (33). We sought to identify which of the three rhCMV strains included in our inoculum replicated most efficiently in maternal plasma of group 1 animals and which strain(s) crossed the placental barrier to result in congenital rhCMV infection. We performed single-genome amplification (SGA) and sequence analysis of the rhCMV UL128 exon 1 region, which had sufficient nucleotide variability to distinguish between the three viral stocks (Fig. S5A). Between 90% and 100% of plasma variants had sequence homology with rhCMV UCD52, whereas the remaining plasma variants were identified as rhCMV UCD59 (Fig. S5B). To confirm our findings, we performed deep-sequencing analysis on two of the three maternal plasma samples. In our analysis, we identified 20 regions across the genome, 100–280 bp in length, for which there were five or more reads and that included at least two single nucleotide polymorphisms (SNPs) that could be used to distinguish between rhCMV 180.92, UCD52, and UCD59 (Fig. S5E). Similar to SGA, deep sequencing revealed UCD52 as the dominant strain in the plasma of animal 369-09 in all 20 genes examined. However, data from the second animal, 174-97, indicate plasma variants to be a mixed population of UCD52 and UCD59 variants as exactly one-half of the gene loci examined (10 of 20) were either UCD52 or UCD59 dominant. This discrepancy is likely a result of limited depth of genome coverage obtained by nonspecific deep-sequence analysis as many of the genes were almost equally represented by both rhCMV UCD52 and UCD59 variants, or alternatively may be due to the increased sensitivity of this assay. The absence of rhCMV 180.92 by SGA, and limited detection by deep sequencing in each of these monkeys is not surprising as only 5% of the variants within this fibroblast-adapted stock encode a full-length genome whereas the remaining 95% lack a portion of the ULb′ region encoding numerous immune-modulatory genes that are important for in vivo replication (29). Accordingly, the group 1 dam (274-98) that was infected with rhCMV 180.92, but not UCD52 or UCD59, had only rhCMV 180.92 sequences detected in maternal plasma, of which 86% of the circulating virus encoded the full-length genome (Fig. S5 C and D). We also performed SGA on DNA extracted from the AF of all three group 1 dams that received a swarm of rhCMV strains which revealed the presence of only rhCMV UCD52 (Fig. S5B). These findings suggest that congenital rhCMV infection in rhesus monkeys may result from a restricted number of transmitted viruses, or alternatively that epithelial-tropic UCD52 may have a fitness advantage for systemic and intrauterine virus replication that should be further explored.

Fig. S5.

Epithelial-tropic rhCMV strains are dominant in the maternal plasma and amniotic fluid (AF) of CD4⁺ T-cell–depleted rhesus monkeys. Single-genome rhCMV variants were amplified from DNA extracted from the maternal plasma and AF using primers specific for a 350-bp region of exon 1 from the UL128 locus. (A) Single-nucleotide polymorphisms (SNPs) located within the UL128 exon 1 region between each of the challenge virus strains used in this study (rhCMV UCD59, 180.92, and UCD52) are depicted in a highlighter plot. (B) rhCMV variants isolated from the plasma and AF of CD4⁺ T-cell–depleted group 1 females inoculated with all three rhCMV strains were identified by single-genome amplification and sequence analysis. (C) Identification of plasma rhCMV variants isolated from the group 1 female receiving only rhCMV 180.92 (274-98) by single-genome amplification. (D) The proportion of rhCMV 180.92 variants with a full-length (ULb′) or deleted Ulb′ region (WT) from DNA extracted from the plasma of the group 1 female receiving only rhCMV 180.92 was determined by variant-specific quantitative RT-PCR analysis. (E) The frequency of each rhCMV strain in the plasma of 369-09 and 174-97 at the peak of viremia was identified in 20 regions across the genome for which deep-sequence analysis yielded five or more reads. The dominant strain at each locus is highlighted in green.

Maternal rhCMV-Specific Cellular Immunity.

One mechanism by which depletion of CD4⁺ T cells could have impacted fetal outcome following acute rhCMV infection, is through its effect on both the cellular and humoral arms of adaptive immunity. CD4⁺ T cells are important for generation of the pathogen-specific memory CD8⁺ T cells, yet the role of maternal CD8⁺ T-cell function in congenital CMV transmission is unknown. To determine whether maternal CD8⁺ T-cell function is correlated with protection against severe congenital rhCMV disease in rhesus monkeys, we compared the CD8⁺ T-cell TNFα, IFNγ, and IL-2 intracellular cytokine response to a pool of overlapping rhCMV-phosphoprotein 65 (pp65) and IE1 peptides in group 1 and group 2 dams (Fig. S6). At 3 wk postinfection, concurrent with the timing of fetal loss, only low levels of rhCMV pp65-specific CD8⁺ T-cell responses were observed in either group (Fig. 3). In contrast, IE1-specific responses were present and generally higher magnitude in group 2 compared with group 1 dams, with TNFα-secreting cells being the most abundant (Fig. 3). The observed differences in CD8⁺ T-cell cytokine production raise the possibility that the absence of maternal CD4⁺ T cells in group 1 females may have caused a blunting effect, albeit modest, on the early rhCMV-specific CD8⁺ T-cell response, which could contribute to the adverse pregnancy outcomes following primary rhCMV infection. Unlike the rhCMV-specific CD8⁺ T-cell responses, rhCMV-specific CD4⁺ T-cell responses of both CD4⁺ T-cell–depleted and nondepleted seronegative dams were mostly undetectable during acute infection (Fig. S7).

Fig. S6.

Gating strategy for CD4⁺ and CD8⁺ intracellular cytokine-producing T cells. The gating strategy for identifying cytokine-producing CD4⁺ and CD8⁺ T cells from maternal PBMCs isolated from whole blood at various time points throughout pregnancy included time gating; FSC singlets; viability gating; CD3⁺ T cells; CD4⁺ versus CD8⁺ T-cell surface markers; and TNFα⁺/IFNγ⁺/IL-2⁺ intracellular markers. Data shown represent gating rhCMV IE1-stimulated PBMCs from a non-CD4⁺ T-cell–depleted group 2 dam at 3 wk postinfection.

Fig. 3.

Intracellular cytokine production by CD8⁺ T cells in CD4⁺ T-cell–depleted and immunocompetent, acutely rhCMV-infected pregnant monkeys. Flow cytometry analysis of maternal rhCMV-pp65 (Top) or IE1-specific (Bottom) intracellular cytokine CD8⁺ T-cell responses in group 1 and group 2 females at 3 wks postinfection. Data are expressed as absolute number of cytokine-producing CD8⁺ T cells per microliter of whole blood after subtraction of background of unstimulated cells.

Fig. S7.

Intracellular cytokine production by CD4⁺ and CD8⁺ T cells in CD4⁺ T-cell–depleted and immunocompetent, acutely rhCMV-infected pregnant monkeys. Flow cytometry analysis of rhCMV-specific maternal CD4⁺ (A) and CD8⁺ (B) intracellular cytokine T-cell responses in CD4⁺ T-cell–depleted group 1 and immunocompetent group 2 females following stimulation of isolated PBMCs with rhCMV-pp65 (Top) or IE1 (Bottom) peptides. Data are expressed as absolute number of cytokine-producing CD4⁺ (A) and CD8⁺ (B) T cells per microliter of whole blood after subtraction of background of unstimulated cells. Identification numbers for each pregnant female are displayed below the x axis.

Maternal rhCMV-Specific Humoral Immunity.

Previous studies have indicated that, although virus-specific antibody production in pregnant women with primary CMV infection does not differ between transmitting and nontransmitting females, the presence of high-avidity CMV-specific IgG responses is associated with reduced risk of congenital HCMV transmission (34). Thus, we measured both the kinetics of maternal rhCMV-specific IgM and IgG antibody responses (Fig. 4 A and B) and IgG avidity (Fig. 4C) in the group 1 and group 2 dams. Peak rhCMV-specific IgM antibody responses appeared by week 2 postinfection in most animals, but were 0.5–1 log lower in group 1 females (Fig. 4A). The kinetics of the acute maternal rhCMV-specific IgG responses were more variable between and within study groups than the IgM responses; however, similar magnitude peak titers were reached by week 3 postinfection in all dams (Fig. 4B). The IgG avidity index was also similar in both CD4⁺ T-cell–depleted and nondepleted dams at weeks 3 (P = 0.857) and 12 postinfection (P = 0.8), yet were lower than that of chronically rhCMV-infected monkeys at both time points (P = 0.03, P = 0.056; Fig. 4C).

Fig. 4.

The appearance of rhCMV-neutralizing antibodies is delayed in acutely rhCMV-infected, CD4⁺ T-cell–depleted females. (A and B) rhCMV-specific IgM and IgG antibody endpoint titers in plasma from group 1 (red) and group 2 (blue) females were determined by ELISA. Dotted lines represent the mean IgM and IgG endpoint titer of three chronically infected rhesus macaques. (C) rhCMV IgG antibody avidity was measured in group 1 (red) and group 2 (blue) females. The IgG avidity index of five chronically infected rhesus monkeys (♦) is shown as a reference for high avidity. (D–F) EC₅₀ titers were measured for IgG antibodies specific to rhCMV gB (D), gH/gL (E), and the pentameric complex (F) in plasma from group 1 (red) and group 2 (blue) females. The limit of detection as indicated by the lower dotted lines is the average of week 0 time points plus 2 SD. The mean IgG antibody EC₅₀ titer from six chronically infected animals is represented by the upper dotted line. (G and H) Neutralizing activity of maternal group 1 (red) and group 2 (blue) plasma against rhCMV 180.92 in rhesus fibroblasts and rhCMV UCD52 in monkey epithelium. Dotted lines indicate the mean neutralizing antibody titers of three chronically infected rhesus monkeys.

Preexisting CMV-neutralizing antibodies, particularly those directed against the gH/gL/UL128-131 pentameric glycoprotein complex (gH/gL-PC), in the sera of naturally seropositive women are hypothesized to contribute to the partial protection against congenital HCMV transmission (35). In our study, all group 2 dams had detectable gB and gH/gL-PC specific antibody responses by week 3 postinfection, and two of the three group 2 dams, including the single nontransmitter, also had antibodies specific for the gH/gL complex as early as 2 wks postinfection (Fig. 4 D–F). In contrast, these responses were not present in group 1 dams at 2 or 3 wks postinfection. Of the two surviving CD4⁺ T-cell–depleted group 1 females, gB binding antibodies did not appear until 5–8 wks postinfection (Fig. 4D), and only one group 1 dam developed antibodies to the gH/gL-PC over the course of pregnancy, emerging at week 5 postinfection (Fig. 4F). We next determined the neutralizing capacity of antibodies in maternal plasma at early time points after rhCMV inoculation in both fibroblast and epithelial cell lines using the autologous challenge virus strains rhCMV 180.92 and UCD52, respectively. Neutralizing responses detected in both fibroblasts and epithelial cells developed more rapidly and were higher magnitude in group 2 dams than group 1 dams at week 3 postinfection (Fig. 4 G and H). The results from these studies suggest that maternal production of rhCMV-neutralizing antibodies against gB and gH/gL-PC within the first 3 wks following rhCMV inoculation is impacted by the presence or absence of maternal CD4⁺ T-cell immunity and could play a significant role in fetal outcome following rhCMV infection independent of maternal health.

Discussion

To our knowledge, our study is the first to report intrauterine CMV transmission in a nonhuman primate animal model. In this model, congenital rhCMV transmission was observed in CD4⁺ T-cell–depleted and immunocompetent, rhCMVsn dams following i.v. rhCMV inoculation during the early second trimester of pregnancy. As in humans, congenital rhCMV infection was diagnosed via the detection of viral DNA in AF, and confirmed by detection of virus in aborted tissue or infant saliva and urine collected within the first few weeks of life.

In our study, the rate of congenital rhCMV transmission was 66% (two of three) among immunocompetent dams. Although the two infected infants in this group showed no congenital rhCMV-associated sequelae, only a small percentage (10–15%) of congenitally infected human newborns display clinical signs of HCMV infection at birth (1). The rhCMV transmission rate in rhCMVsn, CD4⁺ T-cell–depleted dams was 100% and resulted in a high rate (three of four) of spontaneous abortion and a single live-born infant with detectable sequelae of infection. Although congenital CMV is a known cause of fetal loss in humans (36) and is frequently found in the tissues of stillborn fetuses (37), two of the four CD4⁺ T-cell–depleted dams required euthanasia shortly after the discovery of fetal loss introducing the confounding factor of poor maternal health contributing to adverse fetal outcome. Poor CD4⁺ T-cell immunity in many clinical settings is strongly associated with enhanced susceptibility to CMV infection (26–28), which was true for both pregnant dams and their fetuses in this study. However, it is evident from our study that congenital rhCMV transmission can occur in immunocompetent dams following primary maternal rhCMV infection. To avoid the confounding effect of maternal health in this model, future experiments could minimize the viral titer used or exclude CD4⁺ T-cell depletion. Also included in our model was a group of CD4⁺ T-cell–depleted, naturally rhCMVsp dams. Among this group, a single infant had detectable virus in saliva during the first week of life yet had undetectable viral DNA in AF. Further investigation into whether congenital rhCMV transmission can occur in rhCMVsp dams will be critical, as approximately one-half of all congenital HCMV infections in developed regions are a result of secondary maternal infection.

It is evident from our study that CD4⁺ T cells modulate maternal immune responses during acute infection and may be important determinants of fetal outcome. During acute infection, CD4⁺ T-cell cytokine production and the interaction between CD4⁺ T cells and antigen-presenting cells stimulate proliferation of the cytotoxic CD8⁺ T-cell population, which in turn eliminates virus-infected cells. Impaired CD4⁺ T-cell functionality in primary human and rhesus CMV infection are associated with higher viral loads and prolonged virus excretion (15, 38, 39). Thus, as expected, CD4⁺ T-cell–depleted dams exhibited higher plasma virus loads than immunocompetent dams, which could have directly contributed to increased rates of placental transmission. Despite having more replication, the CD8⁺ T-cell response to rhCMV-specific proteins was dampened in CD4⁺ T-cell–depleted dams at 3 wks postinfection when fetal loss was observed, most notably TNFα production. In humans, pregnancy outcome is largely dependent on the balance of cytokines, particularly TNFα, which is expressed in almost every cell type found at the fetal–maternal interface (40). High levels of TNFα have been directly linked to fetal loss (41), but this cytokine is also required for the recruitment of macrophages that participate in maintaining pregnancy (40). Thus, it may be of interest to investigate the role of virus-specific CD8⁺ T-cell cytokine production and function in rhCMV placental transmission, or to more closely examine these responses during severe and asymptomatic congenital rhCMV infection.

Antibody production and maturation following acute rhCMV infection are also dependent on CD4⁺ T-cell help. We did not, however, observe notable differences in the levels of rhCMV-reactive IgM and IgG antibodies or in IgG antibody avidity between CD4⁺ T-cell–depleted and immunocompetent dams. Instead, we observed reduced rhCMV glycoprotein-specific antibody responses to both gB and the pentameric complex, and a delay in antibody neutralization function in fibroblast and epithelial cells at 2 and 3 wks postinfection in animals that received the CD4⁺ T-cell–depleting antibody. Although our findings are supportive of previous studies that implicate a correlation between weakly neutralizing antibodies and enhanced rates of congenital HCMV transmission (42), it remains to be determined whether delivery of strongly neutralizing rhCMV antibodies are sufficient for protection against rhCMV transmission and disease, and which glycoprotein complex(es) should be targeted. However, recent analysis of HCMV transmitting and nontransmitting women identified an association between the presence of neutralizing antibodies against the pentamer complex and reduced congenital infection (35). This, together with our findings that viruses encoding the full-length ULb′ region, which includes UL128–UL131 genes required for epithelial cell entry, replicated most efficiently in vivo and were the only viruses detected in AF stresses the importance of exploring passive immunization as a strategy for maternal vaccination and highlights the need to determine whether viruses lacking the pentameric complex can establish fetal infection.

The current guinea pig model of congenital CMV infection is a mainstay in the field, with the distinct advantage that these small animals are more high-throughput and less resource-heavy than large-animal models. Furthermore, the structural resemblance of the guinea pig placenta to that of humans and longer gestational period compared with other rodent models contribute to its value as a suitable model to test novel vaccine and therapeutic strategies (43). However, preclinical vaccine studies that have shown significant reduction in the rate of congenital infection and pup mortality in guinea pigs have not often predicted the outcome of human trials, a likely reflection of the evolutionary differences of both the species and the species-specific CMV (7–10). Our development of a rhesus monkey model of congenital CMV transmission that is more physiologically similar to humans in reproductive and placental biology (44) and uses a more genetically related CMV to that of humans than small-animal models (17, 18) provides the field with a highly relevant tool that may better predict preclinical efficacy of vaccine candidates. In addition, given the availability of vast resources to evaluate the rhesus immune response, the rhesus monkey congenital CMV model may allow more rigorous analysis of the maternal immune correlates of protection against congenital CMV infection, including the role of cell-mediated immunity that could inform a protective vaccine.

Materials and Methods

Animals and Procedures.

The animal protocol was approved by the Harvard Medical School Area Standing Committee on Animals and the Duke University Medical Center Institutional Animal Care and Use Committee. Harvard Medical School has an approved Animal Welfare Assurance on file with the Office for Laboratory Animal Welfare (Assurance number A3431-01). Indian rhesus macaques were housed at the New England Primate Research Center and maintained in accordance with institutional and federal guidelines for the care and use of laboratory animals (45). Paired mating, pregnancy screening, sample collection, experimental CD4⁺ T-cell depletion, and rhCMV inoculation were performed as described in SI Materials and Methods.

Diagnosis of Congenital rhCMV Infection and Genomic Analysis.

Congenital rhCMV infection was diagnosed following the detection of rhCMV DNA in maternal AF by QPCR and was further confirmed by immunohistochemistry staining of placental tissue using an anti-rhCMV IE1 polyclonal rabbit sera and assessment of virus load in fetal tissue or infant saliva and urine specimens by QPCR. Details for each diagnostic assay are provided in SI Materials and Methods. SGA and deep-sequence analysis were used to determine the dominant rhCMV strain replicating in maternal plasma and AF samples. Primer sets, conditions, and the method of analysis are outlined in SI Materials and Methods.

Humoral and Cellular Immune Analysis.

Maternal antibody kinetics, avidity, and specificity were measured using whole virion or MVA ELISAs and are further described in SI Materials and Methods. Neutralization EC₅₀ titers obtained from telo-RFs and MKEs were calculated using the method of Reed and Muench following manual counting of rhCMV-infected cells stained with a fluorescently labeled rhCMV IE1 monoclonal antibody. Cell maintenance and assay conditions are reported in SI Materials and Methods. Characterization of maternal peripheral CD4⁺ and CD8⁺ T cells was performed by FLOW cytometry using the fluorescently labeled monoclonal antibodies listed in Table S2. Gating strategies for each population are outlined in Fig. S2. Additional gates used to analyze CD4⁺ and CD8⁺ intracellular cytokine T-cell responses are reported in Fig. S7. Detailed protocols are available in SI Materials and Methods.

Table S2.

Fluorescently labeled antibodies used in flow cytometry analysis

Marker	Fluorophore	Clone	Manufacturer	Experiment	Staining
CCR7	FITC	150503	R&D Systems	Phenotype/ICS	Surface
CD95	PE	DX2	BD Biosciences	Phenotype/ICS	Surface
CD28	PerCP-Cy5.5	L293	BD Biosciences	Phenotype/ICS	Surface
CD4	PE-Cy7	L200	BD Biosciences	Phenotype	Surface
CD20	APC	L27	BD Biosciences	Phenotype	Surface
CD3	AF700	SP34-2	BD Biosciences	Phenotype/ICS	Surface
CD8	APC-Cy7	RPA-T8	BD Biosciences	Phenotype/ICS	Surface
CD45	VD450	D058-1283	BD Biosciences	Phenotype	Surface
CD4	BV605	OKT4	BioLegend	ICS	Surface
IFNγ	PE-Cy7	B27	BD Biosciences	ICS	Intracellular
IL-2	APC	MQ-117H12	BD Biosciences	ICS	Intracellular
TNFα	eFluor450	Mab11	eBioscience	ICS	Intracellular

ICS, intracellular cytokine staining.

SI Materials and Methods

Animals.

Before enrollment in our study, seven extended specific pathogen-free (SPF) females were confirmed to lack rhCMV-specific IgM and IgG antibodies by whole-virion ELISA (Antibody Kinetics and Avidity). rhCMV-seronegative males and females were then pair-housed and screened every 3 wk for pregnancy via ultrasound. A similar strategy was used for monitoring pregnancy in separately housed rhCMV-seropositive females from the standard SPF colony at New England Primate Research Center, which is free of simian type D retrovirus, STLV1, SIV, and herpes B virus infection. Females who remained infertile for greater than 3 mo of pair housing were given 3–5 d of 3 mg/kg clomiphene beginning at day 5 of the menstrual cycle (with day 1 being the first day of menstruation as determined by daily vaginal swabbing) to stimulate ovulation. Gestational dating was performed on sonography by measuring the gestational sac size (average of three dimensions) and crown rump length of the fetus. At week 7 of gestation, four rhCMV-seronegative and three rhCMV-seropositve dams were administered an i.v. 50 mg/kg dose of recombinant rhesus anti-CD4⁺ T-cell–depleting antibody (CD4R1 clone; Nonhuman Primate Reagent Resource). During week 8 of gestation, three of the four CD4⁺ T-cell–depleted rhCMV-seronegative females and three additional non-CD4⁺ T-cell–depleted rhCMV-seronegative females received an i.v. inoculation with 2 × 10⁶ TCID₅₀ rhCMV 180.92 (24) in one arm, and 1 × 10⁶ TCID₅₀ each of rhCMV UCD52 and UCD59 (25) in the other arm (diluted in serum-free RPMI). The fourth CD4⁺ T-cell–depleted, rhCMV-seronegative dam was inoculated i.v. with 2 × 10⁶ TCID₅₀ rhCMV 180.92 only. Pan urine collection, saliva sampling via salivette, and blood draws were performed on all dams weekly for the first 4–6 wks after rhCMV infection and then biweekly to monthly throughout the course of study. Breast milk was collected by manual massage during the first week postpartum. Amniotic fluid (AF) was also collected at weekly to biweekly intervals for up to 6 wks following rhCMV inoculation by amniocentesis. When available, placental and fetal products were obtained from the cage and fixed in formalin for hematoxylin–eosin staining and immunohistochemistry or frozen. Fetal growth was monitored by measuring biparietal diameter and femur length during weekly to biweekly fetal sonography over the course of gestation. Sonography was also used to screen for signs of congenital rhCMV-associated sequelae. Additionally, cranial sonograms were performed on infants after birth to identify any brain lesions. Infant urine, saliva, and blood were collected as described for dams during infant examination.

Tissue Processing and Staining.

Standard immunoperoxidase staining for rhCMV was performed on formalin-fixed, paraffin-embedded sections of multiple tissues. Formalin-fixed, paraffin-embedded sections were deparaffinized in xylene, rehydrated in graded alcohol, and subsequently blocked with 3% hydrogen peroxide in PBS. Pretreatment involved microwaving for 20 min in 0.01% citrate buffer, followed by 20 min of cooling at room temperature. Following pretreatment, an avidin–biotin block (Invitrogen) and a 10-min Dako Protein Block were conducted on all sections. A wash of Tris-buffered saline (TBS) with 0.5% Tween 20 followed each step. Sections were incubated with anti–rhCMV-IE1 polyclonal rabbit sera kindly provided by P.A.B. at a 1:1,600 dilution for 30 min at room temperature. Slides were then incubated with biotinylated goat anti-rabbit IgG (Vector Laboratories) at a 1:200 dilution for 30 min at room temperature. This was followed by 30-min incubation at room temperature with Vectastain ABC Standard (Vector Laboratories). Immunolabeling was visualized using diaminobenzidine and counterstained with Mayer’s hematoxylin. Irrelevant, isotype-matched primary antibodies were used in place of the test antibody as negative controls in all immunohistochemical studies. Positive control tissues consisted of archived rhesus macaque lung and testis from an rhCMV-seropositive animal.

Characterization of T Cells by Flow Cytometry.

Characterization of maternal peripheral CD4⁺ and CD8⁺ T cells was performed by mixing 100 µL of EDTA-anticoagulated blood with a pool of fluorescently labeled monoclonal antibodies including CCR7, CD95, CD28, CD4, CD20, CD3, CD8, and CD45 (Table S2). After a 30-min incubation at 4 °C, cells were washed with PBS supplemented with 2% FCS and pelleted at 100 × g for 5 min. Red blood cells were then lysed by adding 1× BD lysis buffer for 40 min at room temperature. Intact cells were centrifuged at 150 × g for 5 min, washed once with PBS containing 2% FCS, and resuspended in 100 µL of PBS with 2% FCS. Cells were fixed by adding 15 µL of 2% formaldehyde and processed by flow cytometry. Gating strategies are outlined in Fig. S2.

Assessment of Virus Load.

rhCMV virus load was measured in plasma, AF, breast milk, saliva, urine, and tissues by quantitative rhCMV-specific PCR following DNA extraction. Urine and mouth washes were concentrated using Ultracel YM-100 filters (Amicon) and frozen for subsequent DNA extraction. DNA was extracted from urine using the QIAmp RNA minikit (Qiagen) and from AF, breast milk, saliva, and plasma by using the QIAmp DNA minikit. DNA was extracted from 10–25 mg of snap-frozen tissue after overnight Proteinase K digestion using the DNeasy Blood and Tissue kit (Qiagen). rhCMV DNA was quantitated by real-time PCR using the 5′-GTTTAGGGAACCGCCATTCTG-3′ forward primer, 5′-GTATCCGCGTTCCAATGCA-3′ reverse primer, and 5′-FAM-TCCAGCCTCCATAGCCGGGAAGG-TAMRA-3′ probe, which amplify and detect a 108-bp region of the rhCMV IE1 gene (26). Between 5 and 10 ng of DNA was added as template to 50 µL of 1× PCR mixture containing 300 nM of each primer, 100 nM probe, 2 mM MgCl₂, 200 µM each of dATP, dCTP, and dGTP, 400 µM dUTP, 0.01 U/µL Amperase UNG, 0.025 U/µL Taq polymerase, and reaction buffer that includes passive reference dye ROX. PCR conditions consisted of an initial 2-min cycle at 50 °C followed by 10 min at 95 °C and 45 cycles of denaturation at 95 °C for 15 s and combined annealing/extension at 60 °C for 1 min. Data are expressed as copies per milliliter for plasma, AF, and breast milk, and as copies per microgram input DNA for tissues, saliva, and urine. rhCMV DNA was quantitated using a standard consisting of a plasmid containing the entire rhCMV-IE1 gene. rhCMV DNA was detected with a linear dynamic range from 10⁰ to 10⁶ copies in the presence of genomic DNA.

Single-Genome Amplification.

DNA was extracted from maternal plasma and AF using the QIAamp MinElute Virus Spin Kit (Qiagen) and added to 96-well PCR plates at a concentration that resulted in amplification of less than 30% of the total number of replicate reactions (46). Each well contained 1× High Fidelity PCR buffer, 2 mM MgSO₄, 100 nM of forward and reverse primers, and 2 U of High Fidelity Taq polymerase in a 25-µL volume (Invitrogen). In the first round of amplification, forward primer 5′-CGGTTGTTGCATATTTTGAA-3′ and reverse primer 5′-AAGCGAACGGCGAGATCAAC-3′ produced a 554-bp fragment corresponding to regions of UL128 exon 1 and UL130 exon 2 following the PCR parameters: 94 °C for 2 min, 40 cycles of denaturation at 94 °C for 30 s, an annealing step at 55 °C for 30 s, and extension for 50 s at 72 °C with a final extension at 72 °C for 10 min. In the second round of amplification, 5 µL from each well of the first reaction was used as template and run under the same PCR conditions using nested forward primer 5′-AAGCAGTAAGATGCAAACGAT-3′ and reverse primer 5′-AGCAGCGAAACCAGCTCCA-3′ for a final 368-bp fragment. Products were purified following the Wizard SV Gel and PCR clean-up kit protocol (Promega) and sent for sequence analysis.

rhCMV Deep Sequencing from Plasma.

From each animal, 0.5 mL of blood plasma was centrifuged at 3,000 × g (4 °C, 5 min) with subsequent filtration of the supernatant through a 0.45-μm filter (Millipore) to remove residual host cells. A mixture of nucleases consisting of DNAfree DNase (Ambion), Baseline Zero DNase (Epicentre), RNase A (Epicentre), and Benzonase (Sigma-Aldrich) was then used to digest nucleotides in the plasma for 90 min at 37 °C, while rhCMV DNA was protected by the viral capsid. Viral DNA was then isolated using the QIAamp MinElute virus spin kit (Qiagen) according to the manufacturer’s instructions, but omitting carrier RNA. Approximately 1 ng of DNA was then subjected to simultaneous fragmentation and adaptor ligation with the Nextera XT DNA sample preparation kit (Illumina) to generate deep-sequencing libraries. Sequencing on the Illumina MiSeq (Illumina) resulted in ∼16.9 million reads. Sequencing data were analyzed using CLC Genomics Workbench 7.1 (CLC bio) and Geneious, version 7.1.7. Low-quality (<q20) and short reads (<90 bp) were removed, and the remaining reads were mapped against the complete genome sequence of rhCMV strain 180.92 reference (DQ120516) to assess coverage across the genome. Trimmed paired reads from 180.92 stock virus was reiteratively assembled to the complete genome sequence of rhCMV strain 180.92 reference to generate a consensus sequence for 180.92 stock virus. Trimmed paired reads from UCD59 stock virus were repetitively assembled by de novo assembly, and then assembled to the complete genome sequence of rhCMV strain UCD52 to generate a consensus sequence for UCD59. SNPs differentiating the three strains were identified by alignment of the consensus sequences of each strain. To quantify the presence of the three different rhCMV stocks (180.92, UCD52, and UCD59) that each animal was infected with, trimmed reads were mapped against genes with high coverage and the total number of individual reads with SNPs differentiating each strain was calculated.

Quantitation of Wild-Type and Full-Length rhCMV 180.92.

The primer–probe sets and plasmid standard used in this study to distinguish between wild-type and full-length rhCMV 180.92 were designed by Assaf et al. (29).

The primer–probe set that amplifies the rhCMV 180.92 wild-type, truncated ULb′ region consists of the 5′-TTCAGGTGAATGGAGTGGTTTCGG-3′ forward primer, 5′-GCTTGACGAGGATGTCTTCGAAGTGT-3′ reverse primer, and 5′-FAM-TTTGCCCAG-ZEN-GATGGGTGCCGCATCT-IABkFQ-3′ probe. The second primer–probe set that amplifies the B segment of ULb′ of both wild-type and full-length rhCMV 180.92 variants includes the 5′-TTGTCGCAGTAAGAACGGTGGTGA-3′ forward primer, 5′-TTAATACCGACGCGGGTTCACAGT-3′ reverse primer, and 5′-FAM-AGTGCTGTT-ZEN-GGCAGTGGTGGTATTGT-IABkFQ-3′ probe.

DNA extracted from plasma using the High Pure Viral Nucleic Acid Extraction Kit (Roche) was added to a 96-well PCR plate in triplicate at a concentration of 10⁴ to 10⁵ copies per reaction and included 1× Platinum Quantitative PCR Supermix-UDG (Invitrogen), 5 mM MgSO₄, 0.25 µM forward and reverse primer sets, 0.2 µM probe, and 50 nM ROX reference dye in 50-µL total volume. PCR conditions were set for 1 cycle each of 50 °C for 2 min and 95 °C for 2 min followed by 40 cycles at 95 °C for 15 s and 60 °C for 30 s on an Applied Biosystems 7300 Fast Real-Time PCR system. Standard curves were generated from serial dilutions of the plasmid standard ranging from 10² to 10⁹ copies per reaction where 100 copies was the limit of detection.

Intracellular Cytokine Staining.

Cryopreserved peripheral blood mononuclear cells (PBMCs) isolated from maternal whole blood were thawed and rested overnight at 37 °C in 5% CO₂ at a maximum concentration of 2 × 10⁶ cells per mL . The next day, PBMCs were divided into three reaction tubes, each containing 1 × 10⁶ cells or less, and treated for 6 h at 37 °C in 5% CO₂ with 1 μg/mL pool of 15-mer overlapping rhCMV-pp65 or IE1 peptides in the presence of 2 μg/mL mouse anti-human CD49d antibody (BD Biosciences), or with CD49d antibody alone. After 1 h, Golgi plug (BD Biosciences) was added to inhibit protein transport. Following the 6-h stimulation, PBMCs were left overnight at 4 °C and then stained extracellularly with CCR7, CD95, CD28, CD3, CD4, and CD8 antibodies (Table S2). Cells were incubated with Aqua Vital dye, fixed and permeabilized with BD Cytofix/Cytoperm (BD Biosciences), and stained intracellularly with antibodies specific for IFNγ, IL-2, and TNFα (Table S2). The gating strategy for CD4⁺ and CD8⁺ cytokine-producing cells is outlined in Fig. S7. Data shown are the absolute numbers of CD4⁺ or CD8⁺ T cells that intracellularly expressed IFNγ, IL-2, or TNFα after subtraction of background levels from unstimulated PBMCs of the same monkey and time point postinfection, and were calculated using absolute lymphocyte numbers obtained from corresponding complete blood counts. PBMCs collected from chronically rhCMV-infected rhesus monkeys stimulated with 10 μg/mL Staphylococcus aureus enterotoxin type B (List Biological Laboratories) were included in each experiment as a positive control.

Virus and Cell Culture.

Telo-RF cells were maintained in Dulbecco’s modified Eagle medium (DMEM) containing 10% FCS, 2 mM l-glutamine, 50 U/mL penicillin, 50 μg/mL each streptomycin and gentamicin, and 100 µg/mL geneticin G418 (Invitrogen). Monkey kidney epithelial (MKE) cells were cultured in DMEM-F12 supplemented with 10% FCS, 2 mM l-glutamine, 1 mM sodium pyruvate, 50 U/mL penicillin, 50 μg/mL each streptomycin and gentamicin, and 5 mL of epithelial growth cell supplement (ScienCell). Virus infections were performed with similar media but with 5% FCS for 4 h at 37 °C and 5% CO₂. Virus was harvested when cells showed 90% cytopathic effects (CPE) by cell scraping. For harvest, infected cells were pelleted by low-speed centrifugation, and the supernatant was placed on ice. Cell pellets were then resuspended in infection media and subjected to three rounds of freeze/thaw cycles. Following centrifugation, supernatants were combined, passed through a 0.45-μm filter, overlaid onto a 20% sucrose cushion, and ultracentrifuged at 70,000 × g for 2 h at 4 °C using an SW28 Beckman Coulter rotor. Virus pellets were resuspended in DMEM containing 10% FCS and titered in telo-RFs using the TCID₅₀ method of Reed and Muench.

Antibody Kinetics and Avidity.

Virus-specific IgM and IgG antibody kinetics were measured in maternal plasma by whole-virion ELISAs. Plates were incubated overnight at 4 °C with 880 PFU/mL of filtered rhCMV 180.92 virus diluted in PBS with Mg²⁺ and Ca²⁺ (PBS+), washed, and then blocked for 2 h with blocking solution (PBS+, 4% whey protein, 15% goat serum, 0.5% Tween 20). Following the 2-h block, threefold serial dilutions of plasma (1:30 to 1:30,000) were added to the wells in duplicate for 1.5 h. Plates were then washed twice and incubated for 1 h with anti-human IgM (Southern Biotech) or anti-monkey IgG (Rockland) HRP-conjugated antibodies at a 1:000 or 1:10,000 dilution, respectively. After three washes, SureBlue Reserve TMB Microwell Peroxidase Substrate (KPL) was added to the wells for 7 min, and the reaction was stopped by addition of TMB stop solution (KPL). Plates were read at 450 nm. The lower threshold for antibody reactivity was considered to be 3 SDs above the average optical density calculated for rhCMVsn samples at the highest plasma dilution (1:30).

Maternal IgG avidity assays were performed similarly to the ELISA described above. Briefly, plates were coated with 1.4 × 10⁴ PFU/mL of filtered rhCMV 180.92 virus for 24 h at 4 °C. Plates were washed, blocked for 2 h, and then incubated for 1.5 h with a 1:300 dilution of maternal plasma in duplicate. After two washes, PBS+ was added to one set of wells while 7 M urea was added to the second set of wells for 5 min. Plates were washed three times before incubation with the same anti-monkey HRP-conjugated antibody. Following addition of the substrate and stop solution, avidity indexes were calculated by dividing the optical density at 450 nm of wells treated with urea by the optical density of wells treated with PBS+ only. Graphed data represent the average avidity index of four experiments.

Antibody Binding Specificity.

MVA vectors encoding for rhesus gB, gH/gL, and gH/gL-PC were constructed as previously described (47, 48). For rhCMV antigen production, BHK cells seeded in T175 flasks were infected for 16–20 h with multiplicity of infection 5 of each MVA vector. Cells were harvested in 5 mL of glycine–saline buffer (0.04 M glycine, 0.15 M NaCl, 0.01 M NaOH, pH 9) when 80–90% CPE was observed. Cell suspensions were then incubated on ice for 4 h, Dounce-homogenized, and cell lysates were clarified at 1,500 × g at 4 °C. Supernatants were aliquoted and stored at −80 °C.

Maternal IgG antibodies specific for rhesus gB, gH/gL, and gH/gL-PC were evaluated by ELISA. Briefly, EIA/RIA microtiter wells (Fisher Scientific) were coated overnight with rhCMV antigen, washed three times with PBST (PBS, 0.1% Tween 20), and then blocked for 2 h with PBS containing 1% BSA. Plasma samples serially diluted (1:200 to 1:625,000) in PBST were added to the wells for 2 h after which the plate was washed three times and incubated with a 1:2,000 dilution of anti-monkey IgG HRP-conjugated antibody (KPL). Finally, wells were developed with 1-Step Ultra TMB-ELISA Substrate Solution (Pierce) and hydrogen peroxide stopping solution. Absorbance was measured at 450 nm. ELISA titers were reported as the serum dilution yielding 50% maximum absorbance (EC₅₀) using a four-parameter logistic (4PL) nonlinear regression model, with the aid of a commercially available statistical program (GraphPad Prism).

Neutralization Assays.

Telo-RF and MKE cells were seeded into 96-well plates and incubated for 2 d at 37 °C and 5% CO₂ to achieve 100% confluency. After 2 d, serial dilutions (1:10 to 1:30,000) of heat-inactivated rhesus plasma were incubated with 8.35 × 10⁴ PFU rhCMV 180.92 or 4.8 × 10⁴ PFU rhCMV UCD52 in a 50-µL volume for 45 min at 37 °C. The virus/plasma dilutions were then added in duplicate to wells containing telo-RF or MKE cells, respectively, and incubated at 37 °C for 2 h. After washing, cells were incubated at 37 °C for an additional 3 h. Infected cells were then fixed for 20 min at 20 °C with 1:1 methanol/acetone, rehydrated in PBS+ three times for 5 min, and processed for immunufluorescence with 0.6 mg/mL mouse anti–rhCMV-IE1 monoclonal antibody provided by Dr. K. Fruh (Oregon Health and Science University, Portland, OR) followed by a 1:500 dilution of goat anti-mouse IgG-Alexa Fluor 488 antibody (Millipore). Nuclei were stained with DAPI for 5 min (Pierce) and imaged under a Nikon Eclipse TE2000-E fluorescent microscope equipped with a CoolSNAP HQ-2 camera at 10× magnification. ImageJ software was used to automatically count cells from a single field of view, and the 50% inhibitory dose was calculated as the sample dilution that caused a 50% reduction in the number of infected cells compared with wells treated with virus only using the method of Reed and Muench.