HIV-exposed uninfected infants: elevated cord blood Interleukin 8 (IL-8) is significantly associated with maternal HIV infection and systemic IL-8 in a Kenyan cohort

Successful implementation of global programs for the prevention of mother-to-child transmission (PMTCT) of HIV has decreased perinatal HIV infections from 500,000 to 15,000 cases per year worldwide in the last decade (UNAIDS, 2015). However, it is now evident that this has created a large and expanding number of HIV exposed but uninfected (HEU) infants. These infants experience increased rates of morbidity and mortality compared with their HIV unexposed (HU) counterparts. A recent meta analysis of 22 studies from low and middle income countries showed a 70% increase in mortality among HEU during the first 2 years of life, and several individual studies report a 3- to fourfold increase in mortality over HU [1, 2].

The preponderance of current evidence suggests that the immune systems of HEU children differ in fundamental ways from HU children including changes in the numbers and proportions of leukocyte subsets present from the first months of life [3‐5]. At birth cord blood T cells from HEU children can mount HIV-1 specific responses, indicating exposure to circulating maternal HIV antigen [6, 7]. In response to a range of stimuli, HEU cord blood mononuclear cells (CBMC) produce different cytokine profiles and reduced T cell helper responses compared to HU CBMC [8, 9]. However, there is controversy in the field as to the cause of the observed immune changes in HEU infants, the relative contribution of maternal health, and relative differences observed in HIV unexposed infants from the same communities. It is plausible that chronic in utero exposure to the maternal environment of persistent HIV infection affects the developing neonatal immune system, resulting in the immune alterations observed at birth. Equally plausible is that exposure to antiretroviral medications, poor nutrition, or exposure to multiple environmental antigens other than HIV, could stimulate a similar profile in HIV unexposed infants living in similar conditions.

Here we designed a retrospective study using careful specimen selection to systematically minimize the above factors known or suspected to confound interpretation of HEU data. Specifically we drew from specimen and data archives established prior to ARV interventions and purposefully selected mother-infants pairs with a clinical medical record indicating no infections or hospitalizations during the study duration in order to minimize the effects of environmental pathogens. Moreover we selected HEU infants and HU control infants of similar birthweight and gestational age and chose to evaluate infant cord blood as a window into prenatal immunity. Here we demonstrate cord blood plasma T cell cytokines and chemokines from HIV uninfected infants born to HIV infected women not taking ARV are markedly different from HIV unexposed infants. HIV positive maternal T cell cytokines were markedly lower than HIV uninfected women at delivery, and despite this, HEU cord plasma contained significantly elevated levels of the pro-inflammatory and chemotactic cytokines IL-6 and IL-8 and reduced levels of an interferon gamma (IFN-γ)-induced chemokine, CXCL11 compared to HU infants.

We hypothesized that maternal HIV, either directly via viral load/viral antigens or indirectly via altered plasma cytokines, would set a different immune stage for HEU infants than for HU infants. IL-6 and IL-8 levels present in cord blood of HEU infants were significantly elevated compared both to their HIV-infected mothers and to their HIV-unexposed community peers. Our study provides a direct link between maternal HIV infection and elevated levels of prenatal IL-6 and IL-8 measured in cord blood from infants who were born HIV uninfected and who did not acquire HIV through vaginal delivery and breastfeeding during the first year of life and who were unexposed to ARV. Our models demonstrate maternal IL-8 directly mediated ~ 25% of the effect of HIV infection on cord blood IL-8, with ~ 75% of the effect unmeasured in this study, indicative of alternative pathways or cellular sources of neonatal IL-8 induction and regulation. In contrast, maternal HIV infection did not mediate infant IL-6 levels via maternal IL-6, leaving all of the effect of maternal HIV infection on neonatal IL-6 induction unexplained in our models.

We have extended the finding that IL-8 is not limited to placental transfer in healthy women to that of HIV-infected-HEU mother-infant pairs [16, 17]. IL-8 is a major T cell effector in newborns and can be produced by endothelial cells, tissue macrophages, and cells of monocytic origin [18]. The IL-8 detected in cord blood here is plausibly derived from maternal or infant origin: from viral antigen-stimulated endothelial cells lining the umbilical cord, tissue macrophages present in the syncytiotrophoblast layer of the placenta or monocytic cells of infant ontogeny [19].

The onset of active labor and vaginal delivery is associated with elevated levels of IL-6 in both maternal and infant circulation [20], however we recorded a ~ 50-fold increase in the IL-6 index of cord blood to delivery blood observed in HEU compared to HU mother-infant pairs, the biggest difference amongst the mediators measured. Infection and inflammation during pregnancy are hypothesized to be associated with induction of placental cytokines, including IL-6, and increased fetal exposure to pro-inflammatory cytokines can result in increased risk of infant neurologic injury, even in the absence of HIV [21]. In addition to maternal infection/inflammation, direct transfer of microbes secondary to HIV infection may occur from the maternal bloodstream across the placenta into the fetal circulation, such as has been observed with increased gut microbial translocation [22]. Alternatively direct or indirect microbial LPS stimulation may induce placental cytokine production and the release or production of downstream mediators following Toll-like receptor signalling [23].

The elevated levels of IL-8 we observed in HEU but not in HU cord blood could result from a combination of increased production or decreased down-regulation of IL-8. As part of the pro-inflammatory antiviral immune response, intracellular regulators of IL-8 include IFN-γ, a cytokine notably present at low levels in newborns [24, 25]. Thus newborns with low levels of IFN-γ might be unable to down-regulate IL-8. Additionally, HIV tat has been shown to signal production of IL-8 [26], and ongoing genital tract viral replication during pregnancy could result in placental HIV tat. Prolonged exposure of fetal and neonatal T lymphocytes to elevated IL-8 likely leads to intracellular signaling of gene pathways associated with cell survival, proliferation, and metabolism [27]. Hypothetically, neutrophils and other granulocytes programmed to respond to IL-8 along a chemotactic gradient also might be unable to respond to the signals if there is no functional gradient present, disabling the intended effect of chemotaxis and instead resulting in an immune ‘confusion’ or ‘paralysis’ until levels of circulating IL-8/IL-6 return to homeostasis. The consequence of a lack of recruitment of neutrophils to the sites of tissue damage and infection is consistent with clinical descriptions of HEU morbidity and mortality.

Our analyses demonstrated a consistent pattern of significantly increased levels of chemotactic cytokines present in cord blood compared to delivery blood in HEU infants compared to HU infants (Fig. 2). The increased relative concentrations of IL-8, CCL3, CCL4, CCL20 in HEU circulation compared to maternal circulation suggests a coordinated response to inflammatory signaling. GM-CSF, although present at levels below that found in maternal circulation, is significantly increased in HEU compared to HU infants. IL-10 levels are increased in HEU infants, although still lower than levels found in maternal circulation. IL-10 is in part regulated by IL-6 levels, thus it is feasible higher circulating IL-6 levels release the inhibition on IL-10 production in HEU infants. Regardless the source of mediators detected in HEU cord blood, the finding of significantly elevated cytokines compared to both HIV unexposed control cord blood plasma as well as paired maternal blood collected at delivery suggests controlled HIV infection during pregnancy drives the fetal immune response.

This study had several strengths. Access to stored specimens from the pre-ARV era allowed us to eliminate the direct and indirect effect of ARVs on maternal and neonatal cytokine production. Longitudinal HIV testing within the parent cohorts allows us to rule out individuals with subsequent acquisition of HIV. While the inclusion of Kenyan HIV-uninfected mother-infant dyads is a strength of the study, allowing us to examine cytokine responses in populations of women living in similar socio-economic conditions, it raises a potential weakness considering the time difference in collection of samples from the two cohorts. Further research is needed to determine the cellular source of the elevated IL-8 and the mechanism of induction/or release from negative feedback for production of this cytokines. Additionally, because we selected normal birthweight and gestational age infants for both HEU and HU infants, we cannot assess the role of maternal and fetal systemic inflammation on birth anthropometry.

The majority of immune function studies of HEU compared to HU infants have evaluated cellular responsiveness following antigenic stimulation in vitro. Our studies focused on the prenatal cytokine environment in which presumably the cellular response would arise. In a study of plasma cytokines from HEU and HU infants born to HIV+ ARV-treated and healthy HIV− pregnant Brazilian women, HEU and HU cord blood plasma IL-8 levels were low and similar in HEU infants delivered via Caesarian section compared to HU infants delivered vaginally [28]. It remains an important public health goal to identify the effects of chronic maternal infections on fetal and infant immunity, correlations with infant health, and response to maternal antiviral treatments.

We conclude that in utero stimuli stemming from infection generated an immune profile consistent with fetal/infant immune activation, particularly cytokines associated with macrophage activation, above what is expected at parturition. The weak association with maternal viral load raises the concern that HEU infants exposed to ARV may still be at risk for increased altered immune profile at birth.