Arsenic-Associated Oxidative Stress, Inflammation, and Immune Disruption in Human Placenta and Cord Blood

As exposure

The concentration of metabolites of iAs in urine was used as a biomarker of exposure because it reflects the ingested dose of iAs from all sources (Vahter et al. 2006). Although the half-life of iAs in the body is relatively short, the daily exposure through water and food results in fairly stable steady-state concentrations in urine. In the larger MINIMat cohort, we observed that maternal urinary As (U-As) is tightly correlated with blood As levels (r² = 0.83, p < 0.001; unpublished data). As in hair and nail samples, used sometimes as biomarkers of As exposure over several months, were not considered in this highly exposed population because of risk of external exposure contamination, for example, bathing and washing in contaminated water. Because As easily crosses the placenta (Concha et al. 1998), the concentration in maternal urine during pregnancy was used as a proxy marker of fetal exposure.

Urine was collected into trace element–free plastic cups, transferred to 24‐mL polyethylene bottles, and stored at −70°C in the laboratory. The mothers’ U-As, defined as the sum of iAs and iAs metabolites in urine (iAs + methylarsonic acid + dimethylarsinic acid), was determined using hydride generation atomic absorption spectrophotometry (Vahter et al. 2006). The limit of detection (LOD) was 1.3 ± 0.27 μg/L. To compensate for variation in dilution, urine samples were adjusted for specific gravity measured by a digital refractometer (RD712 Clinical Refractometer; EUROMEX, Arnhem, Holland). Adjustment by specific gravity has been shown to be less affected by body size, SES, and As exposure than the more commonly used creatinine adjustment (Nermell et al. 2008).

Processing of placental tissue for immunostaining

Placental tissue was collected from mothers immediately after delivery at the clinic. A 2 × 2-inch piece from the fetal portion of the placenta was cut and immersed in 10% buffered formalin (pH 7.0) kept at room temperature (RT). Smaller pieces were cut from the placental specimen and divided into two sets. One set was fixed, embedded in paraffin, sectioned by Microtome (Leica, Nussloch, Germany) at 3‐mm thickness, mounted on glass slides treated with Vectabond reagent (Vector Laboratories, Burlingame, CA, USA), dried overnight at 37°C, and kept at RT before use for immunostaining. The second set of tissues was immersed in 15% sucrose buffer overnight and, after removal of excess fluid, was snap-frozen in liquid nitrogen to retain the antigenic properties.

Immunohistochemical detection of cell surface markers and proinflammatory cytokines

Paraffin-embedded tissue sections were deparaffinized and microwave-treated in target retrieval solution (pH 6.0–6.2) for antigen retrieval. After blocking endogenous peroxidase activity and nonspecific staining, the slides were incubated sequentially with primary antibodies (overnight at RT), antibody conjugate, and avidin-biotin complex at 37°C for 30 min, each with intermittent washing. Positive staining was developed with substrate 3,3′-diaminobenzidine (Sigma-Aldrich, St. Louis, MO, USA). Sections were counterstained with hematoxylin, dehydrated, and mounted. Frozen tissue blocks were sectioned with a cryomicrotome (Leica) at 4–5 μm thickness. Staining for CD64 and cytokines was performed in the cryostat sections as described previously (Raqib et al. 1995). Primary antibodies used were anti-human: CD3 (pan T cell), CD8 (cytotoxic/suppressor T cells), CD64 (FcgRI, monocytes/macrophages), CD68 (phagocytes), myeloperoxidase (MPO; neutrophils), interleukin-1β (IL-1β), and iterferon-γ (IFNγ) (all above from DakoCytomation, Glostrup, Denmark); leptin ob (Y-20) (Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA); tumor necrosis factor-α (TNFα) (BD Biosciences, Sparks, MD, USA); IL-6 (BD Pharmingen, San Diego, CA, USA); IL-10 (Mabtech, Cincinnati, OH, USA); and 8-oxoguanine (8-oxoG) (Chemicon International, Temecula, CA, USA).

Image analysis of immunostaining

Immunohistochemical staining of specific markers in the placental specimens was analyzed with a microscope (Olympus BX51; Olympus Optical Co. Ltd., Tokyo, Japan) connected to a CCD camera (DCF295; Leica Microsystems CMS GmbH, Wetzlar, Germany) and the image analysis system Leica Qwin Runner (version 3). Special software (Tissue-includer), written in the high-level language QUIPS, was used to examine each video microscopic digital image. The color and morphology of the hematoxylin counterstained cells were set as a standard for positive and negative cells, and the positive staining of markers in placental tissue sections was defined by Leica software (Sarker et al. 2009). The acquired video microscopic image was divided into 512 × 512 pixels, and each pixel was expressed in square micrometers (area) after calibration with the current magnification. For each placental tissue section, all fields (average, 400 ± 25 fields) were investigated at 400× magnification, and the average of all fields was used for quantification of immunostaining in each tissue section. The results were expressed as {(total positively stained area) × (total mean intensity [1–256 levels per pixel] of the positive area)} / total cell area, in units referred to as acquired computerized image analysis (ACIA) scores [see Supplemental Material, “Computerized Image Analysis for Detection of Immunostaining,” and Supplemental Material, Figure 1 (doi:10.1289/ehp.1002086)].

For T-cell (CD3⁺, CD8⁺) staining, the computerized image analysis technique could not be applied because the numbers of cells in the total tissue section were very small, so ACIA scores were very low. T-cell phenotypes in the placental sections were assessed manually using an ocular micrometer fitted into the eyepiece of the microscope (Olympus Optical Co. Ltd.), and data were calculated as number of cells per 100‐μm² field area. Two trained individuals counted 400 fields separately, and the average was used.

Cytokines assay in cord blood plasma

Cord blood specimens were collected in heparin-coated sterile vials (Becton Dickinson, Rutherford, NJ, USA) at the subcenter health clinics at delivery and were transported on ice to the laboratory. Plasma was separated and stored at −70°C until analyzed by Bio-Plex 200 reader (BioRad Laboratories, Hercules, CA, USA). Cytokine levels in plasma were determined using Bio-Plex Pro Human Cytokine 18-Plex Panel [IL-1β, IL-2, IL-4, IL-5, Il-6, IL-7, IL-8, IL-10, IL-12 (p70), IL-13, IL-17, granulocyte colony-stimulating factor, granulocyte-macrophage colony-stimulating factor, IFNγ, monocyte chemotactic protein-1, macrophage inflammatory protein-1β, TNFα, and human interferon-inducible protein-10] according to the manufacturer’s instructions. The LOD was approximately 10 pg/mL for these cytokines. For cytokine measurements with values below LOD, the LOD divided by square root of 2 was used.

Maternal As exposure and placental immune markers

Localization of the various markers in the placenta was generally same for all subjects. We generally found increased expression of the various cytokines and reduced cell counts/expression at high exposure levels [median U-As > 60 μg/L at GW30; see Supplemental Material, Figure 2 and “Localization of immune cells and inflammatory markers in the placenta” (doi:10.1289/ehp.1002086)].

Maternal As exposure and placental immune cells

Evaluation of associations between As exposure during pregnancy (U-As at GW8 and GW30) and the phenotypic expression of immune cells in the placenta (CD3⁺, CD8⁺, CD64⁺, CD68⁺, MPO) by regression analysis showed a significant negative association between CD3⁺ cell frequency in the placenta and U-As at GW8 but not at GW30 (Table 2, Figure 1A,B). The association remained significant after adjusting for age, SES, and TC (Table 2). Placental CD8⁺ cells decreased with increasing U-As at GW8, but the trend was not significant (Table 2, Figure 1A). The CD64⁺ cells were significantly negatively associated with U-As at GW8 but not at GW30 (Table 2, Figure 1C,D). We found no significant association between maternal U-As at GW8 or GW30 and placental expression of CD68 (phagocytes) and MPO [neutrophils; see Supplemental Material, Table 1 (doi:10.1289/ehp.1002086)]. We found no significant effect of micronutrient supplementation on placental or cord blood outcome variables and no difference in As exposure among supplementation groups. We found no significant association between the sex of the infant and placental and cord blood immune parameters or As exposure.

Open in a separate window

Figure 1

(A and B) Association between U-As and frequencies of placental CD3⁺ and CD8⁺ cells, using quartiles of U-As concentrations at GW8 (A; n = 130, 32–33 in each quartile; median U-As, 26, 46, 115, and 341 μg/L, respectively) and GW30 (B; n = 130, 32–33 in each quartile; median U-As, 27, 48, 121, and 335 μg/L, respectively). (C and D) Association between U-As and expression of CD64⁺ and CD68⁺ cells, using quartiles of U-As exposure at GW8 (C) and GW30 (D). (E and F) Association between U-As and 8-oxoG and leptin expression, using quartiles of U-As exposure at GW8 (E) and GW30 (F). (G and H) Association between U-As and TNFα, IFNγ, and IL-1β expression, using quartiles of U-As concentrations at GW8 (G) and GW30 (H). Data are means ± SE. Ln, natural log.

*p < 0.05 for U-As modeled as a continuous variable in multivariable-adjusted linear regression analysis (Table 2).

Maternal As exposure and cytokines and oxidative damage in placenta

In regression analyses (Table 2), we found significant positive associations between U-As at both GW8 and GW30 and the placental expression of 8-oxoG (Figure 1E,F), IL-1β, TNFα, and IFNγ (Figure 1G,H). Placental expression of leptin was significantly associated with U-As at GW30 but not at GW8 (Figure 1E,F). We found no significant associations between IL-6 or IL-10 and U-As [see Supplemental Material, Table 1 (doi:10.1289/ehp.1002086)]. In multivariable adjusted linear regressions analysis (adjusting for age, SES, and TC), U-As at both GW8 and GW30 was significantly associated with placental 8-oxoG and IL-1β; U-As at GW8, with TNFα and IFNγ; and U-As at GW30, with leptin (Table 2).

Placental 8-oxoG was significantly associated with several placental cytokines (leptin, IL-1β, TNFα, and IFNγ) (Table 3). To investigate whether the estimated effects of As on placental cytokines may have been mediated via oxidative stress, we performed multivariable adjusted linear regression with placental cytokines (leptin, IL-1β, TNFα, and IFNγ) as dependent variables and U-As as independent variable, controlling for 8-oxoG and potential confounders (Table 3). All the associations between placental cytokines and U-As at GW8 did not remain significant after adjustment for 8-oxoG. The association between U-As at GW30 and placental leptin decreased slightly but remained significant after adjustment for 8-oxoG. The association between U-As at GW30 and IL-1β also decreased and was no longer significant after adjusting for 8-oxoG. Coefficients for the association of U-As at GW30 with placental IFNγ and TNFα were decreased markedly after adjustment for 8-oxoG (in addition to age, SES, and TC), but estimates for associations of 8-oxoG with placental IFNγ and TNFα remained positive and significant when adjusted for U-As at GW30.

In multivariable adjusted linear regression, placental expression of IL-6 and IL-10 was significantly associated with the number of days of maternal urinary tract infection during pregnancy (β-coefficient = 0.45, p = 0.044) and acute respiratory infection (β-coefficient = 0.50, p = 0.017), respectively, after adjustment for U-As and other potential confounders. We observed no significant associations between mothers’ other morbidities (fever and diarrhea) and other markers in the placenta and cord blood.