B2 cells contribute to hypertension and natural killer cell activation possibly via AT1-AA in response to placental ischemia

Abstract

Preeclampsia, new onset hypertension during pregnancy, is associated with activated T helper cells (Th) and B cells secreting agonistic autoantibodies against the angiotensin II type 1 receptor (AT1-AA). The reduced uterine perfusion pressure (RUPP) model of placental ischemia recapitulates these characteristics. We have shown that Th-B cell communication contributes to AT1-AA and symptoms of preeclampsia in the RUPP rat. B2 cells are classical B cells that communicate with Th cells and are then transformed into memory B cells. We hypothesize that B2 cells cause hypertension, natural killer (NK) cell activation, and complement activation during pregnancy through the production of AT1-AA. To test this hypothesis, total splenic B cells and B2 cells were isolated from normal pregnant (NP) or RUPP rats on gestational day (GD)19 and adoptively transferred into GD12 NP rats. A group of recipient rats was treated with a specific inhibitor peptide of AT1-AA. On GD19, mean arterial pressure was measured, tissues were collected, activated NK cells were measured by flow cytometry, and AT1-AA was measured by cardiomyocyte assay. NP recipients of RUPP B cells or RUPP B2 cells had increased mean arterial pressure, AT1-AA, and circulating activated NK cells compared with recipients of NP B cells. Hypertension in NP recipients of RUPP B cells or RUPP B2 was attenuated with AT1-AA blockade. This study demonstrates that B cells and B2 cells from RUPP rats cause hypertension and increased AT1-AA and NK cell activation in response to placental ischemia during pregnancy.

NEW & NOTEWORTHY This study demonstrates that placental ischemia-stimulated B2 cells induce hypertension and circulating natural killer cell activation and angiotensin II type 1 receptor production in normal pregnant rats.

INTRODUCTION

Preeclampsia (PE), new onset hypertension beyond the 20th week of pregnancy, is a leading cause of maternal and fetal demise and affects 5%–8% of pregnancies worldwide (1–3). PE severity is associated with the time of onset of symptoms, which can be divided into early onset (EO) and late onset (LO). EO-PE is more severe and is associated with insufficient spiral artery remodeling, which leads to placental ischemia resulting in hypertension, decreased renal function, and widespread endothelial cell dysfunction (3–5). LO-PE is associated with trophoblast dysfunction leading to hypoxia and resulting in the onset of PE symptoms (6). Both EO-PE and LO-PE are associated with an inflammatory phenotype (7) consisting of endothelial dysfunction, activated T helper (Th)1/Th17 cells, dysregulation of the complement system, activated natural killer cells, and activated B cells producing autoantibodies against the angiotensin II type 1 receptor (AT1-AA) (4, 8–10). These characteristics are recapitulated by the reduced uterine perfusion pressure (RUPP) model of placental ischemia (11). Our laboratory has shown the importance of immune factors in contributing to the hypertension seen in response to placental ischemia with the use of the RUPP model (9, 12, 13). Moreover, we have shown a direct role of CD4⁺ T cells from RUPP rats and women with PE to cause hypertension and AT1-AA, which is attenuated by blockade of T cell-B cell communication (9, 12, 14–19). However, we have not examined the direct role of B cells secreting AT1-AA to cause hypertension during pregnancy.

AT1-AA was discovered in patients with PE in 1999 (8), and it was later established that AT1-AA isolated from women with PE induced hypertension in pregnant mice (20). Moreover, AT1-AA in pregnant rats causes hypertension, mitochondrial dysfunction, antiangiogenic factors, increased endothelin-1 expression, and increased uterine artery and renal artery resistance (21, 22). AT1-AA has also been implicated in activation of the classical complement cascade in PE (23). Importantly, AT1-AA has been found in women with a history of PE 7 yr postpartum (24, 25), thereby suggesting long-term immune memory mechanisms involved in the production of AT1-AA during PE that exists well into the postpartum period.

Although the role of AT1-AA in PE has been well described (26), there are limited studies that have investigated the role of B cells or subsets of B cells in the production of AT1-AA during pregnancy and whether or not that leads to hypertension during pregnancy. B cells are the antibody-producing cells of the immune system and can be classified into B1 and B2 cells. Specifically, B1 cells are innate-like B cells that typically produce low-specificity antibodies through T cell-independent B cell activation. Jensen et al. (27) showed that CD19⁺CD5⁺ B1 cells isolated from the placentas of women with PE can produce AT1-AA in vitro, yet no studies have investigated whether classical B2 cell are also capable of producing AT1-AA in vivo or in vitro. B2 cells are the classical B cells that produce high-specificity antibodies through T cell-dependent B cell activation and result in long-lived memory B cells that can produce antibodies years following the initial exposure (28). One of the primary functions of B cells is to produce antibodies that mark target cells for destruction. Two important methods of antibody-induced cell death are via natural killer (NK) cell activation through CD16 or complement protein deposition into the cell membrane (29, 30). With NK cell activation and complement activation being associated with PE, we asked the question if B cell production of AT1-AA in response to placental ischemia is a stimulus for either NK cell or complement activation during pregnancy.

Our laboratory has shown that B cell depletion with rituximab attenuates hypertension and AT1-AA associated with placental ischemia (31). Moreover, we have also shown that blockade of interactions between CD40L on T cells and CD40 on B cells prevents hypertension and AT1-AA associated with the adoptive transfer of RUPP T cells into normal pregnant (NP) rats (17), further suggesting that T cell-dependent B cell activation is an important step in the production of AT1-AA during PE. Although B1 cells are suspected to produce AT1-AA in culture, no studies have examined the role of B cells or B2 cells stimulated in response to placental ischemia to cause AT1-AA, NK cell, or complement activation and hypertension in a pregnant animal. Therefore, we sought to examine the role of total B cells and specifically B2 cells in producing AT1-AA during pregnancy and hypothesized that classical B2 cells contribute to hypertension in response to placental ischemia by producing AT1-AA.

METHODS

The techniques and data that support the findings of this study are available from the corresponding author on reasonable request.

Animals

Twelve-week-old timed-pregnant Sprague–Dawley rats (200–250 g) were purchased from Harlan Laboratories (Indianapolis, IN) for use in this study. Animals were housed in a temperature-controlled room (23°C) with a 12:12-h light-dark cycle. All experimental procedures executed in this study were in accordance with the National Institute of Health Guide for the Care and Use of Laboratory Animals. All experimental procedures were approved by the Institutional Animal Care and Use Committee of the University of Mississippi Medical Center and were in accordance with National Institutes of Health guidelines.

RUPP Procedure

The RUPP procedure was performed on gestational day (GD)14 under 2% isoflurane anesthesia delivered by a vaporizer (Ohio Medical Products, Champaign, IL). Briefly, a midline incision was made, and a constrictive silver clip (0.203-mm diameter) was placed on the abdominal aorta above the iliac bifurcation. To correct for compensatory blood flow, a silver clip (0.100 mm) was placed on each bilateral uterine arcade at the ovarian end. Following the RUPP procedure, animals received carprofen (5 mg/kg) immediately after surgery and 24 h after to account for postoperative pain. The RUPP procedure reduces uteroplacental blood flow by ∼54% (32).

Carotid Catheterization and Continuous Conscious Blood Pressure Measurement

On GD18, carotid catheters made from V3 tubing (Scientific Commodities, Lake Havasu City, AZ) were inserted and tunneled to the back of the neck and exteriorized for blood pressure measurements in animals under isoflurane anesthesia delivered by a vaporizer. On GD19, rats were placed in individual restrainers and allowed to equilibrate for 1 h before continuous conscious mean arterial pressure (MAP) was measured for 30 min (Cobe III Transducer CDX Sema). Immediately following blood pressure measurements, animals were euthanized under anesthesia and tissues were collected. Pups and placentas were weighed from each dam, and the average pup or placenta weight for the animal was reported as one data point. Blood and tissue were used in assays to determine levels of immune cells or proteins involved in tissue damage (Fig. 1).

Figure 1.

In this study, pregnant Sprague–Dawley rats underwent reduced uterine perfusion pressure (RUPP) surgery on gestational day (GD)14 and carotid catheterization on GD18. On GD19, normal pregnant (NP) and RUPP blood pressure was measured, tissues were collected, and spleens were harvested for B cells and B2 cells. The same day as tissue harvest, isolated B cells were adoptively transferred into GD12 pregnant Sprague–Dawley rats. Some of the rats that received RUPP B cells or RUPP B2 cells underwent mini-osmotic pump surgery to administer “n7aac” peptide to inhibit agonistic autoantibodies against the angiotensin II type 1 receptor (AT1-AA). All animals underwent carotid catheterization on GD18, and blood pressure was measured and tissues were collected on GD19.

B Cell Isolation

On GD19, spleens were collected from NP or RUPP rats and placed in ice-cold PBS (pH 7.0), homogenized in Petri dishes in RPMI medium with 10% FBS, and filtered through a 100-μm cell strainer. This homogenate was then layered over a Ficoll-Hypaque cushion (Lymphoprep, Accurate Chemical & Scientific, Westbury, NY) and spun at 300 g to isolate splenic leukocytes. Total B lymphocytes were isolated using the MagCellect Rat B Cell Isolation Kit according to the manufacturer’s protocol (R&D Systems, Minneapolis, MN). The MagCellect kit uses a negative selection technique to isolate untouched B cells.

B2 Cell Isolation

B2 cells were isolated by negative selection following total B cell isolations. Isolated total B cells were incubated with CD43 Microbeads (Miltenyi Biotech, San Diego, CA) according to the manufacturer’s instructions. B2 cells were then separated using LS magnetic columns (Miltenyi Biotech). Isolated CD43⁻ B2 cells were verified by flow cytometry.

Determination of Isolated B Cells

Flow cytometry analysis was used to verify B cell isolation. Following isolation, B cells were incubated for 10 min with Vioblue-conjugated anti-CD3 (Miltenyi Biotech), PE-Vio770-conjucated anti-CD68 (Miltenyi Biotech), APC-Vio770-conjugated anti-CD45 receptor (Miltenyi Biotech), and VioBright FITC-conjugated anti-CD43 (Miltenyi Biotech) for 10 min at 4°C. Flow cytometry was performed on a Miltenyi MACSQuant Analyzer 10 and analyzed using FlowLogic software (Innovai, Sydney, NSW, Australia). Lymphocytes were gated in a forward and side scatter plot and doublets were then excluded. Singlet lymphocytes were then gated using fluorescence minus one controls (33). B lymphocytes were considered CD3⁻CD68⁻CD45R⁺ cells; B2 lymphocytes were considered CD3⁻CD68⁻CD45R⁺CD43⁻ cells (Fig. 2).

Figure 2.

The gating strategy used for B cells and B2 cells. The purity of B cells was above 80% for total B cell adoptive transfer, and the purity of B cells was ∼98% for adoptive transfer. Lymphocytes were gated in a forward (FSC) and side scatter (SSC) plot, and then doublets were excluded. Singlet lymphocytes were then gated using fluorescence minus one (FMO) controls. Cells were gated using CD3 (VioBlue), CD68 (PE-Vio770), CD45R (APC-Vio770), and CD43 (FITC). B cells were considered CD3⁻CD68⁻CD45R⁺ cells; B2 lymphocytes were considered CD3⁻CD68⁻CD45R⁺CD43⁻ cells.

Adoptive Transfer of B or B2 Lymphocytes

Total B cells or B2 cells (1 × 10⁶ cells suspended in 100 μL of sterile saline) were injected intraperitoneally into NP rats on GD12. There were three groups of rats, which were designated as follows: NP + NP B cell, NP + RUPP B cell, or NP + RUPP B2 cell groups.

Administration of AT1-AA Inhibiting Peptide

AT1-AA inhibiting peptide (“n7aac”; 144 μg/day) was administered on GD14 to a group of NP + RUPP B cell rats and a group of NP + RUPP B2 cell rats via a miniosmotic pump inserted in the peritoneal cavity as previously described (34).

Determination of Circulating and Placental NK Cell Populations Using Flow Cytometry

Circulating and placental NK cell populations isolated on GD19 from all groups were quantified by flow cytometry. At the time of harvest, whole blood and placentas were collected. Peripheral blood mononuclear cells were isolated by centrifugation on a cushion of Ficoll-Hypaque (Lymphoprep, Accurate Chemical & Scientific) according to the manufacturer’s instructions (manufacturer info). For flow cytometric analysis, 1 × 10⁶ cells were incubated for 10 min at 4°C with antibodies against rat NK cell activation structures (ANK61) or rat NK cell antibody (ANK44) (Abcam, Cambridge, MA). ANK61 binds to the killer cell activation structure that is expressed on all NK cells, whereas ANK44 is expressed only on stimulated, cytotoxic NK cells (35). After cells were washed, they were labeled with secondary FITC (Abcam) antibody for 10 min at 4°C. As a negative control for each individual rat, cells were treated exactly as described earlier except that they were incubated with isotype control antibodies conjugated to FITC alone. Subsequently, cells were washed, fixed in FACS buffer, and analyzed for single staining by flow cytometry using MACSQuantify (Miltenyi Biotech).

Lymphocytes were gated in the forward and side scatter plots. ANK61⁺ cells were designated NK cells, and ANK44⁺ cells were designated cytotoxic NK cells (35). The percentage of positive-stained cells above the negative control was collected for individual rats, and mean values for each experimental group were calculated.

Determination of AT1-AA Activity

AT1-AA was isolated from serum and analyzed using isolated cardiomyocytes as previously described (14, 34). AT1-AA was quantified by the chronotropic responses to angiotensin II type 1 receptor-mediated stimulation of cultured neonatal rat cardiomyocytes. Chronotropic responses were measured and are expressed as changes in beats per minute (ΔBPM).

Determination of Circulating, Placental, and Renal Complement Component C1q

Plasma and placentas collected from all groups were measured for complement component C1q using a commercially available ELISA Kit from Novus Biologicals (Littleton, CO) according to the manufacturer’s protocols. The sensitivity of the assay is 0.47 ng/mL, interassay variability is <5.23%, and intra-assay variability is <4.82%. Plasma samples were diluted 1:80,000 as per the manufacturer’s recommendation. For placental and renal samples, ∼100 mg tissue was homogenized in 900 μL of ice-cold PBS (pH 7.4) as recommended by the manufacturer. Tissue homogenates were then centrifuged for 5 min at 5,000 g, and the tissue homogenate supernatant was collected and diluted 1:1,000 for analysis. Placental C1q is reported as milligrams of C1q per milligram of total placental protein. Renal C1q is reported as milligrams of C1q per milligram of total renal protein.

Determination of Complement Components C3 and C3a

Plasma collected from all groups was measured for complement component C3 using a commercially available ELISA kit from Abcam according to the manufacturer’s protocols. The sensitivity of the assay is 2.82 ng/mL, interassay variability is listed as <10%, and intra-assay variability is listed as <10%. Plasma samples were diluted 1:10,000 as per the manufacturer’s instructions.

Plasma collected from all groups was measured for complement component C3a using a commercially available ELISA kit from MyBioSource (San Diego, CA) used according to the manufacturer’s protocol. Plasma samples were diluted 1:10. Renal C3a was determined using the renal homogenate supernatant as described earlier. The renal homogenate supernatant was diluted 1:100. The sensitivity of the assay is 4.69 ng/mL, and the coefficient of variation (CV%) is listed as <10%.

Determination of Circulating Complement Component C4

Plasma collected from all groups was measured for complement component C4 using a commercially available ELISA kit from MyBioSource used according to the manufacturer’s protocol. The detection range of the ELISA is 12.5–800 ng/mL, the sensitivity of the assay is <3.12 ng/mL, CV% is <8%, and the interassay precision is CV% < 10%. Plasma samples were diluted 1:10,000.

Determination of Renal Perforin and Granzymes

Renal levels of NK cell-associated cytolytic proteins (rat perforin, rat granzyme A, rat granzyme B, and rat granzyme K) were measured with commercially available ELISA kits according to the manufacturer’s protocol (MyBioSource). For the rat perforin ELISA, the intra-assay variability is CV% < 15, interassay variability is CV% < 15, and sensitivity is 5.0 pg/mL. For the rat granzyme A ELISA, the intra-assay variability is CV% < 8, interassay variability is CV% < 12, and sensitivity is 0.5 μg/mL. For the rat granzyme B ELISA, the intra-assay variability is CV% < 10, interassay variability is CV% < 10, and sensitivity is 0.38 ng/mL. For the rat granzyme K ELISA, the intra-assay variability is CV% < 4.6, interassay variability is CV% < 7.9, and sensitivity is 15.71 pg/mL.

Statistical Analysis

Statistical analyses were performed using GraphPad Prism 9.1.2 software (GraphPad Software, San Diego, CA). Comparisons between groups were analyzed using one-way ANOVA with a Bonferroni post hoc test. Results are reported as means ± SE and were considered as statistically significant when P < 0.05.

RESULTS

Effect of RUPP B Cells or RUPP B2 Cells on MAP

MAP was determined for each of the following groups of rats: NP (n = 11), RUPP (n = 15), NP + NP B cells (n = 6), NP + RUPP B cells (n = 10), NP + RUPP B cells + n7aac (n = 5), NP + RUPP B2 cells (n = 12), and NP + RUPP B2 cells + n7aac (n = 6) (Fig. 3). MAP was increased in the RUPP group of rats compared with the NP group of rats (118 ± 2 vs. 95 ± 2 mmHg, P < 0.01). Adoptive transfer of NP B cells into NP rats did not increase MAP compared with NP controls (102 ± 3 vs. 95 ± 2 mmHg), but adoptive transfer of RUPP B cells into NP rats induced hypertension compared with animals that received NP B cells (116 ± 3 vs. 102 ± 3 mmHg, P < 0.05). Hypertension associated with adoptive transfer of RUPP B cells into NP rats was attenuated by administration of n7aac peptide (116 ± 3 vs. 99 ± 4 mmHg, P < 0.05). Adoptive transfer of B2 cells from RUPP rats into NP rats increased MAP compared with rats that received NP B cells (112 ± 2 vs. 102 ± 3 mmHg, P < 0.05), which was attenuated with administration of n7aac peptide (98 ± 2 mmHg, P < 0.05).

Figure 3.

Mean arterial pressure (MAP) was increased in response to reduced uterine perfusion pressure (RUPP; n = 15) compared with the normal pregnant (NP) group (n = 11, P < 0.0001), as previously described. Adoptive transfer of NP B cells (n = 6) did not significantly increase MAP, but adoptive transfer of RUPP B cells (n = 10) significantly increased MAP (P < 0.001) compared with recipients of NP B cells. Infusion of agonistic autoantibodies against the angiotensin II type 1 receptor-inhibiting “n7aac” peptide (7AA; n = 6) significantly reduced MAP in recipients of RUPP B cells (P < 0.01). Adoptive transfer of RUPP B2 cells (n = 12) significantly increased MAP compared with recipients of NP B cells (P < 0.05), and infusion of n7aac peptide (n = 6) attenuated the increase in MAP associated with adoptive transfer of B2 cells (P < 0.01). Data were compared using one-way ANOVA and are reported as means ± SE. *Statistically significant (P < 0.05) change compared with the referenced group.

Effect of RUPP B Cells or RUPP B2 Cells on Pup Weight and Placental Efficiency

The RUPP procedure significantly reduced pup weight (1.89 ± 0.07 g, n = 15) compared with the NP group (2.24 ± 0.05 g, n = 11, P < 0.05; Fig. 4A). Adoptive transfer of neither RUPP B cells (2.29 ± 0.05 g, n = 10) nor RUPP B2 cells (2.34 ± 0.07 g, n = 10) into NP rats changed pup weight, but adoptive transfer of NP B cells into NP rats significantly increased pup weight compared with NP rats (2.44 ± 0.05 g, n= 7, P < 0.05). Neither NP + RUPP B cells + n7aac (2.44 ± 0.09 g, n = 5) nor NP + RUPP B2 cells + n7aac (2.24 ± 0.09 g, n = 6) had any change in pup weight compared with NP rats. There were no changes in placental weight compared with the NP group (0.54 ± 0.02 g, n = 11) in RUPP (0.52 ± 0.02 g, n = 15), NP + NP B cells (0.56 ± 0.02 g, n = 7), NP + RUPP B cells (0.54 ± 0.01, n = 10), NP + RUPP B cells + n7aac (0.58 ± 0.04, n = 5), NP + RUPP B2 cells (0.57 ± 0.02 g, n = 12), or NP + RUPP B2 cells + n7aac (0.58 ± 0.03 g, n = 6) (Fig. 4B). The RUPP procedure significantly reduced placental efficiency (0.28 ± 0.01, n = 15) compared with the NP group (0.24 ± 0.01, n = 11, P < 0.05; Fig. 4C). There were no changes in placental efficiency associated with NP + NP B cells (0.23 ± 0.01, n = 7), NP + RUPP B cells (0.24 ± 0.01, n = 10), NP + RUPP B cells + n7aac (0.25 ± 0.02, n = 5), NP + RUPP B2 cells (0.25 ± 0.01, n = 10), or NP + RUPP B2 cells + n7aac (0.28 ± 0.02, n = 6) compared with the NP group.

Figure 4.

The reduced uterine perfusion pressure (RUPP; n = 15) procedure significantly reduced fetal weight (P < 0.01; A) and placental efficiency (P < 0.05; C) compared with the normal pregnant (NP) group (n = 11), as previously described. Adoptive transfer of NP B cells (n = 7) significantly increased fetal weight compared with the NP group (P < 0.05) without affecting placental efficiency. Adoptive transfer of neither RUPP B cells (n = 10) nor RUPP B2 cells (n = 10) changed pup weight or placental efficiency. There were no changes in placental weight (B) associated with the RUPP procedure or with B cell adoptive transfer. The RUPP (n = 15) procedure significantly reduced placental efficiency compared with the NP group (n = 11, P < 0.05). There were no changes in placental efficiency seen in response to B cell adoptive transfer. Data were compared using one-way ANOVA and are reported as means ± SE. 7AA, n7aac peptide. *Statistically significant (P < 0.05) change compared with the referenced group.

Effect of RUPP B Cells or RUPP B2 Cells on Circulating and Placental NK Cells

There were no significant changes in total numbers of circulating NK cells in NP + NP B cells (38.16 ± 9.94% gated, n = 5), NP + RUPP B cells (44.18 ± 6.53% gated, n = 6), or NP + RUPP B2 cells (44.18 ± 6.53% gated, n = 9) compared with the NP group (45.59 ± 7.12% gated, n = 12; Fig. 5A). Circulating cytolytic NK cells were not affected in NP + NP B cells (0.32 ± 0.32% gated, n = 5) but were increased in NP + RUPP B cells (7.60 ± 2.26% gated, n = 6, P < 0.05) and NP + RUPP B2 cells (5.88 ± 1.75% gated, n = 9, P < 0.05) compared with the NP group (Fig. 5B). There were no changes in total placental NK cells in NP + NP B cells (34.63 ± 11.03% gated, n = 4), NP + RUPP B cells (24.72 ± 5.57% gated, n = 6), or NP + RUPP B2 cells (47.53 ± 3.92% gated, n = 8) compared with the NP group (42.73 ± 9.02% gated, n = 11; Fig. 5C). Placental activated NK cells were increased in NP + RUPP B cells (2.32 ± 0.74% gated, n = 7, P < 0.05) compared with the NP group (0.20 ± 0.12% gated, n = 11), but there were no changes seen in NP + NP B cells (0.00 ± 0.00% gated, n = 4) or NP + RUPP B2 cells (1.01 ± 0.22% gated, n = 7) (Fig. 5D).

Figure 5.

There were no changes in total circulating natural killer (NK) cells (A) or total placental NK cells (B) in any of the adoptive transfer groups (n = 5–11). Adoptive transfer of normal pregnant (NP) B cells into NP rats did not change NK cell activity in the circulation (C) or in the placenta (n = 4) (D). Adoptive transfer of reduced uterine perfusion pressure (RUPP) B cells into NP rats significantly increased cytolytic NK cells in the circulation (P < 0.01) and placentas (P < 0.01) compared with either NP or NP + NP B cell rats (n = 6 or 7). Adoptive transfer of RUPP B2 cells into NP rats increased cytolytic NK cells in the circulation (P < 0.05) compared with NP rats (n = 7-8). Data were compared using one-way ANOVA and are reported as means ± SE. *Statistically significant (P < 0.05) change compared with the referenced group.

Effect of RUPP B Cells or RUPP B2 Cells on Circulating AT1-AA

RUPP rats had increased circulating AT1-AA (19.50 ± 1.55 ΔBPM, n = 4) compared with NP rats (0.83 ± 0.62 ΔBPM, n = 4, P < 0.05; Fig. 6), as previously described (31). NP + NP B cells (1.16 ± 0.42 ΔBPM, n = 4) had no changes in circulating AT1-AA compared with the NP group. NP + RUPP B cells had increased AT1-AA (20.33 ± 1.54 ΔBPM, n = 6, P < 0.05) compared with NP and NP + NP B cells. NP + RUPP B2 cells had increased AT1-AA (19.17 ± 1.20 ΔBPM, n = 4, P < 0.05) compared with NP and NP + NP B cells.

Figure 6.

The reduced uterine perfusion pressure (RUPP; n = 4) procedure significantly increased circulating agonistic autoantibodies against the angiotensin II type 1 receptor (AT1-AA; P < 0.001) compared with the normal pregnant (NP) group (n = 4), as previously described. Adoptive transfer of NP B cells into NP rats (n = 4) did not increase circulating AT1-AA compared with NP rats. Adoptive transfer of either RUPP B cells (n = 6, P < 0.001) or RUPP B2 cells (n = 4, P < 0.001) into NP rats significantly increased serum AT1-AA levels compared with NP + NP B cell rats. Data were compared using one-way ANOVA and are reported as means ± SE. *Statistically significant (P < 0.05) change compared with the referenced group.

Effect of RUPP B Cells or RUPP B2 Cells on Circulating and Placental Complement Component C1q

Circulating complement component C1q was not changed between NP (13.2 ± 1.1 mg/mL, n = 6) and RUPP (17.1 ± 2.5 mg/mL, n = 6) groups. Changes in circulating C1q were also not associated with adoptive transfer of NP B cells (19.7 ± 2.5 mg/mL, n = 5), RUPP B cells (24.4 ± 7.2 mg/mL, n = 5), or RUPP B2 cells (14.6 ± 4.6 mg/mL, n = 5) compared with NP controls. In addition, placental C1q was not changed between NP (1.3 ± 0.1 μg/mg protein, n = 6) and RUPP (1.0 ± 0.1 μg/mg protein, n = 5) groups, nor was placental C1q changed in response to adoptive transfer of NP B cells (1.0 ± 0.2 μg/mg protein, n = 5), RUPP B cells (1.3 ± 0.2 μg/mg protein, n = 5), or RUPP B2 cells (0.8 ± 0.1 μg/mg protein, n = 5) compared with NP controls.

Effect of RUPP B Cells or RUPP B2 Cells on Circulating Complement Components C3 and C3a

RUPP rats had increased complement component C3 (1,064 ± 70 µg/mL, n = 5, P < 0.05) compared with NP rats (799 ± 49 µg/mL, n = 5; Fig. 7A). There were no changes in C3 associated with adoptive transfer of NP B cells (959 ± 68 µg/mL, n = 5), RUPP B cells (863 ± 50 µg/mL, n = 6), RUPP B cells + n7aac (866 ± 40 µg/mL, n = 5), RUPP B2 cells (774 ± 47 µg/mL, n = 5), or RUPP B2 cells + n7aac (728 ± 28 µg/mL, n = 4) compared with NP controls.

Figure 7.

A: the reduced uterine perfusion pressure (RUPP; n = 5) procedure significantly increased circulating complement component C3 (P < 0.05) compared with the normal pregnant (NP) group (n = 5). Adoptive transfer of NP B cells (n = 5), RUPP B cells (n = 6), or RUPP B2 cells (n = 5) did not change circulating complement component C3. B: the RUPP (n = 5) procedure significantly increased circulating complement component C3a (P < 0.05) compared with the NP group (n = 7). Adoptive transfer of NP B cells (n = 5), RUPP B cells (n = 6), or RUPP B2 cells (n = 5) did not change circulating complement component C3a. Data were compared using one-way ANOVA and are reported as means ± SE. 7AA, n7aac peptide. *Statistically significant (P < 0.05) change compared with the referenced group.

The RUPP procedure significantly increased circulating complement component C3a (44.1 ± 6.0 ng/mL, n = 5) compared with the NP group (28.7 ± 3.9 ng/mL, n = 7, P < 0.05; Fig. 7B). There were no changes in circulating C3a associated with adoptive transfer of NP B cells (26.78 ± 2.56 ng/mL, n = 5), RUPP B cells (32.90 ± 3.29 ng/mL, n = 6), RUPP B cells + n7aac (21.09 ± 1.45 ng/mL, n = 5), RUPP B2 cells (20.82 ± 2.27 ng/mL, n = 5), or RUPP B2 cells + n7aac (27.24 ± 2.77 ng/mL, n = 4).

Effect of RUPP B Cells or RUPP B2 Cells on Circulating Complement Component C4

The RUPP procedure did not change circulating complement component C4 (892 ± 52 µg/mL, n = 5) compared with the NP group (913 ± 73 µg/mL, n = 5). Adoptive transfer of NP B cells did not increase complement component C4 (1,140 ± 123 g/mL, n = 5). Moreover, adoptive transfer of neither RUPP B cells (1,014 ± 118 µg/mL, n = 5) nor RUPP B2 cells (947 ± 77 µg/mL, n = 5) changed circulating complement component C4. There were no changes in complement C4 in NP + RUPP B cells + n7aac (947 ± 77 µg/mL, n = 5) or NP + RUPP B2 cells + n7aac (863 ± 166 µg/mL, n = 4).

Effect of RUPP B Cells or RUPP B2 Cells on Renal Complement Activation or Granzyme Proteins

The RUPP procedure did not change renal C1q (0.6 ± 0.1 µg/mg protein, n = 5) compared with the NP group (0.9 ± 0.2 µg/mg protein, n = 5; Fig. 8A). Adoptive transfer of NP B cells (0.4 ± 0.02 µg/mg protein, n = 5, P < 0.05), RUPP B cells (0.5 ± 0.04 µg/mg protein, n = 6, P < 0.05), or RUPP B2 cells (0.5 ± 0.04 µg/mg protein, n = 6, P < 0.05) reduced renal C1q levels compared with the NP group. There was no change in renal C3a in the RUPP group (43.51 ± 3.60 ng/mg, n = 5) compared with the NP group (36.40 ± 3.28 ng/mg, n = 5).

Figure 8.

A: the reduced uterine perfusion pressure (RUPP; n = 5) had no change in renal C1q deposition compared with the normal pregnant (NP) group (n = 5). NP + NP B cells (n = 5), NP + RUPP B cells (n = 6), and NP + RUPP B2 cells (n = 6) had decreased renal C1q compared with the NP group (P < 0.05). There was no change in renal C3a between NP (n = 5) and RUPP (n = 5) groups. B: there was no change in renal C3a between NP (n = 5) and RUPP (n = 5) groups. There was no change associated with adoptive transfer of RUPP B cells (n = 6), but adoptive transfer of RUPP B2 cells (n = 6) significantly increased renal C3a (P < 0.05) compared with recipients of NP B cells (n = 5). C: there was no change in renal perforin in any of the groups (n = 4-6). D: the RUPP group (n = 4) had significantly increased granzyme A (P < 0.05) compared with the NP group (n = 4). Adoptive transfer of RUPP B cells (n = 6) or RUPP B2 cells (n = 6) significantly increased renal granzyme A (P < 0.05) compared with recipients of NP B cells (n = 5). E: the RUPP (n = 4) procedure significantly increased granzyme B (P < 0.05) compared with the NP group (n = 4). Adoptive transfer of RUPP B cells (n = 6) or RUPP B2 cells (n = 6) significantly increased renal granzyme B (P < 0.05) compared with recipients of NP B cells (n = 5). F: the RUPP (n = 4) procedure significantly increased renal granzyme K (P < 0.05) compared with the NP group (n = 5). There was no change associated with adoptive transfer of NP B cells (n = 5), RUPP B cells (n = 6), or RUPP B2 cells (n = 6). Data were compared using one-way ANOVA and are reported as means ± SE. 7AA, n7aac peptide. *Statistically significant (P < 0.05) change compared with the referenced group.

There was no change in renal C3a associated with adoptive transfer of RUPP B cells (44.57 ± 4.91 ng/mg, n = 6) compared with NP B cells (27.56 ± 3.67 ng/mg, n = 5); however, adoptive transfer of RUPP B2 cells significantly increased renal C3a (51.85 ± 5.51 ng/mg, n = 6, P < 0.05) compared with NP B cell recipients (Fig. 8B).

There were no changes in renal perforin between any of the groups (NP: 3.78 ± 0.57 ng/mg, n = 5; RUPP: 4.05 ± 0.63 ng/mg, n = 4; NP B cells: 2.57 ± 0.34 ng/mg, n = 5; RUPP B cells: 3.77 ± 0.64 ng/mg, n = 6; RUPP B cells + n7aac: 4.13 ± 0.60 ng/mg, n = 5; RUPP B2 cells: 4.46 ± 0.74 ng/mg, n = 6; and RUPP B2 cells + n7aac: 4.56 ± 0.67 ng/mg, n = 4) (Fig. 8C).

The RUPP group had significantly higher renal granzyme A (1.75 ± 0.09 ng/mg, n = 4) compared with the NP group (1.05 ± 0.10 ng/mg, n = 4, P < 0.05; Fig. 8D). Adoptive transfer of RUPP B cells (1.28 ± 0.10 ng/mg, n = 6) or RUPP B2 cells (1.33 ± 0.15 ng/mg, n = 6) had increased renal granzyme A compared with recipients of NP B cells (0.73 ± 0.12 ng/mg, n = 5, P < 0.05). Renal granzyme A was not reduced with n7Aac and was 1.21 ± 0.18 ng/mg (n = 5) in the RUPP B cell + n7aac group and 1.57 ± 0.04 ng/mg (n = 4) in the RUPP B2 cell + n7aac group compared with their untreated counterparts.

The RUPP group had increased renal granzyme B (16.14 ± 0.93 pg/mg, n = 4) compared with the NP group (8.65 ± 1.16 pg/mg, n = 4, P < 0.05; Fig. 8E). Adoptive transfer of either RUPP B cells (10.97 ± 0.57 pg/mg, n = 6) or RUPP B2 cells (11.57 ± 2.20 pg/mg, n = 6) had significantly increased renal granzyme B compared with recipients of NP B cells (5.34 ± 1.09 pg/mg, n = 5, P < 0.05). Renal granzyme B was not reduced with n7Aac and was 10.57 ± 1.67 pg/mg (n = 5) in the RUPP B cell + n7aac group and 13.51 ± 0.55 pg/mg in the RUPP B2 cell + n7aac group (n = 4) compared with their untreated counterparts.

The RUPP group had increased renal granzyme K (14.24 ± 1.79 ng/mg, n = 4) compared with the NP group (8.77 ± 1.13 ng/mg, n = 5, P < 0.05; Fig. 8F). There was no change in renal granzyme K in any of the other groups (NP B cells: 6.80 ± 0.99 ng/mg, n = 5; RUPP B cells: 8.39 ± 0.72 ng/mg, n = 6; RUPP B cells + n7aac: 9.34 ± 1.89 ng/mg, n = 5; RUPP B2 cells: 8.01 ± 1.12 ng/mg, n = 6; and RUPP B2 cells + n7aac: 10.14 ± 0.46 ng/mg, n = 4).

DISCUSSION

There is a limited amount of research that has investigated the role of B cells or memory capable B2 cells in the pathophysiology of hypertension in PE. Our study demonstrates an important role for B cells as well as for classical memory B2 cells in the production of AT1-AA and in causing hypertension in response to placental ischemia. Moreover, treatment of RUPP B cell recipient rats with the AT1-AA inhibitor peptide n7Aac attenuated hypertension, supporting a role for AT1-AA to be secreted by total B cells and importantly by B2 cells to cause hypertension during pregnancy. Furthermore, our study shows that RUPP B cells and B2 cells activate NK cells, possibly by AT1-AA. However, it appears that other mechanisms outside of B cells and AT1-AA are responsible for activation of the complement cascade.

B cells are divided into two main groups: B1 cells and B2 cells. B1 cells are CD43⁺ and stem from a unique fetal liver progenitor cell (36). B1 cells predominate in early life but make up <5% of total B cells in adults and roughly 50% of peritoneal cavity B cells (37, 38). As a whole, B1 cells seem to function as “innate-like” lymphocytes that produce nonspecific antibodies independent of T cells and help to maintain tissue homeostasis and the response to infection (39, 40). B1 cells are further divided into CD5⁺ B1a cells and CD5⁻ B1b cells (41). B1a cells are CD5⁺ cells and have mostly been investigated in relation to autoimmune pathogenesis; however, their role in normal immune function is not as clear (27, 42, 43). Some peritoneal B1 cells produce IL-10 and may help to quell rampant chronic inflammatory responses (44). B1 cells have been indicated in the pathophysiology of PE as producers of AT1-AA (27). However, AT1-AA has been seen 7 years postpartum (25), which suggests that mechanisms in long-lived memory B cell survival in the development of AT1-AA are associated with classical B2 cell activation.

B2 cells are CD43⁻ and are classical B cells that function with T cells to help to produce high-specificity IgG and result in immune memory (45). B2 cells stem from the common lymphoid progenitor cell and make up the majority of B cells in the body (46). B2 cells are critical in pathogen defenses and use antigen memory to protect the body from repeat infections (47). After a B2 cell is activated by a CD4⁺ T cell, it undergoes class-switch recombination and then proliferates and transforms into antibody-secreting plasma cells and memory B cells (45, 48). B2 cells have been implicated in autoimmune disorders and inflammation (49, 50, 51). However, no studies have tested the ability of classical B2 cells to produce AT1-AA in response to placental ischemia or for B2 cells to cause hypertension and immune activation during pregnancy.

Antibody production is the primary function of B cells. Antibodies are proteins that bind a specific antigen on a pathogen or infected cell and promote its clearance through destruction by the immune system. Antibody-induced clearance can occur through Fc recognition by NK cells or CD8 T cells, complement activation, and opsonization (52). Natural antibodies are low-affinity, polyreactive Ig that are created independently of T cells and are under the subclasses IgA, IgM, and IgG, but most natural antibodies are IgM (53). In normal pregnancies, B cells have dynamic changes in antibody production. IgM, IgA, and IgG concentrations in serum swing throughout normal pregnancies (54). Antibodies are divided into five classes (IgA, IgD, IgE, IgG, and IgM), each with unique effector functions (45). IgA mostly functions to aid in immune responses in mucosal membranes (55). IgD antibodies are seen on the surface of mature B cells (56), and IgE is seen in allergic responses (57). IgM is on the surface of mature B cells and can be secreted in a pentameric form from innate-like B1 cells as a polyreactive antibody (58). IgG antibodies can be categorized into subtypes IgG1−IgG4, are differentiated by structural differences in hinge regions, and are numbered in order of decreasing serum concentration. IgG1−IgG4 are very similar in sequence but can have unique effector functions (52). IgG1 is induced against soluble or membrane-bound proteins, IgG2 is induced against bacterial capsular polysaccharides, IgG3 is very effective at activating proinflammatory signals like activation of NK cells or the complement system and is mainly against viral antigens, and IgG4 is usually induced along with IgE against allergens (52). Repeated activation of memory B cells by the same antigen leads to the production of a large quantity of high-affinity, monospecific class-switched IgG antibodies (the secondary immune response) (45). B cell dysregulation is known to be involved in multiple autoimmune disorders including system lupus erythematosus (59), where B cells produce autoantibodies. B cells producing autoantibodies have been implicated in the pathologies of pregnancy including recurrent miscarriage and PE (27, 60, 61). Autoantibodies are antibodies that recognize self-antigens and can arise from B cells that were not recognized at immune checkpoints, molecular mimicry, and from hypermutation in germinal centers after B cell activation (62, 63). AT1-AA is an IgG3 antibody (64); therefore, it can activate multiple inflammatory pathways such as NK cells or complement and AT1-AA during pregnancy, which has been associated with hypertension, increased circulating tumor necrosis factor-α, increased endothelin-1, and mitochondrial reactive oxygen species (65).

Hubel et al. (24) and Rieber-Mohn et al. (25) have shown that AT1-AA circulates in women with previous PE for years postpartum. T cell-B2 cell communication is important in long-term antibody production (66), and its presence for years postpartum supports the hypothesis that B2 cells play important roles in the production of AT1-AA during pregnancy and postpartum. Here, our data demonstrate that B cells and memory capable B2 cells (the majority of the total B cell population) in response to placental ischemia are able to induce hypertension in NP recipient rats. Furthermore, administration of n7aac peptide, a specific inhibitor to AT1-AA, prevents hypertension associated with adoptive transfer of RUPP total B or B2 cells. Moreover, the isolation of AT1-AA from the serum of rats that received RUPP B or B2 cells indicates that B2 cells are able to make AT1-AA in pregnant rats, thus supporting the idea that memory B cells play an important role in AT1-AA in PE. B1 cells constitute <5% and are a small population of total B cells, and thus we believe that the response seen in recipients of RUPP B cells is due to the predominance of the B2 subset within the total population of B cells.

Our laboratory has previously demonstrated that cytolytic NK cells are elevated in the RUPP rat (34) and in response to adoptive transfer of RUPP CD4⁺ T cells (13). In this study, we showed activation of circulating NK cells in recipient rats by RUPP B cells and RUPP B2 cells. Yet, B2 cells were not able to significantly increase placental cytolytic NK cells, potentially due to the removal of B1 cells. B1 cells were implicated in the pathophysiology of PE, and both B1 and B2 cells may be necessary to fully activate placental NK cells (27). Both antibodies and cytokines are important in NK cell activation (29), and our data indicate a role of other cell types secreting cytokines, such as CD4⁺ T cells, in conjunction with B1 and B2 cells to activate placental NK cells in PE. We found increased renal granzyme A and granzyme B in response to RUPP, RUPP B cells, and RUPP B2 cells. Cytolytic granzymes can be used to represent killer cell activation in tissues, which includes CD8⁺ T cells and NK cells (67), and may be indicators of tissue damage by immune cells. We have previously shown that renal NK cells are activated in RUPP rats (68), and this study further supports that by demonstrating increases in renal granzymes A, B, and K. Moreover, in this study, we demonstrated that B2 cells are important in renal killer cell activation by showing increased granzyme A and granzyme B in the kidneys of RUPP B and B2 cell recipient rats. NK cells are important in the destruction of cells marked by antibodies. Following an antibody binding its antigen, formation of the immune synapse occurs by recognition of the antibody’s FC region by CD16 on the surface of NK cells, resulting in their activation and subsequent release of perforin and granzymes that kill the target cell (69, 70). Uterine NK cells normally have very low cytotoxic activity, but shifts in uterine NK cells to a cytolytic phenotype have been associated with PE and in response to placental ischemia. Activated NK cells have been associated with hypertension and growth restriction in the RUPP model (71–78). Circulating NK cells are able to invade the aortic wall of organs and secrete interferon-γ and other cytokines to attract other immune cells (79). Interferon-γ then serves to activate monocytes, which produce IL-12, which activates NK cells and functions as a positive feedback loop (80) leading to vascular stiffening, tissue destruction, and contributing to hypertension. This could be the mechanism for renal and vascular dysfunction in response to RUPP and RUPP B cells. Interestingly, in our study, we did not see activated placental NK cells and, likewise, did not see reduced fetal weight. Together, this suggests the importance of activated placental NK cells in fetal growth restriction and may suggest the importance of other factors outside of AT1-AA as mechanisms for placental NK cell activation to cause fetal demise during PE.

The RUPP procedure reduced fetal weight and placental efficiency, as previously described (75, 81). Adoptive transfer of neither RUPP B cells nor RUPP B2 cells resulted in reductions in fetal weight or placental efficiency, but, surprisingly, adoptive transfer of NP B cells increased pup weight without altering placental efficiency. This may be due to B cell stimulation of angiogenic factors during pregnancy. Aurora et al. (82) reported decreased immune complex-dependent airway vasculature remodeling in B cell-deficient mice, which suggests a role for B cells in contributing to vascular angiogenesis, a plausible mechanism for improved fetal weight occurring in recipients of NP B cells. Nevertheless, more research is needed to investigate a role for B cells in angiogenesis in pregnancy to draw any conclusions.

In this study, we found no changes in circulating complement in response to adoptive transfer of RUPP B cells. Complement C1q in the circulation, placenta, or kidneys was not increased in response to RUPP, but overall the numbers seen with this assay were very low in all groups. Thus, these data may not offer much insight into the role of B cell or AT1-AA activation of complement during PE. Circulating complement components C1q and C4 were reduced in women with PE compared with NP women (83); this was hypothesized to be due to complement C1q deposition into tissue and C4 cleavage, suggesting classical complement activation. The authors also implicated AT1-AA as a potential mechanism to classical complement activation in PE (83). However, we found no changes in complement C4 in our study. Penning et al. (23) then showed that while NP women and women with PE had glomerular C1q deposition, women with PE had increased C4d deposited in glomeruli. This would further suggest classical complement activation in the kidney during PE as well. In our study, we also noted increased circulating C3 and C3a in the RUPP group but not in response to adoptive transfer of B cells. Interestingly, we found no change in renal C3a in RUPP rats but noticed increased renal C3a in recipients of RUPP B2 cells compared with recipients of NP B cells. This suggests that B2 cells may be able to contribute to renal inflammation. C3a has been reported as both increased and not changed in PE, but C3a may be dependent on EO-PE versus LO-PE (83–86). A study in mice has also suggested that hypertension in response to AT1-AA could be in part due to C3a activation (87). Although adoptive transfer of RUPP B cells did not appear to activate complement, there is still literature to support the connection between AT1-AA and complement activation in PE and in the RUPP rat, which was not present with activated B cells secreting AT1-AA alone. Therefore, this suggests the importance of additional mechanisms of complement activation in RUPP rats and in women with PE. Although the results of this study identified classical B2 cells as important contributors to the pathophysiology of PE, one limitation to the study is that we did not include sham-operated rats as a control. Sham rats are routinely used when studying the effects of RUPP on fetal programming of hypertension and adult diseases. However, we have previously shown that sham-operated animals have no change in blood pressure, fetal morbidity, or inflammation compared with NP controls, and, therefore, we do not include them routinely as it may be seen as an unnecessary procedure for the pregnant rat (88).

In conclusion, our data demonstrate an important role for total B cells and B2 cells to contribute to the hypertension seen in PE as well as a need for further investigation into the role of B cells in PE. Our data suggest that B2 cells are able to produce the AT1-AA and contribute to renal tissue damage as observed with increased renal complement C3a, granzyme A, and granzymes B in response to placental ischemia. As B2 cells are associated with immunological memory, B2 cells could produce AT1-AA during pregnancy and postpartum, thus contributing to cardiovascular and renal disease in women with previous PE. Moreover, selective targeting of naïve B2 cells or blasting (IgG secreting) B2 cells could prevent the production of AT1-AA while leaving existing memory B cells intact to aid in maternal immunity. Future studies could aim to investigate selective depletion of B2 cells in an effort to prevent AT1-AA production in pregnancy or postpartum. Collectively, these data suggest that B2 cells contribute to the hypertension and inflammation seen in PE and indicate B2 cells as another potential therapeutic target for the management of PE.

Perspectives and Significance

Management of PE tends to include antihypertensive therapy (Ca²⁺ channel blockers, β-blockers, and vasodilators), magnesium sulfate to prevent seizures, and corticosteroids to aid in fetal lung development. Inhibitors of the renin-angiotensin-aldosterone system are contraindicated in pregnancy because of their teratogenic effects on fetal kidney development. Thus, there is a dire need for more options to manage the hypertension seen in women with PE. This study adds to the breadth of knowledge connecting inflammation and hypertension in PE. Both innate and adaptive immune systems have been implicated in the pathophysiology of PE, and inhibitors of immune mediators could be effective additions to the medications available in the clinic. Women with systemic lupus erythematosus and Sjögren’s syndrome are able to continue anti-inflammatory therapy during pregnancy, which gives credence to the addition of immune modulators in PE. Women with previous PE have an increased risk for cardiovascular disease and PE in later pregnancies; interruption of immune memory mechanisms could reduce these risks. Potential targets could include interruption of T cell-B cell communication, specific B cell depletion, or specific inhibition of AT1-AA. As a whole, this study shows an important role for B cells and inflammation to contribute to the pathophysiology of PE.

DATA AVAILABILITY

Data will be made available upon reasonable request.

GRANTS

This work was supported by National Institutes of Health Grants RO1HD067541 (to B.L.), P20GM121334 (to B.L. and L.M.A.), 1U54GM115428 (to L.M.A.), and 1R01HL151407 (to D.C.C.) and by American Heart Association Early Career Award 19CDA34670055 (to L. M.A.).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

B.L. conceived and designed research; O.T.H., E.D., N.C., J.L., N.I., and T.W.T. performed experiments; O.T.H., E.D., L.M.A., N.C., D.C.C., T.W.T., and B.L. analyzed data; O.T.H., E.D., L.M.A., N.C., and B.L. interpreted results of experiments; O.T.H. prepared figures; O.T.H. and B.L. drafted manuscript; O.T.H., E.D., L.M.A., N.C., S.F., T.I., and B.L. edited and revised manuscript; O.T.H., E.D., L.M.A., N.C., J.L., N.I., D.C.C., T.W.T., S.F., T.I., R.D., G.W., and B.L. approved final version of manuscript.