High levels of placenta growth factor in sickle cell disease promote pulmonary hypertension

Pulmonary hypertension (PH) occurs in 10% to 30% of patients with sickle cell disease (SCD)^1-3  and is associated with a 17% 2-year mortality in adults.¹  Factors implicated in SCD PH include endothelial dysfunction, pulmonary vasoconstriction, and remodeling, all mechanistically associated with chronic hemolysis, hypoxia, hemostatic activation, and inflammation.^4-6  Nitric oxide (NO) and endothelin-1 are opposing pulmonary vasoactive factors that regulate pulmonary vascular tone.^7,8  Hemolysis in SCD results in quenching of NO by extracellular hemoglobin and reduces availability of NO-synthase substrate.^6,8 

Endothelin-1, a potent pulmonary vasoconstrictor, is normally induced from endothelial cells by hypoxia-mediated up-regulation of hypoxia inducible factor-1α (HIF-1α), and its levels are elevated in PH.⁹  Indeed, endothelin-1 receptor antagonists are used for the treatment of primary PH. Endothelin-1 levels are also significantly elevated in patients with SCD,^10,11  and endothelin-1 receptor antagonists have been recently found to be beneficial in sickle-Antilles-hemoglobin-D mice.¹² 

We recently showed the mechanism of endothelin-1 induction via a novel hypoxia-independent up-regulation of HIF-1α in cultured human pulmonary microvascular endothelial cells by placenta growth factor (PlGF).¹³  Bone marrow erythroid cells produce PlGF, and PlGF levels are significantly increased in patients with chronic hemolytic anemias, SCD, and β-thalassemia, as part of the compensatory erythroid hyperplasia response.^14,15  This study was designed to test the hypothesis that chronically elevated PlGF in SCD contributes to PH.

The mouse PGF cDNA was cloned downstream of the β-globin promoter to replace green fluorescent protein (GFP) in the lentiviral vector sβ-GFP,¹⁶  to generate sβ-PlGF (Figure 1). We have shown this transcription cassette to express specifically in erythroid cells.¹⁷  Virus was produced using transient cotransfection of 293T cells and titered as described.¹⁷ 

Animal studies were performed using Animal Committee–approved protocols at Cincinnati Children's Research Hospital. Lineage⁻Sca-1⁺c-kit⁺ (LSK) cells were obtained by labeling normal mouse (BoyJ; CD45.1) marrow cells with fluorescein isothiocyanate–labeled lineage-specific antibodies, immunomagnetic depletion, staining with anti–Sca1-phycoerythrin, anti–c-kit-allophycocyanin, streptavidin–fluorescein isothiocyanate, and 7-amino-actinomycin D and sorting viable LSK cells¹⁷ ; LSK cells were transduced with 20 to 40 IU virus/cell for 24 hours; 10 000 LSK cells/mouse were transplanted into lethally irradiated C57BL/6 (CD45.2) recipients.¹⁷  Formalin-fixed, paraffin-embedded lungs were stained with anti–smooth muscle actin antibody (Sigma-Aldrich) at a dilution of 1:5000 and Alexa Fluor-488 as described.¹⁸ 

Blood was centrifuged at 4°C at 500g for 15 minutes to separate plasma. PlGF and endothelin-1 were measured using Quantikine PlGF kit (R&D Systems) and endothelin-1 kit (Assay Designs) according to the manufacturer's directions.

Cardiac magnetic resonance (MR) imaging was performed on a 7-Tesla Bruker MR scanner using a custom made single-turn solenoid radiofrequency coil with concurrent MR-compatible ECG and respiratory movement monitoring/gating system (model-1025; SA Instruments). Cine imaging was performed in the short axis using a segmented FLASH sequence: slice thickness = 1.0 mm, matrix size = 256 × 256, in-plane resolution = 117 × 117 μm². TE/TR = 3/5.2 ms, flip-angle = 20°, segments = 1. Fifteen to 20 cine frames were acquired during the cardiac cycle with a temporal resolution of TR ms; and images processed using a custom-built software (MICE, Cincinnati Children's Hospital Medical Center) programmed in IDL (IDL6.2, ITT Visual Information Solutions). Right ventricle (RV) end-diastolic and end-systolic volumes were measured. Animals were exsanguinated for plasma, and hearts isolated for wet weights of RV free-wall mass separated from the left ventricle wall and septum.

Adult patients with any genotype of SCD were enrolled in a protocol approved by the Institutional Review Board, National Heart, Lung, and Blood Institute (www.clinicaltrials.gov identifier NCT00011648), screened for PH by echocardiography as described,¹  and linked plasma samples were stored at −80°C.

Data were analyzed in GraphPad Prism Version 5.0 (GraphPad Software) using Student t test, Mann-Whitney test, Kruskal-Wallis test, and Spearman correlations.

We first characterized the plasma PlGF and endothelin-1 levels and the cardiac status of Berkeley sickle (Berk-SS) mice.¹⁹  Like patients with SCD, Berk-SS mice had significantly elevated circulating levels of PlGF and endothelin-1 and developed RV hypertrophy and dilatation (Figure 1A-C), consistent with an earlier report of hypertrophy associated with high pulmonary artery pressures in Berk-SS mice.⁸ 

To study the specific contribution of elevated PlGF to PH and remove the confounding factors of anemia, hypoxia, or hemolysis, all of which are thought to contribute to PH in SCD, we overexpressed PlGF from bone marrow erythroid cells in normal mice to levels seen in Berk-SS mice. This was achieved by transplanting genetically modified normal murine hematopoietic stem cells with an erythroid-specific sβ-PlGF vector into lethally irradiated normal mice. An analogous sβ-GFP vector was used as a control (Figure 1D). Donor cell engraftment occurred by 8 weeks. Within 8 weeks after engraftment (16 weeks after transplantation), sβ-PlGF mice had elevated PlGF and endothelin-1 levels, compared with sβ-GFP mice, which were comparable with Berk-SS mice. This was associated with RV hypertrophy (Figure 1E-G), similar to that seen in Berk-SS mice. Cardiac magnetic resonance imaging, performed on a subset of 4 PlGF mice, confirmed the presence of RV hypertrophy (Figure 1H-I) and increased RV end-diastolic volume (data not shown) in the PlGF mice; lung sections showed increased smooth-muscle actin around the arterioles (Figure 1J), findings consistent and associated with PH.²⁰  A separate temporal analysis showed endothelin-1 levels rose soon after increased systemic PlGF concentrations; by 10 weeks, cardiac catheterization showed increased RV pressures, which were associated with RV hypertrophy (supplemental Figure 1, available on the Blood Web site; see the Supplemental Materials link at the top of the online article).

We corroborated these results in 123 patients with SCD. Plasma PlGF levels in adults with SCD were more than 8 times that of normal controls (Figure 2A), consistent with an earlier report.¹⁵  Patients with SCD were divided into 3 groups: those with normal peak pulmonary artery pressures peak pulmonary artery pressure (PPAP; defined as tricuspid regurgitant jet velocity [TRV] < 2.5 m/s; n = 56), with mildly elevated TRV (2.5-2.9 m/s; n = 42) and with significantly elevated TRV (TRV ≥ 3 m/s; n = 25). PlGF correlated significantly with PPAP as indicated by TRV group (Figure 2C-D) and TRV as a continuous variable (both P < .001). PPAP has been shown to correlate with mean pulmonary artery pressures measured by cardiac catheterization in patients with SCD.¹  Consistent with release from hyperplastic erythroid cells,^15,21,22  PlGF levels significantly correlated with markers of hemolysis, high lactate dehydrogenase and bilirubin, red blood cell count, and hemoglobin. Supporting our earlier results showing that PlGF promotes a proinflammatory response, PlGF levels also correlated with C-reactive protein and erythrocyte sedimentation rate (Figure 2B).^13-15  In a multivariable logistic regression, PlGF was linked to TRV ≥ 3 m/s, independent of age and ferritin. Hemoglobin and glomerular filtration rate lost significance in a model with endothelin-1. This inseparability of markers of hemolysis, hemoglobin level, erythropoietic rate, and PlGF release in multivariable model is consistent with our concept that these pathways are intrinsically linked mechanistically (supplemental Table 1).

In conclusion, whereas endothelin-1 and PlGF can be up-regulated in association with various pathways, our approach using normal mice that overexpress PlGF unambiguously shows that elevated systemic PlGF concentrations induce endothelin-1 and result in PH. We also show that SCD patients with the highest TRV have the highest PlGF and endothelin-1 levels, PlGF levels were highest in the most anemic patients, presumably from an erythropoietic response,¹⁵  findings recently confirmed by others,²³  although the reticulocyte relationship was not significant in our cohort, possibly the result of the confounding effect of high erythropoietin levels by hydroxyurea use,²⁴  chronic transfusion, and renal insufficiency. Our results suggest that factors related to hemolysis, iron overload, and inflammation may contribute to PlGF regulation in SCD, with consequent stimulation of endothelin-1 release, promoting pulmonary vasoconstriction. Additional effective therapies are needed for PH in SCD, which is associated with a 10-fold higher risk for early mortality.¹  Our data suggest that, in conjunction with strategies that improve NO bioavailability, blockade of PlGF signaling may provide an alternative target to ameliorate PH in SCD or in other chronic hemolytic anemias, which have also been shown to be associated with PH.^25-27 

The authors thank Mary K. Hall for protocol management, Darlene Allen and Mariana Hildensheim for clinical data management, Jeff Bailey and Victoria Summey of the Comprehensive Mouse Core, Cincinnati Children's Research Foundation, for assistance in mouse irradiation and injections, Anastacia Loberg and Kristy Lauderback for assistance with mouse breeding and murine experiments, Dr Jeffrey Molkentin for providing us the PlGF and GFP encoding adenovirus constructs, Dr Clint Joiner, Cancer and Blood Institute, Cincinnati Children's Research Foundation, for critical review of the manuscript, and Dr Susan Wert for advice and assistance with immunohistochemistry.

This work was supported in part by National Institutes of Health/National Heart, Lung, and Blood Institute (1RO1 70135 and U54 HL-06-008; P.M). A.T., L.M., X.W., and G.J.K. were supported by the Division of Intramural Research, National Heart, Lung, and Blood Institute.

National Institutes of Health

Contribution: N.S., V.K.K., and P.M. designed the mouse experiments; N.S. cloned the PlGF vector, conducted and led all of the mouse experiments, and analyzed the data; T.H. cloned the GFP vector; J.W. performed and analyzed cardiac MRI data; W.G. supervised analysis and interpreted MRI data; W.C.N. and M.W.P. performed RV pressure determinations and instructed on RVH determination methods; A.T. and G.J.K. designed the clinical experiments; A.T., L.M., and X.W. performed PlGF and endothelin-1 assays; A.T., L.M., X.W., and G.J.K. analyzed clinical data; and N.S., P.M., and G.J.K. wrote the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Gregory J. Kato, Pulmonary and Vascular Medicine Branch, National Heart, Lung, and Blood Institute, 10 Center Dr, MSC 1476, Bldg 10CRC, Rm 5-5140, Bethesda, MD 20892-1476; e-mail: gkato@mail.nih.gov; or Punam Malik, Division of Experimental Hematology/Cancer Biology and Division of Hematology-Oncology, Cincinnati Children's Hospital Medical Center, 3333 Burnet Ave, Cincinnati, OH 45229-3039; e-mail: Punam.Malik@cchmc.org.