Novel drug targets for ductus arteriosus manipulation: Looking beyond prostaglandins

Abstract

Forty years ago, non-steroidal anti-inflammatory drugs were first reported to decrease systemic prostaglandin levels and promote ductus arteriosus (DA) closure. And yet, prolonged patency of the DA (PDA) remains a significant clinical problem, complicated by imperfect therapies and wide variations in treatment strategy. There are few pharmacology-based tools available for treating PDA (indomethacin, ibuprofen, acetaminophen), or for maintaining DA patency (PGE₁) as is needed to facilitate corrective surgery for ductus-dependent congenital heart defects. Unfortunately, all of these treatments are inefficient and are associated with concerning adverse effects. This review highlights novel potential DA drug targets that may expand our therapeutic repertoire beyond the prostaglandin pathway.

Introduction

The ductus arteriosus (DA) is a muscular artery that connects the fetal aorta and pulmonary artery, providing an integral shunt of the fetal circulation, essential for maintaining fetal wellbeing. In utero, the DA diverts approximately 90% of the right ventricular output away from the high-resistance pulmonary circulation and into the systemic umbilical-placental circulation where gas exchange can occur with maternal blood¹. Postnatally, the DA normally closes within 72 hours after birth to facilitate perfusion of the newly-inflated lungs. Failure to close results in a persistent left-to-right shunt termed patent ductus arteriosus (PDA). PDA affects up to 80% of premature infants² and accounts for 5–10% of all congenital heart defects in term infants³. Hemodynamically significant PDA is associated with significant morbidities including neurodevelopmental impairment, intraventricular hemorrhage, pulmonary hemorrhage, respiratory distress syndrome, bronchopulmonary dysplasia, necrotizing enterocolitis (NEC), spontaneous intestinal perforation, and congestive heart failure⁴. On the other hand, in ductus-dependent heart lesions, which account for 20–25% of all congenital heart defects, it is necessary that patency of the neonatal DA be maintained in order to preserve life-saving pulmonary or systemic circulation prior to corrective surgery⁵.

The fetal DA has intrinsic tone and requires dilating factors to maintain its patency in utero. Nitric oxide (NO) and prostaglandin E₂ (PGE₂) induce vessel relaxation by activating cGMP/PKG and cAMP/PKA signaling respectively. While NO and PGE₂ are typically considered primary mediators of DA dilation, other factors, including potassium (K⁺) channels, adenosine, and atrial naturetic peptide, also play a role^6–9. At birth, the DA constricts in response to the acute increase in oxygen (O₂) tension coupled with a loss of vasodilators. O₂ may act through several different pathways to exert its contractile effects on the DA⁴, including cytochrome P450-mediated induction of endothelin 1¹⁰ and reactive oxygen species-inhibition of voltage-gated potassium (K⁺) channels (Kv1.5 and Kv2.1)¹¹. O₂ may also inhibit ATP-sensitive K⁺ (K_ATP) channels, resulting in membrane depolarization, activation of voltage dependent Ca²⁺ channels and increased intracellular Ca²⁺ accumulation¹².

Current Therapies

There are few pharmacology-based tools currently available for modulating DA tone^13–15. Introduced as a PDA therapy in the mid 1970s^16–18, the nonsteroidal anti-inflammatory drug (NSAID), indomethacin, facilitates DA closure by inhibiting cyclooxygenase COX-1 (PTGS1) and COX-2 (PTGS2) enzymes that are required for PGE₂ synthesis. A potent vasoconstrictor, indomethacin became the PDA drug of choice for many years, despite its association with renal dysfunction, platelet abnormalities, spontaneous intestinal perforation, and NEC^19–21. In the mid 1990s, ibuprofen emerged as an alternative to indomethacin, demonstrating equal efficacy, but with a decreased risk of NEC^22–24. However, use of either of these non-selective NSAIDs is still not ideal given their propensity to induce generalized vasoconstriction^{20, 21, 25}. In addition, indomethacin and ibuprofen are ineffective in approximately 30% of patients²⁶, exposing such infants to increased risk of renal dysfunction, spontaneous intestinal perforation, and NEC with no therapeutic benefit. Recently, acetaminophen has been touted as the optimal choice for PDA management, because it is less likely to cause peripheral vasoconstriction and can be administered to infants with contraindications to NSAIDs²⁷. Like the others, acetaminophen is a COX inhibitor, although it may not be as effective at blocking PGE₂ synthesis in immature vessels, potentially limiting its effectiveness in premature infants²⁸.

The ability of PGE₁ and PGE₂ to relax the ex vivo fetal lamb DA, first demonstrated in 1973 by Coceani and Olley²⁹, led to initiation of several clinical trials to evaluate the efficacy of PGE₁ in maintenance of DA patency in infants with ductus-dependent congenital heart defects^30–32. Today, PGE₁ is still the only pharmacology-based therapy used to keep the DA open, despite its association with peripheral vasodilation, apnea, fever, and other physiologic disruptions¹³. Prolonged treatment produces even more severe side effects, including cortical hyperostosis, gastric-outlet obstruction, and pseudo-Bartter syndrome³³. In addition, while PDA therapies can be administered orally or intravenously, PGE₁-mediated DA dilation requires continuous intravenous infusion, another significant limitation. Despite the inefficient nature of these current tools, relatively little effort has been expended over the last four decades to identify novel targets for pharmacology-based DA manipulation.

Towards Specific DA-targeted Therapies

Lack of progress towards novel DA drug development is not entirely unexpected given the low probability of success for drug development programs in general. A recent study³⁴, the largest of its kind, reports that from 2005–2015 only 8.5 % of all industry-sponsored drug development programs ended in FDA approval. Moreover, the rate at which new drugs are validated and brought to market is staggeringly slow, with the average drug spending 8.14 years in clinical trials³⁴. However, drugs with the highest likelihood of success target proteins with these key characteristics: druggability, enrichment in the tissue of interest, and demonstrated relevance to disease^{35, 36} (Figure 1). Targets with some precedent for binding a drug-like molecule are considered part of the “druggable genome”³⁷. Most modern drugs act on receptors, enzymes, and ion channels/transporters. However, as the pharmaceutical industry devises new ways to target moieties such as protein-protein interactions and nucleic acids, the druggable genome will expand to include non-traditional targets. Accordingly, there will be room for expansion of the potential targets for DA therapy, beyond the PGE-pathway.

Figure 1. Characteristics of ideal drug targets.

DA drug targets with the highest likelihood of receiving FDA approval are those that are druggable, enriched in the DA, and have an established link to PDA (denoted by asterisk in the central, overlapping region of the diagram)

Traditional and non-traditional potential DA drug target candidates listed in Table 1, include novel targets identified as enriched in the DA, using array and RNA-seq analyses to identify the DA’s unique transcriptional identity (reviewed by Yarboro et al, this issue)^38–48. In the majority of these studies, the DA was compared to the ascending aorta, a natural choice given their close proximity and the shared neural crest origin of each vessel’s smooth muscle cells⁴⁹. While these analyses are helpful, looking at DA-specific targets should be expanded to include a comparison to renal, mesenteric, and cerebral vessels, given the potential that novel drugs could, as do all current DA drugs, have adverse effects on these vascular beds⁵⁰. While it is unlikely that any candidate will be exclusively expressed in the DA, those that are significantly enriched lend themselves to the possibility of targeting them with a relatively low “Goldilocks” dose of drug – one that would have just the right effect on the DA, but would be insufficient to act on other tissues given low expression of the target in those tissues²⁶.

Table 1.

Representative List of Potential DA Drug Targets

	Receptors	Enzyme	Channel/Transporter	Non-traditional Targets
Enriched in the DA	Prostaglandin: PTGER4 Endothelin: EDNRA EDNRB Angiotensin: AGTR1 AGTR2 Glutamate: GRIK2 GRM3 Fibroblast Growth Factor: FGFR2 Roundabout Guidance: ROBO2	Prostaglandin Related: PTGIS PDE1B PDE1C PDE3B PDE4B PDE5A CYP1B1 Rho-associated: ROCK1 Smooth Muscle Related: MYLK CAMK1D ECM Related: ADAMTS9	ATP-gated K+ Channel: ABCC9 KCNJ8 Voltage-gated K+ Channel: Kv1.2 Kv1.5 Kv2.1 Ca2+-activated K+ Channel: KCNMA1 KCNMB1 Chloride Channel: CLCA1 CLCA2 GABRA1 TRP Ca2+ Channel: TRPM3	Transcription Factors: TFAP2B HOXB5 FOXF1A SOX9 HIF2a RUNX1 SCX DLX1 DLX2 Smooth Muscle Related: ACTA2 ACTG2 CNN1 DES MYH11 TAGLN Bone Morphogenic Protein: BMP2 BMP4 GREM1 ECM Related: FN1 TNN COL8A2 COL9A1 Cell Adhesion Related: CDH3 ICAM3 ICAM4 ICAM5 THBS1 Ligands: END1 SLIT2
Mouse models	Ptger4−/− Notch2−/− Notch2/Notch3 double−/−	Ptgs2−/− Ptgs1/Ptgs2 double−/− Hpdg−/− Lox−/−	Sloco2a1−/−	Tfap2b−/− Myocd−/− Myh11−/− Jag1−/− Brg1−/− Bmp9−/− with loss of Bmp10 Nfe2−/− Itga2b−/−
Human Syndromes	TGFBR1/2 (Loeys-Dietz)	PTPN11 (Noonan) CHD7 (CHARGE) PTEN (VACTERL) EP300 (Rubinstein-Taybi) RAB23 (Carpenter)	ABCC9 (Cantu) KCNJ8 (Cantu) CACNA1C (Timothy)	TFAP2B (Char) SMADIP1 (Mowat-Wilson) TBX5 (Holt-Oram) CREBBP (Rubinstein-Taybi) TBX1 (DiGeorge) FLNA (Periventricular heterotopia) ZIC3 (Visceral heterotaxy) SEMA3E (CHARGE) FLBN (Larsen)
SNPs	Increased Risk of PDA: AGTR1 (rs5186) TGFBR2 (rs934328) Increased Odds of Response to Indomethacin: CYP2C9 (rs2153628)	Decreased Risk of PDA: PTGIS (rs493694, rs693649)	N/A	Increased Risk of PDA: TFAP2B (rs987237) TRAF1 (rs1056567, rs3761846) PRDM6 (rs879255278) CREBBP (rs130021) Decreased Risk of PDA: ESR1 (rs2234693) IFN- γ (rs2430561)

Candidates are classified as traditional (receptor, enzyme, ion channel or transporter) or non-traditional drug targets and are organized by protein category and type of evidence linking them to PDA (mouse model, human syndrome, or non-syndromic SNP)

Additionally, several transcriptome studies analyzed changes in gene expression in preterm vs term-gestation vessels. This is important to consider given that PDA most often occurs in the setting of prematurity and that immature DAs are less responsive to indomethacin and often lack the smooth muscle machinery to maintain DA closure even if initial constriction is achieved⁵¹. Therefore, targeting pathways involved in fetal DA maturation may be key to developing therapies with a higher success rate in preterm infants. For instance, fetal DAs are significantly less responsive to oxygen-induced constriction than term-gestation vessels^52–54. Therefore, novel PDA therapies that can compensate for a diminished response to O₂ by promoting endothelin signaling or blocking K⁺ channels might be more effective at closing premature vessels.

Insights from Animal Studies

Animal models of PDA have been informative regarding disease pathways that could be therapeutically targeted^{4, 55}. Interestingly, PDA is the most common cardiac anomaly in dogs. Studies show a sex-linked genetic predisposition, with certain breeds and female dogs being at higher risk^55–57. A similar sex-linked predisposition has been reported in humans⁵⁸ although this is not supported by more recent analyses⁵⁹. Similarly, Brown Norway rats have a high prevalence of PDA, making them a popular experimental tool for identifying gene alterations in the setting of disease⁴⁵. Finally, mouse models of PDA highlight the role of specific genes in DA development and function. Some of the latter have impaired smooth muscle development (Tfap2b^−/−, Myocd^−/−, Myh1^−/−), while others have impaired prostaglandin signaling (Ptgs1/2^−/−, Ptger4^−/−, Slco2a1^−/−, Hpdg^−/−), which may seem counterintuitive given the primary role of PGE₂ in DA dilation. However, these animals suggest an additional role for prostaglandin signaling in maturing the contractile program of the ductus⁶⁰. Still other animal models lack various genes that are key in development pathways, including Notch (Notch2/3^−/−, Jag1^−/−), BMP (BMP9/10^−/−), chromatin remodeling (Brg1^−/−), platelet homeostasis (Nfe2^−/−, Itga2b^−/−), or elastin and collagen crosslinking (Lox^−/−)⁶¹. These animals not only give molecular insight into disease-causing mechanisms, they also provide established in vivo platforms for pre-clinical validation of second-generation PDA drugs.

Insights from Human Studies

Results of animal studies are not always faithful predictors of patient outcomes⁶², highlighting the need for human genetic studies⁶³. A representative set of the more than 100 single-gene mutation syndromes featuring PDA that have now been identified^{55, 64, 65} is listed in Table 1. Many of these syndromes also feature complicated congenital heart defects, making it difficult to parse out whether PDA is a primary lesion or the result of other structural and hemodynamic anomalies. Therefore, identifying gene variants associated with non-syndromic PDA may be more informative for sporadically occurring PDA, which makes up the vast majority (90%) of all PDA cases⁶⁶. Single nucleotide polymorphisms (SNPs) associated with PDA²⁶ or response to indomethacin⁶⁷ have also been reported (Table 1), although it has been difficult to confirm these findings in replication cohorts^{68, 69}.

Proof of Concept

Interestingly, targets of current PDA therapeutics fulfill all three of the key characteristics of successful drug targets illustrated in Figure 1. Indomethacin, ibuprofen, and acetaminophen all inhibit COX enzymes encoded by PTGS1 and PTGS2 genes. These enzymes catalyze PGE₂ synthesis, which subsequently regulates DA tone by binding to the DA-enriched receptor, PTGER4. Mice lacking these enzymes or the receptor have a lethal PDA phenotype^{44, 70–73}, demonstrating relevance of these targets to this disease. Therefore, it may be reasonable to assume that novel DA drug targets with the best chance of having clinical success will conform to a similar pattern. As one example, K_ATP channels possess all three key characteristics listed above. These channels are hetero-octameric complexes of pore-forming inward rectifier K⁺ channel subunits (Kir6.1 or Kir6.2) and regulatory sulfonylurea receptor subunits (SUR1, SUR2A, or SUR2B). Channel isoforms have distinct pharmacological and physiological characteristics and are expressed in a tissue-specific manner (Table 2), making them potentially ideal targets for DA-selective therapies⁷⁴.

Table 2.

Isoform-specific expression of K_ATP channel subunits

Subunits	Tissue
Kir6.1/SUR2B	Vascular Smooth Muscle and Endothelium
KIR6.1/SUR2B	Visceral Smooth Muscle
Kir6.2/SUR2A	Skeletal Muscle
Kir6.1/SUR2	Cardiac Conduction System
Kir6.2/SUR1,2	Cardiomyocytes
Kir6.2/SUR1	Pancreatic β Cells, Brain

A regulatory link between K_ATP channels and DA tone is becoming increasingly apparent. Vascular K_ATP channels, formed of the Kir6.1/SUR2B isoform combination, are developmentally regulated and significantly enriched in the DA compared to other vessels^{38–42, 45, 48}. Animal studies have shown that the DA constricts in response to glibenclamide^{12, 75}, a non-specific K_ATP channel inhibitor commonly used to treat diabetes. Conversely, fetal or neonatal exposure to diazoxide, a K_ATP channel activator, inhibited DA constriction in rats⁷⁶, and was reported to cause DA reopening in hypoglycemic infants⁷⁷. Most notably, 50% of patients with Cantu syndrome, caused by activating mutations in Kir6.1 or SUR2B, have a clinically significant PDA that is often resistant to indomethacin therapy, thus requiring surgical ligation to achieve closure⁷⁸.

Capitalizing on the druggability and therapeutic potential of K_ATP channels, several groups, including our own, are working to develop subunit-specific activators and inhibitors^{79, 80}. Using a thallium flux-based high-throughput screen, we identified a very potent novel K_ATP channel agonist (VU063), with preferential activity towards the pancreatic/neuronal channel isoform (Kir6.2/SUR1)^{74, 79}. We performed pressure myography assays on mouse DAs as previously described^{41, 81} to test the specificity and vasoactive potential of this compound (Figure 2). As expected, the SUR2-preferring K_ATP channel activator, pinacidil, dilated the DA in a dose-dependent manner, but VU063 had no effect on DA tone. Thus, while VU063 would, just like diazoxide, be conducive for treating hyperinsulinemia, unlike diazoxide it would not induce off-target vascular effects. Along these lines, it is likely that future studies will identify small molecule antagonists/agonists that preferentially target the vascular channel isoform (Kir6.1/SUR2B), which will allow for a more specific way to modulate DA tone.

Figure 2.

Isolated mouse DAs were mounted in microvessel perfusion chambers for use in pressure myography assays. Vessels were treated with increasing concentrations of pinacidil or VU063. Changes in lumen diameter were measured and plotted as a percent change from the initial lumen diameter reading under baseline (bl) conditions. Pinacidil dilated the DA, while VU063 had no effect.

Perspective and Prospects

In the 45 years since the first reports of indomethacin-induced DA closure^{16–18, 29, 82, 83}, relatively little headway has been made in developing novel approaches for therapeutic modulation of DA tone. As a consequence, there has been little progress in establishing a standard of care over the subsequent decades, leading to significant variability in treatment strategies amongst neonatal care centers and even between providers in a single care center^{84, 85}. As we learn more about the DA, it is becoming clear that PDA is a deceptively complex condition that will likely benefit more from personalized medicine rather than a one-pathway-fits all approach to therapeutics. In lieu of redundant retrospective case analyses, a greater emphasis should be placed on multidisciplinary efforts to identify DA-modulating pathways outside of the prostaglandin signaling cascade and to develop specific pharmacological modifiers of those targets.

Moving towards this goal, increased access to human tissue repositories⁸⁶ coupled with the ability to perform large-scale GWAS and other genome-wide studies will uncover new PDA-associated genes. Furthermore, these studies will provide important pharmacogenetic information that could be used at the bedside to prospectively determine the likelihood of an individual patient responding to indomethacin, thereby shielding non-responders from unnecessary exposure⁸⁷. Moreover, emerging technologies including the generation of tissue-engineered ductus vessels⁸⁸ and the creation of immortalized human DA cell lines for use in high-throughput small molecule screens will provide novel tools to identify and validate second generation DA drugs, setting the stage for rapid advances towards novel DA therapeutic development.