The molecular biology of pelvi-ureteric junction obstruction

Over recent years routine ultrasound scanning has identified increasing numbers of neonates as having hydronephrosis and pelvi-ureteric junction obstruction (PUJO). This patient group presents a diagnostic and management challenge for paediatric nephrologists and urologists. In this review we consider the known molecular mechanisms underpinning PUJO and review the potential of utilising this information to develop novel therapeutics and diagnostic biomarkers to improve the care of children with this disorder.


Introduction
Antenatally detected hydronephrosis is a major clinical dilemma for paediatric nephrologists and urologists (incidence of 1 in 200) [1]. This condition has become more prevalent in recent years as antenatal scanning has become more sensitive and widely used. Approximately one in seven neonates with antenatally detected hydronephrosis has pelvi-ureteric junction obstruction (PUJO) [2][3][4], making PUJO one of the most common causes of congenital urinary tract obstruction, with an incidence of one in 1000 to one in 2000 live births [3][4][5]. Interestingly, males are affected approximately threefold more frequently than females by this condition [4]. The reason for this difference is unknown.
The major challenge for clinicians is deciding which of these children, who are largely asymptomatic, require a pyeloplasty to relieve the obstruction. This is because twothirds of children with PUJO do not sustain renal damage or need surgery, and their hydronephrosis spontaneously improves [8][9][10].
Currently, serial ultrasound and invasive isotope studies are performed to guide surgical management of PUJO [4]. However, the ability of these diagnostic modalities to accurately detect obstruction, identify children at risk of functional deterioration and predict the need for surgery is questionable. Additionally, there remains debate regarding the parameters which indicate clinically significant obstruction [9,[11][12][13].
In general a pyeloplasty is performed for [6]: & differential renal function deterioration (differential function of <40% or a fall of >10% on serial MAG3 renograms) & significant hydronephrosis with a renal pelvis anteroposterior diameter of >3 cm on ultrasound scan & increasing hydronephrosis with an increasing anteroposterior diameter on serial ultrasound scan & symptomatic children.
Our current understanding of the natural history of PUJO as well as our ability to distinguish which children require surgery is inadequate. Available diagnostic tests cannot accurately discern between children with PUJO that will resolve spontaneously and those with PUJO that will persist, causing functional impairment. Consequently, despite radiological monitoring, there is a risk of loss of function in the affected kidney while the patient is under observation [14].
In this review we discuss the currently known molecular mechanisms underlying intrinsic PUJO and whether this information could contribute to the future development of novel therapies and diagnostic biomarkers.

Anatomy of the upper urinary tract
The PUJ is a region of gradual transition from the funnelshaped renal pelvis to the proximal ureter [15] (Fig. 1). It is a physiologic sphincter [16] that is characterised by prominent luminal folds with increased muscle thickness capable of creating a high-pressure zone to regulate urine flow. Similar to the adjacent renal pelvis and ureter, the PUJ comprises three main layers: the inner urothelium, middle smooth muscle and outer adventitia [15]. Smooth muscle contraction propels urine from the renal pelvis to the bladder [17], coordinated by submucosal and intra-muscular nerve plexi [18] and modulated by autonomic innervation involving a range of neurotransmitters that include acetylcholine, noradrenaline, substance P, neurokinin A, calcitonin gene-related peptide, neuropeptide Y, vasoactive intestinal peptide and nitric oxide (NO) [17].

Embryology of the ureter and PUJ
Understanding the normal embryology of PUJ formation is vital when considering where development may proceed incorrectly in congenital abnormalities such as PUJO. The kidney develops from metanephric mesoderm as far along the nephron as the distal tubules. The collecting duct onwards, including the major and minor calyces, renal pelvis and ureter has a different embryological origin, arising from the ureteric bud [19,20]. Thus, the PUJ does not represent an embryological fusion site, rather it is derived exclusively from the ureteric bud. The important molecular pathways that form the ureter and PUJ are shown in Fig. 2 and Table 1 [15, [26][27][28]. Briefly, the ureteric bud, consisting of a simple epithelial layer extending into loose mesenchyme, arises from the mesonephric duct during the fifth week of gestation in humans [26]. Epithelial cell proliferation and differentiation then results in the formation of the transitional epithelium. Epithelial paracrine and mesenchymal autocrine signalling stimulates the formation of smooth muscle cells from mesenchyme, which begins at 12 weeks of gestation in humans [26,29]. Mouse models have implicated a number of signalling molecules in this process of proliferation, aggregation, differentiation and orientation of smooth muscle cells as they encircle the urothelial tube (Fig. 2, Table 1). A second phase of smooth muscle differentiation that particularly affects the renal pelvis and proximal ureter occurs in postnatal mice (equivalent to the second trimester of gestation in humans) and is regulated by calcineurin and angiotensin II signalling [30, 31].  . These findings are noted when the PUJ is excised at pyeloplasty and therefore represent late features of PUJ obstruction (Fig. 3). Although the time course of PUJ disease progression is unknown in humans, genetic mouse models of hydronephrosis show abnormalities of peri-urothelial mesenchymal organisation as early as embryonic day (E) 12.5 (approximately equivalent to 35 days of gestation in humans) [24] and smooth muscle cell differentiation at E15.5 (approximately equivalent to 12 weeks of gestation in humans) [23]. One week postnatally (approximately equivalent to humans at birth) mice with Id2 haploinsufficiency show smooth muscle irregularity and hypertrophy at the PUJ [38], features which are common to human PUJO. The possible mechanisms underlying this pathology are described later in this review.

Modelling PUJO to understand its molecular biology
Adult and neonatal rodent models of complete and partial unilateral ureteric obstruction (UUO) have been extensively used to investigate the molecular biology of congenital obstructive nephropathy. Neonatal models are particularly helpful because rodent nephrogenesis continues for 1 week postnatally and nephron maturation over the subsequent week. Thus, at birth and 1 week of age, rodent kidney development is equivalent to humans at the second Fig. 2 Embryological signalling pathways of the PUJ. The ureteric bud arises from the mesonephric duct and initially consists of only a simple epithelial layer extending into loose mesenchyme. Epithelial cell proliferation and differentiation to form transitional epithelium leads to luminal obliteration, which at the end of the embryonic period is corrected by physiologic recanalisation of the ureter. Epithelial paracrine and mesenchymal autocrine signalling stimulates the proliferation and differentiation of the mesenchyme into smooth muscle cells (SMC) which aggregate and orientate so as to encircle the epithelial tube. Specifically, the urothelium secretes SHH which activates the PTCH1 receptor on adjacent mesenchyme, thereby stimulating mesenchymal proliferation. Mesenchymal cells (MC) express TBX18, a T-box transcription factor, which enables the correct localisation and aggregation of the former around the urothelium. The mesenchymal cells also express BMP4 which acts in an autocrine manner to upregulate TSHZ3 and MYODC. MYODC enables differentiation of SMC by increasing the transcription of genes encoding smooth muscle contractile proteins. DLGH1, expressed by the urothelium and SMC, is responsible for the correct orientation of SMC around the urothelial tube. In postnatal mice (equivalent to second trimester of gestation in humans), increased urine production matches the development of the renal pelvis and is accompanied by a second phase of muscle differentiation that particularly affects the renal pelvis and proximal ureter, regulated by calcineurin and angiotensin II signalling. The timeline refers to days of gestation (E embryonic day) in mouse models. MD Mesonephric duct, UB ureteric bud, MC mesenchymal cells. See Table 1 for description of factors active in the pathways involved in ureteric development trimester of gestation and birth, respectively [11]. This gives a window in which surgery can be performed on the animals to mimic in utero obstruction in humans. Adult obstructive models show a broadly similar pathologic progression to neonatal models with the exception that neonatal obstruction impedes normal maturation and growth of the kidney and leads to early nephron loss. The renal pathologic findings in neonatal and adult UUO models and the timescale of their development are presented in Fig. 4 [39-47].
A comprehensive review comparing neonatal models with human disease confirms their validity for investigating obstructive nephropathy and will not be further discussed in this review [48].

Proposed molecular mechanisms underpinning PUJO
In the following subsections we highlight some of the molecular steps that may lead to the development of intrinsic PUJO and subsequent obstructive nephropathy. Data have been obtained from both adult and neonatal models of complete and partial ureteric obstruction alongside evaluation of tissue obtained at pyeloplasty for human PUJ obstruction.

Neurogenic factors
Light microscopy studies have revealed reduced innervation within the muscular layer of the PUJ in human specimens Fig. 3 Pathologic features of intrinsic PUJO. Reduced luminal mucosal folds, excess collagen deposition, depletion of nerves within the muscular layer, abnormal muscle fibre arrangement, inflammatory infiltrate and both muscle hypertrophy/hyperplasia and muscle atrophy/hypoplasia are seen at the PUJ in human PUJO excised at pyeloplasty for PUJO [33]. This is associated with reduced expression of molecular markers, including glial cell line-derived neurotrophic factor (survival factor for neurons), protein gene product 9.5 (general neuronal marker), and nerve growth factor receptor protein, in the muscle layers of the stenotic PUJ compared to controls. Although it is speculated that these neuronal changes may contribute to the pathogenesis of PUJO, there is as yet no evidence to confirm or refute this notion. Conflicting changes in synaptophysin (e.g. major synaptic vesicle protein p38) expression in terms of both amount (increased and decreased) and distribution (localisation to the nucleus) are reported in PUJO compared to controls and are of uncertain significance. S-100 (schwann cell marker) and neurofilament (neuronal protein) expression   (3) altered urinary shear stress. The latter two mechanisms are likely to be the primary inducers of obstructive renal injury [48], causing dysregulation of many cytokines, growth factors, enzymes and cytoskeletal proteins ( Table 3), resulting in early renal haemodynamic changes followed by structural and functional alterations to the entire nephron. Figure 5 highlights the major mechanisms of renal injury in PUJO. Following a short initial increase in renal blood flow related to local vasodilator production [48], the intrarenal renin-angiotensin-aldosterone system (RAAS) is activated causing pre-and post-glomerular vasoconstriction and a resultant fall in renal blood flow (RBF), medullary oxygen tension and glomerular filtration rate (GFR) [11,48,64,80,[88][89][90]. Proximal tubular hypoxia and necrosis in neonatal rats with UUO suggest that vasoconstriction causes segment-specific ischaemic injury [91]. Accordingly, angiotensin II receptor, type 1 (AT1 receptor) inhibition improves tubular function by increasing RBF and GFR [92].
Reduced urine production and continuing urine drainage by venous and lymphatic systems together with tubular and renal pelvis dilatation result in a subsequent decline in renal pelvic pressure [48, 89, 93], which may be a compensatory mechanism to limit damaging increased intra-renal pressure [93].

Cytokines in the stenotic PUJ
Transforming growth factor-beta (TGF-β) expression is noted in human stenotic PUJ compared to normal controls [94]. Furthermore, the smooth muscle regulators endothelin-1 (smooth muscle constrictor) and adrenomedullin (smooth muscle relaxant) have been shown to be increased and decreased, respectively, in stenotic PUJ disease [95].
Analysis of paediatric renal pelvis tissue proximal to the PUJO for cytokines that show altered renal expression in nephropathy demonstrates increased TGF-β and reduced macrophage inflammatory protein-1alpha (MIP-1α). In contrast, epidermal growth factor (EGF), monocyte chemotactic peptide 1, interferon-γ-inducible protein 10 and RANTES (regulated on activation normal T-cell expressed and secreted) mRNA expression are unchanged, suggesting that TGF-β and MIP-1α play important roles in the development of PUJO [88, 96].

Intra-renal cytokines
Increased intra-renal angiotensin II activates nuclear factor kappa B and ROCK (rho-associated coiled-coil-forming protein kinase), leading to cytokine release and interstitial macrophage infiltration and activation. Intra-renal selectins, integrins, intercellular-adhesion molecule 1, vascular cell adhesion molecule 1, interleukin 1, monocyte chemoattractant peptide 1, colony stimulating factor 1 and osteopontin expression are all involved in macrophage stimulation [11,48,88,97]. Therefore, it appears that renal signals initiate and maintain the injurious inflammatory response to PUJO. Accordingly, both selectin and β2-integrin knockout mouse models show reduced macrophage infiltration into the obstructed kidney after UUO [43,44].

Profibrotic processes
Tubulointerstitial fibrosis is the final common pathway for many chronic kidney disorders, including obstructive uropathy, and is instigated by altered cytokine expression (Table 4). Activated resident interstitial myofibroblasts [98], expressing α-smooth muscle actin (boosts cell contractility) [99], aggregate, proliferate and produce extracellular matrix. Extracellular matrix consisting of collagens I, III and IV, fibronectin, laminin and proteoglycans accumulates due to increased synthesis and reduced degradation [74, 100,101]. Myofibroblasts amplify fibrosis by producing cytokines, including TGF-β1 and TNF-α [11]. Parenchymal damage and renal dysfunction results, such that in children with PUJO the extent of fibrosis significantly correlates with differential renal function [102].
Nitric oxide is an endogenous vasodilator that protects against tubulointerstitial fibrosis and proximal tubular oxidant injury in obstructive nephropathy [79,84,123]. Animal models [111,124,125] and human studies of PUJO show altered endothelial NO synthase (eNOS) and inducible NO synthase (iNOS) expression/activity together with reduced NO production. Lower eNOS expression/activity is associated with worse creatinine clearance, reduced differential renal function [90,126] and increased fibrosis [90,126], oxidant injury and apoptosis [67, 79].

Cellular apoptosis
Apoptosis affects podocytes and endothelial and epithelial cells within the kidney, leading to loss of glomeruli, peritubular capillaries and tubules [11].  [11].

Tubular function impairment
Ureteric obstruction leads to reduced renal expression of the V2 (vasopressin) receptor [130], renal sodium and urea transporters [131][132][133] and aquaporins [134][135][136]. Aquaporins are a family of transmembrane proteins normally expressed by mammalian kidney [137] and urothelium [138,139] that mediate water movement across the cell membrane along an osmotic gradient [140]. Reduced renal aquaporin expression  Tables 3 and 4 in experimental UUO is noted within 24 h of complete obstruction [134]. Similarly, renal aquaporins are downregulated in children undergoing pyeloplasty, and in both human and animal models this reduction is associated with polyuria and reduced concentrating ability following relief of obstruction [141][142][143].

Genetic mechanistic clues in PUJO
Phenotypes similar to PUJO have been noted in numerous transgenic mouse models. Many genes involved in ureteric smooth muscle proliferation and differentiation are implicated, supporting a primary myogenic aetiology. Importantly, one of these genes has been implicated in human disease ( Table 2).
Mutations in TBX18, the gene coding for T Box protein 18, have been reported in association with congenital anomalies of the kidney and urinary tract (CAKUT). In particular, a heterozygous TBX18 truncating mutation (c.1010delG) showing autosomal dominant inheritance has been described across four generations of a family with CAKUT, and predominantly PUJO [56]. The transcription factor TBX18 is necessary for normal smooth muscle cell proliferation, differentiation and localisation around the developing urothelial stalk [24]. TBX18 also directs epithelial proliferation and when absent leads to an abnormally short ureteric bud [28].
In the majority of patients, however, PUJO is a polygenic disorder without an obviously inherited genetic component [11].

Potential therapeutic molecular targets in PUJO
Human and animal studies have highlighted a number of potential therapeutic targets that could be manipulated to alleviate the nephropathy sustained secondary to PUJO. Several drugs targeting these pathways have been assessed in rodent UUO models as described below, however, to our knowledge none of these therapies have been trialed in childhood human PUJO.

Angiotensin-converting enzyme and AT1 receptor inhibitors
In adult rodent UUO models angiotensin-converting enzyme (ACE) inhibitors and AT1 receptor inhibitors given prophylactically (for the duration of obstruction) are beneficial in alleviating nephropathy. Specifically, they reduce TGF-β [121,144] and TNF-α [106] expression, as well as macrophage infiltration and tubulointerstitial fibrosis [84, 105,145]. Additionally, AT1 receptor inhibitors improve tubular function by improving RBF and GFR and attenuating the reduction in sodium transporter and aquaporin 2 (AQP2) expression, thus reducing polyuria and natriuresis [92,112].
ACE inhibitors reduce both AT1 and AT2 receptor stimulation [146] and indirectly increase NO levels via bradykinin generation [84]. This may explain why they confer additional benefits, particularly anti-inflammatory, compared to AT1 receptor inhibitors [97]. Unfortunately, inhibition of angiotensin during either the period of nephrogenesis (first 10 days after UUO) or renal maturation (second 10 days after UUO) in neonatal partial UUO exacerbates renal injury in both the obstructed and contralateral kidney [147,148]. Such studies highlight the importance of these pathways in normal kidney development and maturation.
However, it is important to remember that ACE inhibitors and AT1 receptor inhibitors are frequently used in children with chronic kidney disease, in whom they significantly reduce proteinuria [149] despite not significantly alleviating the natural decline in excretory function [150,151]. They are largely well tolerated, with no apparent effect on growth and development and a low incidence of side effects such as hyperkalaemia, hypotension and renal injury [149].
Statins are commonly used and usually well tolerated in adults. Side effects of treatment include hepatic dysfunction, diabetes mellitus, benign proteinuria, peripheral neuropathy, myalgia and rhabdomyolysis [158]. A 10-year follow-up study of children (≥8 years) treated with statins for familial hypercholesterolaemia demonstrated that few discontinue therapy due to side effects and that there were no serious adverse reactions [159]. In that same study, growth, puberty and educational parameters were also unaffected compared to controls [159].

TGF-β modulation
Prophylactic TGF-β receptor inhibition is renoprotective in adult rodent UUO models, reducing apoptosis, macrophage infiltration, fibrosis, proximal tubular atrophy and atubular glomeruli formation [117,160]. Similarly, anti-TGF-β antibody treatment increases NOS expression while reducing apoptosis and fibrosis [110]. Conversely, prophylactic TGF-β receptor inhibition in neonatal mouse UUO causes widespread renal necrosis, exacerbating the injury in the obstructed kidney and highlighting the differing responses to signalling cascades during renal development [117].
Anti-TGF-β antibody treatment (GC1008) has been trialled in human oncological disease and was generally well tolerated. However, side effects included gingivitis, fatigue and skin rashes, including keratoacanthoma and squamous cell carcinoma development (melanoma patients only). GC1008 treatment has not progressed beyond phase II clinical trials as drug development was discontinued by the manufacturer [161].

COX-2 inhibition
In adult rodent bilateral ureteric obstruction COX-2 inhibition alleviates AQP2 and sodium transporter downregulation and improves post-obstructive polyuria, which would appear to be beneficial [69]. Conversely, other studies have demonstrated that both genetic COX-2 knockout and prophylactic COX-2 inhibition in adult rodent UUO models increase tubular injury, apoptosis and fibrosis, thereby negating potential use in the clinical setting [70,162].
Chronic celecoxib (COX-2 inhibitor) use in children demonstrates a similar frequency of adverse events to nonselective non-steroidal anti-inflammatory drugs, which are most frequently gastrointestinal side effects [163].

Other potential therapeutic options
Other potential therapeutic pathways include those that are able to increase the vasoactive molecule NO, as this has been shown to reduce tubulointerstitial fibrosis in adult rodent UUO models [84]. Although both ACE inhibitors and statins increase NO bioavailability, this is an indirect effect at the expense of drug-related side effects.
Dietary nitrate supplementation is a novel therapeutic option which directly targets the NO pathway, increasing NO generation via nitrite production. Nitrite also has cytoprotective effects independent of NO by influencing mitochondrial function [164], and when administered during rodent ischaemia reperfusion studies reduces renal injury [165].
Despite former concerns associating nitrates with methaemoglobinaemia and carcinogenesis, the nitratenitrite-NO pathway is increasingly implicated in a protective role in humans [166]. Further investigation of dietary nitrate supplementation as a potential therapy in obstructive nephropathy is warranted.

Urinary biomarkers
Identifying early urinary biomarkers in PUJO may be beneficial for the diagnosis, management and prognosis of this condition. Such biomarkers would enable timely detection of children with 'damaging' hydronephrosis who require surgery to protect renal function, while avoiding surgery in those with 'safe' hydronephrosis.

Urinary biomarkers in animal studies
There is little data on urinary biomarkers from animal studies. Proteomics using a rat UUO model demonstrated increased urinary and renal levels of alpha-actinin-1 and moesin at 1 week which corresponded with histological evidence of tubular injury. Following 3 weeks of UUO urine and renal levels of vimentin, annexin A1 and clusterin were significantly elevated, corresponding with substantial renal interstitial fibrosis [167].

Urinary biomarkers in human studies
Many urinary cytokines, growth factors, chemokines, tubular enzymes and tubular transport proteins have been investigated Decreased 1 year post-operative [176] TGF-β Pyeloplasty CMP Increased pre-operative Se 82%/Sp 86% [182] Generally, the primary group measured is children undergoing pyeloplasty; these children are then compared to healthy controls and/or conservatively managed children with PUJO (CMP). The exception in the studies listed in the table is labelled PUJO*, which includes children with conservatively managed PUJO split into 'functional' (t1/2 of renogram < 0 min) and 'obstructed' (t1/2 of renogram > 20 min). In these studies voided urine from children undergoing pyeloplasty was only obtained 42 months post-operative a ALP, Alkaline phosphatase; Ca19-9, carbohydrate antigen 19-9; CyC, cystatin-C; HO-1, heme oxygenase-1; γGT, gamma-glutamyl transferase; IP-10, interferon-γ-inducible protein 10; KIM-1, kidney injury molecule-1; MIP-1α, macrophage inflammatory protein-1α; NAG, N-acetyl-beta-Dglucosaminidase; NGAL, neutrophil gelatinase-associated lipocali; OPN, osteopontinn, RANTES, regulated on activation normal T-cell expressed and secreted b Where applicable sensitivity (Se), specificity (Sp) and accuracy (Ac) of the test at best threshold value from receiver operating characteristic curve analysis is presented c To detect differential renal function (DRF) of <40% out of all hydronephrosis cases d To detect pyeloplasty cases out of all hydronephrosis cases e To detect pyeloplasty cases out of all cases in children undergoing pyeloplasty for PUJO. Studies with conservatively managed PUJO as a comparator are most useful to identify biomarkers able to aid selection of patients for surgery. Potential urinary biomarker proteins measured in bladder urine samples are presented in Table 5.
Finding a suitable biomarker test with high sensitivity, specificity and predictive value is challenging [88], not least because these markers are excreted in health as well as disease, show significant intra-and inter-patient variation and may be affected by patient age, the presence of urinary tract infection and other renal disorders [174,183].
A recent systematic review of urinary and serum biomarkers included 14 studies which reported data on 380 surgically managed PUJO patients, 174 conservatively managed patients and 213 controls [184]. This review reported a wide-range of sometimes conflicting results, and the authors were unable to draw any firm conclusions, attributing this to differences in study design with heterogeneous age groups, various or absent control groups and often short durations of follow-up [184].
More successfully, proteomics of neonatal urine has identified a panel of 51 peptides which distinguish obstruction severity. When implemented in a prospective blinded study it had an accuracy of 94% to predict future need for surgery in newborns with PUJO [185]. However, beyond 1 year of age the sensitivity and specificity of this proteome profile diminished significantly [186].
A single biomarker able to guide selection of patients for pyeloplasty has not yet been identified, indicating a panel of biomarkers may be necessary to achieve this.

Conclusions
Managing children with asymptomatic intrinsic PUJO is a significant challenge for clinicians. Animal and human studies have expanded our understanding of the molecular mechanisms involved in the aetiology of obstruction and in particular the progression of the renal insult. Upregulation of the RAAS and TGF-β expression are fundamental to renal injury, which is attenuated in animal models by therapeutic inhibition of these pathways. Much, however, remains to be learned in order to identify molecular markers and targets useful in the day-to-day diagnosis and management of this condition.