Hedgehog signaling pathway gene variant influences bronchopulmonary dysplasia in extremely low birth weight infants

Shaili Amatya· Sharina Rajbhandari· Sandeep Pradhan· Van Trinh· Umesh Paudel· Lance A.Parton

Abstract Background Genome wide association study identified hedgehog interacting protein gene ( HHIP) variants with chronic obstructive pulmonary disease and asthma.Loss of HHIP, a key regulator of the hedgehog signaling pathway, leads to impaired lung morphogenesis and lethality in animal models, through unimpeded sonic hedgehog expression blocking mesenchymal-expressed fibroblast growth factor 10 (FGF10).Since bronchopulmonary dysplasia (BPD) is also associated with altered lung development and worsens with stimuli including mechanical ventilation, reactive oxygen species, and infl ammation, HHIPand FGF10may be candidate genes.Methods This was an observational, cohort study including extremely low birth weight infants that who developed BPD and those who did not.DNA was isolated from buccal swabs and subjected to allelic discrimination, using specific HHIPand FGF10probes.Protein levels were measured in tracheal aspirates.Student's ttest, Chi-square, Ztest and logistic regression were used.Results Demographic characteristics did not differ except that birth weight (715 ± 153 vs.835 ± 132 g) and gestational age (25 vs.26 weeks) were less in babies with BPD.HHIPvariant rs13147758 (GG genotype) was found to be independently protective for BPD (odds ratio 0.35, 95% confidence interval 0.15-0.82, P= - 0.02).Early airway HHIP protein levels were increased in infants with BPD compared to those without [median (interquartile range) 130.6 (55.6-297.0) and 41.2 (22.1-145.6) pg/mL, respectively; P= 0.05].The FGF10single nucleotide polymorphisms were not associated with BPD.Conclusion HHIP, as a regulator of lung bud formation, affects BPD susceptibility, and may be valuable in understanding the specific mechanisms for this disease as well as for identifying therapeutic targets in the era of personalized medicine.

Keywords Gene · Infants · Lung · Preterm · Single nucleotide polymorphism

Introduction

Bronchopulmonary dysplasia (BPD) is one of the leading causes of morbidity and mortality in extremely low birth weight (ELBW) infants.Historically, BPD was thought to result from prematurity and exacerbated by invasive treatment such as mechanical ventilation [1, 2].The new spectrum BPD is characterized by alveolar simplification, pulmonary vascular pruning, and altered extracellular matrices; and, is infl uenced by antenatal as well as postnatal factors [2].Interactions among multiple signaling pathways involving various pulmonary cells types including epithelial, endothelial, smooth muscle, fibroblasts, and immune cells, are implicated in various stages of lung development as well as lung repair [3].Genetic factors affecting these multiple signaling molecules may also alter susceptibility to BPD in the developing lung.Bhandari et al.reported that 53% of the variability in the incidence of BPD was attributable to genetic differences based on a multicenter retrospective study of 450 preterm twin pairs [1].Another study of preterm infants by Lavoie et al.showed 79-82% heritability for BPD from genetic variations [4].As a chronic lung disease, BPD appears to have polygenic associations rather than just a single gene polymorphism, similar to the adult counterparts of chronic obstructive pulmonary disease and asthma [5].

There are numerous candidate gene studies that have identified genes and pathways involved in lung development, infl ammation, tissue injury and repair that are associated with BPD [6].However, there is a persistent knowledge gap regarding important gene pathways involved in distant lung development contributing toin BPD pathogenesis.

In the developing lung, the different signaling pathways that affect lung morphogenesis likely play important roles in the susceptibility of preterm infants to BPD.Among them, the hedgehog signaling pathway is important during lung bud formation [7, 8].Hedgehog interacting protein (HHIP) is a hedgehog signaling inhibitor of the sonic hedgehog (SHH) ligand.The SHH signaling in the distal lung bud mesenchyme inhibits fibroblast growth factor 10 (FGF10) expression.FGF10 expressed in the distal mesenchyme binds to the receptor, fibroblast growth factor receptor 2b (FGFR2b) to activate signaling and induce lung bud branching.Thus, at distal tips, HHIP levels reduce SHH signaling and induce FGF10 for branching morphogenesis [9, 10].These signaling pathways of SHH and FGF10 are essential during lung generation and interact through WNT (wingless-related integrated site) pathways to promote FGF10 expression and alveolar repair [3].HHIPtargets are also involved in extracellular matrix and cellular proliferation [11].HHIP may also play a role in angiogenesis [12].HHIPis localized to chromosome 4q31 and is highly expressed in lung tissue [13- 15].Therefore,HHIPandFGF10are potential candidate genes for BPD, where normal growth of the immature, developing lung is interrupted, and replaced with aberrant repair, resulting in pulmonary tissue remodeling.HHIPis well studied in genome wide association study (GWAS) associated with respiratory function as well as chronic obstructive pulmonary disease, asthma and lung cancer in adults [16- 18].

We hypothesized that given the role ofHHIPandFGF10in hedgehog signaling pathways and their importance in lung development and during lung injury and alveolar repair, their gene variants modify BPD susceptibility in preterm infants.We chose thirteen candidate single nucleotide polymorphisms (SNPs) ofHHIPgene and five SNPs ofFGF10.These SNPs were chosen as they have been associated with lung function and disease in adults [10, 13, 16- 22].

Methods

Study population

This is an ongoing longitudinal observational cohort study of ELBW infants born with birth weights equal to or less than 1000 g, who were admitted to the regional neonatal intensive care unit at Maria Fareri Children's Hospital at Westchester Medical Center, Valhalla, New York between 2012 and present.Informed parental consent was obtained from mother or father if the parents were married.Demographic data and samples such as buccal swabs and tracheal aspirates were collected during the infants' hospital stays.The study patients were divided into two cohorts: one group that developed BPD and another group that did not develop BPD (No BPD).The diagnosis of BPD was determined by oxygen dependence at 36 weeks corrected gestational age; which included moderate and severe BPD as per the National Institutes of Health BPD consensus [23].This current study BPD definition uses supplemental oxygen rather than the mode of respiratory support as suggested by evidence-based approach by Jensen et al.[24].The study was approved by the Institutional Review Board at New York Medical College and Westchester Medical Center.

DNA extraction

After informed consent, buccal swabs were collected from the infants during their hospital stays.DNA extractions from the buccal swabs were performed using the QIAamp DNA extraction mini kit (Qiagen, Germany).

Genotyping

CandidateFGF10andHHIPSNP genotyping was done using specific TaqMan probes and advanced genotyping master mix (Thermofisher, USA).Real-time polymerase chain reaction was performed on the Bio-Rad CFX96 (Bio-Rad, USA).

Tracheal aspirates analysis

Tracheal aspirates were collected from a sample of the study ELBW infants from birth until 7 days of life.The tracheal aspirates were centrifuged (× 500gat 4 ℃ for 10 minutes); the resulting supernatant × 10,000 g at 4 ℃ for 10 minutes), and the final supernatants were collected and stored at - 80 ℃ until batch analysis.Specific protein quantifications for FGF10 and HHIP were done by sandwich enzyme-linked immunosorbent assays (FGF10-human fibroblast growth factor 10 ELISA kit-MBS451995; HHIP-human HHIP/HIP ELISA kit-LS-F26976-LSBio; LifeSpan BioSciences, Inc., USA).

Statistical analysis

Correlations of gene polymor phisms between the two cohorts were examined by Chi-square.Demographic data of continuous variables were analyzed by Student'sttest or Mann-Whitney test, with Bonferroni corrections for multiple comparisons; and, categorical variables by Chi-square or fisher exact tests.Allele frequencies were tested by Z tests.Multiple logistic regression analyses were done to test for independent association of the gene polymorphism in patients who developed BPD.Airway protein comparative analysis was performed using Mann-Whitney rank sum test.The candidate SNP genotype frequency was tested for Hardy Weinberg equilibrium.APvalue ＜ 0.05 was determined to be significant.Statistical analysis was performed using Sigma plot version 13 and SPSS version 26.

Sample size estimation

The sample size was calculated as total sample needed 246, based on the minor allele genotype frequency for the significant SNP rs13144758 in the BPD cohort, for alpha 0.05 and beta 0.8.

Linkage disequilibrium

The linkage disequilibrium among the SNPs of theHHIPandFGF10were analyzed using JMP Genomics 9.0 (SAS software v9).

Results

Demographics

Demographic and maternal characteristics of the infants who developed BPD (n= 109) and who did not develop BPD (n= 176) are shown in Table 1.Mean birth weights with standard deviations and median gestational ages with interquartile ranges (IQR) of the BPD group were significantly lower than those with No BPD [706 ± 148, 821 ± 127 g; and 25 (24, 26), 26 (25, 27) weeks, respectively].In addition, more in the BPD cohort were male, received postnatal steroids, had patent ductus arteriosus (PDA), periventricular leukomalacia (PVL) and retinopathy of prematurity (ROP), when compared to infants who did not develop BPD.

Table 1 Demographic and clinical characteristics for cohort with/without BPD

HHIPgenotyping

Of the 13 SNPs tested, rs13147758 was found to have a significantly different genotype distribution between the BPD and No BPD groups (P= 0.01).The significant SNP frequency was in Hardy Weinberg equilibrium (P＞ 0.05).This SNP GG genotype was found to be independently associated with a decreased risk for BPD [odds ratio (OR) 0.35; 95% confidence interval (CI) 0.15-0.82,P= - 0.02], after multiple logistic regression analysis correcting for birth weight, male gender, and PDA (Table 2).The analysis was not corrected for other factors that were known to be associated with BPD patients, e.g., use of postnatal steroids, PVL and ROP.The other SNPs tested were: rs1489759, rs1355601, rs923783, rs1542725, rs949567, rs923782, rs1828591, rs1512288, rs1032295, rs13141641, rs1980057 and rs12504628-and these had no significant associations (data not shown).The genotype distributions for the statistically significantHHIPSNP rs13147758 were not different among the racial groups (data not shown).

Table 2 Genotype distribution for HHIPSNP rs13147758 for infants with and without BPD

FGF10genotyping

There were no statistical differences in the genotype distributions among any of the fiveFGF10SNPs (rs2973644, rs900379, rs980510, rs1011814, rs10512833) tested for the groups that had BPD compared to the group that did not have BPD (data not shown).

Airway HHIP and FGF10 protein expression

Airway HHIP protein levels were increased in infants who developed BPD (n= 21) compared to those who did not (n= 10) [median (IQR) 130.6 (55.6, 297.0) and 41.2 (22.1, 145.6) pg/mL, respectively;P= 0.05] (Fig.1).There were no significant differences in airway FGF10 levels between the BPD (n= 15) and No BPD groups (n= 8) [median (IQR) 156.2 (9.5, 348.9) and 84.0 (30.6, 143.6) pg/mL, respectively;P= 0.56] (data not shown).

Fig.1 HHIP airway expression is increased in BPD.Box and whisker plot of HHIP airway expression (pg/mL) in BPD ( n= 21) compared to No BPD ( n= 10) patients from tracheal aspirate collections taken within 7 days of life demonstrates an increased HHIP expression in those with BPD ( P= 0.05).HHIPhedgehog interactive protein, BPDbronchopulmonary dysplasia

Linkage disequilibrium

The linkage disequilibrium (LD) among theHHIPandFGF10SNPs tested for BPD was analyzed.Among theHHIPSNPs, the statistically identified variants rs13417758 and rs1512288 did not show strong LD.The rs12504628 and rs1980057 had square of the correlation coefficient (r2 ) 0.63 or D prime (D') 0.89.TheHHIPSNPs rs1980057 and rs1542725 have a D' of 0.86 with anr2 0.24.These SNPs are not in strong LD, likely due to low allele frequencies.Similarly,HHIPSNPs rs1542725 and rs13141641 are also not in strong LD as D' is 0.87 butr2 is 0.2, which suggests a weak LD.These variants are about 28 kb and about 46 kbdownstream from rs13147758, but still upstream of the gene.Thus, any functionality of these variants is difficult to determine from their locations.The LD plot of the 13HHIPSNPs located on chromosome 4 is shown (Fig.2).Among theFGF10SNPs, there were no significant LDs noted (data not shown).

Discussion

In this candidate SNP study, we studied cohorts of ELBW infants with and without BPD.As expected, the infants with BPD had lower gestational ages and birth weights.As hypothesized, we found aHHIPgene polymorphism infl uencing the BPD susceptibility in ELBW infants.We report that theHHIPSNP rs13147758, GG (minor allele homozygous) genotype was associated with decreased susceptibility to BPD with an odds ratio of 0.3 (95% CI 0.12-0.80), after correcting for birth weight, sex, and PDA.

The SNP rs13147758 lies 107 kb 5′ of the start site ofHHIPand it is located on the intron of chromosome 4q31.This SNP has been associated with adult lung diseases and pulmonary functions; however, it is unclear what regulatory effects explain these associations [25].The role of HHIP in lung morphogenesis and subsequently lung injury and repair make this a candidate gene for BPD pathogenesis.

In adults,HHIPhas been implicated in chronic obstructive pulmonary disease (COPD) and lung cancer by GWAS [17, 18].The homozygote G variant of rs13147758 was associated with better lung function with higher forced expiratory volume in 1 second (FEV1) level when compared with wild type [19].It was also found to be associated with less lung function decline in COPD patients, higher FEV 1 /forced vital capacity ratios, and decreased airfl ow obstruction in COPD patients [20].However, the precise mechanism for its contribution to adult pulmonary disease pathogenesis remains unclear.

Hedgehog interacting protein is involved in lung development as an inhibitor of hedgehog signaling pathway [8].Abnormal lung development may predispose to COPD by impairment of reserve among smokers [5, 19].Cross-talk between HHIP and WNT pathway genes may affect lung remodeling and injury repair processes in chronic lung disease [3].Although the pathogenesis of BPD is different from COPD,HHIPplays a major role in epithelial-mesenchymal interactions during development and would be an important candidate gene of lung development and remodeling in BPD.To our knowledge, the association between a genetic variant ofHHIPand BPD has not been published.We speculate that thisHHIPgene G variant (minor allele) of rs13147758 may preserve the developmental lung processes in preterm infants.

Fig.2 HHIPlinkage disequilibrium plot.The 13 HHIPSNPs located on chromosome 4 analyzed in this study are shown in this LD plot, with the extent of LD increasing according to the color order of blue ＜ grey ＜ red.HHIPhedgehog interactive protein, SNPsingle nucleotide polymorphism, LDlinkage disequilibrium

Zhou et al.identifiedHHIPSNP rs13147758 in the chromosome 4q31 locus, also deemed as the COPD susceptibility locus which is close toHHIPgene and likely to affect the transcription of theHHIPgene [10].The linkage disequilibrium is well documented with this particular SNP and the others in the area [10].In our study, we tested other nearby SNPs, but did not find any other significant associations.Regardless, due to the strong linkage disequilibrium of rs13147758 with others, the overall genetic effect could be shared with other SNPs in the loci as well.Xu et al.[21] found that the minor allele of rs13147758 was protective in the Chinese Han population (OR 0.546; 95% CI 0.332-0.897;P= 0.017).Moreover, animal models have implicatedHHIPhaploinsufficiency in emphysema that was associated with increased oxidative stress.This emphysema was prevented in HHIP ± mice by treatment with the antioxidant N-acetyl cysteine [26].This opens the possibility of potential targeted therapeutics in infants who are at risk of BPD.

In COPD,HHIPgene expression is decreased in lung tissue and associated with over activation of the hedgehog pathway that contributes to uncontrolled cellular proliferation [11].In our study, we report airway HHIP protein expression in ELBW infants in the first 7 days of life.We found that airway HHIP levels were significantly increased in infants who later develop BPD when compared to infants who do not develop BPD.We speculate that due to disruption of normal development in lung remodeling and subsequent BPD, there may be elevated HHIP protein expression.The expression and interactions of and between HHIP over longer periods past the first week of life is unknown.Although HHIP is expressed in the mesenchyme, the protein expression in respiratory fl uid such as sputum of adult patients with COPD has been measured [25].We speculate that as HHIP is involved in lung tissue injury and alveolar repair and may be expressed in respiratory fl uid at the time of injury.

We also investigated downstream in the hedgehog signaling pathway during lung morphogenesis by studyingFGF10gene variants that are associated with airway branching [27].We found no significant associations with theFGF10variants investigated.In past studies,FGF10knockout mice result in lung hypoplasia and lethality [28].However, Danopoulos et al.highlighted the functional difference in mouse and human as FGF10-induced epithelial branching in the mouse but not in human explants [29].In this current study, we found that airway FGF10 protein expression was not significantly different between the BPD and No BPD groups.Interestingly, Benjamin et al.noted that FGF10 was decreased in BPD.This was, however, in older infants with BPD who died; and FGF10 was measured in lung tissue [28].Other FGFs such as FGF 2, 3, 4, 7, 18 along with FGFR2 and 4-given their role in extracellular matrix remodeling have also been studied in association with neonatal lung disease and BPD, and only the SNP ofFGFR4was associated with BPD [30].This highlights the importance of different signaling pathway associations with BPD susceptibility.

The limitation to our study is that this was a longitudinal prospective cohort giving rise to evolutionary variation in clinical practice that may affect the phenotype.The definition of BPD based on oxygen supplementation was used for simplicity, however, the imprecise nature of BPD definition itself imparts limitation to the generalization to the study.Moreover, we had variable sample size due to differences in the SNP call rate and we did not account for any epigenetic effects.There may be variability in the level of protein expression in the tracheal aspirates as it is collected between birth and 7 days of life, with unknown response to exogenous surfactant.Moreover, the protein expression data were not linked with specific genotypes due to a small sample size from the study cohort.

In conclusion, this candidate SNPs study is the first study to demonstrate aHHIPgene variant that is associated with BPD in human preterm infants, independent of their prematurity.The airway expression of HHIP protein was also demonstrated to be increased in ELBW infants who progress to BPD.The tissue-specific transcriptional activity of theseHHIPgene variants and their interactions with other developmental gene pathways may be valuable in uncovering the specific mechanisms for these effects, as well as identifying therapeutic targets in this era of personalized medicine.

Author contributionsAS contributed to substantial contributions to conception and design, acquisition of data, or analysis and interpretation of data, data analysis, and drafting the article or revising it critically for important intellectual content.RS contributed to substantial contributions to conception and design, acquisition of data, or analysis and interpretation of data.PS contributed to data analysis.TV contributed to substantial contributions to conception and design, acquisition of data, or analysis and interpretation of data.PU contributed to drafting the article or revising it critically for important intellectual content.PLA contributed to substantial contributions to conception and design, acquisition of data, or analysis and interpretation of data, drafting the article or revising it critically for important intellectual content, and final approval of the version to be published.

FundingNo extramural financial support was received for this study, intramural support was obtained from Boston Children's Hospital Physicians.

Compliance with ethical standards

Ethical approvalThe study was approved by the Institutional Review Board of New York Medical College and the Clinical Research Institute of Westchester Medical Center.

Conflict of interestThe authors have no confl ict of interest to disclose.

Data availability statementThe datasets generated during and analyzed during the current study are not publicly available as they are being utilized for additional investigations, but are available from the corresponding author on reasonable request.