Hedgehog interacting protein (HHIP) represses airway remodeling and metabolic reprogramming in COPD-derived airway smooth muscle cells

Although HHIP locus has been consistently associated with the susceptibility to COPD including airway remodeling and emphysema in genome-wide association studies, the molecular mechanism underlying this genetic association remains incompletely understood. By utilizing Hhip+/- mice and primary human airway smooth muscle cells (ASMCs), here we aim to determine whether HHIP haploinsufficiency increases airway smooth muscle mass by reprogramming glucose metabolism, thus contributing to airway remodeling in COPD pathogenesis. The mRNA levels of HHIP were compared in normal and COPD-derived ASMCs. Mitochondrial oxygen consumption rate and lactate levels in the medium were measured in COPD-derived ASMCs with or without HHIP overexpression as readouts of glucose oxidative phosphorylation and aerobic glycolysis rates. The proliferation rate was measured in healthy and COPD-derived ASMCs treated with or without 2-DG. Smooth muscle mass around airways was measured by immunofluorescence staining for α-smooth muscle actin (α-SMA) in lung sections from Hhip+/- mice and their wild type littermates, Hhip+/+ mice. Airway remodeling was assessed in Hhip+/- and Hhip+/- mice exposed to 6 months of cigarette smoke. Our results show HHIP inhibited aerobic glycolysis and represses cell proliferation in COPD-derived ASMCs. Notably, knockdown of HHIP in normal ASMCs increased PKM2 activity. Importantly, Hhip+/- mice demonstrated increased airway remodeling and increased intensity of α-SMA staining around airways compared to Hhip+/+ mice. In conclusion, our findings suggest that HHIP represses aerobic glycolysis and ASMCs hyperplasia, which may contribute to the increased airway remodeling in Hhip+/- mice.

Immunofluorescence staining. Lung sections from Hhip +/+ and Hhip +/mice at the age of 8 months (N = 4-5 for each group) were stained with anti-α-SMA antibody (#M085129-2, Dako, Santa Clara, CA) at 1:400 dilution. Airways around 300-600 μm in diameter were chosen for quantification of α-SMA intensity surrounding airways by pixel intensity measurements using ImageJ software. The intensities (size of α-SMA positive staining area) were then normalized to the area of airways.
Trichrome staining. Paraffin-embedded lung sections from Hhip +/+ and Hhip +/mice at the age of 8 months with or without CS exposure for 6 months were used for Masson's trichrome staining. To quantify small airway remodeling in CS-exposed mice, ImageJ software (Version 1.8.0, NIH) was utilized as previously reported 30 to quantify the size of the deposition area of ECM proteins around the small airways (airways having a mean diameter between 300 and 699 μm were included). The positive areas for trichrome staining were then normalized to corresponding small airway area and statistically analyzed. The normal distribution of data was examined Scientific Reports | (2021) 11:9074 | https://doi.org/10.1038/s41598-021-88434-x www.nature.com/scientificreports/ using Shapiro-Wilk normality test. Since these data were normally distributed, one-way ANOVA for multiple comparisons were then used to compare the effects of genotype and treatment.

Cigarette smoke (CS) exposure in mice.
Female Hhip +/and littermate mice (approximately 10 weeks old) 20 were exposed to mixed main-stream and side-stream CS from 3R4F Kentucky Research cigarettes for 5 days/week in Teague TE 10z Chambers (Total Suspended Particulates approximately 100 to 200 mg/m 3 and CO levels around 6 ppm). As a control, mice were exposed to filtered air for the same duration. At the end of the 6-month exposure period, lung mechanics were then measured in mice exposed to CS or air.
Measurement of lung mechanics. Respiratory mechanics including respiratory system resistance and newtonian resistance in unchallenged Hhip +/+ (N = 14) and Hhip +/-(N = 19) mice were measured as previously described 21 . The reduced expression of hhip in Hhip +/mice has also been confirmed previously 20 .
Pyruvate kinase activity measurement. The pyruvate kinase activity was measured by the colorimetric method based on the manufacturer's instructions (Pyruvate Kinase Activity Assay Kit, Sigma). Briefly, 1X10 6 cells were homogenized with pyruvate kinase assay buffer followed by absorbance measurement at a wavelength of 570 nm. The results were calculated based on the standard curve generated using the same kit. Western blotting. Cells were washed with cold phosphate-buffered saline (PBS) twice and lysed in ice-cold RIPA buffer (Cell Signaling Technology, Danvers, MA) with added protease inhibitor cocktail (Bimake, Houston, TX), phosphatase inhibitor cocktail (Bimake, Houston, TX) and 1 mM PMSF (Sigma-Aldrich, Burlington, MA). Equal amounts of protein were separated by 4-15% gradient SDS-PAGE (Bio-Rad, Hercules, CA) and electrotransferred to polyvinylidene fluoride membrane. Membranes were blocked with TBST containing 5% (wt/vol) non-fat dry milk for 1 h and incubated overnight at 4 °C with primary antibodies diluted in TBST with 5% (wt/vol) non-fat dried milk and 0.1% (vol/vol) Tween-20. After incubation with appropriate horseradish peroxidase-conjugated anti-mouse or anti-rabbit IgG (GE Healthcare, Burlington, MA) secondary antibody for 1 h at room temperature followed by thorough washing in TBST, the membrane was developed for immunoreactive bands detection by G:BOX image developing system (Syngene Frederick, MD) with an enhanced chemiluminescent or supersignal west femto chemiluminescent substrates (Thermo Fisher Scientific, Grand Island, NY). Blot images in TIFF format were captured by G:Box image software and then ranged in Powerpoint software without any inappropriate contrast modification. Band densities were quantified by ImageJ (NIH) software with α-tubulin as the internal loading control for normalization.
Reverse transcription quantitative real-time PCR (RT-qPCR). Total cellular RNA was extracted with RNeasy (Qiagen, Germantown, MD) kit according to the manufacturer's instructions. cDNA was synthesized using high-capacity cDNA reverse transcription kit (Applied Biosystems, Grand Island, NY). RT-qPCR was performed using Taqman gene expression master mix and QuantStudio 12 K Flex (Thermo Fisher Scientific, Grand Island, NY). At least three independent experiments were performed, and each sample was assayed in triplicates to determine the mRNA levels measured by QuantStudio 12 K Flex software v1.2.2 with comparative Ct method with GAPDH as the reference gene and using the formula 2 -ΔΔCt .
Measurements of mitochondrial respiration. Extracellular Flux Analyzer (Seahorse Bioscience, Santa Clara, CA) was used to measure the oxygen consumption rate (OCR), an indicator of mitochondrial respiration, in primary human lung ASMCs in a 24-well plate. Briefly, cells were seeded directly into XF24 plates, and mitochondrial respiration was measured in Agilent Seahorse XF Base Medium (Agilent Technologies, Santa Clara, CA) supplemented with D-glucose (25 mM) and pyruvate (10 mM). ASMCs were seeded at a density of 3 × 10 4 cells/ well, and OCR was measured using the Seahorse XF Cell Mito Stress Test program. During the Seahorse assay, oligomycin (2 mM), carbonyl cyanide-4-(trifluoromethoxy) phenylhydrazone (2 mM) and antimycin/rotenone (1 mM of each) were sequentially added into each well for measurements of basal respiration, ATP production, maximum respiration, and proton leak (complex I driven), respectively. www.nature.com/scientificreports/ were attached to the well (around 4-6 h following seeding), one of six plates was collected after supernatant removal, and stored with attached ASMCs at − 80 °C (Day 0, which indicates the seeding day). The remaining five 96-well plates were then collected subsequentially in the following 7-8 days using the same method and stored at -80 °C. After collecting all six plates, DNA contents of each well from six 96-well plates were measured using the CyQUANT NF Cell Proliferation Assay Kit (#C35006, Invitrogen, Grand Island, NY). DNA content readings from day 1 to day 7 or 8 were then normalized to day 0 readings (the first plate collected) for statistical quantification to reflect the relative cell number fold changes as several days of culture. All primary human ASMCs from Lonza were cultured in SmGM-2 BulletKit medium (CC-3182, Lonza), which contains 1.0 g/L glucose, 5% FBS, and growth supplements including human epidermal growth factor, Insulin, human fibroblast growth factor-beta (FGF-2) as well as gentamicin/amphotericin-B. Cultures were incubated at 37 °C in a humidified 5% CO 2 /air atmosphere, and the medium was changed every other day.
Statistical analysis. All statistical analyses were performed using PRISM software (GraphPad Software, Inc. San Diego, CA). Unpaired Student's t-test were used for comparisons of two groups with normally distributed data, respectively. The normal distribution of data including α-SMA staining and trichrome staining in Hhip +/+ and Hhip +/mice with or without CS was confirmed using Shapiro-Wilk normality test. We also performed one-way ANOVA for multiple comparisons of means to compare the effect of genotype and treatment under each given condition. Means were considered significantly different if P < 0.05.

A metabolic shift to aerobic glycolysis in COPD-derived ASMCs is essential for COPD-derived ASMC hyper-proliferation.
To examine the metabolic reprogramming from oxidative phosphorylation (OXPHOS) to aerobic glycolysis in COPD-derived ASMCs, seahorse mitostress assay was performed, demonstrating decreased basal oxygen consumption rate (OCR) in COPD-derived ASMCs compared to normal ASMCs ( Fig. 1A and B). Furthermore, lactate levels, indicating aerobic glycolysis, were increased in COPDderived ASMCs compared to normal ASMCs (Fig. 1C). These results suggested the reprogramming of glucose metabolism from OXPHOS to aerobic glycolysis (Warburg effects) in COPD-derived ASMCs. To determine whether COPD-derived ASMCs rely on glycolysis to provide sufficient energy for their proliferation, we applied glycolysis inhibitor 2-Deoxy-D-glucose (2-DG), a competitive glucose surrogate, in both COPD-derived and healthy ASMCs. Interestingly, COPD-derived ASMCs demonstrated a more significant reduction of glycolysis upon 2-DG treatment compared to normal ASMCs (Fig. 1D). A previous study has identified the hyperproliferation of COPD-derived ASMCs in response to FBS 26 , our results embedded in Fig. 1E and F also support the hyperproliferation rate of COPD-derived ASMCs compared to normal ASMCs with the presence of FBS. In addition, COPD-derived ASMCs showed decreased cell proliferation rates when treated with 2-DG (0.5 mM), in contrast to normal ASMCs ( Fig. 1E and F). Furthermore, 2-DG treatment has minimal effects on OCR in COPD-derived ASMCs (Supplemental Fig. 1A and B), suggesting reduced cell proflieration upon 2-DG treatment likely result from reduced glycolysis (Fig. 1D). Our results indicate that glycolysis inhibition is more effective for decreasing the proliferation of COPD-derived ASMCs which have higher levels of active glycolysis compared to healthy ASMCs ( Fig. 1E and F).

HHIP attenuated metabolic reprogramming towards aerobic glycolysis in COPD-derived
ASMCs. We previously found decreased expression of HHIP in human COPD lungs (18) in COPD-derived ASMCs compared and healthy ASMCs ( Fig. 2A), significantly reduced expression of HHIP mRNA levels was noted in COPD-derived ASMCs compared to normal ASMCs from female subjects, but not in male subjects). Current HHIP antibodies were unable to detect endogenous HHIP, future experiments are warranted to measure HHIP in normal and COPD-derived ASMCs at protein levels. Given important roles of glycolysis in COPD ASMCs, we examined possible regulation of glycolysis by HHIP. Indeed, overexpression of HHIP in COPDderived ASMCs led to modest but significant reductions of lactate levels (Fig. 2B), decreased cell proliferation (Fig. 2C), without significant changes of mitochondrial OCR (Fig. 2D), suggesting HHIP overexpression altered aerobic glycolysis and cell proliferation rate without affecting mitochondrial OXPHOS in AMSCs. The overexpression efficiency was confirmed by RT-qPCR (Fig. 2E) and Western blot (Fig. 2F). These results suggest that HHIP may inhibit COPD-derived ASMC hyperproliferation and attenuated the elevated aerobic glycolysis detected in these cells.
HHIP inhibits the activity of PKM2 that promotes aerobic glycolysis in COPD-derived ASMCs. To identify potential mechanisms by which HHIP represses the metabolic reprogramming in COPD-derived ASMCs, we measured the activity of PKM2, a critical rate-limiting enzyme for glycolysis in muscle cells. The knockdown of HHIP significantly increased the pyruvate kinase activity in normal ASMCs, which were abolished by transfection of si-PKM2 (Fig. 3A). However, the PKM2 protein levels were not altered by either HHIP overexpression or knockdown in healthy or COPD-derived ASMCs, respectively (Fig. 3B). To determine the role of PKM2 in regulating glycolysis of COPD-derived ASMC, si-RNA mediated PKM2 knockdown with or without HHIP knockdown was applied to COPD-derived ASMCs followed by measurements of  (Fig. 3C) and increased basal OCR (Fig. 3D). The knockdown efficiency of PKM2 was confirmed by Western blot (Fig. 3E). The knockdown efficiency of HHIP was confirmed by Taqman RT-qPCR (Fig. 3F). In summary, HHIP may repress airway remodeling by inhibiting glycolytic reprogramming-induced ASMC hyperproliferation, which could be partly through inhibiting PKM2 activity, thus preventing airway remodeling in COPD pathogenesis.
Hhip is expressed in murine lung ASMCs. A recently published single-cell RNA-Seq data in murine lungs has indicated the expression of HHIP is enriched in ASMCs 31 . Consistent with this finding, our results demonstrate positive β-gal signals, indicative of Hhip expression in Hhip +/mice, are present in cells with positive α-smooth muscle actin staining (α-SMA) in the proximal airway, suggesting that Hhip is expressed in ASMCs or myofibroblasts. Interestingly, the location of HHIP is specifically in airway smooth muscle cells but not vascular smooth muscle cells (Fig. 4), suggesting exclusive localization of Hhip in ASMCs.  www.nature.com/scientificreports/ and B). Additionally, compared to Hhip +/+ mice, Hhip +/− mice showed significantly increased respiratory system resistance (Max Rrs) and newtonian resistance (Max Rn) ( Fig. 5C and D) which are indicatives of greater total and central airway resistance respectively, consistent with the greater airway smooth muscle mass in Hhip +/mice. Moreover, collagen thickness around airways, as indicated by Trichrome staining 32 , were increased in CS-exposed Hhip +/mice compared to CS-exposed Hhip +/+ mice ( Fig. 5E and F), suggesting haploinsufficiency of Hhip led to increased collagen deposition around airways in mice exposed chronically to CS. These results provide additional evidence supporting an inhibitory effect of Hhip on airway remodeling.

Discussion
As the third leading cause of global death, COPD susceptibility is influenced by both environmental factors and genetic determinants. Airway remodeling, including an increased amount of ASMCs around airways, is inversely correlated with lung function 4 and positively related to COPD severity 5 . Although GWAS has identified that www.nature.com/scientificreports/ chromosome 4q31 locus is significantly associated with COPD susceptibility and lung function [7][8][9] , the role of HHIP in airway remodeling is unknown 20,21 . In murine models, we demonstrate that Hhip +/mice spontaneously develop airway remodeling at 8 months of age as indicated by increased α-SMA intensity around airways and increased airway resistance compared to WT littermates. In primary human ASMCs, we found that metabolic reprogramming from OXPHOS to aerobic glycolysis contributes to cell proliferation in COPD-derived ASMCs [25][26][27] . Our previously published paper has shown that the Hhip mRNA levels in murine lungs were not altered by CS exposure for six months 20 . Here the expression of HHIP showed a trend toward reduction in female COPD-derived ASMCs. However, the interpretation of this finding requires caution due to the limited subjects used in our study. Importantly, HHIP was shown to attenuate aerobic glycolysis and cell hyperproliferation in COPD-derived ASMCs. Airway remodeling in COPD, including the structural changes of airways, could be induced by increased levels of growth factors and cytokines 33,34 . Histological examination of airways and high-resolution CT are usually used to assess and quantify the extent of airway remodeling, which contributes to irreversible airway narrowing in COPD pathogenesis 35 . Growth factors such as TGF-β initiate airway remodeling by stimulating smooth Previously, increased airway smooth muscle mass has been shown in COPD patients at Gold stage III and IV compared to healthy individuals 5 . Importantly, studies suggest that increased muscle mass likely contributes to increased lung resistance observed in COPD 37 . Since airway remodeling in COPD leads to lung function reduction that is not fully reversed by corticosteroids and β-agonists 38 , the development of novel therapeutic interventions based on a better mechanistic understanding of airway remodeling is urgently needed. Recent reports on the genetic determinant of airway remodeling on COPD illuminate novel insights into the development of airway remodeling. Genetic variants nearby HHIP and MECR (mitochondrial trans-2-enoyl-CoA reductase) were previously implicated in small airway obstruction even in non-smokers 12 . Interestingly, MECR is a crucial gene involved in the last step of mitochondrial fatty acid synthesis. This finding further suggests that metabolic changes in airways may genetically determine airway obstruction. Cellular metabolic homeostasis is essential for maintaining cell function, including proliferation and differentiation 39 . Glucose present in normal cells usually undergoes oxidative phosphorylation to generate ATP under normoxic conditions. However, hyperproliferative cells, such as cancer cells, rely on aerobic glycolysis rather than glucose OXPHOS to maintain a high cellular proliferation rate despite sufficient oxygen supply. Such metabolic reprogramming is known as the Warburg effect 40 . Though aerobic glycolysis is relatively less efficient in producing adenosine 5′-triphosphate (ATP), it facilitates the uptake and incorporation of nutrients into the biomass needed for new cells 41 . Recent studies reported imbalanced cellular bioenergetics in COPD patients. For example, disrupted lipid metabolism was detected in serum and sputum samples from human COPD patients 42,43 , as well as in CS-induced murine emphysema models 44 consistent with dysregulated metabolism occurring during COPD development 42 . In murine models, CS increased fatty acid oxidation in airway epithelial cells 45 . In healthy human subjects, long-term smoking led to altered metabolites in airway basal cells, as assessed by global metabolic profiling 46 . Recently, a Warburg-like effect was reported in COPD-derived ASMCs and was speculated to contribute to the airway remodeling occurring in COPD 12 . In particular, Teperino et al. have reported that activation of the hedgehog pathway promoted Warburg-like metabolism in muscle 47 , highlighting the essential role of hedgehog pathway in regulating glycolytic metabolism. However, inhibitory effects of Hhip on the Hedgehog pathway relies on the competitive binding with three types of Hedgehog ligands present in the culture condition. Therefore, whether Hedgehog pathway is involved in the regulation of Hhip in the glycolysis in airway smooth muscle cells needs further investigation. Herein, using murine genetic models targeting the most replicated COPD GWAS gene-HHIP, we demonstrated increased airway remodeling in Hhip +/mice associated with increases in total and central airway resistance increase. Importantly, the metabolic shift of the Warburg-like effect from oxidative phosphorylation to aerobic glycolysis in COPD-derived ASMCs is ameliorated by HHIP, providing the first evidence for contributions of HHIP to the repression of metabolic reprogramming towards aerobic glycolysis. Notably, although the Warburg effect was observed in COPD-derived ASMCs as indicated by increased lactate levels and decreased OCR compared to normal ASMCs, transient overexpression of HHIP attenuates increased lactate levels modestly without influencing OCR in COPD-derived ASMCs, indicating incomplete rescue of metabolic reprogramming in COPD-derived ASMCs by resuming HHIP levels, possibly due to insufficient genetic changes over a short period against decades of accumulated metabolic.
Interestingly, expression of HHIP showed a reduced trend in COPD-derived ASMCs in female subjects. However, a larger sample size of ASMCs from healthy and COPD subjects from both genders is needed in future studies to explore sex-genetic interaction in determining HHIP expression and/or lung function since the number of primary cells used for these in vitro experiments are limited. Besides, whether differential hormone levels in male and female patients play essential roles in people carrying susceptibility genetic loci during COPD pathogenesis needs further investigation. Of note, transient restoration of HHIP levels by overexpression in www.nature.com/scientificreports/ COPD-derived ASMCs is insufficient to reprogram these cells back to the normal-like metabolic condition of ASMCs, suggesting chronic metabolic shift in COPD-derived ASMCs may result from multiple genetic changes combined over time. However, we indeed found female Hhip +/mice showed increased airway obstruction and airway remodeling compared to age-matched WT mice, supporting the importance of HHIP to maintain airway homeostasis. Future evaluations on the stress-induced airway obstruction such as under methacholine challenge in Hhip +/mice may facilitate a better understanding of the regulation of HHIP on airway function. One possible mechanism by which HHIP modulates glycolysis in ASMCs is through the rate-limiting glycolytic enzyme-PKM2, which catalyzes the final step in glycolysis to produce pyruvate. PKM2 is selectively expressed in highly proliferative cells 48 . It has been shown that PKM2 activity is necessary for sustaining proliferative signaling in cancer cells 49 . By interacting with HIF-1α, PKM2 promotes the Warburg effect in www.nature.com/scientificreports/ macrophages 50,51 . In our current study, we found that HHIP reduces PKM2 activity in ASMCs, which is essential for metabolic reprogramming from OXPHOS towards aerobic glycolysis in COPD-derived ASMCs. Furthermore, we demonstrate that expression levels of PKM2 was not regulated by Hhip. Future studies are warranted to determine how HHIP regulates pyruvate kinase activity, possibly by influencing PKM2 phosphorylation, or interacting with PKM2, or affecting the conformational changes of PKM2, i.e., the transition between tetramer and dimer related to its activity in the lung cancer cell lines. Furthermore, whether PKM2 contributes to CSinduced airway remodeling in Hhip +/mice requires additional investigations on compound genetic deficient mice in the CS exposure model. Other than PKM2, other molecular mediators responsible for the effect of HHIP on metabolic reprogramming in ASMCs, such as p-AKT 52 , p-AMPK 47 , mTOR, and HIF-1α 53 require additional studies. In the future, to comprehensively evaluate the effect of Hhip in modulating airway resistance, provocation/induced bronchoconstriction experiments, such as methacholine responsiveness assessment, are needed in Hhip +/and/or Hhip conditional knockout mice.
In summary, we linked the metabolic reprogramming to the function of genetic determinants in remodeling airways during COPD pathogenesis. Though further direct evidence in vivo is still needed, our current data from mixed models including primary human normal and COPD-derived ASMC cultures as well as Hhip +/mice suggested HHIP protected the lung from airway remodeling possibly by repressing ASMC hyperproliferation associated with metabolic reprogramming towards aerobic glycolysis. Our results highlight a novel activity for HHIP in airway remodeling and provide important insights into the molecular mechanisms by which HHIP may protect lung from airway remodeling and eventually lung function reduction during COPD pathogenesis, which may ultimately shed light on novel potential therapeutic approaches to limit airway remodeling in COPD pathogenesis.