Substance P Promotes Diabetic Corneal Epithelial Wound Healing Through Molecular Mechanisms Mediated via the Neurokinin-1 Receptor

Among various pathological conditions by diabetes, ocular complications have been a leading cause of blindness in the world, including diabetic retinopathy, cataract, and various ocular surface disorders (1). The most recognized diabetic changes in the cornea, i.e., clinical diabetic keratopathy, include impaired corneal sensation, superficial punctate keratitis, and persistent corneal epithelium defects (2). The uncontrolled impairment of corneal wound healing increases the susceptibility of corneal ulcer, microbial keratitis, and even perforation. Diabetic corneal pathology always exhibits epithelial basement membrane abnormalities, reduced hemidesmosome density, and delayed wound healing (3,4). However, hyperglycemia also directly impairs the cellular metabolism and causes abnormal changes of corneal epithelium, such as Akt-, epidermal growth factor receptor (EGFR)–, and Sirt1-mediated cell responses to environmental challenges (5–7). Moreover, excess oxidative stress, resulting from enhanced accumulation of reactive oxygen species (ROS) and impaired antioxidant capabilities in response to hyperglycemia, has been postulated as an important pathological mechanism, whereas the reduction of ROS attenuated the progression of various diabetes complications (8–13), including in the cornea (5,14).

Substance P (SP), released predominantly by peripheral terminal, is an 11–amino acid neuropeptide that acts as a neurotransmitter mediating nociceptive transmission. It mainly functions through the interaction with neurokinin receptors, members of the tachykinin subfamily of G-protein–coupled receptors, among which the neurokinin-1 (NK-1) receptor shows a preferential affinity for SP. SP is currently known as a neuroimmunomodulatory regulator of the immune system (15). In the cornea, SP has been detected in the nerve fibers of naïve cornea (16,17) and in antigen-presenting cells in herpetic stromal keratitis (18). Moreover, SP causes the mobilization of bone marrow–derived stem cells to participate in corneal wound healing (19,20). Interestingly, recent evidences have shown that SP activates the EGFR, mitogen-activated protein kinases (MAPKs), extracellular signal–regulated kinases (ERKs), and phosphoinositide 3-kinase–Akt (PI3K-Akt) signaling pathways, as well as promotes the healing of inflamed colonic epithelium (21) and possesses antiapoptotic effects in colonocytes (22), dendritic cells (23), tenocytes (24), neutrophils (25), and bone marrow–derived stem cells (26).

The cornea is one of the most densely innervated tissues in the body, containing nerve fibers derived from the trigeminal ganglion. Corneal nerve fibers exert important trophic influences and contribute to the maintenance of corneal epithelium homeostasis, whereas the dysfunction of corneal innervation produces an impairment of corneal epithelial wound healing, known as neurotrophic keratitis, as caused by for instance herpetic viral infections, trigeminal nerve damage, or diabetes (27–29). Although tear fluid impairment is also involved (30), the neurotrophic deficits may play a major role in the pathogenesis of diabetic keratopathy, as the density of corneal nerve fibers and corneal sensation decreased in diabetic patients (31–33). However, although being the major sensory neurotransmitter and neuropeptide released from corneal nerve fibers, SP was shown not to be significantly decreased in diabetic cornea (34). Nevertheless, paradoxically, topical SP application promoted corneal epithelial wound healing in diabetic animals and humans when combined with insulin-like growth factor-1 (IGF-1) or epidermal growth factor (EGF) per group (29). The promoting mechanisms of SP on diabetic corneal epithelial wound healing remain elusive. In this study, we sought to demonstrate the protective mechanism of SP against diabetic corneal epithelial wound healing using streptozotocin-induced type 1 diabetic mice and high glucose–treated corneal epithelial cells.

Adult male C57BL/6 mice were purchased from the Beijing HFK Bioscience Co. Ltd. (Beijing, China). All animal experiments were carried out in accordance with The Association for Research in Vision and Ophthalmology Statement for the Use of Animals in Ophthalmic and Vision Research. The mice underwent induction of type 1 diabetes with an intraperitoneal injection of 50 mg/kg streptozotocin (Sigma-Aldrich, St. Louis, MO) in ice-cold citrate–citric acid buffer (pH 4.5) for 5 days, and control mice received equal amount of buffer. Blood glucose levels were monitored with a OneTouch Basic glucometer (LifeScan; Johnson & Johnson, Milpitas, CA). In the current study, diabetic mice were used after 12 weeks of final streptozotocin injection, at which point the HbA_1c values were 10.83 ± 0.74% (94.75 ± 7.93 mmol/mol), whereas normal mice had HbA_1c values of 4.25 ± 0.1% (22.5 ± 1 mmol/mol). For topical SP application, 5 μL SP (1 mmol/L; Calbiochem, San Diego, CA) in distilled water was dropped on the corneal surface by a 10 μL tip, six times daily per eye for 4 days (for the measurement of corneal sensitivity) or 4 days prescrape and 3 days postscrape (for the measurement of corneal epithelial wound healing). For NK-1 receptor inhibition, NK-1 receptor antagonist L-733,060 (6.6 μg in 5 μL distilled water; Sigma-Aldrich) was injected subconjunctivally 3 days before corneal sensitivity measurement and corneal epithelium scrape in normal mice, or 24 h before corneal epithelium scrape in diabetic mice according to our preliminary experiments. Control mice were treated with distilled water vehicle.

Corneal sensation was measured bilaterally by using a Cochet-Bonnet esthesiometer (Luneau Ophtalmologie, Chartres Cedex, France) in unanesthetized control, diabetic, SP-treated diabetic mice, and the NK-1 receptor antagonist–injected mice before the scrape of corneal epithelium. The testing began with the maximal length (6 cm) of nylon filament and shortened by 0.5 cm each time until the corneal touch threshold was found. The longest filament length resulting in a positive response was considered as the corneal sensitivity threshold, which was verified four times.

Normal, diabetic, and SP-treated diabetic mice were anesthetized by an intraperitoneal injection of xylazine and ketamine followed by topical application of 2% xylocaine. The entire corneal epithelium including limbal region (marked with 3 mm trephine) was scraped with Alger Brush II corneal rust ring remover (Alger Co., Lago Vista, TX) and subsequently applied with ofoxacin eye drops to avoid infection. Usually, one eye was wounded at a time in each animal. The defects of corneal epithelium were visualized at 24, 48, and 72 h by instilling 0.25% fluorescein sodium and photographed under slit lamp (BQ 900; Haag-Streit, Bern, Switzerland). The staining area was analyzed by using ImageJ software and calculated as the percentage of residual epithelial defect.

Mouse corneal epithelial cell line (TKE2) was provided by Dr. Tetsuya Kawakita of Keio University (Tokyo, Japan) (35). Human corneal tissues were handled according to the tenets of the Declaration of Helsinki. Primary human corneal epithelial cells (HCECs) were established from limbal explants of donor corneas according to a previous report (36). For the analysis of Akt, EGFR, and Sirt1 signaling activation, and the staining of ROS, glutathione (GSH), and mitochondria, both cells were starved overnight in bovine pituitary extract–free keratinocyte serum-free medium (Invitrogen, Carlsbad, CA) and subsequently incubated in 30 mmol/L glucose or mannose (osmotic control) for 3 days with or without 1 μmol/L SP.

For the proliferation analysis, the TKE2 cells were starved overnight in bovine pituitary extract–free keratinocyte serum-free medium, treated with high glucose for 3 days in the absence or presence of 1 μmol/L SP, and measured using 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) colorimetric assay. For the migration analysis, the cells were cultured in high glucose until confluence, subsequently wounded with a micropipette tip, and incubated with or without SP for 24 h. Digital images of wound closure were used for quantitative assessment of migration by using ImageJ software. Each assay was conducted in at least triplicate.

Eyeballs were snap frozen in Tissue-Tek optimum cutting temperature compound (Sakura Finetek, Tokyo, Japan). Frozen corneal sections were fixed in 4% paraformaldehyde for 15 min, permeabilized with 0.1% Triton X-100 for 10 min, and blocked with normal serum for 1 h at room temperature. The samples were stained with primary antibodies overnight at 4°C, washed, and incubated with fluorescein-conjugated secondary antibodies at 37°C for 1 h (antibody information as listed in Supplementary Table 1). All staining was observed under a confocal microscope or an Eclipse TE2000-U microscope (Nikon, Tokyo, Japan) after counterstaining with DAPI.

Total RNA was extracted from mouse corneal epithelium using NucleoSpin RNA kits (BD Biosciences, Palo Alto, CA). cDNAs were synthesized using the PrimeScript First Strand cDNA Synthesis Kit (TaKaRa, Dalian, China). Real-time PCR was carried out using SYBR Green reagents and the Applied Biosystems 7500 Real-Time PCR System (Applied Biosystems, Foster City, CA). The specific primers used are listed in Supplementary Table 2. The cycling conditions were 10 s at 95°C followed by 45 two-step cycles (15 s at 95°C and 1 min at 60°C). The quantification data were analyzed with the Sequence Detection System software (Applied Biosystems) using GAPDH as an internal control.

Total protein was extracted from the lysed samples of mouse corneal epithelium, cultured TKE2 cells, and HCECs in RIPA buffer. Samples (total protein concentration: 40 μg for mouse corneal epithelium and 45 μg for cultured cells) were run on 12% SDS-PAGE gels and then transferred to a polyvinylidene fluoride membrane (Millipore, Billerica, MA). The blots were blocked by nonfat dry milk for at least 1 h and incubated with primary antibodies (Supplementary Table 1) in Tris-buffered saline with Tween for 1 h at room temperature. The blots were washed three times and incubated with a horseradish peroxidase–conjugated secondary antibody (Amersham Biosciences, Piscataway, NJ). Finally, the blots were visualized via enzyme-linked chemiluminescence using the ECL kit (Chemicon, Temecula, CA).

For the observation of mitochondrial structure, superoxide generation, and membrane potential, the cells were preloaded with 100 nmol/L MitoTracker Green (Beyotime, Haimen, China) for 30 min, 5 μmol/L MitoSOX Red reagent (Beyotime), and 5 µg/mL 5,5′,6,6′-tetrachloro-1,1′,3,3′-tetraethyl-benzimidazole-carbocyanide iodine (JC-1; Beyotime), respectively, for 15 min at 37°C. The fluorescence was observed and captured using a Nikon confocal microscope.

For the observation of intracellular ROS and GSH staining, fresh corneal cryostat sections and cultured cells were loaded with 10 μmol/L fluorescence probe 2,7-dichlorodihydrofluorescein diacetate, acetyl ester (DCHF-DA; Molecular Probes, Eugene, OR) and 50 μmol/L monochlorobimane (Sigma-Aldrich), respectively, for 30 min at 37°C. The staining was observed and captured using a Nikon confocal microscope. For the measurement of ROS generation, 50,000 cells were harvested and incubated with 5 μmol/L DCHF-DA for 20 min at 37°C. For the measurement of GSH content, 50,000 cells were harvested and freeze-thawed with liquid nitrogen and a 37°C water bath three times. The supernatant was collected and mixed with the provided working buffer and NADPH. The total GSH content was quantified by comparison with known GSH standards according to the manufacturer’s instructions (Beyotime). The ROS and GSH fluorescence intensity was measured using a Multi-Mode Microplate Reader (SpectraMax M2; Molecular Devices, Menlo Park, CA).

Data in this study were representative of at least three different experiments and presented as the means ± SD. Statistical analysis was performed using SPSS 17.0 software (SPSS, Chicago, IL) and one-way ANOVA. Differences were considered statistically significant at P < 0.05.

To assess the effects of SP on diabetic corneal epithelial wound healing, entire corneal epithelium was scraped in age-matched normal mice and diabetic mice with or without topical SP application for 7 days. Punctate fluorescence staining showed that SP was detected in corneal epithelium after topical application, which suggests the applied SP penetrated the apical tight junction barrier in diabetic mice (Supplementary Fig. 1). The corneal epithelial healing rate exhibited a significant difference from 48 h after corneal epithelium scrape (Fig. 1A). The defect size of corneal epithelium in SP-treated diabetic mice (48 h: 28.78 ± 11.78%; 72 h: 4.44 ± 2.29%, n = 6) was significantly improved from that of diabetic mice (48 h: 63.59 ± 7.85%; 72 h: 22.73 ± 9.85%, n = 6) and reached the equal level of normal mice (48 h: 19.59 ± 5.67%; 72 h: 3.08 ± 2.17%, n = 6) (Fig. 1B). Moreover, the attenuated corneal sensitivity in diabetic mice was also restored by topical SP application, although still being lower than that of normal mice (n = 8 per group) (Fig. 1C). In addition, compared with diabetic mice, more inflammatory cell infiltration was found underneath the corneal epithelium margin of SP-treated diabetic mice at 48 h, whereas it was reduced at 72 h after corneal epithelium scrape (Supplementary Fig. 2). The results suggest that topical-applied SP penetrates into corneal epithelium and promotes corneal epithelial wound healing in diabetic mice, accompanied by recovery of corneal sensitivity, and an early inflammatory and resolution response.

To assess the effect of SP on corneal epithelial wound healing in vitro, mouse corneal epithelial cells were treated with high glucose with equal concentration of mannose as osmotic control. Subsequently, the confluent cells were wounded and treated with or without SP for another 24 h to analyze the migration rate. The results showed that high glucose treatment caused significant delay of corneal epithelial cell migration, whereas SP addition improved the migration capacity of high glucose–treated cells to the same level of normal cells (n = 3 per group) (Fig. 2A and B). In addition, SP promoted the proliferation rate of corneal epithelial cells that was impaired by high glucose treatment for 3 days (n = 3 per group) (Fig. 2C). The experiments were performed three times with similar results.

To elucidate the mechanism underlying the promotion of SP on corneal epithelial wound healing, we investigated the effects of SP on the activation of Akt, EGFR, and Sirt1 of diabetic corneal epithelium. Representative phosphorylated Akt (p-Akt), p-EGFR, and Sirt1 staining is shown in Fig. 3A. The expression levels of p-Akt, p-EGFR, and Sirt1 were significantly upregulated in diabetic corneal epithelium after topical SP application for 4 days (n = 3 per group) (Fig. 3B). Furthermore, SP also upregulated the expression levels of p-Akt, p-EGFR, and Sirt1 in both mouse TKE2 cells and primary HCECs, which were impaired by high glucose treatment (Supplementary Fig. 3). The results suggest that SP application in both diabetic mice and cultured corneal epithelial cells reactivates Akt, EGFR, and Sirt1 altered by hyperglycemia, which may in part explain the promoting mechanisms of SP in diabetic corneal epithelial wound healing.

To evaluate the effect of SP on the regulation of hyperglycemia-induced oxidative stress in corneal epithelium, corneal sections were loaded with fluorescence probe DCHF-DA and monochlorobimane for the detection of intracellular ROS and GSH. Representative results showed that a significant increased ROS and reduced GSH staining were detected in diabetic mouse corneal epithelium than in that of normal mice. However, a weak ROS and strong GSH staining of corneal epithelium was found after topical SP application in diabetic mice (Fig. 4A). Moreover, the expression of major intracellular free radical scavengers in corneal epithelium, including manganese superoxide dismutase (MnSOD), catalase, NAD(P)H: quinone oxidoreductase 1 (NQO1), thioredoxin (TXN), and heme oxygenase 1 (Hmox1) at the mRNA transcription level, were recovered from the diabetic group after topical SP application (n = 4 per group) (Fig. 4B). In addition, immunofluorescence staining and Western blot revealed that the protein levels of NQO1, catalase, and MnSOD in diabetic corneal epithelium partially recovered after topical SP application (n = 4 per group) (Fig. 4C and D). The results suggest that topical SP application attenuates hyperglycemia-induced oxidative stress in diabetic corneal epithelium, at least via the mechanism of reducing ROS accumulation and increasing intracellular GSH content and antioxidant gene expression.

Mitochondrial dysfunction plays an important role in the progress of various diabetes complications. To evaluate the effects of SP on the mitochondrial dysfunction in corneal epithelium induced by hyperglycemia, cultured corneal epithelial cells were treated with high glucose in the presence or absence of SP for 3 days. Exposure to elevated glucose caused increased ROS accumulation and reduced GSH content of corneal epithelial cells in vitro (Fig. 5A and B), similar to the diabetic corneal epithelium in vivo. However, SP treatment significantly reduced the oxidative stress caused by high glucose in corneal epithelial cells, as showed by decreased ROS accumulation and increased GSH content (n = 3 per group) (Fig. 5A and B). The results were also repeated by using the primary HCECs (Supplementary Fig. 4). Furthermore, compared with normal or mannose-treated cells, high glucose–treated cells assumed apparent mitochondrial superoxide staining (MitoSOX staining in Fig. 5C), accompanied by a significant change of mitochondrial structure (MitoTracker staining in Fig. 5C) and loss of mitochondrial membrane potential (red to green fluorescence of JC-1 staining in Fig. 5C). However, the addition of SP attenuated the generation of mitochondrial superoxide and promoted the recovery of mitochondrial structure and membrane potential that was impaired by high glucose treatment in corneal epithelial cells (Fig. 5C).

To assess if the NK-1 receptor mediates the improvements of SP on diabetic corneal epithelial wound healing, the NK-1 receptor–specific antagonist L-733,060 was injected before topical SP application in diabetic mice. In unwounded mouse corneal epithelium, the staining density of p-Akt, p-EGFR, and Sirt1 was attenuated in antagonist-injected SP-treated diabetic mice, as compared with the diabetic mice treated with SP alone (Fig. 6A). In corneal epithelium–scraped mice, the antagonist injection before SP application fully reversed the promotion of SP on diabetic corneal epithelial wound healing, with 28.95 ± 3.89% epithelial defect in antagonist-injected SP-treated mice as compared with only 4.44 ± 2.29% defect in mice treated with SP alone and with 22.73 ± 9.85% defect in untreated diabetic mice at 72 h (n = 5 per group) (Fig. 6B). Moreover, at 72 h postscrape, a stronger staining intensity of p-Akt and proliferation marker Ki-67 was found in the migrating area of SP-treated diabetic corneal epithelium than in that of either untreated or antagonist-injected diabetic corneal epithelium (Fig. 6C). The results suggest that the NK-1 receptor mediates the reactivation of Akt, EGFR, and Sirt1 by SP, and that the promotion of SP on diabetic corneal epithelial wound healing is also NK-1 receptor mediated.

To assess the effects of SP-NK-1 receptor signaling inhibition on corneal epithelium, L-733,060 was injected subconjunctivally in normal mice. After 3 days of antagonist injection, the unwounded corneal epithelium assumed a similar attenuation of p-Akt, p-EGFR, and Sirt1 staining as that in diabetic mice (Fig. 7A), accompanied by decreased corneal sensitivity (n = 5 per group) (Fig. 7B). After 72 h of corneal epithelium scrape, the antagonist-injected mice showed a significant delay of corneal epithelial wound healing, with 31.77 ± 5.07% epithelial defect in antagonist-injected mice as compared with only 3.08 ± 2.17% defect in control mice (n = 5 per group) (Fig. 7C), and attenuated p-Akt and Ki-67 staining in the migrating area of corneal epithelium as compared with that in normal mice (Fig. 7D), which was similar to the pathological changes in diabetic corneal epithelium (Fig. 7C and D). Similar results were also obtained with the injection of another NK-1 receptor antagonist spantide I (data not shown). The results show that local injection of NK-1 receptor antagonist in normal mice causes similar pathological changes in corneal epithelial wound healing and corneal sensitivity as that seen in diabetic mice, suggesting that normal activation of SP-NK-1 receptor signaling plays a critical role in the homeostasis of corneal epithelium and corneal sensation.

The cornea is one of the most densely innervated parts of the human body, and as such, its sensory nerves play not only a prominent role in nociception but also in providing trophism to the corneal tissue. In diabetes, corneal sensitivity, nerve fiber density, and epithelial wound healing are reduced significantly (27,31,32). However, the mechanisms are not completely understood. In the current study, we found that SP promoted epithelial wound healing, stimulated the reactivation of Akt, EGFR, and Sirt1, as well as attenuated oxidative stress in diabetic corneal epithelium. Furthermore, the study shows that the impairment of SP-NK-1 receptor signaling causes changes of normal corneal epithelium similar to those of diabetic mice. The results suggest that SP-NK-1 receptor signaling regulates the activation of multiple signaling pathways that are needed for corneal epithelial wound healing, whereas this regulation is impaired in diabetic corneal epithelium and rescued by SP via an autoregulatory mechanism (37,38). Moreover, SP restored the corneal sensitivity of diabetic mice close to the same level of normal mice, whereas local inhibition of SP-NK-1 receptor signaling caused decreased corneal sensitivity in normal mice similar to that in diabetic mice. The results suggest that the impairment of SP-NK-1 receptor signaling may also be involved in the attenuation of corneal sensitivity in diabetes, which is supported by the fact that the NK-1 receptor exists in peripheral nerve (39). Taken together, SP, secreted by corneal sensory nerve fibers, may be the key neurotransmitter and neuropeptide that mediates corneal nociception transmission and provides trophism to the corneal epithelium. Furthermore, as for the treatment of diabetic keratopathy, many growth factors, cytokines, and various agents have been evaluated for their capacity to accelerate corneal wound healing (40). However, SP and its functional derivative (FGLM-amide), with the advantages of smaller molecules and higher efficiency, have been shown to be effective for the treatment of persistent corneal epithelial defects in clinical studies (40–42). In addition, we found that SP promoted the regeneration of nerve fibers in diabetic corneal epithelium and accelerated trigeminal neuronal growth in vitro that was impaired by high glucose (Supplementary Fig. 5).

The NK-1 receptor, the preferred receptor of SP, mediates a variety of physiological and pathophysiological responses (22,23,43). Although previous studies have confirmed that SP enhances corneal epithelial migration when combined with IGF-1 or EGF (41,44,45), the exact mechanism remains unclear. Here we show that SP reactivates Akt, EGFR, and Sirt1 and promotes ROS scavenging capacity that are impaired by hyperglycemia. The results suggest a molecular basis for the synergistic effects of SP and IGF-1 or EGF on the enhancement of diabetic corneal epithelial wound healing (46,47). Even more interestingly, we found that the local inhibition of the SP-NK-1 receptor signaling in normal mice causes pathological changes of the corneal epithelium similar to those of diabetic corneal epithelium. These results suggest that SP-NK-1 receptor signaling may be involved in the maintenance of corneal epithelium homeostasis and also in the protection from hyperglycemia stress, whereas the impairment of SP-NK-1 receptor signaling may explain the fragility of diabetic corneal epithelium in vivo.

Prolonged hyperglycemia always causes the perturbation of catabolic pathways and the overproduction of ROS in the mitochondria, which in turn plays a critical role in the development of diabetes complications (48), including diabetic keratopathy (5). In the current study, we confirmed that the mitochondrial superoxide level was upregulated in high glucose–treated corneal epithelial cells, accompanied by changes of mitochondria structure and loss of mitochondrial membrane potential. Interestingly, we found that SP attenuates the dysfunction of the mitochondria by high glucose. Moreover, SP also elevates the intracellular GSH level, the main antioxidant in the cells. In addition, although increased Nrf2 expression was detected in diabetic corneal epithelium (data not shown), the expressions of Nrf2 downstream antioxidant genes, including MnSOD, catalase, NQO1, TXN, and Hmox1, were downregulated in diabetic corneal epithelium, whereas they were upregulated after SP application, which suggests that additional regulatory mechanisms may exist between Nrf2 and its downstream antioxidant gene expressions in diabetes (49,50). All things considered, the improvement of oxidative stress by SP via the improved mitochondrial function, elevated GSH level, and upregulated expression of antioxidant genes may also play an important role in the protection of corneal epithelium in diabetes.

In conclusion, our study demonstrates, for the first time, that SP promotes diabetic corneal epithelial wound healing while simultaneously triggering the reactivation of the Akt, EGFR, and Sirt1 signaling of importance for that healing, as well as rescuing corneal sensation, improving mitochondrial function, and decreasing ROS accumulation caused by hyperglycemia. Furthermore, we show that local inhibition of SP-NK-1 receptor signaling abolishes the promotion of SP on corneal epithelial healing in diabetic mice and causes diabetic corneal pathological changes in normal mice. Thus, the SP-NK-1 receptor signaling may play a critical role in the maintenance of corneal epithelium homeostasis, and SP signaling may, through the NK-1 receptor, contribute to the promotion of diabetic corneal epithelial wound healing by the rescued activation of Akt, EGFR, and Sirt1, the improvement of mitochondrial function, and the increased ROS scavenging capacity of corneal epithelium.

Acknowledgments. The authors thank Yangyang Zhang, Wenjie Sui, Qian Wang, Hua Gao, Suxia Li, and Zhaoli Chen (Shandong Eye Institute) for their help with the animal experiments, statistical analysis, and human tissue collection.

Funding. This work was partially supported by the National Basic Research Program of China (2012CB722409) and the National Natural Science Foundation of China (81170816 and 81200665). Q.Z. is partially supported by the Shandong Provincial Excellent Innovation Team Program and Taishan Scholar Program (20081148). P.D. is partially supported by the J.C. Kempe and Seth M. Kempe Memorial Foundations, the Swedish Society of Medicine, the Cronqvist and KMA Foundations, and the National Swedish Research Council (521-2013-2612, Q.Z. coapplicant).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. L.Y. and G.D. contributed to sample testing, data analysis, and study design. X.Q., M.Q., Y.W., and H.D. contributed to sample testing and data analysis. P.D., L.X., and Q.Z. contributed to study design, data analysis, and manuscript preparation. L.X. and Q.Z. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.