Introduction
Interstitial lung disease (ILD) is a heterogeneous disease in terms of its variety, particularly in regard to its clinical course and treatment [
1]. Among ILDs, idiopathic pulmonary fibrosis (IPF) has a poor prognosis and requires a unique treatment strategy using antifibrotic drugs, while immunosuppressive therapy, which is frequently administered for non-IPF ILDs, is not recommended for IPF [
2]. Accurately diagnosing IPF is therefore an important task for respiratory physicians.
The diagnosis of IPF requires evidence of a typical “usual interstitial pneumonia (UIP)” pattern either through radiology or pathology, as well as exclusion of the possibility of other non-IPF ILDs [
2]. However, accurately diagnosing IPF can sometimes be difficult because non-IPF ILDs, including collagen vascular disease-related ILD and other fibrotic ILDs, can exhibit a UIP-like pattern [
3‐
5]. Even for experts in ILD, diagnosing IPF with a high certainty is challenging and is an active research topic [
6,
7]. In this context, biomarkers that can easily distinguish IPF from non-IPF ILDs would be helpful. In Japan, Krebs von den Lungen-6 (KL6) is one of the most popular biomarkers for the clinical management of ILDs. However, as several researchers have pointed out, this molecule is more suitable for evaluating disease behavior and prognosis than differentiating ILDs [
8,
9], which suggests a need for another biomarker for the purpose of differential diagnosis.
We recently performed a microarray analysis using bleomycin-induced myofibroblasts and steady-state fibroblasts directly isolated from mouse lungs [
10]. In the transcriptome data, interestingly, we found that gremlin-1, a secreted glycoprotein and an antagonist of bone morphogenetic protein-4 (BMP4), was significantly upregulated in the myofibroblasts of the fibrotic lungs. Consistent with our data, previous studies have also suggested that gremlin-1 is upregulated in human lungs with ILDs, particularly IPF [
11,
12]. Based on these results, we therefore sought to explore the possibility of serum gremlin-1 as a novel biomarker for ILD.
Discussion
In this study, we demonstrated that the protein of gremlin-1 was upregulated in fibrotic lungs, particularly at myofibroblasts, CD163-positive M2-like macrophages, and bronchiolar/alveolar epithelium in IPF lungs. The serum gremlin-1 concentration was significantly higher in patients with IPF than those with non-IPF ILD, suggesting a pathophysiological role of this molecule in IPF.
Gremlin-1 is an endogenous BMP4 antagonist, and the antagonizing regulation employs extracellular and intracellular pathways [
13]. This molecule is essential during respiratory development for normal airway patterning and for the differentiation of distal epithelial cells [
14]. In contrast, gremlin-1 has been studied about the associations with abnormal lung conditions, such as hypoxia, pulmonary hypertension, and above all, lung fibrosis [
13]. In fact, in our mouse microarray dataset, gremlin-1 was upregulated while BMP4 was downregulated in the myofibroblasts of fibrotic mouse lungs (Fig.
2a). The similar expression pattern of gremlin-1 and BMP4 is reproduced even in IPF lung-derived (myo)fibroblasts [
11], implying a common role for myofibroblasts in fibrotic lungs in inhibiting BMP4 signaling, probably via gremlin-1 autocrine activity. BMP4 is thought to inhibit fibroblast proliferation and differentiation into myofibroblasts [
11,
15]. In fact, a previous experimental study using noggin-insufficient mice revealed that the upregulation of BMP signaling could reduce the severity of bleomycin-induced lung fibrosis [
16]. Thus, gremlin-1 upregulation and BMP signaling downregulation in myofibroblasts would support their own presence, leading to the persistence of lung fibrosis.
Based on our immunostaining results, the origin of lung gremlin-1 appears to be not only myofibroblasts but also the bronchiolar/alveolar epithelium and CD163-positive M2-like macrophages. Consistent with these results, a recent bulk RNA-seq data analysis of human cells also showed that
GREM1 mRNA expression in fibrotic lungs is significantly upregulated, even in alveolar type 2 epithelial cells (approximately 14.7-fold) and alveolar macrophages (approximately as much as 116.2-fold) compared with normal control cells [
17]. Another study showed that gremlin-1 overexpression in lung epithelial cells induced mRNA upregulation of transforming growth factor-β1 (TGF-β1), a potent fibrosis-promoting factor. In fact, epithelial gremlin-1 overexpression could lead to lung fibrosis, even in vivo [
18]. In contrast, it is well known that M2-like macrophages also play an important role during fibrogenesis, particularly as a primary source of TGF-β [
19]. Gremlin-1 secretion may be an additional role for M2-like macrophages and might further activate the surrounding fibroblasts. Collectively, gremlin-1 upregulation in epithelium and macrophages could cooperatively contribute to the pathogenesis of lung fibrosis.
The increase in serum gremlin-1 concentration was significantly prominent in IPF compared with non-IPF ILDs. Although the exact mechanism underlying this gap is unclear, it could be mostly due to the difference in the quantity of myofibroblasts or fibroblastic foci, which are significantly more numerous in IPF than in non-IPF ILD lungs or in UIP pattern than in non-UIP pattern [
4,
20,
21]. This may partly explain our finding that UIP compatibility on chest HRCT was associated with serum gremlin-1 concentrations, particularly even in patients with non-IPF ILD (Fig.
4b, c). In addition, based on our pathology findings, the severity of M2-like or M2-polarized macrophage infiltration might affect the serum gremlin-1 value. Compared with normal lungs, the significantly greater infiltration of M2-like macrophages has been reported in IPF lungs and in NSIP lungs [
22]. As we have also suggested, M2-like macrophages are increased in the lungs of dermatomyositis-associated ILD [
23]. Actually, our patients with dermatomyositis-associated-ILD showed relatively high serum gremlin-1 concentrations greater than our cutoff value 10.4 ng/mL. In contrast, compared with myofibroblasts and macrophages, we could not find apparent differences in the pattern of gremlin-1 protein expression in epithelial cells between IPF and non-IPF ILDs. Accordingly, the gap in serum gremlin-1 concentrations among patients with ILD appears to mainly reflect the total amount of myofibroblasts and perhaps that of M2-like macrophages.
A previous pathology study evaluated the utility of gremlin-1 for the differential diagnosis of ILD and reported that the mRNA and protein expression of gremlin-1 in lung samples was more increased in IPF than in NSIP [
12], which is consistent with our results. However, the authors also showed that gremlin-1 expression assessed by lung immunostaining was negatively correlated with the patients’ pulmonary function, which was not apparent in our serum data. This might be explained by the difference in selected samples (lung specimens versus serum) and by a statistical limitation due to the relatively low number of participants in both studies. Although further verification with larger cohorts is essential, our data would broaden the possibility of gremlin-1 as a novel differential diagnosis biomarker using “serum”, which can be easily and non-invasively evaluated.
There have been numerous potential IPF biomarkers identified for evaluating disease severity and prognosis, including KL6, SP-A/D, matrix metalloproteinases, and osteopontin [
8,
9,
24]. We recently reported that latent TGF-β binding protein-2 is a candidate for reflecting the process of fibroblast-to-myofibroblast differentiation [
10]. However, biomarkers for the differential diagnosis between IPF and non-IPF ILDs are not frequently reported. Thus far, matrix metalloproteinases and their combination with other markers, such as SP-D and osteopontin, seem promising for this purpose [
25,
26]. Although we could not directly compare these biomarkers with gremlin-1 in the same cohort, in spite of single molecule, the AUC value of gremlin-1 can be comparable to that of previous reports. To further increase the AUC value, new combinations of gremlin-1 with such candidate biomarkers would be worth trying.
In summary, gremlin-1 was upregulated in fibrotic lungs, particularly in IPF, and serum concentration measurements may be useful for improving the diagnostic certainty of IPF versus non-IPF ILDs. Our results again highlight the importance of gremlin-1 in the pathogenesis of IPF.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.