Plasma Sphingolipids Associated with Chronic Obstructive Pulmonary Disease Phenotypes

Methods

Results

Discussion

This study of more than 250 subjects from the COPDGene cohort is the largest and the only combined targeted and nontargeted metabolomic study of plasma sphingolipid species in smokers with and without COPD; it is the first to link specific sphingolipids with distinct COPD phenotypes. We identified novel sphingolipid subclasses such as gangliosides and sphingomyelins as potential biomarkers inversely associated with severe emphysema and trihexosylceramide as a potential biomarker positively associated with frequent COPD exacerbations. All these associations were noted even after adjusting for covariates of age, smoking status, airflow obstruction, sex, and BMI; most were replicated in an independent cohort using an orthogonal mass spectrometry approach in a different laboratory.

Sphingomyelins are integral components of the plasma membrane, typically enriched in lipid microdomains of caveolae and clathrin-coated pits (22). Sphingomyelins are produced from ceramides via sphingomyelin synthase, or they can generate ceramides through hydrolysis by sphingomyelinases (Figure 5). In addition to rapidly generating ceramides as second messengers, sphingomyelins may themselves play a role in cell migration, apoptosis, autophagy, and cell survival/proliferation (22). We identified 10 specific sphingomyelins (e.g., SM.d18.1.N16.0) that were inversely associated with emphysema, whereas others (e.g., SM.d18.1.N18.0) were not. These species-specific effects could only be studied with mass spectrometry; however, we also noted strong correlations among the whole subclass of sphingomyelins. A recent report (5) found no statistically significant association between HRCT-measured emphysema severity at baseline and sphingomyelin, measured with less accurate spectrophotometric assays (23). However, low baseline plasma sphingomyelin was associated with worse COPD (lower FEV₁), whereas high levels were associated with more rapid progression of emphysema. Our data support the finding that worse COPD is associated with lower plasma sphingomyelins (5), but the apparent paradoxical observation of higher plasma sphingomyelins with progression of emphysema cannot be validated by our experimental design. Such observation might be explained if low sphingomyelin levels reflect low lung mass and progression of emphysema occurs in lungs that are not yet severely emphysematous. This could be explored by careful kinetics of specific sphingomyelins accompanied by concomitant HRCT measurements of the lung.

Plasma ceramides may be generated by activated plasma sphingomyelinases (24) (Figure 5) and could signal outside-in (25) and/or be further metabolized to sphingosine (by plasma ceramidases). To the extent that plasma levels reflect cellular metabolism, the decrease in both sphingomyelins and ceramides with subsequent increases in ceramide metabolites suggests that low sphingomyelin levels are not due to a blockage in sphingomyelin synthase (which would result in ceramide accumulation) but due to increased catabolism coupled with decreased availability of its building block precursor, ceramide (Figure 5). In such conditions, supplementation of sphingomyelin may fuel ceramide production and catabolism, by a “revved up” system. Plasma sphingomyelins might also reflect the burden or quality of circulating (shed) cellular plasma membranes, such as apoptotic microparticles or exosomes. Clearly, these unbiased metabolomic results raise important mechanistic questions that can be explored in models of COPD.

Ceramides (especially Cer.d18.1.N16.0), gangliosides (especially GM3.d18.1.N16.0), and monohexosylceramides (especially monohexcer.d18.1.N16.0) also showed an inverse correlation with emphysema. Although ceramides have been known to play a role in the pathogenesis of emphysema for several years (10, 26), the involvement of gangliosides and monohexosylceramides has not been described. An up-regulation of ceramide production has been typically associated with models of emphysema, by causing endothelial and epithelial cell apoptosis (3) or by inhibiting alveolar macrophage clearance of apoptotic cells (13). In this cross-sectional sampling, the inverse correlation of plasma ceramides with emphysema appears surprising. However, if our global metabolomics assessment of plasma sphingolipids reflects pulmonary cellular events, worse emphysema may be linked to increased activity of both sphingomyelinase and recycling (from gangliosides) pathways, followed by ceramide consumption/degradation, rather than accumulation (Figure 5), possibly via plasma ceramidases. Because plasma S1P was not increased and tended to be negatively correlated with emphysema and COPD exacerbations, unlike sphingosine, it is possible that the sphingolipid metabolism is accelerated toward ceramide production and degradation but inhibited at the level of sphingosine kinase activity (Figure 5). The reduced production of S1P may be associated with decreased cellular survival and proliferation and increased apoptosis of alveolar cells (15). Much less is known about the significance of decreased gangliosides (or glycoceramides) in the lung, reports showing association with diabetes or hearing loss (27). Although our earlier genomic analysis hinted at this involvement (4), this is the first report of decreased monosialodihexosylganglioside GM3.d18.1.N16.0 as putative biomarker of emphysema. Similarly, the role of monohexosylceramides in the lung is unknown, but they were shown to bind to surfactant protein D (28) and to decrease on activation of acid sphingomyelinase and acid β-glucosidase1, as a sign of ceramide production via recycling from gangliosides during inflammation (29). These results may embolden future investigations of lung sphingolipid metabolomic signatures and the functional role of the gangliosides and monohexosylceramides in COPD.

The strongest positive association with COPD exacerbations was for trihexosylceramides (trihexcer.d18.1.N22.0 and trihexcer.d18.1.N24.0), a lipid class with unknown lung effects; one case report of Fabry disease (characterized by overabundance of trihexosylceramides) showed association with COPD (30). Other sphingolipids such as ceramide (d18.1.N18.0) and S1P were inversely associated with COPD exacerbations. This, along with negative correlations of plasma sphingomyelins and ceramides with S1P, suggests a rapid flux of sphingosine-to-ceramide-to glycosylated ceramides metabolism, with potential disruption of ceramide-to-S1P metabolism (Figure 5) in those with COPD exacerbations. Alternatively, whether tissues or cells may differentially contribute to the abundance of plasma metabolites in these individuals could be explored by comparing plasma with lung tissue metabolomics signatures. Because these samples were collected at a time remote from an actual exacerbation, metabolomic signatures are unlikely to reflect corticosteroid or antibiotic treatment or acute infections. Rather, they may indicate specific changes associated with developing COPD exacerbations. For example, shunting ceramide metabolism toward trihexosylceramides may render the lungs more susceptible to inflammation, due to defects in lymphocyte function, as learned from a mouse model of Fabry disease (31).

To further translate biomarker findings to disease pathogenesis, we explored alterations of gene–metabolite associations with disease phenotypes. We identified a negative association of sphingosine levels with the expression of CYR61 only in subjects with severe COPD. Although the significance of this finding remains to be tested, the extracellular matrix protein CCN1/Cyr61 is activated by S1P via receptor 2 (32), which also activates rho kinase to cause lung damage during cigarette smoke exposure (33, 34). Other disrupted gene–metabolite associations of interest in those with severe COPD exacerbations were between ceramide and alkaline ceramidase, which degrades unsaturated long chain ceramides (35); ORMDL1, which regulates de novo ceramide biosynthesis, recently involved in asthma (36); and 3-ketodihydrosphingosine reductase, which reduces 3-ketodihydrosphingosine to dihydrosphingosine (37). Another intriguing association was that between circulating ceramide and COL4A3BP, also known as CERTL, a gene encoding for ceramide transporter protein Goodpasture antigen-binding protein, which mediates the nonvesicular intracellular transport of ceramides (38). Superimposing potential sphingolipid biomarkers with genetic signatures helped map the directionality of sphingolipid metabolism in individuals with COPD exacerbations (Figure 5). Using similar approaches may strengthen the associations of sphingolipid metabolism and chronic bronchitis or obstructive ventilatory defects in COPD, which were weak at the plasma metabolite levels, but in conjunction with genomic findings suggested a potential metabolic flux toward higher S1P and de novo pathway activation, respectively (Figure E3).

In summary, this study demonstrates that specific sphingolipids, such as ceramides, sphingomyelins, and gangliosides, may be biomarker candidates of COPD phenotypes, such as emphysema and frequent COPD exacerbations. Our study’s strengths include the large number of highly phenotyped subjects, the multiple specific sphingolipid measurements, and independent replication. Limitations of our study include lack of healthy nonsmoking subjects, of longitudinal studies, and of lung tissue, bronchoalveolar lavage fluid, or sputum analyses. The latter would have allowed us to probe if plasma findings parallel those in the lung and to validate a small study that suggested sputum sphingolipids are associated with COPD (6). Our results in a much larger cohort of patients with COPD corroborate those reported in the sputum, in particular that there are abnormalities in sphingolipid metabolism and glycosphingolipid changes in COPD. The complementary nature of these studies conducted in different compartments may help interpret our results in plasma, which indicate an overall accelerated metabolism/catabolism of ceramide at the systemic level, in contrast to the increased ceramides and dihydroceramides recovered in the sputum in COPD. This may be due to active transfer and use of plasma sphingolipids in the lung with release of ceramides in sputum; alternatively, it may be due to different dynamics of sphingolipid metabolism in circulating cells and endothelium compared with airway macrophages or epithelial cells that may contribute to sputum sphingolipid content.

Our results strengthen the evidence implicating ceramide in emphysema and suggest a previously unsuspected role for ceramide metabolism in COPD exacerbations. Furthermore, our data suggest that additional subclasses of sphingolipids such as sphingomyelins, gangliosides, monohexosylceramides, and trihexosylceramides should be investigated, to validate their role as biomarkers and as potential contributors to COPD pathogenesis.