A vicious cycle between meibomian intracrine activity and inflammation.

(A) Laser-microdissection of eyelid tarsal plates of Hsd3b6+/+ and Hsd3b6−/− mice. (B) Volcano plot of differentially expressed genes between Hsd3b6+/+ and Hsd3b6−/− eyelid tarsal plates. n = 4 biologically independent samples for both genotypes. (C) Bubble plot showing GO terms for upregulated DEGs in Hsd3b6−/− tarsal plates. (D) Heatmap showing genes categorized as inflammatory response in (C). n = 4 per genotype. (E) Effects of eyelid subcutaneous injection of DHT on Tnf and Ccl3 expression in Hsd3b6−/− tarsal plates (n = 5 mice). Left eyelids were treated with vehicle (Veh). (F) Experimental timeline for subcutaneous treatment of LPS for testing its effect on eyelid inflammatory cytokine gene expression and meibomian-gland intracrine activity. (G) Changes in expression of the inflammatory cytokine genes Il1b, Il6, Tnf, and Ccl3 in (F). Values were determined by qRT-PCR (n = 4 for both Veh- and LPS-treated group). (H) Relative meibomian 3β-HSD activity in (F) (n = 4 per group). Values were determined by radioisotopic tracing of 3β-HSD activity using in vitro isolated tarsal plates. (I) Schematic illustration of decreased intracrine activity and inflammation, which form a vicious cycle in the eyelid. Data in (E) were analyzed using paired two-sided t-test; in (G), two-way analysis of variance (ANOVA) with Tukey’s post hoc test; and in (H), unpaired two-sided t-test. Values represent means ± SEM; *P < 0.05.

CD38 mediates NAD depletion-driven vicious cycle in the meibomian gland.

(A) Schematic of the NAD salvage pathway and NAD-dependent enzymatic activity of Hsd3b6. (B) Hsd3b6−/− versus Hsd3b6+/+ expression of genes involved in the NAD salvage pathway. The color of circles indicates fold change and the size indicates P-value. Data are based on Fig.1B. (C) Representative immunofluorescence images of CD38 in Hsd3b6+/+ and Hsd3b6−/− eyelid coronal sections. The sections were visualized by DAPI staining and bright field microscopy. (D) The number of CD38 positive cells around meibomian gland acini in Hsd3b6+/+ and Hsd3b6−/− mice. n = 8 eyelids per genotype. (E) Meibomian gland NAD levels in Hsd3b6+/+ (n = 5) and Hsd3b6−/− (n = 7) mice. (F) Meibomian gland NAD levels in Hsd3b6−/− mice after eyelid subcutaneous injection of 78c (n = 4 mice). Left eyelids were treated with Veh. (G) The number of CD38 positive cells around meibomian gland acini in Hsd3b6−/− mice after eyelid subcutaneous injection of DHT (n = 7 mice). Left eyelids were treated with Veh. (H) Meibomian gland NAD levels in Hsd3b6−/− mice after eyelid subcutaneous injection of DHT (n = 3 mice). Left eyelids were treated with Veh. (I) Relative mRNA levels of Cd38, Sirt1, Parp1, Nampt, Nmnat1, Nmnat2, and Nmnat3 in the tarsal plate of wildtype mice after eyelid injection of Veh (n = 3-4) or LPS (n = 4). (J) Meibomian NAD⁺ levels in wildtype mice after treatment of Veh (n = 6) or LPS (n = 4). (K) Effects of 78c on meibomian NAD⁺ levels in wildtype mice treated with LPS (n = 5). Left eyelids received Veh without 78c after eyelid injection of LPS. (L) Effects of 78c on meibomian 3β-HSD activity in wildtype mice treated with LPS (n = 5). The eyelid tarsal plates from LPS-treated mice were incubated in culture with or without 78c. Data in (D), (E), (I) and (J) were analyzed using unpaired two-sided t-test; in (F), (G), (H), (K) and (L), paired two-sided t-test. Values are means ± SEM; *P < 0.05, **P < 0.01, ***P < 0.001.

Increased CD38-driven eyelid intracrine-activity decline in aged mice.

(A) Human eyelid tissue sections stained for CD38. Shown are a pair of serial sections from 85-year-old and 41-year-old male eyelid specimens, stained for CD38 and HSD3B1. Dashed lines delineate HSD3B1-positive meibomian acini. CD38-immunopositive cells are pointed by arrowheads. (B) The number of CD38 positive cells around meibomian acini in human eyelids (41-year-old male, 34-year-old female, 85-year-old male and 85-year-old female). (C) Representative immunofluorescence images of CD38 in aged (24 mo) and young (6 mo) eyelid sections. The sections were visualized by DAPI staining and bright field microscopy. (D) The number of CD38 positive cells around meibomian acini in young (2-6 mo) and aged (20-24 mo) mice (n = 7 for each group). (E) Age-associated gene expression changes in the NAD-salvage pathway. Tarsal plates of aged and young mice were subjected to gene-specific qRT-PCR. n = 6-12 per group. (F) Immunoblotting of CD38 in young (2 mo) and aged (20 mo) eyelid tarsal plates (n = 4 for each group). β-Actin was used as a loading control. (G) Quantification of ecto-enzymatic NMNase activity in isolated tarsal plates. The rate of conversion from D4-NMN to D4-NAM was quantified for young (2-6 mo, n = 6), aged (21 mo, n = 7), and 78c-treated aged (21 mo, n = 3) mouse tarsal plates. (H) Meibomian gland NAD levels in young (2-5 mo) and aged (20 mo) mouse eyelids after in vitro incubation with 78c or Veh (n = 6 for each group). (I) 3β-HSD activity in the eyelid tarsal plates incubated in culture with or without 78c (n = 6 eyelids per condition, 24-25 mo). Statistics in (D), (E), and (F), unpaired two-sided t-test; in (G), one-way ANOVA followed by Bonferroni’s test; in (H), two-way ANOVA followed by Tukey’s test; and in (I), paired two-sided t-test. Values are means ± SEM; *P < 0.05, **P < 0.01, ***P < 0.001. mo, months old.

Age-associated meibomian gland phenotype after 78c-NMN in vivo treatment.

(A) Schematic of 78c and NMN treatment. 78c was subcutaneously injected into lower eyelid and NMN was administered via eye-drop. Left eye served as a treatment control. (B) Meibomian gland NAD levels in aged (21 mo) mice treated with Veh (n = 4), 78c (n = 4), 78c plus NMN (n = 5), or NMN (n = 5). Mice received two doses of Veh or 78c 20 h and 4 h prior to euthanasia. One dose of NMN was given by instillation at 1 h before euthanasia. (C-D) Meibomian gland morphology in the lower eyelids of aged (24 mo) mice treated with 78c (C) or 78c plus NMN (D). Left eyes were treated with Veh. The graph shows the size of the meibomian gland area (n = 10 mice per condition). Treatment was performed for 7 w, with administration of 78c once every 3 d and NMN three times every 2 d. (E) RNA-seq data showing 78c+NMN-versus Veh-treated eyelid tarsal plate transcriptome, presented in MA-plot format (n = 3 biologically independent samples for each group; each sample was generated using a pool of 3 lower eyelids) for aged and young mice. Plots in red and blue represent transcripts significantly upregulated and downregulated by 78c+NMN treatment, respectively (Padj < 0.1). Treatment was performed for 4 w. (F) Bubble plot showing GO terms for upregulated (top) and downregulated (bottom) DEGs in 78c+NMN treated tarsal plates of aged mice. Data in (B) were analyzed using paired or unpaired two-sided t-tests with FDR correction; in (C) and (D), paired two-sided t-test. Values are means ± SEM; *P < 0.05, **P < 0.01.

Reduced inflammation due to 78c+NMN treatment, a potential virtuous cycle.

(A) Representative immunofluorescence images of CD38 in Veh-treated or 78c+NMN treated eyelids of aged mice. The upper graph shows the number of CD38 positive cells around meibomian acini in each condition (n = 6 mice, 25 mo). The lower graph indicates RNA-seq data for Cd38 mRNA in the tarsal plate in Fig. 4E. Treatment was performed for 4 w. (B) Representative immunofluorescence images of CD38 in Veh-treated or 78c+NMN treated eyelids of Hsd3b6−/− mice. The graph shows the number of CD38 positive cells around meibomian acini in each condition (n = 5 mice, 5-10 mo). Treatment was performed for 4 w. (C) Venn diagram showing the overlap between downregulated genes by 78c+NMN treatment and upregulated genes upon Hsd3b6 deletion. The shared DEGs were subjected to GO term enrichment analysis. (D) A schematic model of local circular circuits in the eyelids. A conversion from a pathologic vicious cycle into a therapeutical virtuous cycle may underlie 78c+NMN’s effects on MGD. Data in (A, upper) and (B) were analyzed using paired two-sided t-test; in (A, lower), unpaired two-sided t-test. Values are means ± SEM; *P < 0.05.

Eyelid CD38 expression and co-labelling with a macrophage marker CD11b, related to Fig. 2.

Representative immunofluorescence images of CD38 (red) and a macrophage marker CD11b (green) in Hsd3b6−/− eyelid.

Human female eyelid sections, immunolabeled for CD38, related to Fig. 3.

Anti-CD38 and anti-HSD3B1 immunohistochemistry using a pair of serial sections from 85-year-old and 34-year-old female eyelid specimen. The dotted boxes indicate the region of magnified view. CD38-immunopositive cells are pointed by arrowheads.

Two different monoclonal antibodies for the mouse CD38 — clones EPR21079 and E9F5A — exhibit similar staining patterns in the tarsal plate of aged mice, related to Fig. 3.

The graph shows the number of CD38 positive cells around meibomian acini in young (2 mo, n = 4) and aged (23 mo, n = 3) mice when using clone E9F5A. Data are means ± s.e.m. *P < 0.05, unpaired t-test.