Monocytes in type 1 diabetes families exhibit high cytolytic activity and subset abundances that correlate with clinical progression

Abstract

Monocytes are immune regulators implicated in the pathogenesis of type 1 diabetes (T1D), an autoimmune disease that targets insulin-producing pancreatic β cells. We determined that monocytes of recent onset (RO) T1D patients and their healthy siblings express proinflammatory/cytolytic transcriptomes and hypersecrete cytokines in response to lipopolysaccharide exposure compared to unrelated healthy controls (uHCs). Flow cytometry measured elevated circulating abundances of intermediate monocytes and >2-fold more CD14⁺CD16⁺HLADR⁺KLRD1⁺PRF1⁺ NK-like monocytes among patients with ROT1D compared to uHC. The intermediate to nonclassical monocyte ratio among ROT1D patients correlated with the decline in functional β cell mass during the first 24 months after onset. Among sibling nonprogressors, temporal decreases were measured in the intermediate to nonclassical monocyte ratio and NK-like monocyte abundances; these changes coincided with increases in activated regulatory T cells. In contrast, these monocyte populations exhibited stability among T1D progressors. This study associates heightened monocyte proinflammatory/cytolytic activity with T1D susceptibility and progression and offers insight to the age-dependent decline in T1D susceptibility.

Fig. 1.. Monocytes of patients with ROT1D and their healthy autoantibody negative siblings are hyper-responsive to TLR4 ligation compared to uHC.

The analysis included uHC [(A) n = 22; age, 15.9 ± 2.7 years; (B and C) n = 23; age 15.7 ± 2.9 years], stage 0 LRS [(A) n = 31; age, 10.9 ± 3.3 years; (B and C) n = 31; age, 10.8 ± 3.4 years; all autoantibody negative], stage 0 HRS [(A) n = 33; age, 11.4 ± 3.6 years; (B and C) n = 35; age, 11.3 ± 3.6 years; all autoantibody negative], and patients with ROT1D [(A) n = 22; age, 11.2 ± 3.7 years; HbA1c, 7.3 ± 1.0%; (B and C) n = 23; age, 11.0 ± 3.8 years; HbA1c, 7.3 ± 1.0%]. Monocytes were negatively selected from PBMCs. A total of 10⁶ monocytes per well were incubated in RPMI complete medium for 24 hours with LPS (10 ng/ml). The conditioned medium was recovered, and cytokines (IL-1β, IL-6, and TNF-α) were measured in duplicate by ELISA [(A to C), respectively]. Each dot represents a single participant. Horizontal bars indicate the mean ± SE. Significance was assessed with a two-tailed t test. Cytokine secretion levels between patients with ROT1D and the healthy normoglycemic sibling cohorts were not different (P > 0.05).

Fig. 2.. Monocytes of patients with ROT1D and their healthy autoantibody negative siblings exhibit proinflammatory transcriptomes compared to uHCs.

The analysis included uHC (n = 14; age, 15.3 ± 2.8 years), stage 0 LRS (n = 19, age, 11.7 ± 2.8 years, all autoantibody negative), stage 0 HRS (n = 15; age, 12.9 ± 3.4 years, all autoantibody negative), and patients with ROT1D (n = 14; age,12.4 ± 2.8 years; HbA1c, 7.5 ± 1.4%). RNA extracted from monocytes negatively selected from cryopreserved PBMCs was analyzed with the Affymetrix U133plus 2.0 array. This array interrogates 54,613 transcripts/variants. Because low intensity data are often unreliable, only probe sets where ≥75% of the samples had a log₂ intensity > 4.5 relative fluorescence units (RFU) (n = 31,571) were retained for analysis. Among the six pairwise comparisons, 826 probe sets met thresholds: |log₂ ratio| > 0.5, FDR < 20%. The number of transcripts meeting thresholds for each pairwise comparison: ROT1D versus HRS = 215, ROT1D versus LRS = 9, ROT1D versus uHC = 309, LRS versus uHC = 206, LRS versus HRS = 122, and HRS versus uHC = 356. (A) Hierarchical clustering of the 826 probe set union. (B) Heatmap illustrating the mean expression levels of 177 commonly regulated unique probe sets (|log₂ ratio| > 0.5; FDR < 0.2; χ² P < 1.0 ×10⁻⁵) among sorted monocyte subsets described by Schmidl et al. (23), and the 826 differentially expressed transcripts by ROT1D, HRS, LRS, and uHC monocytes. (C) Pearson correlation coefficients for the 177 commonly regulated unique probe sets. (D) GO biological processes associated with the 826 probe set union. (E) Expression levels of well-annotated transcripts that met thresholds among the six-pairwise comparisons.

Fig. 3.. scRNA-seq analysis of monocyte-enriched PBMCs of uHC, LRS, HRS, and patients with ROT1D.

The analysis included uHC (n = 2; age, 15.9 ± 2.6 years), stage 0 LRS (n = 2; age, 10.7 ± 0.2 years, all autoantibody negative), stage 0 HRS (n = 2; age, 13.6 ± 3.8 years, all autoantibody negative), and patients with ROT1D (n = 2; age, 9.4 ± 1.5 years; HbA1c, 7.6 ± 0.8%). (A) t-SNE plot shows 38,594 sequenced cells distributed into 13 clusters. Colors denote cluster identity as determined by the unsupervised clustering algorithm, and each dot represents an individual cell. The putative cluster identities, based on expression of established cell lineage markers, are indicted. The number of successfully profiled single cells per cluster were: CD14⁺ monocytes (n = 12,542), NK cells (n = 5154), naïve CD4⁺ T cells (n = 4973), memory CD4⁺ T cells (n = 3782), naïve CD8⁺ T cells (n = 3168), NKT cells (n = 2262), B cells (n = 2013), CD16⁺ monocytes (n = 1373), memory CD8⁺ T cells (n = 1324), γδ T cells (n = 894), CD4⁺ T_reg cells (n = 522), CD1c dendritic cells (n = 481), and plasmacytoid dendritic cells (n = 106). (B) Violin plots illustrate expression distribution of lineage-defining marker genes across the 13 clusters. (C) A heatmap illustrating the scaled expression of up to 15 discriminative gene sets (adjusted P value < 3.5 ×10⁻¹²) for each of the 13 clusters. The color bar illustrates the z-score distribution from −2.5 (purple) to 2.5 (yellow). DC, dendritic cell.

Fig. 4.. scRNA-seq profiling of circulating monocytes.

(A) UMAP plot shows 13,915 monocytes were clustered into ten subsets. Colors denote cluster identity as determined by unsupervised clustering algorithm, and each dot represents an individual cell. Numbers of successfully profiled single cells per subset are as follows: cluster 0 (n = 2798), cluster 1 (n = 2451), cluster 2 (n = 2302), cluster 3 (n = 1872), cluster 4 (n = 1480), cluster 5 (n = 1300), cluster 6 (n = 702), cluster 7 (n = 440), cluster 8 (n = 397), and cluster 9 (n = 173). (B) Expression pattern of marker genes (CD14, CD16/FCGR3A, GNLY, PRF1, KLRD1, and KLRB1) are indicated by color intensity. (C) Dot plot shows relative expression of selected marker genes belonging to each cluster. (D) Gene module scores for clusters 0 to 9, calculated using the AddModuleScore function in Seurat, were used to quantify the average expression level of the 97 mono4/NK-like transcripts at the single-cell level for each cohort. Tabulated are the P values for module score differences between cohorts that reached significance. n.s., not significant.

Fig. 5.. All scRNA-seq clusters of ROT1D monocytes express elevated levels of cytolyic and cytokine/chemokine transcripts.

(A) Expression levels of cytolytic transcripts differentially expressed between monocytes of uHC, LRS, HRS, and patients with ROT1D. Differential expression analysis was conducted for the top 100 transcripts in cluster 9. Shown are transcripts that meet a threshold of P < 0.05 in any one of the six possible pairwise comparisons. (B) Expression levels of selected cytokines and chemokines differentially expressed between monocytes of uHC, LRS, HRS, and patients with ROT1D. The cytokines and chemokines shown met an adjusted P value of <0.05 in one or more of the six possible comparisons between cohorts. Shown are monocytes of all clusters merged and each individual monocyte cluster (0 to 9). (C) Chemokine/cytokine gene enrichment scores for clusters 0 to 9, calculated using the AddModuleScore function in Seurat, were used to quantify the average expression level of the 14 cytokine/chemokine transcripts at the single-cell level for each cohort.

Fig. 6.. Cross-sectional flow cytometry analyses show patients with ROT1D have higher circulating percentages of intermediate monocytes.

The analysis included uHC (n = 11; age, 14.0 ± 2.8 years), stage 0 LRS (n = 14; age, 12.1 ± 3.8 years; all autoantibody negative), stage 0 HRS (n = 14; age, 12.4 ± 3.2 years; all autoantibody negative), patients with ROT1D (n = 25; age, 12.3 ± 3.0 years; HbA1c, 7.8 ± 1.3%), and patients with LST1D (n = 15; age, 18.7 ± 8.0 years; HbA1c, 8.5 ± 1.0%). (A) Representative flow cytometry profiles for identification of classical, intermediate, and nonclassical monocytes. Cryopreserved PBMCs were thawed and stained. A negative gating strategy was used to identify CD3⁻, CD19⁻, and CD56⁻ lymphocytes. Monocytes were differentiated from NK cells by gating on HLA-DR^hi and segregated into CD14⁺CD16⁻ (classical), CD14⁺CD16⁺ (intermediate), and CD14^dimCD16⁺ (nonclassical) populations. (B) Percentages of classical, intermediate, and nonclassical monocytes in the uHC, LRS, HRS, ROT1D, and LST1D cohorts (left to right). (C) Percentages of NK cells in the uHC, LRS, HRS, ROT1D, and LST1D cohorts. The differences between groups were assessed by analysis of variance (ANOVA).

Fig. 7.. Cross-sectional flow cytometry analyses show patients with ROT1D have higher circulating percentages of mono4 monocytes.

The analysis included uHC (n = 11; age, 14.0 ± 2.8 years), stage 0 LRS (n = 14; age, 12.1 ± 3.8 years, all autoantibody negative), stage 0 HRS (n = 14; age, 12.4 ± 3.2 years; all autoantibody negative), patients with ROT1D (n = 25; age, 12.3 ± 3.0 years; HbA1c, 7.8 ± 1.3%), and patients with LST1D (n = 15; age, 18.7 ± 8.0 years; HbA1c, 8.5 ± 1.0%). (A) Representative flow cytometry profiles for identification of KRLD1^highKLRB1^high and KLRD1^highPRF1^high monocytes. (B) Percentages of KLRD1^highKLRB1^high monocytes among total monocytes among the uHC, LRS, HRS, ROT1D, and LST1D cohorts. (C) Percentages of KLRD1^highKLRB1^high monocytes among classical, intermediate, and nonclassical monocytes (left to right) among the uHC, LRS, HRS, ROT1D, and LST1D cohorts. (D) Percentages of KLRD1^highPRF1^high monocytes among total monocytes among the uHC, LRS, HRS, ROT1D, and LST1D cohorts. (E) Percentages of KLRD1^highPRF1^high monocytes among classical, intermediate, and nonclassical monocytes (left to right) among the uHC, LRS, HRS, ROT1D, and LST1D cohorts. Each dot represents a different individual. Horizontal bars indicate the mean ± SE. The differences between groups were assessed by ANOVA.

Fig. 8.. The relationship between the percentages of intermediate and nonclassical monocytes at baseline and future β cell function among patients with ROT1D.

(A) Significant inverse relationship exists between baseline percentages of CD14⁺CD16⁺ (intermediate) monocytes and the rate (slope) of C-peptide decline over the 24-month study period (r = −0.45, P = 0.03). The cohort consisted of 22 patients with ROT1D (age, 11.5 ± 3.3 years; HbA1c, 7.9 ± 1.5% at baseline). (B) Significant direct relationship exists between baseline percentages of CD14^dim CD16⁺ (nonclassical) monocytes and the rate (slope) of C-peptide decline over the 24-month study period (r = 0.46, P = 0.03). A truncated linear regression was used to estimate the slope of decrease for each individual. Larger negative slope values imply a faster decline of stimulated C-peptide. (C) Ratio of intermediate monocytes to nonclassical monocytes among the uHC, LRS, HRS, ROT1D, and LST1D cohorts illustrated in Fig. 6. (D) Significant inverse relationship exists between the intermediate-to-nonclassical monocyte ratio at baseline and 2-hour C-peptide AUC percent change from baseline at 6, 12, and 24 months after onset among patients with ROT1D. (E) Significant negative relationship exists between the intermediate-to-nonclassical monocyte ratio at baseline and the rate (slope) of C-peptide decline over the 24-month study period (r = −0.62, P = 0.002).

Fig. 9.. Temporal analysis of monocyte and T_reg populations in stage 0 HRS nonprogressors.

Twenty-two stage 0 HRS were analyzed at two time points. The mean age at the first sampling was 8.2 ± 3.5 years; the second sampling was conducted 5.4 ± 1.3 years later. (A) Percentages of KLRD1^highKLRB1^high monocytes among total, classical, intermediate, and nonclassical monocytes (left to right) at time points 1 and 2. (B) Percentages of KLRD1^highPRF1^high monocytes among total, classical, intermediate, and nonclassical monocytes (left to right) at time points 1 and 2. (C) Percentages of KLRD1^highGZMB^high monocytes among total, classical, intermediate, and nonclassical monocytes (left to right) at time points 1 and 2. (D) Representative flow cytometry profiles showing the gating strategy for resting and activated T_reg cells. Resting and activated CD4+ T_regs were defined as CD45RA⁺FoxP3^low and CD45RA⁻FoxP3^high, respectively. (E) Left to right: Abundance of activated T_reg among total T_reg at time points 1 and 2, ratio of KLRD1^highKLRB1^high monocytes among total monocytes: percentage activated T_reg among total T_reg, ratio of KLRD1^highPRF1^high monocytes among total monocytes: percentage activated T_reg among total T_reg at time points 1 and 2, and ratio of KLRD1^highGZMB^high monocytes among total monocytes:percentage activated T_reg among total T_reg at time points 1 and 2. Each dot represents a different individual. Horizontal bars indicate the mean ± SE. The differences between time points were assessed with a paired t test.