An enzyme that selectively S-nitrosylates proteins to regulate insulin signaling

Abstract

Acyl-coenzyme A (acyl-CoA) species are cofactors for numerous enzymes that acylate thousands of proteins. Here, we describe an enzyme that uses S-nitroso-CoA (SNO-CoA) as its cofactor to S-nitrosylate multiple proteins (SNO-CoA-assisted nitrosylase, SCAN). Separate domains in SCAN mediate SNO-CoA and substrate binding, allowing SCAN to selectively catalyze SNO transfer from SNO-CoA to SCAN to multiple protein targets, including the insulin receptor (INSR) and insulin receptor substrate 1 (IRS1). Insulin-stimulated S-nitrosylation of INSR/IRS1 by SCAN reduces insulin signaling physiologically, whereas increased SCAN activity in obesity causes INSR/IRS1 hypernitrosylation and insulin resistance. SCAN-deficient mice are thus protected from diabetes. In human skeletal muscle and adipose tissue, SCAN expression increases with body mass index and correlates with INSR S-nitrosylation. S-nitrosylation by SCAN/SNO-CoA thus defines a new enzyme class, a unique mode of receptor tyrosine kinase regulation, and a revised paradigm for NO function in physiology and disease.

Keywords: S-nitrosylation, nitric oxide, redox signaling, posttranslational modification, diabetes, nitrosylase, insulin receptor.

Figure 1.. Identification and characterization of a SNO-CoA-Assisted Nitrosyltransferase (SCAN).

(A) Purification of SNO-CoA-binding proteins from bovine liver using SNO-CoA-resin. SNO-thiopropyl- and CoA-resin were used as controls. Eight protein bands identified (including BLVRB) on silver-stained SDS-PAGE gel are indicated by gene name and arrows; n=3. (B) Interaction of SNO-CoA resin with endogenous BLVRB in HEK cell lysate (upper) or with recombinant BLVRB (lower), detected by Western blot. Other resins (amylose resin, thiopropyl Sepharose 6B, SNO-thiopropyl Sepharose 6B, glutathione (GSH)-agarose, glutathione-SNO (GSNO)–Sepharose 4B, CoA–agarose, Acetyl-CoA–agarose and Palmitoyl-Coenzyme A–agarose) were used as controls; n=2. (C) Proteins found in both the BLVRB-dependent nitrosoproteome and the BLVRB cytosolic interactome. (D) Endogenous S-nitrosylation of HO2 (SNO-HO2) in untargeted HEK cells (HEK-WT) and BLVRB-knockout HEK cells (HEK-BLVRB^−/−), each expressing eNOS. Results are from two independent cell lines. (E) In vitro S-nitrosylation of HO2 by BLVRB in the presence of increasing amounts of SNO-CoA. GST (Glutathione S-transferase) is used as control. (F&G) Quantification of SNO-HO2 (from D, N=4) and (from E, n=3). SNO-HO2 level is normalized to expression of HO2 (input). (H) Enzymatic activity of BLVRB to S-nitrosylate HO2; n=3. (I&J) Competition for SCAN binding to SNO-CoA resin by a fixed dose (50μM) of SNO-cysteamine, NADH or NADPH (I); and by varying amounts of NADPH (J); n=2. (K) Reduced binding of mutant SCAN (QTG/NAA) to SNO-CoA resin; n=2. (L) Endogenous SNO-HO2 in HEK-WT cells (WT), HEK-SCAN^−/− cells, wild-type SCAN-re-expressing HEK cells (WT-SCAN) and mutant SCAN-re-expressing HEK cells (QTG-NAA), respectively, each expressing eNOS. (M) Quantification of SNO-HO2 levels in L; n=4. (N) S-nitrosylation of HO2 by recombinant WT-SCAN and SCAN-QTG/NAA protein in vitro with increasing SNO-CoA. (O) Quantification of SNO-HO2 in N; n=5. (P) Identification of SNO sites within SCAN. Amount of SNO-SCAN in HEK cell lines overexpressing wild-type SCAN or three mutated SCAN forms (C109R, C188R or C109/188R); n=3. (Q) Amount of SNO-HO2 in HEK-WT versus four HEK-SCAN^−/− cell lines overexpressing empty vector, wild-type SCAN (SCAN-WT), SCAN-R35G, SCAN-QTG/NAA and SCAN-C109/188R respectively; n=3. (R) S-nitrosylation of HO2 by recombinant WT-SCAN and SCAN-C109/188R protein in vitro with increasing SNO-CoA. (S) Quantification of SNO-HO2 in R; n≥3. (T) Working model of ‘ping-pong’ mechanism utilized by SCAN. The samples of WT-SCAN in Figures O and S are the same. All results are presented as mean ± SD. Two-tailed Student’s t-test was used to detect significance in Fig. 1F. One-way ANOVA with Tukey post hoc was used to detect significance in Fig. 1G, 1M, 1O, 1P and 1S. *, p<0.05; **, p<0.01; and ****, p<0.0001. See also Figures S1 and S2, and Tables S1 and S2.

Figure 2.. SCAN mediates S-nitrosylation of INSRβ/IRS1 and insulin resistance on high fat diet (HFD).

(A) Expression of SCAN in the indicated organs from wild-type mice (SCAN^+/+) and SCAN-knockout mice (SCAN^−/−), vs p97 as loading control; n=2. (B) Expression of SCAN and iNOS in hindlimb skeletal muscle (lateral gastrocnemius) from WT mice fed chow vs HFD for 16-weeks (upper), and from 12-week-old WT vs mutant obese mice (ob/ob; lower). Data are shown for two independent mice, and actin is loading control; n=3 mice. (C&D) Blood glucose (C) and plasma insulin levels (D) in chow-fed vs 16-week HFD-fed SCAN^+/+ and SCAN^−/− male mice; n=15–27 overnight-fasted mice in C and n=7–10 5-hour-fasted mice in D, per group. (E) Insulin tolerance test. Blood glucose in 5 hour-fasted 16-week HFD-fed male SCAN^+/+ and SCAN^−/− mice, immediately before and at the indicated time points after injection of human insulin (1U/kg body weight, i.p.); n=22–27 mice per group. (F) Glucose tolerance test. Blood glucose levels in overnight-fasted 16-week HFD-fed male SCAN^+/+ and SCAN^−/− mice, immediately before and at the indicated time points after injection of glucose (2g/kg body weight, i.p.); n=16–23 mice per group. (G) Glucose uptake in ex vivo soleus muscles from HFD-fed SCAN^+/+ and SCAN^−/− mice, measured with the non-metabolizable glucose analog 2-deoxyglucose, in the absence or presence of insulin (12 nM); n=10 (5 female and 5 male mice). (H) S-nitrosylation of INSRβ (INSR) and IRS1 in skeletal muscle (lateral gastrocnemius) of 5 hour-fasted chow-fed or HFD-fed SCAN^+/+ and SCAN^−/− mice. (I) Quantification of H; n=4 mice per group. (J) SCAN increases SNO-CoA-induced in vitro S-nitrosylation of IRS1 (purified from IRS1-overexpressing HEK cells) compared to GST control (lower). (K) Quantification of J; n=3. (L) Effect of SCAN on insulin signaling activity markers. Phosphorylation of INSR(pTyr1162), IRS1(pTyr608), AKT(pSer473) and AS160(pThr642) in 5-hour fasted SCAN^+/+ and SCAN^−/− mice, 30 min and 60 min after insulin administration (1U/kg body weight, i.p.), or ‘-’ no insulin administration. Data are from two independent mice at each timepoint, and spliced blots for IRS1/p-IRS1 are indicated. All results are presented as mean ± SD. One-way ANOVA with Tukey post hoc was used to detect significance in Fig. 2C, 2D, 2G, 2I & 2K. ITT and GTT in Fig. 2E and 2F were analyzed by two-way (time × treatment) repeated measures analysis of variance followed by Sidak’s multiple comparisons test. *, p<0.05; **, p<0.01; ***, p<0.001; and ****, p<0.0001. See also Figures S3 and S4 and Table S4.

Figure 3.. Insulin induces S-nitrosylation of INSRβ/IRS1 to inhibit signal transduction.

(A) SNO-INSRβ and SNO-IRS1 in overnight-starved wild-type L6 cells (L6-WT) and SCAN-knockout L6 cells (L6-SCAN^−/−) at 1, 30, 60, 120 and 240 min after removal of insulin following 10-minute insulin treatment (100 nM). (B) Quantification of SNO-IRS1 in A, compared to total IRS1; n=3. (C) Quantification of SNO-INSRβ in A, compared to total INSRβ; n=3. (D) SNO-INSRβ and SNO-IRS1 in skeletal muscle (lateral gastrocnemius) from 5-hour fasted SCAN^+/+ and SCAN^−/− mice, 60 min after insulin administration (1U/kg body weight, i.p.) or after no insulin administration (−). (E) Quantification of D; n=3 mice per group. (F) Phosphorylation of INSRβ(pTyr1162), IRS1(pTyr608), AKT(pSer473) and AS160(pThr642) in overnight-starved L6-WT and L6-SCAN^−/− cells at 1, 30, 60, 120 and 240 min after removal of insulin, following 10-minute insulin treatment (100 nM); n=4, quantitation shown in Figures S5I–L. (G&H) Normal (G) and severe (H) insulin tolerance test. Blood glucose level in 5-hour fasted chow-fed SCAN^+/+ and SCAN^−/− male mice, immediately before and at the indicated time points after injection of human insulin (1U/kg body weight in panel G; 2.5U/kg body weight in panel H, i.p.); n=10 mice in panels G and H. (I) SNO-INSRβ in skeletal muscle from 2 representative mice during severe insulin challenge as in H. (J) Quantification of I (n=4 mice per condition). All results are presented as mean ± SD. ANOVA with Tukey post hoc was used to detect significance in Fig.3E and 3J. Fig. 3B, 3C, 3G and 3H were analyzed by two-way (time × treatment) repeated measures analysis of variance followed by Sidak’s multiple comparisons test. *, p<0.05, **, p<0.01, ***, p<0.001 and ****, p<0.0001. See also Figure S5.

Figure 4.. Insulin-stimulated S-nitrosylation of INSRβ/IRS1 is coupled to NOS activity.

(A) Phosphorylation of eNOS(pS1177) in skeletal muscle from 5-hour fasted C57BL/6 mice, 10 min after insulin administration (1U/kg body weight, i.p.). (B) Quantification of A; n=5 mice. (C) Phosphorylation of nNOS(pS1412) in skeletal muscle from 5-hour fasted C57BL/6 mice, 10 min after insulin administration (1U/kg body weight, i.p.). (D) Quantification of C; n=4 mice. (E) Phosphorylation of eNOS(pS1177) in overnight serum-starved L6 cells at the indicated time after removal of insulin, following 10-minute insulin treatment (100 nM). (F) Quantification of E; n=3. (G) Amounts of SNO-INSRβ, SNO-IRS1 and SNO-SCAN in overnight-starved PBS-treated (control) and L-NMMA-treated (100 μM) L6 cells at 1, 30, 60, 120 and 240 min after removal of insulin following 10-minute insulin treatment (100 nM). (H-J) Quantification of SNO-IRS1 (H), SNO-INSRβ (I) and SNO-SCAN (J) from G, respectively; n=3. (K) Phosphorylation of AKT(pSer473) and AS160(pThr642) in PBS-treated (control) and L-NMMA-treated (100 μM) L6 cells at 1, 30, 60, 120 and 240 min after removal of insulin following 10-minute insulin treatment (100 nM); n=3, quantitation shown in Figures S5P–Q. Two-tailed Student’s t-test was used to detect significance in Fig. 4B and 4D. One-way ANOVA with Tukey post hoc was used to detect significance in Fig. 4F. Fig. 4H–4J were analyzed by two-way (time × treatment) repeated measures analysis of variance followed by Sidak’s multiple comparisons test. *, p<0.05; **, p<0.01; ***, p<0.001; and ****, p<0.0001.

Figure 5.. S-nitrosylation of INSRβ by SNO-CoA-dependent SCAN activity regulates insulin signaling.

(A) SNO-INSRβ and SNO-IRS1 in L6-SCAN-WT, L6-SCAN-QTG/NAA and L6-C109/188R cell lines, respectively. (B) Quantification of SNO-INSRβ and SNO-IRS1 in A; n=3. (C) Phosphorylation of IRS1(pTyr608) and AKT(pSer473) in overnight serum-starved L6-SCAN-WT, L6-SCAN-QTG/NAA and L6-C109/188R cell lines, 30 min after a 10-minute insulin treatment (100 nM). (D) Quantification of phosphorylation level of IRS1 and AKT in C; n=3. (E) Four peptides containing single candidate SNO sites (cysteine residues, red) within INSR identified by SNO-RAC-coupled mass spectroscopy. (F) Identification of primary SNO site within INSRβ. SNO-INSRβ with the indicated mutations of candidate SNO sites expressed in HEK cells. (G) Quantification of SNO-INSRβ in F; n=3. (H) Phosphorylation of AKT(pSer473) and AS160(pThr642) in overnight serum-starved INSR-WT and INSR-C1083A expressing L6 cells at 1, 30, 60, 120 and 240 min after removal of insulin, following a 10-minute insulin treatment (100 nM). (I-J) Quantification of phosphorylation of AKT(pSer473) (I) and AS160(pThr642) (J) in H, respectively; n=3. All results are presented as mean ± SD. Two-tailed Student’s t-test was used to detect significance in Fig. 5B and 5D. One-way ANOVA with Tukey post hoc was used to detect significance in Fig. 5G, 5I and 5J. *, p<0.05 and **, p<0.01. See also Figure S6.

Figure 6.. S-nitrosylation of INSRβ is associated with BMI and SCAN expression in human adipose tissue and skeletal muscle.

(A) Expression of SCAN in 14 human subcutaneous adipose samples (Hsad#) with indicated BMI, and quantification (lower panel). GAPDH is used as internal loading control. Expression of SCAN was first normalized with expression of internal control GAPDH, and then versus the average expression level of SCAN in the same group of samples. (B-C) Expression of SCAN in 14 human skeletal muscle samples (B, see Fig. S6E) and 28 human adipose tissue samples (C, see Fig. S6F), plotted against patient BMI. (D) SNO-INSRβ in 14 human subcutaneous adipose samples from patients with the indicated different BMI, and quantification (lower). SNO-INSRβ was first normalized with input of INSRβ, and versus the average SNO-INSRβ level in the same group of samples. (E-F) SNO-INSRβ in 14 human skeletal muscle samples (E, see Fig. S6G) and 26 human adipose tissue samples (F, see Fig. S6H), plotted against patient BMI. (G) SNO-INSRβ in 14 human skeletal muscle samples is plotted against SCAN expression level. (H) SNO-INSRβ in 26 human adipose tissue samples is plotted against SCAN expression level. (I) S-nitrosylation-based inhibition of insulin signaling and insulin resistance. Under healthy conditions, insulin induces S-nitrosylation of INSRβ/IRS1 via eNOS/nNOS-coupled SCAN/SNO-CoA activity, promoting termination of insulin signaling. In obesity, pro-inflammatory cytokines and free fat acids (FFAs) induce sustained S-nitrosylation of INSRβ/IRS1 through iNOS-coupled SCAN/SNO-CoA activity, leading to insulin resistance. Simple linear regression was performed to identify the relationships among BMI, SCAN expression level, and SNO-INSRβ level (using blot data in A, D and Fig. S6E–H). R-Squared (r²) and slope significance p values show the goodness of fit of the regression model. See also Figure S6 and Tables S3 and S4.