Adult mice are unresponsive to AAV8-Gremlin1 gene therapy targeting the liver

Abstract

Objective: Gremlin 1 (GREM1) is a secreted BMP2/4 inhibitor which regulates commitment and differentiation of human adipose precursor cells and prevents the browning effect of BMP4. GREM1 is an insulin antagonist and serum levels are high in type 2 diabetes (T2D). We here examined in vivo effects of AAV8 (Adeno-Associated Viral vectors of serotype eight) GREM 1 targeting the liver in mature mice to increase its systemic secretion and also, in a separate study, injected recombinant GREM 1 intraperitoneally. The objective was to characterize systemic effects of GREM 1 on insulin sensitivity, glucose tolerance, body weight, adipose cell browning and other local tissue effects.

Methods: Adult mice were injected with AAV8 vectors expressing GREM1 in the liver or receiving regular intra-peritoneal injections of recombinant GREM1 protein. The mice were fed with a low fat or high fat diet (HFD) and followed over time.

Results: Liver-targeted AAV8-GREM1 did not alter body weight, whole-body glucose and insulin tolerance, or adipose tissue gene expression. Although GREM1 protein accumulated in liver cells, GREM1 serum levels were not increased suggesting that it may not have been normally processed for secretion. Hepatic lipid accumulation, inflammation and fibrosis were also not changed. Repeated intraperitoneal rec-GREM1 injections for 5 weeks were also without effects on body weight and insulin sensitivity. UCP1 was slightly but significantly reduced in both white and brown adipose tissue but this was not of sufficient magnitude to alter body weight. We validated that recombinant GREM1 inhibited BMP4-induced pSMAD1/5/9 in murine cells in vitro, but saw no direct inhibitory effect on insulin signalling and pAkt (ser 473 and thr 308) activation.

Conclusion: GREM1 accumulates intracellularly when overexpressed in the liver cells of mature mice and is apparently not normally processed/secreted. However, also repeated intraperitoneal injections were without effects on body weight and insulin sensitivity and adipose tissue UCP1 levels were only marginally reduced. These results suggest that mature mice do not readily respond to GREMLIN 1 but treatment of murine cells with GREMLIN 1 protein in vitro validated its inhibitory effect on BMP4 signalling while insulin signalling was not altered.

Fig 1. Effect of AAV8-Gremlin 1 gene transfer on liver and BAT in 30 weeks old male mice (C57BL6/N) after 12 weeks of LFD followed by 18 weeks of HFD.

(A) Representative light microscopy image X10 from liver sections showing hepatic Gremlin 1 protein expression (n = 8 in each group). (B) qRT-PCR of AAV8-Gremlin1 mRNA/18S in liver (optimized sequence) and (C) endogenous Gremlin 1 in BAT. (D) Gremlin 1 levels in serum. (E) Representative light microscopy image X10 from liver sections showing no difference in lipid accumulation (n = 8 in each group). (F) qRT-PCR of inflammation and fibrosis markers (mRNA/18S). (G) Representative light microscopy image X20 from liver sections showing hepatic CD68 and Collagen 1 (n = 8 each group).

Fig 2. AAV8-Gremlin 1 mice on Low Fat Diet (LFD) or High Fat Diet (HFD) were similar to the control group and hepatic overexpression of Gremlin 1 did not alter metabolic response.

Male mice (C57BL6/N) were fed 12 weeks of LFD followed by 18 weeks of HFD, terminated at the age of 30 weeks. (A) Body weights week 1–12 (LFD) and weeks 12–24 (HFD), (B) food intake on LFD or (C) fasting glucose levels. (D) Insulin sensitivity test (ITT, LFD), (E) glucose tolerance test (GTT, LFD), (F) HFD ITT and (G) HFD GTT. Results are means ± S.E.M. n = 12 in each group.

Fig 3. Gremlin 1 IP-injections slightly improved glucose tolerance but had no effect on insulin sensitivity, pyruvate tolerance or whole-body inflammation in 38 weeks old male mice (C57BL6/N) fed 60% HFD.

(A) body weight, (B) food intake, (C) insulin sensitivity test (ITT), (D) hepatic gluconeogenesis measured as pyruvate tolerance test (PTT) (E) glucose tolerance test (GTT), (F) total area under the curve (AUC) and (G) insulin levels at 15 min of GTT. (H) Inflammation marker (SAA-3). Results are means ± S.E.M. n = 8 in each group. Statistical significance was determined using Student’s t test (two-tailed) or ANOVA as appropriate. *P<0.05.

Fig 4. Effect of long-term Gremlin 1 IP injections on WAT and BAT in 38 weeks old male mice (C57BL6/N) fed 60% HFD.

(A) WAT weight and (B) subcutaneous adipose cell size. (C and D) qRT-PCR of markers of being, inflammation and fibrosis in SubQ WAT. (E) Graph presenting the UCP1/β-Tubulin protein expression in SubQ WAT using Western blots and band intensities calculated with ImageJ (n = 8 in each group). (F) Representative light microscopy image X20 from SubQ sections showing UCP1 protein expression (n = 6 in each group). (G) Graph presenting the UCP1/β-Tubulin protein in BAT using Western blots and band intensities calculated with ImageJ (n = 4 in each group). (H) Representative light microscopy image X20 from BAT sections showing UCP1 protein expression (n = 4 in each group). Results are means ± S.E.M. Statistical significance was determined using Student’s t test (two-tailed). *P<0.05.

Fig 5. Kidney weight was slightly increased but morphology was not affected by Gremlin 1 IP injections and no effects were seen in liver morphology in 38 weeks old male mice (C57BL6/N) fed 60% HFD.

(A) Graph presenting kidney weight. (B) Representative light microscopy image X10 from kidney sections stained with Hematoxylin and Eosin stain (H&E) and Periodic Acid Schiff (PAS), (n = 6 each group). (C) Kidney hydroxy-proline content. (D) Representative light microscopy image from liver stained with H&E, F4/80 (10X for both staining), Oil red O and Picrosirius (20X for both staining) (n = 8 in each group). (E) Hepatic hydroxy-proline content. Results are means ± S.E.M. Statistical significance was determined using Student’s t test (two-tailed). *P<0.05.

Fig 6

rGremlin 1 inhibits BMP4-induced pSMAD activation but did not change expression of differentiation and inflammation markers in differentiated 3T3/L1 (A) and in C2/C12 cells (B). The cells were first incubated for 3h with rGremlin 1 followed by 1h BMP4 as shown. (A, B) The figures show representative Western blots using antibodies reactive to p-Smad(1,5,9) upper panels and total Smad(1,5,9) lower panels. (C) qRT-PCR of inflammation and fibrosis markers (mRNA/18S) in 3T3/L1 cells (n = 3 separate experiments).

Fig 7

rGremlin 1 did not inhibit insulin signaling in differentiated 3T3/L1 (A, C) or in C2C12 cells (B, D). The figures show representative western blots of cells incubated for 3 h and 24h as shown with/w/o rec-Gremlin1 (400ng/ml). Insulin stimulation was performed with 10nM insulin for 10 minutes. Blots present results with antibodies reactive to p-ser 473 Akt (A, B, E, F) and p-thr 308 Akt (C, D) (upper panels) and total Akt (lower panels). Effect of rGremlin 1 (with validated correct protein folding) on insulin signaling in human IHH liver cells (E) and differentiated 3T3-L1 (F) cells. Band intensities were calculated using ImageJ and plotted on the right as the ratio of p-Akt intensity normalized to total Akt (n = 3 separate experiments). Results are means ± S.E.M. Statistical significance was determined using Student’s t test (two-tailed). *P<0.05.