Ret kinase-mediated mechanical induction of colon stem cells by tumor growth pressure stimulates cancer progression in vivo

Abstract

How mechanical stress actively impacts the physiology and pathophysiology of cells and tissues is little investigated in vivo. The colon is constantly submitted to multi-frequency spontaneous pulsatile mechanical waves, which highest frequency functions, of 2 s period, remain poorly understood. Here we find in vivo that high frequency pulsatile mechanical stresses maintain the physiological level of mice colon stem cells (SC) through the mechanosensitive Ret kinase. When permanently stimulated by a magnetic mimicking-tumor growth analogue pressure, we find that SC levels pathologically increase and undergo mechanically induced hyperproliferation and tumorigenic transformation. To mimic the high frequency pulsatile mechanical waves, we used a generator of pulsed magnetic force stimulation in colonic tissues pre-magnetized with ultra-magnetic liposomes. We observed the pulsatile stresses using last generation ultra-wave dynamical high-resolution imaging. Finally, we find that the specific pharmacological inhibition of Ret mechanical activation induces the regression of spontaneous formation of SC, of CSC markers, and of spontaneous sporadic tumorigenesis in Apc mutated mice colons. Consistently, in human colon cancer tissues, Ret activation in epithelial cells increases with tumor grade, and partially decreases in leaking invasive carcinoma. High frequency pulsatile physiological mechanical stresses thus constitute a new niche that Ret-dependently fuels mice colon physiological SC level. This process is pathologically over-activated in the presence of permanent pressure due to the growth of tumors initiated by pre-existing genetic alteration, leading to mechanotransductive self-enhanced tumor progression in vivo, and repressed by pharmacological inhibition of Ret.

Fig. 1. High frequency pulsatile mechanical stresses maintain Lgr5+ stem cells physiological rate in colon crypts, in vivo.

a Visualization of Lgr5+ SC in the colon of Lgr5-EGFP mice after WIN treatment for 5 days. The number of SC Lgr5-EGFP positive cells per crypt was calculated with the images taken with a ×40 objective and the percentage of Lgr5-EGFP positive crypts in the colon with images obtained with a ×10 objective. Scale bar is 10 μm. b Quantification of a. Mean number of Lgr5-EGFP positive cells per crypt multiplied by the percentage of positive crypts in the colon. Mann–Whitney test; **p < 0.01, ns not significant. c Visualization of endogenous colonic pulsatile waves using ultrasound. Experimental set-up and timeline of the experiment. Heat maps represent the space- and time-dependent axial displacement along the colonic wall in WT mice colon before and after injection of WIN. d Quantification of the pulsatile activity in WT mice colon before and after injection of WIN. (n = 4 mice, three acquisitions were analyzed for each mouse and for each condition, before and after drug injection). The WIN acquisitions were chosen in a time window between 10 and 25 min after drug injection. Mann–Whitney test; ****p < 0.0001. e Pulsated magnetic field gradient set-up drawing. The force F(t) induced by the 2 cm magnets of 1Tesla magnetization oscillate between the force F generated at z = 2.5 cm and F = 0 at z = −20 cm below the colon, mimicking endogenous pulsatile deformations (see text). The mouse drawing was imported from reference. f Up, ultrasound imaging of mouse colon deformation induced by magnetic stimulation. Representative B-Mode acoustic image of a non-injected colon sample with sur-imposed color-coded 2D map of the maximum displacement amplitude compared to magnetically loaded colon explant after magnetic stimulation. White lines represent colon walls. Down, quantification of the maximum displacement amplitude in WT colon samples injected or non-injected with magnetic liposomes after magnetic stimulation. Mean of the maximum displacement amplitude in magnetized colon explants = 3 ± 1.8 µm (n = 4 mice with three measurements per mouse). Mean of the maximum displacement amplitude in control colon explants = 1 ± 0.3 µm (n = 4 mice with three measurements per mouse). Mann–Whitney test, ***p < 0.001. Note that the p-value between control and Stb-UML is still significant even if we do not take into account the three higher measurements. g Lgr5+ SC in the colon after application of a pulsated magnetic compression mimicking high-frequency colon pulsatile movements for 5 days in the Lgr5-EGFP mouse model. Control: mice injected with UML only without magnet implantation; Stb-UML: mice injected with UML and subjected to 30 min of magnet implantation to stabilize the UML; veh: injected with the vehicle of WIN; WIN: injected with WIN; Lgr5+ cells have been detected with an anti-GFP antibody. N = 3 experiments. Scale bar is 10 μm. h Quantification of g. Mean number of Lgr5-EGFP + SC per crypt multiplied by the percentage of positive crypts per mouse in the colon. Mann–Whitney test; *p < 0.05; **p < 0.01; ns: not significant. Error bars: standard deviation, except for d and f in which it represents the minimum to maximum data values (excluding one outlier for d-WIN - the outlier being taken into consideration in the p-value evaluation) and for which the box represents the difference between the 75th and 25th percentile.

Fig. 2. The mechanotransductive stimulation of pRet drives high frequency pulsatile colonic stresses maintenance of homeostatic Lgr5+ SC number, in vivo.

a Levels of Y1062 Ret kinase phosphorylation after colon magnetization with UML in WT mice, veh: injected with the vehicle of WIN, WIN: implemented with WIN, + Stb-UML: previous conditions with UML stabilized in the colon in the presence (+ pulsed stresses) or in the absence (w/o pulsed stresses) of the pulsated magnetic field designed to produce deformations of pulsatile movements amplitude and frequency (see Fig. 1f), for 48 h. White arrows show crypts with a pRet signal of 1–3 pRet positive cells/crypt, n = 3 experiments. Scale bar is 10 μm. b Quantification of a. c Lgr5+ SC in the colon after application of a pulsated magnetic compression mimicking high frequency pulsatile colonic stresses for 5 days in the Lgr5-EGFP mouse model, with and without Vande treatment. Control: mice injected with UML only without magnet implantation; Stb-UML: mice injected with UML and subjected to 30 min of magnet implantation to stabilize the UML; veh: injected with the vehicle of WIN; WIN: injected with WIN; vehV: implemented with the vehicle of Vande; Vande: treated with Vande. Lgr5+ cells have been detected with an anti-GFP antibody. N = 3 experiments. Scale bar is 10 μm. d Quantification of c. Mean number of Lgr5-EGFP + SC per crypt multiplied by the percentage of positive crypts per mouse in the colon. Mann–Whitney test; *p < 0.05 **p < 0.01; ***p < 0.001; ns not significant. Error bars: standard deviation.

Fig. 3. Permanent tumorous mechanical stresses over-stimulate Lgr5+ SC number and CD133 CSC production, in vivo.

a Lgr5+ SC in Lgr5-EGFP mice after application of a permanent magnetic compression mimicking tumor growth pressure for 1 month. Control: mice without UML injection (n = 7 mice); UML: mice injected with UML without magnet implantation (n = 8 mice); UML Magnet: mice injected with UML plus permanent magnet implantation for 1 month (n = 6 mice). Scale bar is 10 μm. b Lgr5+ SC in Apc;Lgr5-EGFP mice after application of a permanent magnetic compression mimicking tumor growth pressure for 1 month. Control: mice without UML injection (n = 5 mice); UML: mice injected with UML without magnet implantation (n = 2 mice); UML Magnet: mice injected with UML plus permanent magnet implantation for 1 month (n = 5 mice). Scale bar is 10 μm. c Quantification of a. Mean number of Lgr5-EGFP + SC per crypt and per mouse. Mann–Whitney test; **p < 0.01. d Quantitative analysis of b. Mean number of Lgr5-EGFP + SC per crypt and per mouse. Mann–Whitney test: **p < 0.01; ns not significant. e CD133 + CSC after application of a permanent magnetic compression mimicking tumor growth pressure for 1 month in Apc;Lgr5-EGFP mice. Controls: 4 and 9 month-old mice (n = 5 mice); Control 5 month-old mice having gastric tumors: small intestine and colon (n = 3 mice); UML Magnet 1 m: mice injected with UML plus permanent magnet implantation for 1 month (n = 5 mice). Red arrows show CD133 + crypts. Small white frames show similar images with Lgr5-EGFP and nuclei staining. Scale bar is 10 μm. f Quantitative analysis of e. Percentage of CD133 + crypts per mouse. Mann–Whitney test; *p < 0.05. Error bars: standard deviation.

Fig. 4. The mechanical stimulation of SC and CSC multiplication is pRet dependent in Apc heterozygous mice colon, in vivo.

a Levels of pRet+ crypts (up), Lgr5+ SC (middle), and CD133 + CSC (down) in Apc;Lgr5-EGFP mice after application of a permanent magnetic compression mimicking tumor growth pressure for 1 month, with and without Vande. Control Veh: mice without UML injection treated for 1 month with vehicle of Vande (n = 8 mice); UML + Magnet 1 m Veh: mice injected with UML plus permanent magnet implantation and treated with vehicle for 1 month (n = 9 mice); Control Vande: mice without UML injection treated for 1 month with Vande (n = 6 mice); UML + Magnet 1 m Vande: mice injected with UML plus permanent magnet implantation and treated with Vande for 1 month (n = 8 mice). N = 2 experiments. Red arrows show CD133 + stained cells. Small white frames show similar images with Lgr5-EGFP and nuclei stainings. Scale bar is 10 μm. b Quantitative analysis of the pRet signal. Percentage of crypts with ≥4 pRet positive cells per mouse. Mann–Whitney test; **p < 0.01 and ****p < 0.0001. c Quantitative analysis of Lgr5-EGFP signal. Mean number of Lgr5+ SC per crypt and per mouse. Mann–Whitney test two-tailed; **p < 0.01. d Quantitative analysis of CD133 signal. Percentage of CD133 + crypts per mouse. Mann–Whitney test; ***p < 0.001. e CD133 + cancer cell marker in Apc and Apc;N1Cre-ERT2 old mice (16 month-old). CD133 antibody staining in colon crypts of mice treated with vehicle (n = 6 mice) and Vande (n = 7 mice). Red arrows show CD133 + stained cells. Scale bar is 10 μm. (f), Quantification analysis of e. Percentage of CD133 + crypts per mouse. Mann–Whitney test. *p<0.05. Error bars: standard deviation.

Fig. 5. Paneth niche cells of the intestine are mechanically induced in Apc heterozygous mice colon in a Ret mechanosensitive dependent process.

a Lysozyme staining of the colon tissue after application of a permanent magnetic compression mimicking tumor growth pressure for 1 month (n = 8 mice) and 3.5 months (n = 4 mice), compared to control (n = 8 mice) in Apc mice. Small boxes display lysozyme positive signal in red to increase contrast with DAPI in blue. Scale bar is 10 μm. b Lysozyme staining of the colon tissue after application of a permanent magnetic compression mimicking tumor growth pressure for 1 month, treated with the pRet inhibitor Vande in Apc;Lgr5-EGFP mice. Control Veh: mice treated with the vehicle of Vande (n = 8 mice); UML Magnet 1 m Veh: mice injected with UML plus permanent magnet implantation for 1 month and vehicle treated (n = 8 mice); Control Vande: mice treated with Vande for 1 month (n = 6 mice); UML Magnet 1 m Vande: mice injected with UML plus permanent magnet implantation for 1 month and Vande treated (n = 9 mice). Small boxes display lysozyme positive signal in red to increase contrast with DAPI in blue. c Quantification of a. Mean number of lysozyme positive crypts per mice. Mann–Whitney test; *p < 0.05; **p < 0.01. d Quantification of b. Mean number of lysozyme positive crypts per mice. Mann–Whitney test; **p < 0.01; ns not significant. Error bars: standard deviation.

Fig. 6. The pharmacological inhibition of the Ret/β-cat pathway with Vande inhibits mechanical induction of, and spontaneous tumorigenesis initiation in vivo.

a Live imaging of ACFs (orange narrow) at different times of the experiment subjected to the application of a permanent magnetic compression mimicking tumor growth pressure for 1 month, with or without Vande treatment in Apc;Lgr5-EGFP mice. b ACFs number counting at different times until 28 days of the application of permanent magnetic compression mimicking tumor growth pressure application on the colon of Apc;Lgr5-EGFP mice with or without Vande treatment, n = 6–9 mice/condition. N = 2 experiments. Statistical significance determined using the Holm-Sidak method, with alpha = 0.05. Adjusted p (D14, UMLm veh vs. veh) <0.05, adjusted p (D28, UMLm veh vs. veh) <0.01, adjusted p (D14, UMLm vande vs. UMLm veh) <0.05, adjusted p (D28, UMLm vande vs. UMLm veh) <0.01. c Live imaging of ACF (orange arrows) at 28^th day of vehicle or Vande treatment of Apc old mice. d ACF number counting at different times during the 1 month of vehicle or Vande treatment, n = 13–14 mice/condition. N = 3 experiments. Statistical significance determined using the Holm-Sidak method, with alpha = 0.05. Adjusted p (D28) <0.01. e Intestine upper section pictures after 1 month of vehicle or Vande treatment. Intestine tumors are surrounded by a red circle. f Quantification of e. N = 7 mice/condition. Statistical significance determined using Mann–Whitney test. Error bars: standard deviation, except for d, f in which it is standard error to the mean.

Fig. 7. pY1062-Ret is activated in human colon, and other human solid tumors.

a Y1062 phosphorylation of Ret (pRet) in the colon of WT (N=10), dysplastic tubulous adenoma (N = 10), early invasive intra-mucosal adenocarcinoma (N = 10) and invasive (sub-mucosa and beyond) (N = 10) adenocarcinoma of variable grade (N = 10). b Level of activation of pRet as a function of the tumorous grade (N = 10 by stage). Error bars: standard deviation. c pRet in adenocarcinoma invasive primary tumors of pancreas (N = 71), ovary (N = 101), lung (N = 31), endometrium (N = 31), prostate (N = 31), uveal (N = 61), head and neck (N = 61), breast (N = 61), and cervix malignant (N = 51) tumors. All WT controls are N = 10 (see adapted statistical analysis procedure in methods). Numbers correspond to the phosphorylation level order of magnitude in its predominant subcellular location. Scores in black: p < 0.01, in yellow: p = 0.25 tendency, in red: data heterogeneity prevents statistical analysis.