Glycogen synthase kinase 3 inhibition controls Mycobacterium tuberculosis infection

Abstract

Compounds targeting host control of infectious diseases provide an attractive alternative to antimicrobials. A phenotypic screen of a kinase library identified compounds targeting glycogen synthase kinase 3 as potent inhibitors of Mycobacterium tuberculosis (Mtb) intracellular growth in the human THP-1 cell line and primary human monocytes-derived macrophages (hMDM). CRISPR knockouts and siRNA silencing showed that GSK3 isoforms are needed for the growth of Mtb and that a selected compound, P-4423632 targets GSK3β. GSK3 inhibition was associated with macrophage apoptosis governed by the Mtb secreted protein tyrosine phosphatase A (PtpA). Phospho-proteome analysis of macrophages response to infection revealed a wide array of host signaling and apoptosis pathways controlled by GSK3 and targeted by P-4423632. P-4423632 was additionally found to be active against other intracellular pathogens. Our findings strengthen the notion that targeting host signaling to promote the infected cell's innate antimicrobial capacity is a feasible and attractive host-directed therapy approach.

Keywords: Medical Microbiology; Microbiology; Molecular biology.

Graphical abstract

Figure 1

Intracellular screening of kinase inhibitor libraries identified GSK3β as a key host controller of Mtb infection (A) Screening of 313 unique compounds from the PKIS/UNC library at 10 μM concentration against THP-1 cells infected with RFP-expressing intracellular Mtb, using the CellInsight CX5 HCS platform (CX5). Z’ = 0.58 ± 0.12 across 8 plates. Mtb % inhibition, calculated based on the intracellular fluorescence intensity dual normalized to the BDQ positive control (100% inhibition) and the negative DMSO vehicle control (0% inhibition), was plotted as a function of THP-1 cell survival as measured by DAPI-stained nuclei cell counts compared to the negative control. Green circles in upper right quadrant represent compounds with greater than 20% Mtb inhibition and 70% THP-1 survival. (B) Distribution of GSK3β inhibitors based on their intracellular inhibition of Mtb. GSK3 inhibitors from the above library were rescreened against luciferase-expressing Mtb in THP-1 cells using a luciferase assay. Mtb % inhibition was calculated based on the relative luminescence, dual normalized to the Rifampicin positive control (100% inhibition) and the negative DMSO control (0% inhibition). Over half of the tested compounds (22/42) demonstrated at least 20% inhibition of Mtb intracellular growth. Z’ = 0.56. (C) Screening of a GSK3β focused library of compounds (Takeda) at 10 μM against luciferase-expressing Mtb in THP-1 macrophages using a luciferase assay. Mtb % inhibition (as calculated in B) was plotted as a function of THP-1 cell survival as determined by MTT assay. Green circles in upper right quadrant represent compounds with greater than 50% Mtb inhibition and 70% THP-1 survival; compound P-4423632 circled. (D–G) Dose-dependent inhibitory activity of four select GSK3 inhibitors against intracellular Mtb and the corresponding compound structures (inset). THP-1 cells infected with luciferase-expressing Mtb were treated with 2-fold serial dilutions of the indicated compounds. Mtb % inhibition was calculated as in (B). Non-linear regression (variable slope) was used to fit the data of the log (inhibitor) vs. response (±SD) using GraphPad Prism software; N = 3.

Figure 2

Genetic validation of GSK3’s role in restricting Mtb intracellular growth (A) Inhibition of intracellular Mtb by downregulating GSK3 isoforms using siRNA. GSK3α, GSK3β or both were knocked down in THP-1 cells, followed by infection with Mtb. Inhibition % represents the % of Mtb fluorescence area and relative luminescence compared to BDQ control (100% inhibition) and normalized to cells transfected with scrambled siRNA (0% inhibition). Data represent mean ± SEM of 3 biological replicates using high-content analysis (CX5) and luciferase assay; statistics were performed using one-way ANOVA followed by Bonferroni’s post hoc test compared to the scrambled siRNA control (not shown); ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. (B) Confirmation using qPCR of siRNA knockdown of GSK3α and GSK3β RNA levels in THP-1cells following transfection with GSK3α siRNA (red), GSK3β siRNA (blue), and both GSK3α and GSK3β siRNA (purple); siRNA transfection is indicated. Fold expression represent the GSK3 variant expression levels in cells transfected with the indicated GSK3 siRNA compared to transfection with scrambled siRNA control, defined at 1 (not shown). Data represent the mean +SD of a representative experiment. N = 4. Statistics were performed using two-way ANOVA followed by Bonferroni’s post hoc test compared to the scrambled siRNA control; ∗∗∗p < 0.001. (C) Mtb infection of CRISPR-inactivated GSK3α (red) or GSK3β (blue) in THP-1 cells showing the effect of gene disruption with and without 72h P-4423632 treatment. Parental, GSK3α and GSK3β knockout THP-1 cells were infected with RFP-expressing Mtb and treated with 10 μM P-4423632 (right group) or DMSO control (Untreated, left group). Data represent high-content analysis of intracellular fluorescence area dual normalized to untreated (DMSO vehicle control) parental THP-1 cells (defined as 100% Mtb survival) and BDQ-treated parental cells (0% survival). Knocking out GSK3α and GSK3β in THP-1 cells were able to inhibit intracellular survival of Mtb with the greatest effect seen by blocking both GSK3α (by knockout) and GSK3β (by chemical inhibition). Data represent mean ± SEM of 3 biological replicates. Statistics were performed using two-way ANOVA followed by Bonferroni’s post hoc test. ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001. (D) Contribution of P-4423632 inhibition of Mtb in parental, GSK3α and GSK3β backgrounds. Treating GSK3α KO cells with 10 μM P-4423632 further inhibited intracellular survival of Mtb by over 75% of its untreated counterpart, similar to the reduction observed in parental cells; whereas addition of P-4423632 to GSK3β KO cells had less of an effect (∼50% reduction). Reduction % was calculated as 100%-treated/untreated data from (C) for each cell type. Statistics were performed using one-way ANOVA. ∗p < 0.05; ∗∗p < 0.01.

Figure 3

Inhibition of Mtb growth in primary macrophages and modulation of apoptosis (A) Dose-response curve of P-4423632 at 72 h post treatment of hMDMs infected with Mtb (dark red circles) and hMDM % survival (black squares). Data were acquired using the OPERA Phoenix High-Content microscope and analyzed using the Harmony software. Mtb inhibition at 72 h was calculated in relations to non-treated Mtb reflecting the negative % difference of intracellular Mtb fluorescence area compared to the DMSO vehicle control, normalized to the fluorescence area at 2h post infection. The % survival of hMDMs was calculated based on the nuclear count (DAPI stain) of cells compared to the DMSO control. Data represent 3 biological replicates ±SEM of mixed donors of the average of quadruplicate technical repeats. N = 3. Non-linear regression was used to fit the data of the log (inhibitor) vs. response (variable slope) curve using GraphPad Prism 10 analysis software. Shaded area represents the 95% confidence bands of the true curve. (B) Decreased THP-1 apoptosis during Mtb infection is associated with PtpA. THP-1 cells were infected with Mtb WT (dark red) or Mtb ΔptpA (dark blue) at MOI of 6. Apoptotic activity was determined using the AUTOptosis method by monitoring chromatin condensation using DAPI staining. Data were normalized to uninfected cells and are a representative of three separate experiments ±SD; N = 6, ∗∗p values of 0.0022 performed by Mann Whitney non-parametrical test. (C) GSK3β inhibitor increases apoptotic activity in THP-1 cells. THP-1 cells infected with Mtb WT or Mtb ΔptpA at an MOI of 3, with or without treatment with 20 μM P-4423632 and harvested at 48 h post-infection. Apoptotic activity was determined using Annexin V FITC assay. Statistical differences between treated and untreated groups were analyzed by two-way ANOVA followed by Bonferroni’s post hoc test; N = 2, ∗∗∗p < 0.001. (D) CRISPR-KO of GSK3β in THP-1 cells (blue) increased apoptosis in response to infection with Mtb compared to parental THP-1 cells (gray) in relation to non-infected cells. Apoptotic activity ±SD was measured using the AUTOptosis method; N = 6, ∗∗p values of 0.0043 performed by Mann Whitney non-parametrical test. (E and F) Inhibitory effect of GSK3β inhibitor, P-4423632, in PtpA knockout background. (E) Dose dependent activity of P-4423632 against intracellular Mtb (dark red squares) compared to the ΔptpA mutant (dark blue circles), determined by high content analysis of fluorescent Mtb. Mtb % inhibition was calculated by dual normalization to the positive control and negative control ±SD. Non-linear regression (variable slope) was used to fit the data of the log (inhibitor) vs. response using GraphPad Prism software. (F) CFU counts of THP-1 cells infected with WT Mtb or ΔptpA mutant with or without 96 h of 10 μM P-4423632 treatment. Data are a representative of two biological experiments performed in triplicate ±SD, N = 3. Statistical differences between treated and untreated, and WT and mutant were analyzed by two-way ANOVA followed by Bonferroni’s post hoc test; ∗∗∗p < 0.001, ∗∗p < 0.01, ∗p < 0.05.

Figure 4

Modulation of macrophage phosphorylation upon infection and treatment with P-4423632 using antibody microarray analysis (A and B) Volcano plots showing log₂ fold change in normalized signal intensity and log₁₀p values for modulated protein phosphorylation sites after (A) infection with Mtb and (B) infection with Mtb and treatment with P-4423632. Red and blue labels show significantly up-regulated and down-regulated phosphosites, respectively. Green labels show phosphosites whose phosphorylation levels in relation to infection were noticeably modulated by P-4423632. p values were adjusted using the Benjamini-Hochberg method. (C) Scatterplot showing the ratio of log₂ fold change in phosphorylation after infection or infection plus treatment with P-4423632. p values were calculated using an empirical Bayesian variance estimate derived from Cyber-T method. The solid blue line corresponds to a one-to-one ratio of log₂ fold change while the red dotted lines represent the 97.5% quantiles of the data. The 25 proteins with the largest difference in fold change (distance to the blue line) are labeled in black. (D) Heatmap showing log₂ fold change in relation to non-infected macrophages control. The 25 most modulated phospho-sites between infection and drug treatment were monitored in the labeled samples: P = parental THP-1 cells, KO = GSK3β CRISPR knockout in THP-1 cells, WT = infected with WT Mtb, ΔptpA = infected with ΔptpA mutant of Mtb. Row labels include the protein name and the affected phosphorylation site. (E) Network diagram 16 highly modulated proteins from (C) belonging to the GSK3β (blue lines), phagocytosis (green) lines, and apoptosis (red lines), related pathways. Pathways are defined by KEGG with GSK3β related disease specific pathways filtered out. Node and point size in the diagram reflect either the number of pathway groups a protein is present in or the number of proteins present in the pathway group. Visualization and statistical tests for antibody microarray analyses (A–E) were performed using CAT PETR. (F) Western blot analysis of Y279 - Y216 phosphorylation status of GSK3α and GSK3β cells infected with Mtb. P = parental THP-1 cells, KO = GSK3β CRISPR knockout in THP-1 cells, WT = infected with H37Rv WT Mtb, ΔptpA = infected with ΔptpA mutant of Mtb. Hsp27 included as loading control.