USP8 promotes intracellular infection by enhancing ESCRT-mediated membrane repair, limiting xenophagy, and reducing oxidative stress

Abstract

The host ESCRT-machinery repairs damaged endolysosomal membranes. If damage persists, selective macroautophagy/autophagy clears the damaged compartment. Mycobacterium tuberculosis (Mtb) is an intracellular pathogen that damages the phagosomal membrane and targets ESCRT-mediated repair as part of its virulence program. The E3 ubiquitin ligases PRKN and SMURF1 promote autophagic capture of damaged, Mtb-containing phagosomes. Because ubiquitination is a reversible process, we anticipated that host deubiquitinases (DUBs) would also be involved. Here, we screened all predicted mouse DUBs for their role in ubiquitin targeting and control of intracellular Mtb. We show that USP8 (ubiquitin specific peptidase 8) colocalizes with intracellular Mtb, recognizes phagosomal membrane damage, and is required for ESCRT-dependent membrane repair. Furthermore, we show that USP8 regulates the NFE2L2/NRF2-dependent antioxidant signature. Taken together, our study demonstrates a central role of USP8 in promoting Mtb intracellular growth by promoting phagosomal membrane repair, limiting ubiquitin-driven selective autophagy, and reducing oxidative stress.Abbreviation: BMDMs: bone marrow-derived macrophages; CFUs: colony-forming units; DUB: deubiquitinase; ESCRT: endosomal sorting complexes required for transport; LLOMe: L-leucyl-L-leucine methyl ester; MFI: mean fluorescence intensity; MOI: multiplicity of infection; Mtb: Mycobacterium tuberculosis; NFE2L2/NRF2: nuclear factor, erythroid derived 2, like 2; PMA: phorbol 12-myristate 13-acetate; ROS: reactive oxygen species; USP8: ubiquitin specific peptidase 8.

Keywords: Autophagy; Mycobacterium tuberculosis; deubiquitinase; endomembrane damage; oxidative stress.

Figure 1.

RNAi screen identifies DUBs that regulate ubiquitination and intracellular survival of Mtb. (A) schematic of imaging screen. DUBs were silenced in atg5 cKO BMDMs. Non-targeting scrambled (scr) and Tsg101 siRNA were controls. The next day, BMDMs were activated with IFNG, followed by infection with DsRed-expressing Mtb (MOI 5). 24 hpi cells were immunostained for ubiquitin. (B) heat maps show Z-scores of the mean fluorescence intensity (MFI) of ubiquitin around Mtb in the indicated conditions. Data is average from 100-4000 bacteria from 4 independent experiments each using a different siRNA targeting the DUB. Genes in each group were tested and analyzed together. (C) images show Mtb (red) and ubiquitin (green) in BMDMs transfected with indicated siRNA. White boxed areas are magnified. Magnified panel in the Usp8 k/d condition shows examples of Mtb outside or in the periphery of swollen vesicles. 20X images were acquired with automated high content scanning (see methods for details). Scale: 10 µm. (D) schematic of intracellular growth screen. DUBs were depleted in WT BMDMs. Scr siRNA was a negative control. Two days after silencing, BMDMs were infected with Mtb (MOI 5), and colony-forming units (CFUs) were enumerated 4 hpi (D0) and 72 hpi (D3). (E) heat map shows the average of the D3/D0 CFU ratio from the indicated samples. Data are from 2 independent experiments with duplicates. (B, E) *p ≤ 0.04, **p ≤ 0.008, ***p ≤ 0.0004, ****p ≤ 0.0003; one-way ANOVA with Dunnett’s multiple comparisons test. (F) Mtb survival in USP8-depleted BMDMs and scr controls. Data show average from 2 independent experiments with 5 replicates. (G, H) Mtb survival in control and USP8-depleted THP-1 cells (G) and primary human monocyte-derived macrophages (H). Data show mean ± s.e.m from 3 independent experiments. ****p ≤ 0.0001; two-tailed unpaired Student’s t test.

Figure 2.

Mtb ubiquitination is elevated in Usp8-silenced cells independently of PRKN and SMURF1. Control and Usp8-silenced BMDMs were infected with fluorescent mtb (MOI 5) for 24 h. (A) Immunofluorescence (IF) images show mtb (red) and ubiquitin (FK2, green). Images are shown in 3D, and arrows mark the bacteria that are magnified in panels “a” and “b” to illustrate those not colocalized (none), partially (partial), or completely enveloped (full) by ubiquitin. 60× images were acquired and projected in 3D. (B) donut plots show the percentage of ubiquitin-colocalized WT or ΔesxA Mtb. Bacteria (150-300) from each condition were manually analyzed from 3D reconstructions. Data is from 2 independent experiments, p-value is Fisher’s exact test comparing the proportion of bacilli in “none” and “colocalized” (includes both partial and full). (C, D) immunofluorescence (IF) images and corresponding donut plots show colocalization of LGALS3 (red) with WT Mtb or ΔesxA Mtb (green) in control and Usp8-silenced BMDMs 24 hpi. Data are from 2 independent experiments. (E, F) donut plots show percentage of WT or ΔesxA Mtb colocalized with ubiquitin (E) and LGALS3 (F) in control or Usp8-silenced BMDMs treated with LLOMe 24 hpi. Cells were treated with 1 mM LLOMe for 15 min followed by washing and fixation after 2 h. (G) IF images (20×) show colocalization of Mtb (red) and ubiquitin (FK2, green) in control and prkn KO BMDMs upon Usp8-silencing 24 hpi. (H) Corresponding dot plot shows FK2 MFI from > 200 bacteria from 3 independent experiments. (I) 60× if images show Mtb (red) and FK2 (green) in control and smurf1 KO macrophages transfected with control or Usp8 siRNA 24 hpi. (J) Corresponding dot plots show FK2 MFI from 2 independent experiments. (K) 20× images show Mtb (red) and K48-linked ubiquitin (green) in BMDMs transfected with control or Usp8 siRNA 24 hpi. (L) Corresponding dot plot showing K48 MFI from 4 independent experiments, ****p < 0.0001 unpaired Student’s t test. (M, N) Plots show ratio of day 3:Day 0 Mtb CFU in control versus (M) prkn KO or (N) smurf1 KO BMDMs with or without Usp8-silencing. (M, N) values are mean ±s.e.m from 3 independent experiments. (H, J, M, N) *p ≤ 0.04, **p ≤ 0.006, ***p ≤ 0.0005, ****p < 0.00001; Brown-Forsythe and Welch ANOVA tests. (A, C, I) images were acquired using 60× objective lens and are shown in (A, C) 3D in a bounding box with 10 μm graduations and (I) single XY plane. (H, J, L) automated image analysis was used to calculate MFI of FK2 intensity around a bacterium in the image. (G, I, K) Scale bar: 10 μm.

Figure 3.

USP8 is recruited to damaged endomembranes and is involved in ESCRT-III mediated membrane repair. (A, B) control or Usp8-silenced BMDMs infected with Mtb for 24 h were immunostained for CHMP1A. (A) IF images of Mtb (red) and CHMP1A (green). (B) violin plot shows mean fluorescence intensity (MFI) of CHMP1A in association with Mtb. Data shows median and quartiles from 3 independent experiments with > 2000 bacilli,****p < 0.0001; Student’s t test. MFI was calculated from IF images acquired using 20× objective lens in a single plane. (C) schematic for calculating colocalization using 3D images. 3D images were reconstructed from individual z-stacks. Using IMARIS software, parameters were defined to convert Mtb or protein puncta to surface or spots, respectively. Next, parameters were defined to calculate number of spots touching surfaces or other spots to estimate the number of “colocated” or “non-colocated” spots. (D) 3D images showing Mtb (red) and CHMP4B (green) in control or Usp8-silenced cells 24 hpi. (E) plot shows percentage of CHMP4B puncta colocated with Mtb; p < 0.0001, Fisher’s exact test. Data analyzed from 3-dimensional z-stack images acquired using 60× objective lens. (F) IF images of USP8 (red) and CHMP4B (green) after LLOMe treatment for 15 and 30 min. Images are one slice of a 3-dimensional z-stack. (G) plot showing number of CHMP4B, USP8, or colocated puncta normalized to total cell number in untreated or LLOMe-treated cells after 15 and 30 min. Data from 3 independent experiments with 22-29 cells in each sample. *p ≤ 0.03, **p<0.009, ***p = 0.0002, ****p < 0.0001; Fisher’s exact test. (H-L) control or Usp8-silenced cells were treated with LLOMe for 15 min, followed by washout for 15 and 45 min. (H, I, J) cells were immunostained for CHMP4B (green) or (K, L) VPS4. (I, J, L) violin plots show CHMP4B MFI (I, J) and VPS4 MFI (L) of cells in control (I) and Usp8-silenced (J) BMDMs. (K) control or Usp8-silenced cells were treated with LLOMe for 15 min, followed by washout for 15 and 45 min. Cells were immunostained for VPS4 (green). (L) violin plots show VPS4 MFI of control or Usp8-silenced BMDMs. (H, K) images are shown in a single XY plane; cell borders are indicated by white line. (I, J, L) data shows mean from 3 independent experiments. * p = 0.03, **p ≤ 0.005, ***p ≤ 0.0009, ****p < 0.0001; Brown-Forsythe and Welch ANOVA tests. (M) 3D images show USP8 (green) colocalization with WT or ΔesxA Mtb (red) in BMDMs 24 hpi. (N) donut plots show percentage of Mtb colocalizing with USP8 24 hpi. Data from 2 independent experiments with 200-400 bacteria 24 hpi in WT BMDMs. p-value; Fisher’s exact test. (A, D, F, M) boxed areas are enlarged in panels. (A, F, H, K) scale bar: 10 μm. (D, M) images are shown in 3D inside a bounding box with 20 μm graduations. The bacteria are cropped and magnified in 3D (not to scale) to show the degree of colocalization.

Figure 4.

USP8-depletion induces xenophagy which restricts Mtb in macrophages. (A, B) images and corresponding donut plots show colocalization of Mtb with SQSTM1, NBR1, TAX1BP1, and MAP1LC3B in Usp8-silenced BMDMs and controls 24 hpi. (C, D) images and corresponding donut of Mtb and LysoTracker in control and Usp8-silenced BMDMs 72 hpi. (A, C) images are shown in 3D within a bounding box having 20 μm graduations in the XY plane. Arrows show Mtb that are magnified to show different types of colocalization (none, partial and full). These are not to scale; they are cropped in 3D and rotated to show a different angle from the original image. Panels for NBR1 (green) and TAX1BP1 (purple) are from the same cell. (B, D) data from 2 independent experiments where ~ 150-200 bacteria were manually analyzed as not colocalized (none), partially colocalized (partial), or completely enveloped (full) with the host protein. (E) Western blots show SQSTM1 in control and Usp8-silenced BMDMs that were uninfected or Mtb-infected for 24 h. As indicated, cells were treated with chloroquine (CQ; 10 μM) 3 h prior to harvest. Quantification is shown for SQSTM1 ratio based upon band intensity of SQSTM1 normalized to ACTB/β-actin. (F) gene expression of Sqstm1 upon Usp8-silencing in uninfected (light gray) and Mtb-infected cells (dark gray) 72 hpi. Values were normalized to Actb/β-actin and then to scr control. Data is mean±s.e.m from three independent experiments. ***p = 0.0001 unpaired t test. (G) plot shows ratio of Mtb CFU at Day 3 and Day 0 post-infection from WT (Cre⁻) and atg5 cKO, BMDMs transfected with control or Usp8 siRNA. Data is from 4 independent experiments with 5 replicates. *p = 0.01, **p ≤ 0.001, ***p = 0.0001, NS=not significant; Brown-Forsythe and Welch ANOVA test. (B, D) p-value from Fisher’s exact test.

Figure 5.

USP8 depletion induces oxidative stress. (A, B) volcano plots show differentially expressed genes in Usp8-silenced versus control BMDMs that were (A) uninfected or (B) Mtb-infected (MOI 5) 72 h. DEGs with adjusted p-value ≤0.01 and fold change ≥ 1.2 (red) or ≤ 1.2 (blue) are indicated. (C, D) Expression of (C) Slc7a11 and Nqo1, and (D) Serpin9b and Gsta3 was determined by qPCR from uninfected (light gray) or Mtb-infected (dark gray) BMDMs transfected with scrambled or Usp8 siRNA 72 hpi. Data shows fold change relative to scrambled controls. Data are mean ± s.e.m from 3 independent experiments. ^#p = 0.058, *p ≤ 0.03, **p ≤ 0.007 unpaired t test. (E, F) nuclear translocation of NFE2L2 was assessed in uninfected and mtb-infected control and Usp8-silenced BMDMs 72 hpi. (E) IF images show nuclear translocation of NFE2L2 in Mtb (red)-infected macrophages upon Usp8 silencing. Scale bar: 10 μm. (F) violin plot shows Pearson’s colocalization coefficient of NFE2L2 and DAPI in the indicated conditions. Data from 279-453 cells from three independent experiments. (G) images show the range of CellRox MFI as a rainbow heat map in control or USP8-depleted BMDMs that were uninfected (UI) or Mtb-infected for 72 h. (H) violin plot shows CellRox MFI in control and USP8-depleted BMDMs that were either uninfected or Mtb-infected for 72 h. Data from at least 500 cells from 3 independent experiments. (F, H) **p = 0.0006, ****p < 0.0001, one-way ANOVA with Tukey’s multiple comparisons test; US=unstained. (I, J, K) WT (Cre⁻) and atg5 cKO (I), atg7 cKO (J), atg14 cKO (K) BMDMs were treated with control or Usp8 siRNA and infected with Mtb for 72 h in presence or absence of N-acetyl cysteine (NAC). Plot shows ratio of Mtb CFU at Day 3 and Day 0 post-infection. Data is from 2-3 independent experiments. **p ≤ 0.009, ***p ≤ 0.0002, ns=not significant; Brown-Forsythe and Welch ANOVA tests.

Figure 6.

Schematic of the role of USP8 in endomembrane damage, xenophagy, and oxidative stress response during Mtb infection. Under normal conditions, USP8 is recruited to phagosomes that are damaged by intracellular Mtb and is important to coordinate repair. The reduced phagosomal damage limits xenophagy as does USP8 inhibiting SQSTM1. In USP8-depleted cells, Mtb-induced phagosomal damage remains unresolved resulting in recruitment of LGALS3/Galectin-3 and autophagy adaptors (NBR1, TAX1BP1, SQSTM1) and activation of xenophagy. Usp8-silenced cells have increased oxidative stress, which could be caused by accumulation of damaged mitochondria, ubiquitin stress, and/or endosomal dysfunction. Based upon the literature, USP8-deficiency also stabilizes SQSTM1, which in turn facilitates dissociation of KEAP1 with NFE2L2 resulting in its nuclear translocation. The combination of enhanced xenophagy and oxidative stress facilitates infection control in Usp8-silenced cells.