Abstract

   The breast cancer susceptibility gene 1 (BRCA1) is a major tumor
   suppressor gene and is most frequently mutated in hereditary breast
   cancer. BRCA1 plays a critical role in many biological processes,
   especially maintaining genomic stability in the nucleus, yet its role
   in the cytoplasm remains elusive. Here, it is revealed that BRCA1
   maintains a healthy mitochondrial network through regulating
   mitochondrial dynamics, including fission and fusion. BRCA1 deficiency
   causes dysfunctional mitochondrial dynamics through increased
   expression of mitofusin1/2. With mitochondrial stress, BRCA1 is
   recruited to the mitochondrial outer membrane, where it plays an
   essential role in maintaining a healthy mitochondrial network.
   Consequently, BRCA1 deficiency impairs stress‐induced mitophagy through
   blocking ataxia‐telangiectasia mutated (ATM)‐AMP‐activated protein
   kinase (AMPK)‐Dynamin‐related protein 1 (DRP1)‐mediated mitochondrial
   fission and triggers NLRP3 inflammasome activation, which creates a
   tumor‐associated microenvironment, thereby facilitating tumor
   proliferation and metastasis. It is further shown that inflammasome
   inhibition can prevent tumor recurrence and metastasis. This study
   uncovers an important role of BRCA1 in regulating mitophagy and
   suggests a therapeutic approach for fighting this deadly disease.

   Keywords: ATM‐AMPK‐DRP1, BRCA1, inflammasomes, mitochondrial dynamics,
   mitophagy
     __________________________________________________________________

   Breast cancer susceptibility gene 1 (BRCA1) not only negatively
   regulates mitochondrial fusion but also facilitates stress‐induced
   fission and mitophagy. BRCA1 deficiency causes oxidative stress and
   inflammasome activation, which creates a tumor‐associated
   microenvironment, thereby promoting breast cancer development. This
   study uncovers the function of BRCA1 in regulating mitophagy and
   provides a rational approach for treating BRCA1 mutant breast cancer
   with inflammasome inhibitors.

   graphic file with name ADVS-7-1903616-g009.jpg

1. Introduction

   Breast cancer is the most common disease in women, which accounts for
   30% of all new cancers.[qv: 1] The clinical and molecular heterogeneity
   of breast cancer is well known. Based on molecular classification,
   triple‐negative breast cancer (TNBC), which lacks expression of
   estrogen receptor (ER), progesterone receptor (PR), and Her2, accounts
   for ≈12–17% of breast cancer.[qv: 2] Patients with TNBC have a
   relatively poorer prognosis compared with those with other breast
   cancer subtypes; this is due to its aggressive clinical properties and
   lack of established molecular targets for therapy.[qv: 3] Therefore,
   there has been an intense interest in finding new medications that can
   treat TNBC.

   The breast cancer susceptibility gene 1 (BRCA1) is a major breast
   cancer suppressor gene, which encodes a protein critical for
   maintaining DNA integrity and genomic stability. More than 75% of
   tumors developing in women who carry BRCA1 mutations are TNBC.[qv: 4]
   BRCA1 tumor suppressor activity has been attributed to its nuclear
   localization, where it participates in signaling pathways for DNA
   damage repair, transcription regulation, chromatin remodeling, cell
   cycle checkpoint control, and apoptosis.[qv: 5,6] Meanwhile, BRCA1 has
   been identified as a protein that shuttles between the nucleus and the
   cytoplasm.[qv: 7] Nuclear export of BRCA1 could be induced by DNA
   damage in the p53 dependent mechanism.[qv: 8] However, the function of
   BRCA1 in cytoplasmic processes, which may be independent from
   maintenance of genomic stability, is poorly understood.

   Mitochondria are crucial organelles for energy production and cellular
   homeostasis in mammalian cells; therefore, the maintenance of a healthy
   mitochondrial network is critical in the development as well as in the
   response to physiological adaptations and stress conditions throughout
   life.[qv: 9] Mitophagy, a selective autophagic process, plays an
   important role in maintaining mitochondrial function. Mitochondria as
   dynamic organelles are constantly undergoing fission and fusion, which
   are essential for regulation of mitophagy.[qv: 10] Defects in mitophagy
   could lead to pathological conditions, such as neurodegeneration,
   inflammasome activation, and cancer.[qv: 11–13] Recent studies
   indicated that BRCA1 deficiency could impair oxidative phosphorylation
   and decrease ATP production in cardiac and muscle tissues,[qv: 14–16]
   suggesting that BRCA1 is involved in mitochondrial functions. However,
   little is known about how BRCA1 relates to mitophagy in response to
   mitochondrial damage and how defects in mitophagy contribute to
   BRCA1‐associated breast cancer.

   In this study, we seek to determine the mechanism by which BRCA1 is
   involved in mitophagy and its impact on therapeutic treatment of
   BRCA1‐associated breast cancer. Our findings demonstrate that BRCA1
   deficiency impairs mitochondrial function and mitophagy through
   AMP‐activated protein kinase (AMPK)‐mediated mitochondrial fission and
   induces inflammasome activation, which then promotes metastasis of
   Brca1 mutant mammary tumor. This suggests that inflammasome inhibition
   could serve as a therapeutic target for the treatment of
   BRCA1‐associated breast cancer.

2. Results

2.1. BRCA1 Is Essential for Mitophagy

   We previously demonstrated that mice with mammary gland (MG)‐specific
   deletion of Brca1 exon 11 (Brca1^flox/flox; MMTV‐Cre) spontaneously
   developed mammary tumors.[qv: 17] To explore the function of BRCA1 in
   cytoplasm, we carried out a genome‐wide unbiased approach to analyze
   gene expression by RNA sequencing (RNA‐Seq) in both Brca1 mutant (MT)
   and wild‐type (WT) MGs from Brca1^flox/flox; MMTV‐Cre, and
   Brca1^flox/flox mice, respectively. Bioinformatics analysis of the
   whole transcriptome indicates that loss of BRCA1 has a profound impact
   on gene expression networks related to mitochondrial functions (Figure
   [46]1A). Gene set enrichment analysis (GSEA), using the curated gene
   set compilation hallmark of transcripts downregulated in the MG of MT
   mice compared with WT mice, detected enriched genes corresponding to
   mitochondrial organization (Figure [47]1B). This finding suggests that
   BRCA1 mutation might cause dysfunction of mitochondria.

Figure 1.

   Figure 1
   [48]Open in a new tab

   BRCA1 is required for stress‐induced mitophagy. A) Top pathways from
   GSEA upregulated and downregulated genes in Brca1 MT versus WT mammary
   glands by using gene ontology (GO) analysis. B) GSEA plot of enrichment
   in “mitochondrion organization” gene set, significantly downregulated
   in Brca1 MT mammary glands. C) Analysis of mitophagy activity in shCtrl
   and shBRCA1 Hela‐mCherryParkin under CCCP treatment by clearance of
   ATP5B. Left panel: Representative images of cells immunostained to
   ATP5B (green); DAPI, DNA‐binding dye; Scale bar, 20 µm. Right upper
   panel: Western blot for BRCA1. Right lower panel: Quantification for
   ATP5B level after CCCP treatment (more than 30 cells were counted per
   group). D) Representative images of shCtrl and shBRCA1 cells
   immunostained to mtDNA in the absence and presence of OA treatment
   (left panel, scale bar, 20 µm) and quantified for mitophagy (right
   panel) (ten fields counted per group). E) Immunoblot analysis of COXII
   and β‐actin (loading control throughout) in shCtrl and shBRCA1 cells
   exposed to CCCP or oligomycin/antimycin A (OA) treatment. F)
   Quantification of COXII level in (E), which normalized by β‐actin level
   (n = 3 per group). G) Analysis of damaged mitochondria in shCtrl and
   shBRCA1 MDA‐MB‐231 cells after different dosages of CCCP treatment (n =
   3 per group). H) Measurement of cytoplasmic and I) mitochondrial ROS in
   shCtrl and shBRCA1 MDA‐MB‐231 cells treated by CCCP (n = 3 per group).
   Data represent the mean ± SEM and are representative of three
   independent experiments. Significant differences were determined by an
   unpaired two‐tailed t‐test (C and D) or two‐way ANOVA with Tukey's
   multiple comparison test (F–I). *p < 0.05, **p < 0.01, ***p < 0.001,
   ****p < 0.0001.

   Mitophagy is a specialized autophagy for clearing damaged mitochondria,
   and plays a critical role in maintaining mitochondrial functions. A
   major trigger for mitophagy is via the PTEN‐induced putative kinase 1
   (PINK1)/Parkin pathway.[qv: 9] To verify the effect of BRCA1 on
   mitophagy, we first monitored occurrence of mitophagy in Hela cells
   that express mCherry‐Parkin (Hela‐mCherryParkin cells) with
   mitochondrial damage treatment in different ways. We performed
   knockdown (KD) of BRCA1 in Hela‐mCherryParkin cells infected with three
   different lentivirus‐shRNA against BRCA1 (shBRCA1) and found that
   shBRCA1–2 and shBRCA1–3 worked well and could block Parkin‐mediated
   mitophagy after treatment with a chemical mitochondrial uncoupler,
   carbonyl cyanide 3‐chlorophenylhydrazone (CCCP), as measured by
   quantitation of the ATP5B level (Figure [49]1C and Figure S1A–C,
   Supporting Information). To further address the function of BRCA1 in
   mitophagy, we used shBRCA1–3 to silence BRCA1 expression for the
   following experiments. As another indicator of mitophagy, mitochondrial
   DNA (mtDNA) contents were measured by immunofluorescence. As shown in
   Figure [50]1D, mtDNA was eliminated in shCtrl cells after treatment
   with mitochondrial respiration inhibitors, oligomycin and antimycin A
   (OA), whereas BRCA1 KD cells contained more mtDNA, indicating an
   inhibition in mitophagy. Meanwhile, we analyzed mitophagy in shCtrl and
   BRCA1 KD cells by measuring degradation of mitochondrial inner membrane
   protein cytochrome C oxidase subunit II (COXII) after treatment with
   CCCP or OA. The results showed that COXII was more degraded in shCtrl
   cells compared to BRCA1 KD cells (Figure [51]1E,[52]F), suggesting that
   BRCA1 is required for mitophagy.

   To further confirm that BRCA1 is also essential for mitophagy in cells
   with endogenous Parkin, we detected mitophagy in Brca1^flox/flox mouse
   embryonic fibroblasts (MEFs), infected with adenovirus‐green
   fluorescent protein (GFP) (AdGFP) or AdCre. These results showed that
   deletion of Brca1 impaired mitophagy through measurement of
   mitochondrial proteins such as COXII, COXIV, and TOM20 (Figure S1D–F,
   Supporting Information). In addition, we also used the mt‐mKeima assay,
   which was developed recently for detecting mitophagy activity using a
   pH‐sensitive fluorescent protein.[qv: 18,19] mt‐mKeima changes
   excitation peak from green (488 nm) to red (561 nm) when engulfed in
   lysosomes, which allows assessment of mitophagy.[qv: 18] CCCP‐induced
   mitophagy was analyzed in Brca1^flox/flox MEFs with tamoxifen‐inducible
   Cre (Tam‐Cre), in which 4‐hydroxytamoxifen (4‐HT) treatment caused
   BRCA1 deletion (Figure S1G, Supporting Information). The results showed
   that CCCP significantly resulted in spectral shift of Ctrl group, but
   had no effect on 4‐HT group, which was reflected by 561/488 ratio of
   mt‐mKeima (Figure S1H,I, Supporting Information). These results further
   confirmed that BRCA1 plays the vital role in mitophagy.

2.2. Loss of BRCA1 Leads to Accumulation of Damaged Mitochondria via the
Blocking of Mitophagy

   As indicated above, BRCA1 deficiency impairs mitophagy, which suggests
   that more damaged mitochondria accumulate in BRCA1 mutant breast cancer
   cells under mitochondrial stress. To verify this, the damaged
   mitochondrial level was monitored in MDA‐MB‐231 cells by measuring
   mitochondrial membrane potential. The results showed that loss of BRCA1
   caused more damaged mitochondria accumulated in cells even under a low
   dosage of CCCP (5 × 10^−6 m), which had no obvious effect on shCtrl
   cells (Figure [53]1G). Furthermore, shBRCA1 cells produced excessive
   cytoplasmic and mitochondrial reactive oxygen species (ROS), indicating
   accumulation of damaged mitochondria after damage from CCCP (Figure
   [54]1H,[55]I). The results demonstrate that Brca1 deletion causes
   accumulation of damaged mitochondria through inhibiting mitophagy, and
   the loss of BRCA1 impairs mitochondrion‐related signaling pathways in
   the mammary gland (Figure [56]1A,[57]B).

2.3. BRCA1 Is Essential for Autophagosome Formation during Mitophagy

   The initiation of mitophagy is mediated by the stabilization of PINK1
   and translocation of Parkin to the mitochondria.[qv: 20] To address how
   BRCA1 regulates mitophagy with mitochondrial damage, we first evaluated
   whether BRCA1 mediated activation of the PINK1/Parkin pathway. As shown
   in Figure S2A,B in the Supporting Information, CCCP treatment induced
   accumulation of PINK1 and mitochondrial recruitment of Parkin in both
   shCtrl cells and shBRCA1 cells, suggesting that BRCA1 is not involved
   in PINK1/Parkin activation. Parkin, as E3 ubiquitin ligase, leads to
   the conjugation of ubiquitin in various mitochondrial substrates, such
   as mitofusin1 and 2 (MFN1/2), mediating their proteasomal
   degradation.[qv: 21] We found that BRCA1 deficiency had no effect on
   the ubiquitin level of mitochondria or the degradation of MFN1 and MFN2
   after CCCP treatment (Figure S2C,D, Supporting Information). These
   results demonstrate that BRCA1 regulates stress‐induced mitophagy
   independent of PINK1/Parkin activation.

   Mitophagy occurs through general autophagic machinery, so we sought to
   determine whether BRCA1 might regulate mitophagy by mediating general
   autophagic activity. As shown in Figure [58]2A,[59]B, BRCA1 KD
   inhibited p62 degradation under CCCP treatment, but caused greater
   accumulation of the lipidated LC3 form (LC3‐II), indicating that BRCA1
   deficiency blocks CCCP‐induced autophagic flux. Unexpectedly, the
   general autophagy activity induced by rapamycin or hank's balanced salt
   solution (HBSS) was not altered by loss of BRCA1, as indicated by
   lipidation of LC3 and degradation of p62 (Figure [60]2C,[61]D and
   Figure S3A, Supporting Information). This suggests that BRCA1 regulates
   mitophagy irrespective of general autophagic pathway.

Figure 2.

   Figure 2
   [62]Open in a new tab

   Loss of BRCA1 impairs autophagosome formation during mitophagy. A)
   Immunoblot analysis of BRCA1, p62, and LC3 in shCtrl and shBRCA1
   Hela‐mCherryParkin under CCCP treatment. B) Quantification of p62 level
   in (A), which normalized by β‐actin level (n = 3 per group). C)
   Immunoblot analysis of BRCA1, p62, and LC3 in shCtrl and shBRCA1
   Hela‐mCherryParkin under rapamycin (5 × 10^−6 m) treatment. D)
   Quantification of the p62 level in (C), normalized by the β‐actin level
   (n = 3 per group). E) Representative fluorescent images of the puncta
   formation of GFP‐LC3 in shCtrl and shBRCA1 Hela‐mCherryParkin infected
   with GFP‐LC3 retrovirus, with or without CCCP treatment, for 6 h. Scale
   bar, 20 µm. F) Quantification of GFP‐LC3 puncta in (E) (ten fields
   counted per group). G) Representative fluorescent images of the
   colocalization of GFP‐LC3 with mitochondria (labeled by MitoDsRed) in
   Hela‐HA‐Parkin infected with retrovirus‐GFP‐LC3 and
   lentivirus‐MitoDsRed while under CCCP treatment. Hochest33258,
   DNA‐binding dye. Scale bar, 10 µm. H) Pearson's coefficient is shown as
   the quantification of GFP‐LC3 puncta colocalized with mitochondria per
   cell in (G) (ten fields counted per group). Data represent the mean ±
   SEM and are representative of three independent experiments.
   Significant differences were determined by two‐way ANOVA with Tukey's
   multiple comparison test (B, D, and F) or an unpaired two‐tailed t‐test
   (H) *p < 0.05, ****p < 0.0001.

   It was found that the recruitment of autophagosomes to mitochondria
   facilitates subsequent selective autophagy in response to mitochondrial
   stress.[qv: 11] To address why the loss of BRCA1 suppresses autophagic
   flux, we monitored autophagosome formation by quantification of GFP‐LC3
   punctuation (a marker of autophagosome) in shCtrl and shBRCA1 cells,
   expressing GFP‐LC3 after CCCP treatment. GFP‐LC3 puncta were
   dramatically decreased in shBRCA1 cells compared to shCtrl cells, and
   the colocalization of LC3‐GFP and mitochondrion was also reduced in
   BRCA1 KD cells (Figure [63]2E–[64]H). In addition, lysosomes, marked
   with LAMP1‐red fluorescent protein (RFP), were colocalized more
   significantly with shCtrl mitochondria compared to shBRCA1 mitochondria
   (Figure S3B,C, Supporting Information). This finding indicates that
   BRCA1 deficiency inhibits stress‐induced mitophagy through blocking
   autophagosome formation.

2.4. BRCA1 Negatively Regulates Mitochondrial Fusion

   Mitochondria are dynamic organelles, which continually change shape
   through the combined action of fission, fusion, and movement along
   cytoskeletal tracks.[qv: 22] Our results indicate that expression of
   MFN1/2 was elevated in shBRCA1 cells compared to shCtrl cells in the
   absence of CCCP treatment (Figure S2D, Supporting Information),
   suggesting that BRCA1 might regulate mitochondrial dynamics in breast
   cancer cells. To corroborate our hypothesis, the expression of
   mitochondrial dynamics‐related proteins was measured in MDA‐MB‐231
   cells treated with shCtrl or shBRCA1. The results indicate that
   expression of MFN1/2 increases in shBRCA1 cells at both the protein and
   RNA levels through enhancing the promoter activities of Mfn1 and Mfn2
   (Figure [65]3A–[66]C), but other dynamic proteins, such as optic
   atrophy type 1 (OPA1), dynamin‐related protein 1 (DRP1), and
   mitochondrial fission protein 1 (FIS1), showed no significant
   difference (Figure [67]3A). These results suggest that BRCA1 KD
   enhances mitochondrial fusion by promoting MFN1/2 expression. We then
   examined if BRCA1 had an effect on the morphology of mitochondria by
   using MitoTracker Green FM. As shown in Figure [68]3D, shBRCA1 cells
   display a more condensed mitochondrial network, similar to shCtrl cells
   treated with mitochondrial division inhibitor 1 (Mdivi‐1). We further
   quantified BRCA1‐mediated mitochondrial‐morphological changes using an
   automated system for the quantification and classification of
   mitochondrial morphology, Micro‐P.[qv: 23] According to Micro‐P
   analysis, shBRCA1 cells displayed a significantly higher percentage of
   the branched tubule type of mitochondrion with a decrease in the small
   globe type (Figure [69]3E). Because mitochondrial fusion increases
   mitochondrial mass, we investigated this in shCtrl and BRCA1 KD cells
   with or without Mdivi‐1 treatment through flow cytometry. The results
   showed that BRCA1 KD significantly increased mitochondrial mass
   compared to shCtrl cells; and similar increase was also observed in
   shCtrl cells after Mdivi‐1 treatment (Figure [70]3F). In addition,
   electron microscopy revealed that more elongated mitochondria in
   shBRCA1 cells than in shCtrl cells (Figure [71]3G,[72]H).

Figure 3.

   Figure 3
   [73]Open in a new tab

   Loss of BRCA1 increases MFN1/2 expression and enhances mitochondrial
   fusion. A) Immunoblot analysis of BRCA1, MFN1, MFN2, OPA1, DRP1, and
   FIS1 in shCtrl and shBRCA1 MDA‐MB‐231 cells with different BRCA1 shRNA.
   Asterisk (*) indicates MFN2 band. B) The mRNA levels of Mfn1 and Mfn2
   in shCtrl and shBRCA1 MDA‐MB‐231 cells (n = 3 per group). C) The
   promoter activities of Mfn1 and Mfn2 in shCtrl and shBRCA1 MDA‐MB‐231
   cells by measuring luciferase activity. Basic, pGL3‐Basic; Mfn1/Mfn2,
   pGL3‐Basic driven by Mfn1 or Mfn2 promoter (n = 3 per group). D)
   Mitochondrial morphology of shCtrl and shBRCA1 MDA‐MB‐231 cells. shCtrl
   cells were treated by Mdivi1 (50 × 10^−6 m) for 24 h. Upper panel:
   Representative fluorescent images of mitochondrial morphology stained
   by MitoTracker Green, Scale bar, 10 µm. Lower panel: Micro‐P analytic
   images of mitochondrial morphology, Micro‐P algorithm classified
   mitochondria into small globe mitochondrion (blue) and branched
   mitochondrion (red). E) Percentage of small globe subtype and branched
   subtype classified by Micro‐P (20 cells counted per group). F) The
   mitochondrial mass in shCtrl and shBRCA1 MDA‐MB‐231 cells with or
   without Mdivi1 treatment (n = 3 per group). G) Electron microscopy
   analysis of mitochondrial morphology in shCtrl and shBRCA1 MDA‐MB‐231
   cells. Scale bar, 500 nm. H) Quantification for mitochondrial lengths
   by ImageJ (six to seven fields counted per group). I) Immunoblot
   analysis of MFN1 and MFN2 in Brca1^flox/flox; Tam‐Cre MEFs with or
   without 4‐HT (5 × 10^−6 m) treatment. J) Mitochondrial morphology of
   Brca1^flox/flox; Tam‐Cre MEFs with or without 4‐HT treatment, stained
   by MitoTracker Green. Scale bar, 10 µm. K) The mitochondrial mass in
   Brca1^flox/flox; Tam‐Cre MEFs with or without 4‐HT treatment (n = 3 per
   group). Data represent the mean ± SEM and are representative of three
   independent experiments. Significant differences were determined by an
   unpaired two‐tailed t‐test (B, H, and K) or ANOVA with Tukey's multiple
   comparison test (C, E, and F). *p < 0.05, **p < 0.01, ***p < 0.001,
   ****p < 0.0001.

   To further confirm the effect of BRCA1 on mitochondrial fusion, we
   examined MFN1/2 expression and mitochondrial morphology in
   Brca1^flox/flox MEFs with Tam‐Cre along with 4‐HTtreatment. Likewise,
   loss of Brca1 enhanced mitochondrial fusion through increased MFN1/2
   expression (Figure [74]3I–[75]K and Figure S4A, Supporting
   Information). On the contrary, the restoration of BRCA1 expression in
   the BRCA1 mutant HCC1937 cell line decreased MFN1/2 expression and
   changed mitochondrial morphology from tubular networks to fragmented
   puncta (Figure S4B–D, Supporting Information). In addition, we found
   that BRCA1 mutation increased MFN1/2 expression in breast cancer
   patient‐derived xenograft (PDX) models (Figure S4E,F, Supporting
   Information). Excessive mitochondrial fusion could block segregation of
   damaged organelles from the healthy mitochondrial network and cause
   more mitochondrial damage. Therefore, these combined data indicate that
   BRCA1 could regulate mitochondrial fusion and maintain a healthy
   mitochondrial network.

2.5. BRCA1 Is Required for Stress‐Induced Mitochondrial Fission through the
Mediation of AMPK Activation

   For the initiation of mitophagy, the mitochondrial network must be
   divided into smaller mitochondria through fission[qv: 24] and
   degradation of mitofusin induced by PINK1/Parkin could be a means of
   preventing mitochondrial fusion.[qv: 21] Our finding that BRCA1
   deficiency could promote mitochondrial fusion by increasing the level
   of MFN1/2 (Figure [76]3), which suggests that excessive fusion may lead
   to the failure of autophagosome formation and a mitophagy defect in
   BRCA1 deficient cells. However, we also found that mitochondrial damage
   with CCCP promptly induced MFN1/2 degradation in both Ctrl and BRCA1 KD
   cells (Figure S2D, Supporting Information), indicating how BRCA1
   regulates stress‐induced mitophagy is not clear. To address the effect
   of BRCA1 on stress‐induced mitophagy, we examined mitochondrial
   localization of BRCA1 under mitochondrial stress. We found that BRCA1
   translocated onto mitochondria after CCCP treatment (Figure [77]4A) and
   was removed by treatment with proteinase K (Figure [78]4B),
   demonstrating that BRCA1 is sensitive to proteinase K treatment and
   localizes on the outer membrane of mitochondria: this is similar to
   other outer membrane proteins such as TOM20. A subcellular distribution
   pattern of BRCA1 by immunofluorescence confirmed CCCP‐induced
   localization of BRCA1 to mitochondria (Figure [79]4C). To avoid
   nonspecific issue that might occur with antibody, we also generated
   Flag‐tagged BRCA1 MEFs for detecting BRCA1 localization under CCCP
   treatment (Figure S5A, Supporting Information), and the data confirmed
   that CCCP caused mitochondrial translocation of BRCA1 in MEFs, as
   detected by the Flag antibody (Figure S5B,C, Supporting Information).

Figure 4.

   Figure 4
   [80]Open in a new tab

   BRCA1 interacts with AMPKα on the mitochondria. A) The localization of
   BRCA1 on mitochondria in MCF7 cells under CCCP treatment at the
   indicated time points. Total, total cell protein. α‐Tubulin and Lamin A
   are load control for cytoplasmic and nuclear protein, respectively. B)
   BRCA1 localized on outer‐membrane of mitochondria under CCCP treatment.
   MCF cells were treated with CCCP for 1 h. TOM20 and COXIV are
   indicators for mitochondrial outer‐ and inner‐membrane protein,
   respectively. C) Immunofluorescent images of BRCA1 in MCF7 cells with
   or without CCCP treatment. Mitochondria labeled by immunostaining for
   TIM50. Scale bar, 20 µm. D) Network of proteins interacting with BRCA1
   analyzed by STRING. Cluster 7 was highlighted. E,F) CoIP analysis of
   the interaction of BRCA1 and AMPKα1/2 in 293T cells transfected with
   MBP‐BRCA1 and AMPKα1/2‐expressing vectors. G) BRCA1 interacts with
   AMPKα through its RING domain. Upper panel: Six overlapping fragments
   that cover the entire BRCA1 protein. Lower panel: Analysis of the
   interaction between AMPKα1 and BRCA1 fragments. Asterisks indicate the
   positions of BRCA1 fragments. H) Phosphorylation of AMPKα is required
   for its interaction with BRCA1 under CCCP treatment. 293T cells were
   treated or not with CCCP after transfected with MBP‐BRCA1 and
   AMPKα1‐expressing vectors. Cell lysates were treated or not with
   λ‐phosphatase (λPPase) for 30 min before immunoprecipitation. I) CoIP
   analysis of interaction of endogenous BRCA1 and AMPKα1 on mitochondria
   of MCF7 cells. Mitochondrial protein exacted from MCF7 cells with or
   without CCCP treatment, with CoIP analysis performed. Upper panel:
   Immunoprecipitation performed by AMPKα1 antibody. Lower panel:
   Immunoprecipitation performed by BRCA1 antibody. Data are
   representative of at least two independent experiments.

   To identify the potential partners that interact with BRCA1 during
   mitophagy, maltose binding protein (MBP)‐tagged BRCA1 was overexpressed
   in 293T cells and immunoprecipitated by the MBP antibody. We pulled
   down 74 proteins that contained more than four unique identified
   peptides, indicating that these proteins could interact with BRCA1
   (Table S1, Supporting Information). We then generated the
   protein–protein interaction network based on molecular functions by
   STRING. The network shows the five big interaction clusters involved in
   tRNA biosynthesis, pre‐mRNA splicing, protein processing, cytoskeleton,
   and glycogen metabolism, but these clusters have little to do with
   mitophagy (Figure [81]4D). However, cluster 7, containing PRKAA1 and
   PRKAA2, caught our attention. The cluster is involved in two catalytic
   subunits of AMP‐activated protein kinase (AMPK), AMPKα1 and AMPKα2
   (Figure [82]4D). AMPK is a highly conserved metabolic sensor that plays
   a strong role in regulating cell growth, metabolism, and autophagy.[qv:
   25] It is suggested that BRCA1 might regulate stress‐induced mitophagy
   dependent on AMPK signaling activation. To substantiate our hypothesis,
   we first validated the interaction of BRCA1 with AMPKα1 and AMPKα2. The
   MBP‐BRCA1 construct with corresponding candidate‐expressing constructs
   were transfected into 293T cells and the interaction was checked by
   Co‐immunoprecipitation (CoIP) and Western blotting. The results showed
   that MBP‐BRCA1 could interact reciprocally with AMPKα1 and AMPKα2
   (Figure [83]4E,[84]F).

   BRCA1 is a multidomain protein that contains N‐terminal RING domain,
   Coiled‐Coil domain, BRCA1 C‐terminal (BRCT) repeats, nuclear export
   sequence (NES), and two nuclear localization sequences.[qv: 26,27] We
   investigated the ability of AMPKα1 to interact with six Flag‐tagged
   BRCA1 fragments, and found that fragment #1, containing the RING domain
   and NES, interacted with AMPKα1 (Figure [85]4G). Further results
   indicated that CCCP‐induced mitochondrial damage enhanced the
   interaction of BRCA1 with AMPKα1, which is dependent on phosphorylation
   of AMPKα (Figure [86]4H). Furthermore, to validate the interaction in
   the mitochondria, we extracted mitochondrial proteins from MCF7 cells,
   with or without CCCP treatment, and examined the connection between
   endogenous BRCA1 with AMPKα1. As shown in Figure [87]4I, BRCA1 and
   AMPKα1 interacted reciprocally, which was enhanced by CCCP treatment.
   These results suggest that BRCA1 could play a role in AMPK signaling
   pathway.

   A recent study indicated that AMPK could regulate mitochondrial fission
   and subsequent mitophagy via inducing the phosphorylation of MFF
   (mitochondrial fission factor).[qv: 28] Therefore, we investigated
   whether BRCA1 regulated AMPK activation and mitochondrial fission with
   CCCP treatment. We found that BRCA1 KD could impair AMPK activation, as
   reflected by decreased phosphorylation of AMPKα and its known
   downstream targets, such as Acetyl‐CoA Carboxylase (ACC) and MFF
   post‐CCCP treatment (Figure [88]5A). BRCA1 deficiency also impaired
   oligomycin‐induced AMPK activation (Figure S5D, Supporting
   Information). Conversely, ectopic expression of BRCA1 could restore
   phosphorylation of AMPK and MFF and sensitize additional
   phosphorylation with CCCP treatment (Figure [89]5B). However, BRCA1 had
   no effect on AMPK activation under AMPK inducer
   5‐aminoimidazole‐4‐carboxamide riboside (AICAR) treatment (Figure S5E,
   Supporting Information), suggesting that BRCA1 is essential for AMPK
   activation with mitochondrial damage. Because activated AMPK
   phosphorylation of MFF is required for mitochondrial localization of
   DRP1 and mitochondrial fission,[qv: 28] we investigated whether BRCA1
   affected DRP1 localization after CCCP treatment. As shown in Figure
   [90]5C,[91]D, BRCA1 KD inhibited CCCP‐induced recruitment of DRP1 to
   mitochondria. Consequently, it blocked DRP1‐mediated segregation of
   damaged mitochondria, reflected by more compact mitochondria in shBRCA1
   cells on the CCCP treatment, compared with shCtrl cells (Figure S5F,G
   and Movie S1,2, Supporting Information). Mitochondrial fission
   generates smaller mitochondria, which can easily be sequestered by
   autophagosomal membrane; therefore, DRP1 deficiency could impair
   mitophagy under CCCP treatment, reflected by the degradation of
   mitochondrial proteins such as HSP60 and COXII (Figure S5H, Supporting
   Information). These combined data suggest that loss of BRCA1 inhibits
   AMPK‐mediated MFF phosphorylation and DRP1 mitochondrial translocation
   and blocks stress‐induced mitophagy due to defective mitochondrial
   fission.

Figure 5.

   Figure 5
   [92]Open in a new tab

   Loss of BRCA1 impairs mitochondrial localization of DRP1 via blocking
   activation of ATM‐AMPK‐MFF pathway. A) BRCA1 KD suppressed CCCP‐induced
   AMPK activation in 293T cells. Cells were treated by CCCP (10 × 10^−6
   m) for indicated time points. B) Ectopic expression of BRCA1 restored
   CCCP‐induced AMPK activation in BRCA1 mutant HCC1937 cells. The cells
   stably transfected with vector (VEC) or BRCA1‐expressing vector (BRCA1)
   and treated by CCCP for indicated time points. C) Representative
   fluorescent images of the localization of DRP1 on mitochondria in
   shCtrl and shBRCA1 Hela cells after CCCP treatment for 1 h.
   Mitochondria were labeled by staining with TOM20 antibody. DAPI,
   DNA‐binding dye. Scale bar, 10 µm. D) Pearson's coefficient is shown as
   the quantification of DRP1 colocalized with mitochondria per cell in
   (C) (ten fields counted per group). E,F) Inhibition of ATM suppressed
   CCCP‐induced AMPK activation in 293T cells. E) shCtrl and shATM 293T
   cells were treated by CCCP for indicated time points; F) 293T cells
   were pretreated by ATM inhibitor KU‐60019 (5 × 10^−6 m) for 30 min,
   then treated by CCCP for indicated time points. G) Immunoblot analysis
   of cellular localization of BRCA1, ATM, and AMPKα in MCF7 cells before
   and after CCCP treatment for 1 h. Nuc, nuclear protein; Cyto,
   cytoplasmic protein; Mito, mitochondrial protein. H) Immunoblot
   analysis of ATM, AMPKα, and DRP1 levels on mitochondria in shCtrl and
   shBRCA1 MCF7 cells under CCCP treatment. Total, total cell protein.
   Data represent the mean ± SEM and are representative of at least two
   independent experiments. Significant differences were determined by
   two‐way ANOVA with Tukey's multiple comparison test. *p < 0.05, ****p <
   0.0001.

2.6. BRCA1 Is Required for Stress‐Induced AMPK Activation through Mediating
the Localization of ATM‐AMPKα in the Mitochondrion

   It is well known that liver kinase B1 (LKB1) and
   calcium/calmodulin‐dependent protein kinase kinase‐β (CaMKKβ), two
   major upstream kinases, activate AMPK by phosphorylating Thr172 in the
   activation loop of the catalytic α‐subunit.[qv: 25] To investigate the
   upstream kinase of CCCP‐induced AMPK activation, we examined whether
   LKB1 or CaMKKβ were required for CCCP‐induced AMPKα and MFF
   phosphorylation. The results showed that LKB1 or CaMKKβ KD had no
   effect on the phosphorylation of AMPKα and MFF under CCCP treatment
   (Figure S6A,B, Supporting Information). Meanwhile, AMPK activation
   induced by CCCP was unaffected in Hela cells, which lacks LKB1;
   however, it was blocked by shBRCA1 (Figure S6C, Supporting
   Information). These results indicate that AMPK activation induced by
   CCCP is BRCA1‐dependent but was not LKB1‐ or CaMKKβ‐dependent.

   It was shown that ATM, an upstream kinase, could phosphorylate AMPKα
   both in vivo and in vitro.[qv: 29] We then examined whether ATM was
   involved in CCCP‐induced AMPK activation. The data indicated that CCCP
   treatment induced phosphorylation of ATM and AMPK, and inhibition of
   ATM activity by ATM KD or ATM inhibitor (KU‐60019) led to impairment of
   AMPK activation (Figure [93]5E,[94]F), which demonstrated that ATM is
   required for CCCP‐induced AMPK activation. It is known that BRCA1
   facilitates the ability of ATM to phosphorylate downstream substrates
   under DNA damage treatment;[qv: 30] as such, we hypothesized that BRCA1
   might be required for ATM‐mediated AMPK activation. To corroborate
   this, we first investigated the localization of BRCA1, ATM, and AMPKα
   before and after CCCP treatment. As shown in Figure [95]5G, CCCP
   treatment could not only cause translocation of BRCA1 onto mitochondria
   but also increased levels of ATM and AMPKα on the mitochondrion. We
   found that BRCA1 deficiency impaired mitochondrial translocation of ATM
   and AMPKα after CCCP treatment for 0.5 h (Figure [96]5H), so the level
   of DRP1 on the mitochondria decreased in shBRCA1 cells when compared to
   shCtrl cells; this confirms localization of DRP1 as measured by
   immunofluorescence (Figure [97]5C,[98]D). These data indicate that the
   loss of BRCA1 inhibits ATM‐mediated AMPK activation through blocking
   mitochondrial localization of ATM and AMPKα.

2.7. Loss of BRCA1 Leads to Inflammasome Activation

   It has been reported that defective mitophagy leads to the accumulation
   of damaged mitochondria and excessive ROS, which then activates the
   NLRP3 inflammasome.[qv: 12] We found that damaged mitochondria and ROS
   accumulated in BRCA1‐deficient cells (Figure [99]1G–[100]I). Pathway
   enrichment and GSEA analysis indicated that the inflammatory response
   significantly increased in Brca1 mutant MG (Figure [101]1A and Figure
   S7, Supporting Information). These results suggest that loss of BRCA1
   might lead to activation of inflammasome. To confirm this, we first
   used a human monocytic cell line (THP‐1 cells), which represents the
   most commonly used cells to assess the inflammasome activation. We
   found that nigericin, an inflammasome inducer, could dramatically
   promote activation in BRCA1 KD cells, compared to control cells (Figure
   [102]6A). Meanwhile, CCCP‐induced mitochondrial stress enhanced ROS
   production and generated more cleaved caspase 1 and mature
   interleukin‐1β (IL‐1β) in BRCA1 KD cells (Figure [103]6C,[104]E). To
   establish the effect of BRCA1 on inflammasome activation in mammary
   cells, we analyzed IL‐1β secretion and caspase 1 activation in MCF10A
   cells with detectable inflammasome‐associated components, compared to
   several breast cancer cell lines (Figure S8, Supporting Information).
   The results showed that BRCA1 KD led to more ROS production and
   inflammasome activation in MCF10A cells (Figure [105]6B,[106]D,F),
   which indicates that BRCA1 deficiency can trigger inflammasome
   activation.

Figure 6.

   Figure 6
   [107]Open in a new tab

   BRCA1 deficiency promotes inflammasome activation. A) Release of IL‐1β
   from LPS‐primed THP‐1 cells and B) MCF10A cells stimulated with CCCP or
   Nigericin, as measured by ELISA (n = 3 per group). C) Measurement of
   mitochondrial ROS in THP‐1 cells and D) MCF10A cells (n = 3 per group).
   E) Immunoblot analysis of pro‐caspase 1 and pro‐IL‐1β processing in
   LPS‐primed THP‐1 cells and F) MCF10A cells stimulated as indicated.
   Data represent the mean ± SEM and are representative of three
   independent experiments. Significant differences were determined by
   two‐way ANOVA with Sidak's multiple comparison test (A and B) or with
   Tukey's multiple comparison test (C and D). *p < 0.05, **p < 0.01, ***p
   < 0.001, ****p < 0.0001.

2.8. Blocking Inflammasome Activity Ameliorates Brca1 Mutant Mammary Tumor
Recurrence and Metastasis

   IL‐1β promotes tumor progression and counteracts immunosurveillance by
   recruiting myeloid cells, such as myeloid‐derived suppressor cells
   (MDSCs) and tumor‐associated macrophages (TAMs).[qv: 31,32] This
   suggests that BRCA1 mutation could establish a tumor‐associated
   microenvironment for cancer progression via inflammasome activation. To
   verify the hypothesis, we first investigated immunocytes' profiles in
   the Brca1 mutant MG and mammary tumors from Brca1^flox/flox;MMTV‐Cre
   mice, which frequently develop a single mammary tumor.[qv: 33,34] We
   also analyzed the immunocytes' profiles in Trp53 mutant MG and mammary
   tumors from Trp53^flox/flox;MMTV‐Cre mice, since TP53 mutations are the
   second most frequent genetic alteration in breast cancer, observed in
   23% of them, according to the catalog of somatic mutations in cancer
   database.[qv: 35] Although total T cells, including CD4^+ and CD8^+ T
   cells, have no significant difference in these groups, activated CD8^+
   T cells significantly downregulate in Brca1mutant tumor compared to WT
   MG or Trp53 mutant tumor (Figure [108]7A,[109]B). The results indicate
   that inflammasome activation triggered by BRCA1 deficiency leads to
   impairment of CD8^+ T cell activity. To confirm this, we investigated
   whether inhibition of inflammasome activity could reactivate CD8^+ T
   cells and alleviate Brca1 mutant tumor progression. As shown in Figure
   S9A in the Supporting Information, we surgically removed primary
   mammary tumors and observed tumor relapse and metastasis with or
   without inflammasome inhibitor (glibenclamide, GLI) treatment. In
   according with our expectations, GLI treatment inhibited inflammasome
   activity in recurrent tumors as measured by cleaved caspase 1 level
   (Figure [110]7C,[111]D), and the treatment dramatically increased the
   percentage of activated CD8^+ T cells in the total number of CD8^+ T
   cells, which was not affected by the inhibitor (Figure [112]7E and
   Figure S9B, Supporting Information). We found that the treatment
   delayed tumor recurrence (Figure [113]7F) and blocked lung metastasis
   of the Brca1 mutant tumor (15.3%) compared to no treatment group
   (64.2%) (Figure [114]7G and Figure S9C, Supporting Information). Two
   mice injected with GLI had no relapse for up to six months. Our results
   support that inflammasome activation in Brca1mutant tumor impairs CD8^+
   T cell activation and promotes tumor progression.

Figure 7.

   Figure 7
   [115]Open in a new tab

   Excessive inflammasome activation promotes mammary tumor recurrence and
   metastasis. A) Immunocytes profile analysis in Brca1 mutant MG (B.MG),
   Brca1 mutant tumor (B.T), Trp53 mutant MG (TP53.MG), and Trp53 mutant
   tumor (Trp53.T) after normalization with WT MG or in B.T after compared
   with Trp53.T (B.T_Trp53.T) according to the RNA‐Seq. B) GSEA plot of
   enrichment in “activated CD8 T cell” gene set significantly
   downregulated in B.T versus Trp53.T. C) Immunoblot analysis of
   procaspase 1 (proCasp1) processing in recurrent tumors with or without
   GLI treatment. D) Quantification of proCasp1 and Casp1 p20 level in
   (C), normalized by β‐actin level (n [control] = 6, n [GLI] = 5). E)
   CD8^+ T cells and activated CD8^+ T cells in primary and recurrent
   tumors analyzed by flow cytometry with or without GLI treatment (n
   [control] = 5, n [GLI] = 6). F) Relapse time of tumors after removing
   primary mammary tumors from Brca1^flox/flox;MMTV‐Cre mice with or
   without GLI treatment (n [control] = 14, n [GLI] = 13, two mice in GLI
   group show no relapse). G) Representative images of lung with
   hemotoxylin and eosin (H&E) staining sections from Brca1^flox/flox;
   MMTV‐Cre mice after primary mammary tumor removal with or without GLI
   treatment. Arrow indicates metastatic tumor on the lung. Scale bar, 200
   µm. H) M1/M2 macrophage polarization analysis in indicated groups. I)
   Representative fluorescent images of primary and recurrent tumor
   sections with immunostaining for F4/80. Scale bar, 20 µm. J)
   Quantification of F4/80 positive macrophages in recurrent tumors,
   relative to those in primary tumors with or without GLI treatment (n
   [control] = 6, n [GLI] = 6). K–M) The mRNA level of M2 macrophage
   markers in recurrent tumors after normalization by those in primary
   tumors with or without GLI treatment, measured by real‐time PCR (n
   [control] = 6, n [GLI] = 6). N) GSEA plot of enrichment in
   “mitochondrion organization” gene set significantly downregulated in
   breast cancers carrying BRCA1 mutation (BRCA1.T) versus cancers
   carrying TP53 R175H mutation (TP53.T) according to the METABRIC
   database. O) GSEA plot of enrichment in “activated CD8 T cell” gene set
   downregulated in BRCA1.T versus TP53.T. P) M1/M2 macrophage
   polarization analysis in BRCA1.T compared with that of TP53.T. Data
   represent the mean ± SEM. Significant differences were determined by
   two‐way ANOVA with Sidak's multiple comparison test (D) or an unpaired
   two‐tailed t‐test (F, J, K, L, M, and P). *p < 0.05, **p < 0.01, ***p <
   0.001.

   As mentioned above, MDSCs and TAMs are key components of the
   tumor‐associated microenvironment that promotes tumor progression and
   metastasis.[qv: 31,32] MDSCs rather than other myeloid cells, such as
   neutrophils and monocytes/macrophages, significantly increased in Brca1
   mutant MG (Figure [116]7A), but not in Brca1 mutant tumor (Figure S9D,
   Supporting Information). This suggests that MDSCs may play a role in
   the initiation of Brca1 mutant tumor, not in tumor progression. We
   found that although the level of macrophages in Brca1 mutant tumors was
   not different from that in Trp53 mutant tumors (Figure S9E, Supporting
   Information), Brca1 mutant tumors displayed unbalanced M1/M2 macrophage
   polarization compared to other groups (Figure [117]7H). It is well
   known that most TAMs are considered to have an M2 phenotype, while
   playing a crucial role in building an immunosuppressive environment
   that blocks T cell activation.[qv: 36] The results show that more
   polarized M2 macrophages were recruited into Brca1 mutant tumor
   compared with those in WT MG and Trp53 mutant tumors (Figure [118]7H).
   Moreover, we found that GLI treatment reduced the percentage of
   macrophages in relapse tumors, compared with those in primary tumors,
   which were analyzed by macrophage marker F4/80 staining (Figure
   [119]7I,[120]J). The M2 polarization of TAM could be demonstrated by
   analyzing their transcriptional profiling.[qv: 36] Therefore, we
   measured the mRNA levels of M2 macrophage markers such as CD163, CD206,
   and Mgl2 in recurrent tumors and in primary tumors with or without GLI
   treatment. This shows that the levels of these markers significantly
   decrease in recurrent tumors treated with GLI after normalizing those
   levels in primary tumors (Figure [121]7K–[122]M). These data suggest
   that macrophages infiltrating tumor tissues were driven by Brca1 mutant
   tumor‐derived cytokines to acquire a polarized M2 phenotype, which
   impairs CD8^+ T cells activation.

   To verify this phenomenon in clinical samples, we analyzed immunocyte
   profiles in breast cancers developed in patients carrying the
   pathogenic BRCA1 mutations in the Molecular Taxonomy of Breast Cancer
   International Consortium (METABRIC) database.[qv: 37] It was reported
   that nearly all luminal breast cancers carry point mutations in
   TP53,[qv: 38] so we used breast cancer patients with a TP53 R175
   mutation, a hotspot mutation as a control group. This demonstrates that
   mitochondrial function is impaired in BRCA1 mutant tumors, while
   enriched genes correspond to mitochondrial organization (Figure
   [123]7N), and BRCA1 mutant tumor accumulates more M2‐polarized
   macrophages, as well as containing less activated CD8^+ T cells than
   the TP53 mutant tumors (Figure [124]7O,[125]P). These observations show
   that BRCA1 mutation can cause abnormal M2 macrophage recruitment to
   suppress CD8^+ T cell activation in clinical samples.

   The data suggest that blocking excessive inflammasome activity in Brca1
   mutant tumor can postpone tumor recurrence and inhibit tumor metastasis
   through impaired the recruitment of TAMs.

3. Discussion

   BRCA1 is a well‐known tumor suppressor, having been linked to a range
   of cellular processes such as DNA repair, transcriptional regulation,
   chromatin remodeling, and cell cycle checkpoints.[qv: 5,6] While these
   BRCA1 functions are largely attributed to nuclear localization, the
   role of BRCA1 in the cytoplasm remains elusive. Although a recent study
   has indicated that multiple Fanconi anemia pathway genes including
   BRCA1 are required for mitophagy,[qv: 39] its underlying mechanism
   remains poorly defined, and the relationship of defective mitophagy
   with BRCA1‐associated breast cancer development is unknown. In this
   study, we reveal that BRCA1 is located at the mitochondrial membrane at
   a low level but is significantly enhanced under stress conditions.
   Moreover, BRCA1 plays an important role in maintaining mitophagy
   through involvement in mitochondrial fission and fusion. BRCA1
   deficiency blocks mitophagy, leading to the accumulation of damaged
   mitochondria and inflammation which is involved in tumorigenesis and
   cancer metastasis.

   The mitochondrion is a highly complex organelle, which dynamically
   changes its morphology and network through continuously undergoing two
   opposite processes: fusion and fission.[qv: 10] It has been reported
   that deletion of BRCA1 leads to enlarged mitochondria in skeletal
   muscle,[qv: 16,40] indicating that BRCA1 deficiency causes
   dysfunctional mitochondrial dynamics, yet the underlying mechanism has
   not been illustrated. Our data demonstrated BRCA1 negatively regulates
   mitochondrial fusion through inhibiting expression of MFN1/2 at the
   transcriptional level. Both MFN1 and MFN2 are GTPases localized in the
   outer membrane of the mitochondrion. With some other proteins, such as
   OPA1, whose expression is not affected by BRCA1, these proteins take
   part in the fusion process.[qv: 22] Loss of BRCA1 results in a higher
   level of MFN1/2 proteins, accelerating mitochondrial fusion, which can
   block the separation of damaged mitochondria from healthy ones and
   negatively affect mitophagy.

   It has been reported that BRCA1 interacts with ACC, one of the
   downstream targets of AMPK, by pulling down BRCT residues of BRCA1,[qv:
   41] and BRCA1 stabilizes phosphorylation of ACC to affect lipid
   synthesis via this interaction.[qv: 42] In this study, full length of
   BRCA1 was pulled down to screen for interacting proteins. It is
   determined that AMPKα1/2 interacts with BRCA1 via the RING domain, and
   BRCA1 positively regulates AMPK signaling activation under
   mitochondrial damage, not AICAR treatment, demonstrating that a novel
   function of BRCA1 on AMPK signaling.

   Phosphorylated BRCA1[qv: 43] or a number of isoforms of BRCA1[qv: 44]
   are located inside the mitochondria. To elucidate the molecular
   mechanism between BRCA1 and CCCP‐induced AMPK activation, we deciphered
   localization of BRCA1 on mitochondria with or without mitochondrial
   damage. Our data indicated that BRCA1 is located on the outer membrane
   of mitochondria at low levels under normal conditions, as it can be
   degraded with proteinase K treatment similar to other outer membrane
   proteins, as mitochondrial stress promotes more BRCA1 translocation to
   mitochondria. Although BRCA1 has no mitochondrial targeting signal, it
   is known that its main binding partner, BARD1, can translocate on
   mitochondria that depend on BRCT domain.[qv: 45,46]It suggests that
   mitochondrial translocation of BRCA1 may be regulated by its own BRCT
   domain or interaction with BRAD1, which will be addressed in future
   study.

   Furthermore, the increased BRCA1 on the outer membrane of mitochondria
   is correlated with the activation of AMPK and its downstream proteins,
   while the recruitment of DRP1 to the mitochondrial outer membrane plays
   a central role in triggering mitochondrial fission. We found BRCA1 is
   required for the translocation of ATM onto mitochondria to promote
   activation of AMPK under stress condition, which provides a link in the
   recruitment of DRP1 to mitochondria with BRCA1. These findings
   demonstrate that BRCA1‐ATM‐AMPK‐DRP1 signaling together to regulate
   mitochondrial fission.

   Thus, BRCA1 deficiency affects mitochondrial fusion through increasing
   the transcription of MFN1/2 and affects fission through impairing
   stress‐induced AMPK activation and DRP1‐mediated mitochondrial fission.
   These effects can result in elongated mitochondria and inhibit the
   dynamic balance of mitochondria in BRCA1 mutant cells, leading to the
   failure of mitophagy: the initiation of mitophagy requires the
   mitochondrial network to be divided into smaller mitochondria through
   fission.[qv: 24] These data uncover an important function of BRCA1 at
   the mitochondrial outer membrane that regulates mitophagy.

   It has been shown that insufficient clearance of dysfunctional
   mitochondria, which can result from defective mitophagy, has been
   implicated in a series of pathophysiological conditions, including
   inflammasome activation, genotoxic stress, and promotion of
   tumorigenesis.[qv: 12,13] Another important finding is that loss of
   BRCA1 triggers inflammasome activation which then promotes Brca1 mutant
   tumor relapse and metastasis. In this study, upregulated inflammasome
   activation enriches TAMs' recruitment to suppress CD8^+ T cell
   activation in Brca1 mutant mammary tumors; thus, inhibition of
   inflammasome activation could postpone tumor relapse and block lung
   metastasis after surgically removing primary cancers. It is known that
   metastatic dormancy complicates the treatment of breast cancer, as
   surgical excision of primary tumors results in the acceleration of
   metastatic tumors.[qv: 47] Our findings suggest that the interference
   of inflammasome activation could protect BRCA1 mutant breast cancer
   patients from metastatic recurrence after resecting primary tumors.
   These results also provide more evidence to support recent findings
   that anti‐inflammatory treatment reduces the incidence of early
   metastatic relapse in breast cancer patients and decreases tumor growth
   in mice models for dormancy.[qv: 47,48] Interestingly, our result shows
   that Brca1 mutant‐associated tumor microenvironment is different from
   that of Trp53 mutant‐associated type. Activated CD8^+ cells were
   significantly inhibited in Brca1 mutant tumors, but monocytes,
   dendritic cells, and NK cells were dramatically downregulated in Trp53
   mutant tumors (Figure [126]7A). This suggests that inflammasome
   inhibition may be applicable in Brca1 mutant tumor, but not in TP53
   mutant tumor.

   In summary, we provide the following model for the essential role of
   BRCA1 in promoting stress‐induced mitophagy to maintain a healthy
   mitochondrial network at the outer membrane of the mitochondrion
   (Figure [127]8 ). This function occurs through regulating both
   mitochondrial fusion and fission. We demonstrated that 1) BRCA1
   negatively regulates the key components of mitochondrial fusion
   machinery of MFN1/2 expression under normal conditions; 2) BRCA1
   promotes AMPK‐induced DRP1‐MFF activation through mediating
   mitochondrial translocation of ATM and AMPKα under stress conditions;
   3) BRCA1 deficiency impairs stress‐induced mitophagy through blocking
   ATM‐AMPK‐DRP1‐mediated mitochondrial fission. Excessive mitochondrial
   fusion leads to elongated mitochondria and prevents the formation of a
   healthy mitochondrial network in BRCA1 deficient cells, leading to
   blocked mitophagy; 4) BRCA1 deficiency triggers NLRP3 inflammasome
   activation and provides a tumor‐associated microenvironment that
   facilitates tumor proliferation and metastasis. Inhibition of
   inflammasome activity can serve as a therapeutic approach to fight this
   deadly disease.

Figure 8.

   Figure 8
   [128]Open in a new tab

   Schematic model for the mechanism by which BRCA1 is essential for
   mitophagy. BRCA1 is required to maintain mitophagy through involvement
   in mitochondrial fission and fusion. BRCA1 deficiency then blocks
   mitophagy, with the accumulation of damaged mitochondria as well as
   inflammation, involved in tumorigenesis and cancer metastasis.

4. Experimental Section

Animals and Treatments

   All animal experiments were approved by the University of Macau Animal
   Ethics Committee under the protocol (UMAEC‐050‐2015). PDX models of
   breast cancer were purchased from The Jackson Laboratory (TM00096,
   Brca1 WT; TM00089, Brca1 MT). Brca1^flox/flox; MMTV‐Cre mice were
   generated in the previous study[qv: 17] and Trp53^flox/flox; MMTV‐Cre
   mice were generated by crossing Trp53^flox/flox mice[qv: 49] with
   MMTV‐Cre mice. For RNA‐Seq analysis, mammary glands and tumors were
   collected from female Brca1^flox/flox; MMTV‐Cre mice or
   Trp53^flox/flox; MMTV‐Cre mice, and mammary glands from female
   Brca1^flox/flox mice as control group. For tumor recurrence and
   metastasis experiment, Brca1^flox/flox; MMTV‐Cre mice with mammary
   tumors (diameter around 1 cm) were i.p. injected with
   phosphate‐buffered saline (PBS) or NLRP3 inflammasome inhibitor
   Glibenclamide (GLI, 5 mg kg^−1) one day before surgery, and mice
   underwent surgery to remove primary tumors. Then, the mice were
   administrated with PBS or GLI by continuous injection every three days
   for five times, and tumor relapse and metastasis were examined.

Plasmid

   The lentiviral shRNA plasmid pLKO.1 targeting BRCA1 (clone ID
   TRCN0000039833, TRCN0000039837, and TRCN0000010305), DRP1
   (TRCN0000318424), ATM (TRCN0000010299), LKB1 (TRCN0000000408), CaMMKβ
   (TRCN0000002299), and shRNA control plasmid were purchased from Sigma.
   MBP‐BRCA1 expressing vector was constructed in previous study.[qv: 50]
   Mfn1 and Mfn2 luciferase reporters were requested from Dr. Martin
   Lidell. Six BRCA1 fragments expressing vector were requested from Dr.
   Rong Li. pCHAC‐mt‐mKeima and pBMN‐mCherry‐Parkin were gifts from
   Richard Youle (Addgene plasmid # 72342 and # 23956); pRK5‐HA‐Parkin was
   a gift from Ted Dawson (Addgene plasmid # 17613); pBABEpuro GFP‐LC3 was
   a gift from Jayanta Debnath (Addgene plasmid # 22405); pLV‐mitoDsRed
   was a gift from Pantelis Tsoulfas (Addgene plasmid # 44386);
   LAMP1‐mRFP‐FLAG was a gift from David Sabatini (Addgene plasmid #
   34611); pLenti.CMV/TO_PRKAA1 and pLenti.CMV/TO_PRKAA2 were gifts from
   Reuben Shaw (Addgene plasmid # 74446/74447).

Cell Culture and Virus Infection

   Hela, 293T, MDA‐MB‐231, MCF7, THP‐1, and MCF 10A cell lines were
   obtained from the American Type Culture Collection (ATCC). BRCA1 mutant
   HCC1937 cell line and its derivative reconstituted with full length
   BRCA1 were gifts from Dr. Junjie Chen.[qv: 51] All cell lines except
   THP‐1 and MCF 10A cells were grown in dulbecco's modified eagle medium
   (DMEM) (Thermo Fisher Scientific, 11965118) supplemented with 10% fetal
   bovine serum (FBS) (Thermo Fisher Scientific, 26140079). THP‐1 cells
   were cultured with ATCC‐formulated RPMI‐1640 Medium (ATCC, 30‐2001)
   supplemented with 10% FBS and 0.05 × 10^−3 m 2‐mercaptoethanol. MCF10A
   immortalized mammary epithelial cells were cultured with DMEM/F12 (1:1;
   Thermo Fisher Scientific, 11330057) supplemented with 5% horse serum
   (Thermo Fisher Scientific, 16050122), 0.5 µg mL^−1 hydrocortisone
   (Sigma, H0888), 20 ng mL^−1 epidermal growth factor (EGF) (Thermo
   Fisher Scientific, PHG0311L), 10 µg mL^−1 insulin (Sigma, I1882), and
   100 ng mL^−1 cholera toxin (Sigma, C8052). Primary MEFs were isolated
   from E13.5–14.5 Flag‐tagged Brca1, Brca1^flox/flox or
   Brca1^flox/flox;Tam‐Cre embryos using a standard procedure and
   maintained in DMEM with 15% FBS. Lentivirus and Retrovirus infection
   were performed as described.[qv: 52] Adenoviruses expressing Cre
   recombinase and GFP (Ad‐Cre) or GFP alone (Ad‐GFP) were purchased from
   Vector Development Laboratory, Baylor College of Medicine, Houston.

Transfection and Luciferase Assay

   The transfections were carried out with Lipofectamine 3000 (Thermo
   Fisher Scientific, L3000015) according to the manufacturer's
   instructions. For luciferase assay, cells were harvested after 24 h
   post‐transfection and luciferase activity was assayed using a dual
   luciferase reporter assay system (Promega, Madison, WI). Renilla
   luciferase activity was used for normalization.

Immunofluorescence Staining

   The cells were fixed with 4% Paraformaldehyde and then stained with
   corresponding antibodies using methods described previously.[qv: 52]
   Deparaffinized sections from primary and recurrent tumors were cooked
   with Retriever (Electronic Microscopy Science, 62700–10) in Buffer A
   (citrate buffer, pH 5.0) followed by staining with antibody against
   F4/80. Images were acquired using Nikon A1R Confocal System or Olympus
   IX83 Inverted Microscope, and then fluorescent signal intensity, area,
   and colocalization of signals were analyzed by ImageJ software
   (National Institutes of Health (NIH), Bethesda). The antibodies used
   for immunofluorescence staining are listed in Table S2 in the
   Supporting Information.

Mitophagy Treatment and Analysis

   Mitophagy treatment and analysis were performed as described
   previously.[qv: 19] Briefly, mCherry‐Parkin or HA‐Parkin stably
   expressing Hela cells were established (Hela‐mCherryParkin or
   Hela‐HA‐Parkin) and infected with lentivirus shCtrl or shBRCA1 for 48
   h, and then the cells were treated with 10 × 10^−6 m carbonyl cyanide
   m‐chlorophenyl hydrazone (CCCP, Sigma‐Aldrich) or 10 × 10^−6 m
   oligomycin (Sigma‐Aldrich) combined with 5 × 10^−6 m antimycin A
   (Sigma‐Aldrich) for the indicated time points. MEFs were infected with
   Ad‐Cre or Ad‐GFP for 48 h and then treated with 30 × 10^−6 m CCCP for
   the indicated time points. For 24 h treatment, Z‐VAD‐FMK
   (Sigma‐Aldrich) was supplemented to prevent cells from apoptosis
   induced by mitochondrial stress. To calculate the percentage of mtDNA
   stain remaining, the area of total cellular DNA (Area[tDNA]) and
   nuclear DNA (Area[nDNA]) were determined by anti‐DNA antibody and
   4′,6‐diamidino‐2‐phenylindole (DAPI) staining, respectively. mtDNA
   staining area was calculated by using the following formula:
   Area[tDNA]‐Area[nDNA]/n, in which n is cell number in a field, then the
   mtDNA stain remaining after treatment was normalized by mtDNA staining
   area without treatment. For mt‐mKeima assay, mt‐mKeima were stably
   expressed in MEFs by lentivirus transduction, then treated with 4‐HT
   for 48 h to induce BRCA1 KD. After CCCP treatment, mt‐mKeima were
   measured on Nikon A1R Confocal System using dual‐excitation
   measurements at 488 nm (pH 7) and 561 nm (pH 4) lasers with 700/75 nm
   emission filters. The fluorescence intensity were analyzed using
   NIS‐Elements software and the ratio of fluorescence intensity between
   561 and 488 nm was calculated.

Time‐Lapse Microscopy

   To visualize mitochondria, Hela‐HA‐Parkin were infected with
   lentivirus‐MitoDsRed and the fluorescent cells were sorted by flow
   cytometry (BD FACSAria III). Then, the cells were infected with shCtrl
   or shBRCA1 lentivirus and seeded in confocal glass‐bottom 6‐well plates
   (SPL Life Sciences, 30206). Imaging was started immediately after
   addition of CCCP on Olympus IX83 Inverted Microscope at 37 °C and 5%
   CO[2]. The images were acquired every 5 or 10 min for 90 min and then
   were exported as uncompressed AVI sequences at one frame per second.

Immunoprecipitation and Western Blot Analysis

   For western blot analysis, cells were lysed in radioimmunoprecipitation
   assay (RIPA) buffer (10 × 10^−3 m Tris‐HCl pH 8.0, 150 × 10^−3 m NaCl,
   1% Triton X‐100, 1% deoxycholate acid, 0.1% SDS, and protease inhibitor
   cocktail) after PBS washing. For immunoprecipitation, cells were lysed
   with lysis buffer (20 × 10^−3 m Tris‐HCl pH 7.5, 150 × 10^−3 m NaCl, 10
   × 10^−3 m EDTA, 10 × 10^−3 m EGTA, 1% NP‐40, and protease inhibitor
   cocktail) at 24 h post‐transfection, and then the cell lysate was
   immunoprecipitated by corresponding antibodies or control IgG. After
   washing, the precipitates were analyzed by Western blot. For λPPase
   treatment, 1000 units of λPPase (New England Biolabs, P0753), 50 µL 10
   × NEBuffer for Protein MetalloPhosphatases, and 1 × 10^−3 m MnCl[2]
   were added into the cell lysate (400 µL) and incubated at 30 °C for 30
   min. Western blot was carried out by ChemiDoc MP Imaging System
   (BIO‐RAD) with correspondent antibodies. The band intensity was
   quantified using ImageJ, and the numbers under immunoblots represent
   intensity relative to the first band. Relative quantification was
   achieved by normalization to the level of β‐actin. Antibodies for
   immunoprecipitation and western blot are listed in Table S2 in the
   Supporting Information.

Quantitative Real‐Time PCR

   Total RNA was isolated with Trizol reagent (Thermo Fisher Scientific)
   according to the manufacturer's instructions. Reverse transcription was
   performed using QuantiTect Reverse Transcription Kit (QIAGEN, 205313).
   Real‐time PCR reactions were performed using FastStart Universal SYBR
   Green Master (Roche, 4913850001) on QuantStudio 7 Flex Real‐Time PCR
   System (Thermo Fisher Scientific). Relative quantification was achieved
   by normalization to the amount of 18S. Primers used for real‐time PCR
   are listed in Table S3 in the Supporting Information.

RNA‐Seq and Data Processing

   Total RNA from mammary glands and tumors was processed using RNeasy
   mini kit (QIAGEN, 74106), and RNA concentration and integrity were
   measured using the Agilent 2100 Bioanalyzer (Agilent Technologies).
   cDNA libraries were prepared from RNA starting material (RNA integrity
   number values > 7.0), using NEBNext Ultra RNA Library Prep Kit for
   Illumina (New England Biolabs, E7530) according to the manufacturer's
   instructions, and library quality was checked on Agilent 2100
   Bioanalyzer. Total 13 libraries generated from WT MG (two samples),
   Brca1 MT MG (two samples), Brca1 MT tumors (four samples), Trp53 MT MG
   (two samples), and Trp53 MT tumors (three samples) were paired‐end
   sequenced by HiSeq 2500 (Illumina) in Genomics, Bioinformatics and
   Single Cell Core, Faculty of Health Science, University of Macau. The
   quality of the sequencing data was analyzed by using FastQC (version
   0.11.5) and raw reads with low quality were removed using Trim Galore
   (version 0.4.4) prior to analysis of the data. The criteria of removing
   low quality reads were set as: quality Phred score less than 28 and
   reads containing adaptor sequences. All the trimmed reads were mapped
   to reference mouse genome (mm10, GRCm38) by using STAR (version 020201)
   and Subjunc (version 1.5.3), and the mapped counts were extracted using
   featureCount from Subread package (version 1.5.3). Subsequently, read
   count data containing 47594 genes with raw reads were preprocessed by
   filtering out genes with zero read counts across different samples
   within cohort. After filtering, 31880 genes remain in per sample. The
   read count data were normalized to produce transcripts per kilobase
   million (TPM‐counts per length of transcript (kb) per million reads
   mapped), which can be used for downstream differential expression
   analysis. Pathway enrichment analysis was performed using the in‐house
   R code. All data processing were completed in shell high performance
   computing (HPC) command line environment. Raw RNA‐Seq data files are
   available at the NCBI Sequence Read Archive (SRA, PRJNA604672)
   [129]https://www.ncbi.nlm.nih.gov/bioproject/PRJNA604672.

Protein Pull‐Down and Mass Spectrometry Analysis

   BRCA1 pull‐down was performed according to the method described
   previously.[qv: 50,53] Briefly, 293T cells were transfected with
   MBP‐BRCA1 or control MBP vector. Forty eight hours post‐transfection,
   these cells were harvested and lysed with lysis buffer containing 0.1%
   NP‐40, 25 × 10^−3 m Tris‐HCl (pH 7.5), 150 × 10^−3 m NaCl, 10%
   glycerol, 2 × 10^−3 m MgCl[2], 0.1 × 10^−3 m ZnCl[2], 0.1 × 10^−3 m
   EDTA, 1 × 10^−3 m dithiothreitol (DTT), protease inhibitor cocktail,
   and phosphatase inhibitor PhoSTOP (Roche). The lysate was sonicated and
   centrifuged at 14 000 rpm at 4 °C for 10 min. The supernatants were
   added into Amylose resin (New England Biolabs, E8021S) and rotated for
   4 h at 4 °C. After washing four times with lysis buffer, bound proteins
   were resolved by 4–12% NuPAGE, stained with Coomassie blue, and
   analyzed via in‐gel digestion followed by liquid chromatography‐mass
   spectrometry using a linear ion‐trap mass spectrometer (LTQ,
   ThermoElectron). The LTQ was operated in a data‐dependent mode in which
   the seven most abundant peptide molecular ions in every mass
   spectrometry (MS) scan were sequentially selected for fragmentation and
   acquisition of MS^2 spectra. Tandem mass spectra were searched against
   the human protein database using SEQUEST software (ThermoFinnigan). The
   protein–protein interaction network was built up by STRING
   ([130]https://string-db.org/)

Inflammasome Activity Analysis

   THP‐1 cells were differentiated for 3 h with 1 × 10^−6 m phorbol
   12‐myristate 13‐acetate (PMA) (Sigma‐Aldrich), washed, and then primed
   by lipopolysaccharides (LPS) (200 ng mL^−1, Sigma‐Aldrich) for 3 h
   following by CCCP (5 × 10^−6 m) or Nigericin (10 × 10^−6 m,
   Sigma‐Aldrich) treatment for 5 or 1 h, respectively. MCF10A cells were
   primed by LPS (200 ng mL^−1) for 3 h following by CCCP or nigericin
   treatment for 18 h or 5 h, respectively. After treatment, cell lysis
   and cell culture supernatants were collected for analyzing inflammasome
   activity. Cell culture supernatants were assayed for IL‐1β with ELISA
   Ready‐SET‐Go! Kit (Thermo Fisher Scientific, 88‐7261‐86) according to
   the manufacturer's instructions. For measurement of extracellular
   cleaved IL‐1β and caspase‐1 by Western Blot, cell culture supernatants
   were precipitated by the addition of an equal volume of methanol and
   0.25 volumes of chloroform as described previously.[qv: 54] The
   supernatant/methanol/chloroform mixtures were vortexed and then
   centrifuged for 10 min at 20 000 × g. The upper phase was discarded and
   500 µL methanol was added to the interphase. This mixture was
   centrifuged for 10 min at 20 000 × g and the protein pellet was dried
   at 55 °C.

Reactive Oxygen Species and Mitochondrial Membrane Potential Analysis by
Fluorescence‐activated Cell Sorting (FACS)

   For ROS measurement, intracellular H[2]O[2] was measured by H[2]DCFDA
   (10 × 10^−6 m, Thermo Fisher Scientific) and mitochondrial ROS was
   determined by MitoSOX (5 × 10^−6 m, Thermo Fisher Scientific). In
   brief, cells were performed with CCCP (10 × 10^−6 m) for 18 h and then
   incubated with H[2]DCFDA or MitoSOX for 30 min according to
   manufacturer's instructions, then cells were collected and analyzed by
   flow cytometry. For mitochondrial membrane potential analysis, the
   cells were stained with both Mitotracker Green (200 × 10^−9 m, Thermo
   Fisher Scientific) and tetramethylrhodamine ethyl ester perchlorate
   (TMRE) (100 × 10^−9 m, Thermo Fisher Scientific) after CCCP treatment
   for 18 h, then mitochondrial membrane potential per unit of
   mitochondrial mass on a single cell was assessed by flow cytometry.

Mitochondrial Isolation

   MCF7 or MEF cells were treated with CCCP (10 or 30 × 10^−6 m,
   respectively) for the indicated time points. Mitochondrial isolation
   were performed by Mitochondria Isolation Kit according to the
   manufacturer's instructions (Abcam, ab110171). For MCF7 cells, isolated
   mitochondria were treated by proteinase K (0.8 mg mL^−1) alone or
   proteinase K combined with digitonin (0.2 mg mL^−1) and 1% SDS, then
   harvested for immunoblot.

Tumor Digestion and Flow Cytometry

   Tumors were isolated, finely minced, and digested with Digestion I at
   37 °C for 4 h. After cells were spun down, they were treated with
   Digestion II for 5 min. Digested tumor cells were washed with Hanks
   solution and lysed with red blood cell (RBC) Lysis Solution. Cells were
   resuspended in Flow Cytometry Staining Buffer (ThermoFisher Scientific)
   and incubated with antibodies (Table S2, Supporting Information) for 1
   h on ice. Then, the cells were washed and resuspended by Flow Cytometry
   Staining Buffer, and analyzed by flow cytometry (BD FACSAria III).
   Signal threshold definition was defined using all‐stained, unstained
   and isotype controls. Gating strategy is shown in Figure S9B in the
   Supporting Information. Digestion I: DMEM/F12 supplemented with 5% FBS,
   500 ng mL^−1 hydrocortisone, 10 ng mL^−1 EGF, 20 ng mL^−1 cholera
   toxin, 300 U mL^−1 collagenase III, 100 U mL^−1 hyaluronidase;
   Digestion II: 5 mg mL^−1 dispase II, 0.1 mg mL^−1 deoxyribonuclease.

Clinical Data from The METABRIC

   METABRIC breast cancer normalized transcriptome data and mutation file
   with clinical information were downloaded from the cBioPortal
   ([131]https://www.cbioportal.org).[qv: 37,55] Thirty‐six BRCA1 mutant
   cancers with single‐nucleotide variant or indel mutation and 34 TP53
   mutant cancers with R175 mutation were found (Table S4, Supporting
   Information). Thus, a total of 24368 genes were used for GSEA and
   macrophage polarization analysis using TP53 mutant group as control.

Gene Set Enrichment Analysis (GSEA) and Macrophage Polarization Analysis

   GSEA was performed using the R package clusterProfiler.[qv: 56]
   Briefly, the enrichment score (ES) was calculated as the maximum
   deviation from zero of the weighted fraction of genes present minus the
   fraction not present up to a given index in a gene expression matrix
   ordered by phenotype correlation. The statistical significance (nominal
   P value) of the ES of a gene set was estimated using an empirical
   gene‐based permutation test. The normalized enrichment score was
   calculated by creating 1000 permutations of the ES and scaling the
   observed ES by the mean score of the permutations. GSEA was applied to
   selected gene sets to test their enrichment in given datasets including
   mitochondrion organization (GO:0007005), inflammatory response
   (GO:0006954), and activated CD8 T cell.[qv: 57] To analyze macrophage
   polarization, macrophage signature genes were obtained including 56/37
   (mouse/human) M1 marker genes and 33/29 (mouse/human) M2 marker genes
   (Table S5, Supporting Information) according to previous study.[qv: 58]
   Macrophage polarization analysis was performed as described
   previously.[qv: 59] The expression of the marker genes were first
   normalized across patients, and a Student's t‐test score comparing the
   expression of macrophage M2 marker genes to macrophage M1 marker genes
   was considered as M2‐M1 score. The score greater than zero reflects
   samples with M2 macrophage polarization feature, while the score less
   than zero reflects samples with M1 macrophage feature.

Statistical Analysis

   All values are expressed as the mean ± standard error of the mean (SEM)
   of individual samples. Samples were analyzed by using unpaired
   two‐tailed t‐test or one/two‐way analysis of variance (ANOVA) test.
   Values of p < 0.05 were considered as statistically significant.
   Correlation was studied by Pearson's correlation test as indicated.
   Statistical analysis was performed using GraphPad Prism 7.

Conflict of Interest

   The authors declare no conflict of interest.

Author Contributions

   Q.C. and C.‐X.D. conceived and designed the study and wrote the paper.
   Q.C., H.L., J.B., W.H., L.L., C.P., T.M., and J.X. acquired the data.
   Q.C., H.L., J.B., W.H., H.W., L.L., C.P., and T.M. developed the
   methodologies. Q.C., H.L., J.B., W.H., H.W., and L.L. analyzed and
   interpreted the data. H.L., K.M., and X.X. provided mice.

Supporting information

   Supporting Information
   [132]Click here for additional data file.^ (2MB, pdf)

   Supplemental Movie 1
   [133]Click here for additional data file.^ (485.2KB, mp4)

   Supplemental Movie 2
   [134]Click here for additional data file.^ (438.1KB, mp4)

   Supplemental Table 1
   [135]Click here for additional data file.^ (112.4KB, pdf)

   Supplemental Table 2
   [136]Click here for additional data file.^ (173.2KB, pdf)

   Supplemental Table 3
   [137]Click here for additional data file.^ (97.7KB, pdf)

   Supplemental Table 4
   [138]Click here for additional data file.^ (31.3KB, pdf)

   Supplemental Table 5
   [139]Click here for additional data file.^ (28.1KB, pdf)

Acknowledgements