Abstract

   Nucleosome assembly proteins (NAPs) have been identified as histone
   chaperons. Testis‐Specific Protein, Y‐Encoded‐Like (TSPYL) is a newly
   arisen NAP family in mammals. TSPYL2 can be transcriptionally induced
   by DNA damage and TGFβ causing proliferation arrest. TSPYL1, another
   TSPYL family member, has been poorly characterized and is the only
   TSPYL family member known to be causal of a lethal recessive disease in
   humans. This study shows that TSPYL1 and TSPYL2 play an opposite role
   in TGFβ signaling. TSPYL1 partners with the transcription factor FOXA1
   and histone methyltransferase EZH2, and at the same time represses
   TGFBR1 and epithelial‐mesenchymal transition (EMT). Depletion of TSPYL1
   increases TGFBR1 expression, upregulates TGFβ signaling, and elevates
   the protein stability of TSPYL2. Intriguingly, TSPYL2 forms part of the
   SMAD2/3/4 signal transduction complex upon stimulation by TGFβ to
   execute the transcriptional responses. Depletion of TSPYL2 rescues the
   EMT phenotype of TSPYL1 knockdown in A549 lung carcinoma cells. The
   data demonstrates the prime role of TSPYL2 in causing the dramatic
   defects in TSPYL1 deficiency. An intricate counter‐balancing role of
   TSPYL1 and TSPYL2 in regulating TGFβ signaling is also unraveled.

   Keywords: FOXA1, nucleosome assembly protein, TGFBR1, TGFβ signaling,
   TSPYL1, TSPYL2
     __________________________________________________________________

   TSPYL1 partners with FOXA1 and EZH2 to repress TGFBR1 transcription,
   thereby mitigating the level of TGFβ receptors and sensitivity to TGFβ.
   TSPYL1 deficiency enhances the response to TGFβ and stabilizes the
   related protein TSPYL2. TSPYL2 interacts with the SMAD2/3/4 complex to
   regulate transcription and hence executes the context‐dependent TGFβ
   driven events including epithelial mesenchymal transition in BE(2)‐C
   and A549 cells.

   graphic file with name ADVS-11-2306486-g009.jpg

1. Introduction

   Transforming growth factor β (TGFβ) signaling can regulate cell growth,
   epithelial‐mesenchymal transition (EMT), differentiation, and
   homeostasis.^[ [44]^1 ^] The receptor for TGFβ peptide is a heteromeric
   complex of TGFBR1 (also called ALK5) and TGFΒR2, which phosphorylates
   SMAD2 and SMAD3 to form a complex with SMAD4 upon ligand binding. The
   phospho‐SMAD complex is responsible for executing the early
   transcription response in TGFβ signaling. SMAD3 and SMAD4 recognize
   specific DNA sequences to determine gene expression.^[ [45]^2 ^]
   However, the DNA binding affinity of SMAD3 and SMAD4 is relatively
   weak, so interaction of the SMAD complex with other context‐specific
   transcription factors is required for transcription regulation.^[
   [46]^2 , [47]^3 ^] The control of TGFβ signaling in vivo is complex.
   Besides the bioavailability of active TGFβ, the low abundance of cell
   surface receptors and a very high ligand binding affinity allow the
   cell to exquisitely regulate its ligand sensitivity. For example,
   regulated expression of TGFBR1 determines T cell quiescence and
   activation,^[ [48]^4 ^] and late‐stage adult neurogenesis.^[ [49]^5 ^]
   Besides playing a central role in development and homeostasis, TGFβ
   also participates in disease processes such as cancer and organ
   fibrosis.

   TSPYL2 (also named CDA1, CINAP, DENTT, and NP79) belongs to the
   Nucleosome Assembly Protein (NAP) superfamily. NAPs are histone
   chaperones for the specific association of DNA and histones, and they
   also regulate transcription through facilitating the dynamic binding of
   protein complexes to chromatin.^[ [50]^6 ^] Previous data show that
   overexpression (OE) of TSPYL2 augments the phosphorylation of SMAD2,^[
   [51]^7 ^] and SMAD3 upon the addition of TGFβ.^[ [52]^8 ^] Furthermore,
   proliferation arrest induced by overexpressing TSPYL2 is dependent on
   SMAD4 in A549 lung carcinoma cells; whereas knockdown (KD) of TSPYL2
   confers resistance to TGFβ induced proliferation arrest in HaCaT human
   keratinocytes.^[ [53]^7 ^] Pointing toward the in vivo significance of
   TSPYL2 in TGFβ signaling, TSPYL2 protein increases upon stimulation by
   TGFβ in mouse vascular smooth muscle cells,^[ [54]^8a ^] and targeting
   TSPYL2 specifically retards renal extracellular matrix accumulation and
   fibrosis in a diabetic mouse model.^[ [55]^9 ^] However, how TSPYL2
   participates in TGFβ signaling is yet to be established.

   There are 6 TSPYL genes in humans, namely TSPYL1 to TSPYL6, with TSPYL3
   being a pseudogene. TSPYL1, 2, 4, and 5 have ubiquitous expression at
   variable expression levels, while TSPYL6 is expressed exclusively in
   the testes from GTEX datasets ([56]https://www.gtexportal.org/).
   TSPYL1,2, 4, and 5 regulate the expression of specific CYP genes.^[
   [57]^10 ^] Besides, TSPYL2 regulates the expression of neuronal
   genes.^[ [58]^11 ^] Importantly, TSPYL1 is essential for survival.
   Loss‐of‐function mutations of TSPYL1 cause sudden infant death with
   dysgenesis of testes syndrome (SIDDT).^[ [59]^12 ^] Patients presented
   with multiple signs of visceroautonomic dysfunction and died before
   one‐year‐old from cardiac or respiratory arrest. Tspyl1 knockout (KO)
   mice also have retarded growth and most die around the weaning stage,
   but there is no testis dysgenesis.^[ [60]^13 ^] Despite the vital
   importance of TSPYL1, little is known about its function.

   When we examined the importance of TSPYL1 in neural proliferation and
   differentiation in a neuroblastoma cell line BE(2)‐C, intriguingly KO
   of TSPYL1 caused a phenotype consistent with EMT. Here we establish
   that TSPYL1 depletion drives EMT. Mechanistically, TSPYL1 interacts
   with the transcription factor FOXA1 and histone methyltransferase EZH2
   to repress the transcription of TGFBR1. The augmented TGFβ signaling
   upon TSPYL1 depletion stabilizes TSPYL2, which interacts with the
   SMAD2/3/4 complex to activate TGFβ target genes including TGFBR1, and
   mediates the EMT phenotype. Our data provide insight into the
   biological significance of TSPYL1 in controlling TGFβ signaling and in
   counterbalancing the effects of TSPYL2.

2. Results

2.1. TSPYL1 Depletion Drives EMT through TGFβ Signaling

   Since TSPYL1 is critical for the normal functioning of the central
   nervous system as indicated in SIDDT, we explored whether KO of TSPYL1
   in the neuroblastoma cell line BE(2)‐C affected differentiation in
   vitro. KO of TSPYL1 in BE(2)‐C cells resulted in enlarged cells (Figure
   [61]1A). There were enhanced migratory and invasive abilities, together
   with increased transcript levels of SNAI1 and SNAI2 encoding the EMT
   transcription factors SNAI1 and SLUG (Figure [62]S1A–D, Supporting
   Information). These changes are typical during neural crest cell
   migration and differentiation.^[ [63]^14 ^] To establish that the loss
   of TSPYL1 can trigger EMT in epithelial cells, we employed epithelial
   cell lines including lung carcinoma A549 and breast ductal carcinoma
   MCF7. We knocked down TSPYL1 by lentiviral constructs carrying two
   independent short hairpin RNAs (shRNAs). TSPYL1sh#2 targets the 3’UTR
   of TSPYL1 while TSPYL1sh#4 targets the coding region. TSPYL1 KD in A549
   and MCF7 cells adversely affected their proliferation (Figure [64]S2,
   Supporting Information) and in A549, fibroblast‐like morphological
   changes were also observed (Figure [65]1A). Therefore, we performed a
   mechanistic study for EMT in A549 cells. First, we conducted wound
   healing and transwell assays to ascertain EMT behavior. Compared to the
   control group transduced with the transfer vector pLKO.1, TSPYL1 KD
   cells migrated significantly faster across the wound or the transwell,
   while TSPYL1 OE had the opposite effect (Figure [66]1B,C). Invasion
   assay using Matrigel‐coated transwells further indicated that TSPYL1 KD
   triggered cell invasion (Figure [67]1D). These behavioral changes,
   accompanied by reduced cell proliferation, did not increase the
   metastatic potential when the TSPYL1‐depleted A549 cells were injected
   into nude mice (Figure [68]S3, Supporting Information). At the
   molecular level, TSPYL1 KD resulted in reduced E‐cadherin and increased
   mesenchymal marker levels including N‐cadherin, Vimentin, SNAI1, and
   ZEB1 (Figure [69]1E). Conversely, OE of TSPYL1 increased E‐cadherin and
   decreased N‐cadherin, Vimentin, SNAI1, and ZEB1 levels (Figure [70]1F).
   In the genetic rescue, re‐introduction of TSPYL1 in TSPYL1 KD A549
   cells increased the expression of E‐cadherin and decreased the
   expression of N‐cadherin and Vimentin after 3 days. As highlighted, the
   rescue was more efficient for TSPYL1sh#2 than TSPYL1sh#4 KD cells
   (Figure [71]1G). This is expected as TSPYL1sh#2 targets the 3’UTR of
   the endogenous transcripts which is absent in the ectopic cDNA, while
   TSPYL1sh#4 targets the TSPYL1 coding region in both endogenous and
   ectopic transcripts. Taken together, our results confirmed that TSPYL1
   depletion drives EMT in BE(2)‐C and A549 cells. We conclude that TSPYL1
   suppresses EMT.

Figure 1.

   Figure 1
   [72]Open in a new tab

   TSPYL1 depletion drives EMT. A) Representative images of control,
   TSPYL1 knockout (KO) BE(2)‐C clones and knockdown A549 cells. TSPYL1 KO
   BE(2)‐C clones were generated by transfection with sgRNA cloned into
   CRISPR/Cas9 vector PX459. Transfected cells were selected with
   puromycin and single colonies were picked. An image of one
   representative clone was shown. A549 cells were transduced with empty
   vector pLKO.1 or two different TSPYL1sh for 48 h, and images were taken
   7 days post‐transduction (i.e. removal of viruses). Scale bar: 100 µm.
   B) Confluent TSPYL1 knocked down or overexpression A549 cells were
   treated with Mitomycin C for 2 h before wounding. Images were taken
   immediately (0 h) and 16 h afterward. Cell borders were outlined. The
   percentage of migrated area was measured by Image J and presented on
   the right. Data are shown as the mean ± S.D. n ≥ 4 wells, ^* p < 0.05,
   ^** p < 0.01 by unpaired Student's t‐test. Scale bar: 100 µm. C, D)
   A549 cells were seeded onto Transwell directly (C) or with precoating
   of Matrigel (D). Transwells were removed 24 h later. Cells that had
   migrated through the transwell were stained with crystal violet. Images
   were taken under 20X magnification and the cell number was counted.
   Data are shown as the mean ± S.D. n ≥ 5 images, ^** p < 0.01, ^*** p <
   0.001, ^**** p < 0.0001 by unpaired Student's t‐test. Scale bar:
   100 µm. E) The protein levels of epithelial marker E‐cadherin and
   mesenchymal markers as indicated were determined by immunoblotting in
   control or knockdown A549 cells on day 9 after transduction. F) The
   protein levels of epithelial marker E‐cadherin and mesenchymal markers
   as indicated were determined by immunoblotting in A549 cells transduced
   with empty vector pSIN or TSPYL1‐Flag 3 days post‐transduction. G)
   Control or TSPYL1 knockdown A549 cells were transduced with empty
   vector or TSPYL1‐Flag. Protein samples were collected after 72 h for
   immunoblotting of proteins as indicated. The highlighted pair showed
   efficient rescue by TSPYL1‐Flag with TSPYL1sh#2, targeting the
   endogenous but not ectopic TSPYL1 cDNA.

   TGFβ is one of the major inducers of EMT.^[ [73]^15 ^] We reason that
   TSPYL1 controls the sensitivity to TGFβ and TSPYL1 KD results in the
   EMT phenotype due to upregulation of TGFβ signaling. This can be
   evaluated by the level of pSMAD2/3 and quantification of the luciferase
   reporter activity driven by the SMAD binding element.^[ [74]^16 ^] Upon
   TGFβ treatment, TSPYL1 KD A549 cells indeed showed greater
   phosphorylation of SMAD2/3 while OE of TSPYL1 attenuated the level of
   pSMAD2 (Figure [75]2A). Consistently, TSPYL1 KD cells had higher
   reporter activities upon stimulation by TGFβ while TSPYL1 OE cells had
   lower activities (Figure [76]2B). TGFβ activates the transcription of
   direct target genes SMAD7,^[ [77]^17 ^] SERPINE1,^[ [78]^18 ^] and
   SNAI1 via SMAD proteins.^[ [79]^19 ^] In turn, EMT transcription
   factors such as SNAI1 change the expression of epithelial and
   mesenchymal genes.^[ [80]^15a ^] Under TGFβ treatment, TSPYL1 KD
   induced a much higher fold change in the transcript level of the above
   three TGFβ direct target genes and mesenchymal gene CDH2
   (Figure [81]2C) as well as the protein level of EMT markers
   (Figure [82]2D). Furthermore, the specific TGFBR1 kinase inhibitor
   SB‐431542 prevented the changes of EMT markers in TSPYL1 KD cells
   (Figure [83]2E). Therefore, our results verify that TSPYL1 represses
   TGFβ signaling and TGFβ driven EMT.

Figure 2.

   Figure 2
   [84]Open in a new tab

   TSPYL1 knockdown promotes TGFβ driven EMT in A549 cells. A)
   Phospho‐SMAD2/3 protein levels were measured upon TGFβ treatment by
   immunoblotting. Cells were transduced with viruses as indicated on top
   for 48 h, and experiments were performed 3 days post‐transduction in
   all subsequent experiments. A dose of 5 ng mL^−1 TGFβ1 was added to the
   culture medium and protein lysates were collected at 0, 20, and 40 min
   for immunoblotting. B) TGFβ driven transcriptional activities as
   measured by a luciferase reporter containing 4 SMAD binding elements.
   TSPYL1 knockdown or overexpression A549 cells were transfected with
   reporter plasmids for 24 h, serum starved for 24 h and 2 ng mL^−1 TGFβ1
   was added to culture medium for 24 h before luciferase reporter assay.
   Data are shown as the mean ± S.D. n ≥ 3, ^* p < 0.05, ^** p < 0.01 by
   unpaired Student's t‐test. C) The transcript levels of TGFβ target
   genes and mesenchymal marker gene CDH2 were measured by qPCR. Control
   and knockdown cells were serum starved for 24 h and treated with or
   without 5 ng mL^−1 TGFβ1 for 24 h. The mRNA levels of indicated genes
   were analyzed by qPCR. n = 3, ^** p < 0.01, ^*** p < 0.001, ^**** p <
   0.0001 by unpaired Student's t‐test. D) The expression of EMT markers
   was analyzed by immunoblotting upon TGFβ treatment. Control and
   knockdown cells as indicated on top were serum starved for 24 h and
   then treated with or without 5 ng mL^−1 TGFβ1 for another 24 h for
   immunoblotting. E) The expression of EMT markers and TGFBR1 were
   analyzed by immunoblotting. Control and knockdown cells as indicated
   were treated with 10 µm TGFBR1 inhibitor SB‐431542 for 72 h and samples
   were immunoblotted for proteins as indicated.

2.2. TSPYL1 Binds Gene Promoters in TGFβ Signaling Pathway Relevant to EMT

   As TSPYL1 contains a NAP domain, we have tested the hypothesis that
   TSPYL1 is part of the transcription regulatory complex in TGFβ pathway
   gene promoters. We first confirmed that TSPYL1 was associated with the
   chromatin (Figure [85]S4A, Supporting Information) with the chromatin
   purification method.^[ [86]^20 ^] We also made use of our KO cells to
   confirm the specificity of TSPYL1 antibodies by immunocytochemistry
   (Figure [87]S4B, Supporting Information) for subsequent Cleavage Under
   Targets and Release Using Nuclease (CUT&RUN) experiment. Next, we
   examined TSPYL1 genomic localization by CUT&RUN followed by
   Next‐generation sequencing in A549 cells. In agreement with the model
   that TSPYL1 regulates gene expression directly, a majority (43.55%) of
   peaks mapped to the promoter region (here defined as ‐2000 bp to +
   100 bp) (Figure [88]3A). The peaks in the promoter region were assigned
   to 2143 annotated genes (Figure [89]3B). Of these, 19 genes are amongst
   the 160 EMT genes listed under the GeneOntology database (GO: 0001837,
   [90]http://amigo.geneontology.org/) (Figure [91]3C). From KEGG pathway
   analysis, 13 out of the total 2143 genes are components of TGFβ
   signaling, while 5 of the 19 EMT genes are involved in both TGFβ
   signaling and published EMT pathways (Figure [92]3D).^[ [93]^15a ^] The
   TSPYL1 binding peaks in these 5 genes, namely TGFBR1, SMAD2, SMAD7,
   PPP2CA and NOG, were shown (Figure [94]3E). We further checked if the
   transcript levels of these 5 genes were affected upon TSPYL1 KD. Only
   TGFBR1 and SMAD7 showed a consistent and significant increase in
   transcript levels upon KD (Figure [95]3F), while TSPYL1 OE
   significantly reduced the expression of all 5 genes (Figure [96]3G).
   These data pointed toward the importance of TSPYL1 in directly
   suppressing TGFBR1 and SMAD7.

Figure 3.

   Figure 3
   [97]Open in a new tab

   TSPYL1 binds TGFβ pathway genes in EMT in A549 cells. A) Genomic
   distribution of TSPYL1 peaks identified by CUT&RUN‐next generation
   sequencing. Promoter: ‐2000 bp to +100 bp of transcription start site;
   TTS: Transcription Termination Site; UTR: Untranslated Region. B) The
   number of annotated genes from TSPYL1 CUT&RUN peaks according to HOMER
   Version 4.9 ([98]http://homer.ucsd.edu/homer/). C) Venn diagram
   comparing TSPYL1 target genes with EMT genes. The TSPYL1 targets were
   defined as genes with CUT&RUN peaks in the promoter region. The EMT
   gene list was from GeneOntology database (GO: 0001837,
   [99]http://amigo.geneontology.org/). The fold enrichment of peaks in 19
   common genes between TSPYL1 targets and EMT genes were indicated in the
   bar chart. D) TSPYL1 targets were subjected to KEGG pathway analysis
   and the EMT related pathways were selected. The number of genes in
   published pathways related to EMT was shown. The number of common genes
   with the GO EMT gene list is indicated in red. E) Genomic binding of
   TSPYL1 in the promoter region of TGFβ pathway EMT genes as shown by
   integrated genomic view. Peak tracks of IgG (light grey) served as the
   negative control. Scale indicated in the bracket on the left. F, G)
   Transcript levels of TGFβ pathway genes in A549 cells transduced with
   control and TSPYL1 knockdown (F) or overexpression viruses (G). Data
   are mean ± SD, n = 3, ^* p < 0.05, ^** p < 0.01, ^*** p < 0.001, ^****
   p < 0.0001, ns: not significant, by unpaired Student's t‐test.

2.3. TSPYL1 Interacts with FOXA1 to Repress TGFBR1

   We focus on TGFBR1 since its cell surface abundance is crucial in
   controlling the cellular sensitivity to TGFβ.^[ [100]^21 ^] We first
   identified suitable antibodies for ChIP in human cells by
   ChIP‐immunoblotting and further validated the antibody specificity with
   KO cells (Figure [101]S5A,B, Supporting Information). We observed the
   binding of TSPYL1 to the promoter of TGFBR1 in various cell lines,
   including A549, BE(2)‐C, MCF7, non‐transformed renal tubular cell line
   HK‐2, human embryonic stem cell line H9, and peripheral blood
   mononuclear cells from a healthy donor (Figure [102]4A). Furthermore,
   TSPYL1 KD resulted in increased expression of TGFBR1 in A549, BE(2)‐C,
   MCF7, and HK‐2 (Figure [103]3F, [104]4B,C). Although we were unable to
   find suitable antibodies for detecting mouse TSPYL1, we successfully
   knocked down Tspyl1 in primary mouse embryonic fibroblasts (MEF) with
   TSPYL1sh#4 (Figure [105]S2, supporting information), which led to
   increased Tgfbr1 expression (Figure [106]4B,C) and impaired cell
   proliferation (Figure [107]S2, Supporting Information). The findings
   indicate that TSPYL1 directly represses TGFBR1 expression in both
   normal and cancerous contexts.

Figure 4.

   Figure 4
   [108]Open in a new tab

   TSPYL1 interacts with FOXA1 to bind the promoter of TGFBR1. A) Binding
   of TSPYL1 to TGFBR1 promoter in indicated cell lines or cells as
   verified by ChIP‐PCR. IgG was used as the negative control. PBMC:
   peripheral blood mononuclear cells. B, C) Increased transcripts (B) and
   protein (C) of TGFBR1 in TSPYL1 knockdown cells. At 3 days
   post‐transduction, cells were subjected to qPCR and immunoblotting.
   MEF: mouse embryonic fibroblasts. Data are shown as the mean ± SD, n ≥
   3, ^* p < 0.05, ^** p < 0.01, ^*** p < 0.001, ^**** p < 0.0001 by
   unpaired Student's t‐test. D) Visualization of FOXA1 ChIP‐seq profile
   within the genomic region surrounding the promoter of TGFBR1 from the
   website ([109]http://chip‐atlas.org). The binding motif of FOXA1 (in
   red) was predicted with MoLoTool ([110]https://molotool.autosome.org/).
   TSPYL1 CUT&RUN peaks were placed alongside for reference. E) EMSA
   showing shift in the predicted FOXA1 binding motif‐containing oligos
   (lane 2) and supershift in the presence of FOXA1 antibodies (lane 3).
   NE: nuclear protein extract from A549; Competitor: unlabeled oligos. F)
   Confocal images showing the endogenous TSPYL1‐FOXA1 interaction in A549
   wildtype (WT) cells as analyzed by proximity ligation assay (PLA).
   White punctate staining indicated the interaction between TSPYL1 and
   FOXA1 within 40 nm and DAPI was used to reveal the nucleus. The
   quantification of positive signal relative to TSPYL1 knockout (KO) A549
   cells (negative control) was shown on the right. Data are shown as the
   mean ± SD, n ≥ 10 images, ^**** p < 0.0001 by unpaired Student's
   t‐test. Scale bar: 10 µm. G) Co‐immunoprecipitation to determine the
   region of TSPYL1 for interacting with FOXA1. The upper panel is the
   schematic of full length TSPYL1 and its truncated versions. FL: Full
   length; ΔC: C‐terminal truncated; SIDDT: Truncated protein from a frame
   shift allele in Sudden Infant Death with Dysgenesis of Testes syndrome.
   HA‐tagged TSPYL1 FL and its truncation plasmids were transiently
   transfected into HEK293FT cells. At 72 h post‐transfection, cells were
   subjected to immunoprecipitation (IP) with FOXA1 antibodies and
   immunoblot (IB) with indicated antibodies. H) ChIP‐PCR showed the
   dynamic binding of TSPYL1, FOXA1, EZH2; and H3K27me3 mark at TGFBR1
   promoter under TGFβ treatment. A549 cells were treated with or without
   5 ng mL^−1 TGFβ1 for 48 h and subjected to ChIP with TSPYL1, FOXA1,
   EZH2, H3K27me3 and IgG antibodies. TGFBR1 promoter fragments were
   amplified by PCR. I) ChIP‐PCR showed the loss of TSPYL1 binding upon
   FOXA1 knockdown. A549 cells transfected with control siRNA (siCtrl) or
   siRNA targeting FOXA1 (siFOXA1) for 48 h and subjected to ChIP with
   TSPYL1 and IgG antibodies. TGFBR1 promoter fragments were amplified by
   PCR. FOXA1 knockdown was verified by immunoblotting on the left. J)
   FOXA1 knockdown increased the transcript level of TGFBR1, TGFβ target
   genes and mesenchymal marker gene CDH2. A549 cells were transfected
   with siCtrl or siFOXA1 and subjected to qPCR at 48 h post‐transfection.
   Data are shown as the mean ± SD, n = 4, ^** p < 0.01, ^**** p < 0.0001,
   by unpaired Student's t‐test.

   The recruitment of TSPYL1 to target promoters is anticipated to require
   transcription factor partners that recognize specific DNA sequences.
   We, therefore, utilized the ChIP‐atlas database
   ([111]http://chip‐atlas.org/) to search for transcription factors that
   experimentally bind ± 5 kb of TGFBR1 transcription start site in A549
   cells. Our analysis identified 11 transcription factors including
   FOXA1, FOSL2, JUN, NR3C1, USF1, MAX, CTCF, E2F6, CEBPB, JUNB and SMAD3.
   Among them, the loss of FOXA1 has been shown to drive EMT in various
   cancers.^[ [112]^22 ^] To further investigate the role of FOXA1 in
   TGFBR1 regulation, we searched within the FOXA1 ChIP peak using the
   Sequence Motif Location Tool ([113]https://molotool.autosome.org/).
   This analysis identified the putative FOXA1‐binding site located at
   ‐658 to ‐647 bp from the TGFBR1 transcription start site
   (Figure [114]4D). The supershift assay confirmed that this predicted
   site is a bona fide FOXA1 binding site (Figure [115]4E). We further
   searched the ChIP‐atlas database for other FOXA1 target genes in A549
   cells that overlap with TSPYL1 target genes identified in our current
   CUT&RUN and identified 586 genes that are common targets of both FOXA1
   and TSPYL1 (Figure [116]S6A,B; Table [117]S1, Supporting Information).

   Next, we tested whether TSPYL1 co‐operates with FOXA1 in vivo.
   Co‐localization between these endogenous proteins was detected by
   immunocytochemistry (Figure [118]S7A, Supporting Information) and
   further demonstrated by Proximity Ligation Assay (PLA), which detects
   protein molecules located within 40 nm (Figure [119]4F). To map the
   domain in TSPYL1 for interaction with FOXA1, we constructed the
   full‐length HA‐tagged TSPYL1 and its truncated versions. Only the
   full‐length TSPYL1 and the ΔN version containing the full NAP domain
   and C‐terminus interacted with FOXA1, while the truncation causing
   SIDDT did not (Figure [120]4G; Figure [121]S7B, Supporting
   Information).

   To gain insights into the dynamics of TSPYL1/FOXA1 complex binding to
   the TGFBR1 promoter during EMT, we treated A549 cells with TGFβ.
   Intriguingly, FOXA1 but not TSPYL1 remained bound to the TGFBR1
   promoter after 48 h of TGFβ treatment (Figure [122]4H). FOXA1 protein
   stability is enhanced through methylation by EZH2,^[ [123]^23 ^] which
   also serves as the methyltransferase for the repressive histone 3 mark
   H3K27me3. Previous studies have reported that H3K27me3 is associated
   with the suppression of TGFBR1 transcription.^[ [124]^24 ^] In
   agreement with this, we detected the presence of EZH2 binding and the
   H3K27me3 mark before the addition of TGFβ but not afterward
   (Figure [125]4H). Furthermore, TSPYL1 binding was abolished upon FOXA1
   KD (Figure [126]4I), which resulted in the loss of transcriptional
   repression of TGFBR1 and several TGFβ target genes including SMAD7,
   SERPINE1, and SNAI1, as well as SNAI1 target gene CDH2
   (Figure [127]4J). Taken together, the repression of TGFBR1 by TSPYL1 is
   dependent on FOXA1. The proximity of TSPYL1, FOXA1, EZH2, and H3K27me3
   to the TGFBR1 promoter (Figure [128]4D,H) further confirms that TSPYL1
   represses TGFBR1 through partnering with FOXA1 and EZH2.

2.4. TSPYL1 Knockdown Stabilizes TSPYL2 through TGFβ Signaling

   TSPYL2 is upregulated by the addition of TGFβ in certain cell lines and
   in mouse primary aortic smooth muscle cells.^[ [129]^8 , [130]^25 ^]
   Therefore, we examined whether TSPYL2 is elevated upon TSPYL1 KD via
   increased TGFβ signaling. Even though there was only a mild increase or
   no change in transcript level, TSPYL2 protein levels were significantly
   increased upon TSPYL1 KD in both A549 and HEK293FT, indicating
   enhancement of protein stabilization (Figure [131]5A,B). TSPYL2 is
   subjected to proteasome degradation.^[ [132]^26 ^] To further verify
   the enhanced TSPYL2 stability, we treated A549 and HEK293FT cells with
   cycloheximide (CHX) to inhibit protein synthesis and examined the
   half‐life of TSPYL2 in the control and TSPYL1 KD environment. Indeed,
   the rate of TSPYL2 protein reduction by 50% in TSPYL1‐depleted cells
   was slower than that of vector control (Figure [133]5C). Moreover,
   inhibition of TGFBR1 kinase activity prevented the up‐regulation of
   TSPYL2 in TSPYL1 KD A549 and HEK293FT cells, confirming that TSPYL2
   upregulation is dependent on TGFβ signaling in both cell lines
   (Figure [134]5D). TSPYL2 stabilization was impaired in TSPYL1 KD cells
   when SMAD2, but not SMAD3 was KD (Figure [135]5E), indicating the
   dependence on the canonical SMAD pathway. In addition, OE of TSPYL2
   increased transcript and protein levels of TGFBR1 and increased
   phosphorylation of SMAD2/3 (Figure [136]5F), in relation to increased
   expression of mesenchymal markers and reduced epithelial marker
   E‐cadherin upon further culture until morphological changes similar to
   EMT were observable (Figure [137]5G,H). To prove that TSPYL2 mediates
   the EMT phenotype in TSPYL1 KD, co‐KD of TSPYL1 and TSPYL2 was employed
   for the rescue experiments. Criteria for success include the
   attenuation of TGFβ induced SMAD2 phosphorylation (Figure [138]5I) and
   expression of TGFBR1 and CDH2 (Figure [139]5J,K). Importantly, cells
   regained the cobblestone morphology (Figure [140]5L). The data
   demonstrate that TSPYL2 stabilization is caused by TGFβ‐SMAD2 signaling
   and leads to EMT in A549 cells.

Figure 5.

   Figure 5
   [141]Open in a new tab

   Stabilization of TSPYL2 upon TSPYL1 knockdown mediates TGFβ signaling
   and EMT. A) TSPYL2 transcripts increased upon TSPYL1 knockdown in A549
   but not in HEK293FT cells. Cells were infected by lentivirus and
   subjected to qPCR three days after infection. Data are shown as the
   mean ± SD, n = 3, ^* p < 0.05, ^** p < 0.01, ns: not significant by
   unpaired Student's t‐test. B) TSPYL2 protein increased upon TSPYL1
   knockdown in A549 and HEK293FT cells. Cells were infected by lentivirus
   and subjected to immunoblotting 3 days after transduction. C) TSPYL2
   protein was stabilized upon TSPYL1 knockdown. Three days after
   infection, control and TSPYL1 knockdown cells were treated with
   cycloheximide (CHX; 50 µg mL^−1 for A549, 10 µg mL^−1 for HEK293FT
   cells) for the indicated time. Immunoblotting signals of TSPYL2 to
   loading control were quantified with image J and normalized to the
   ratio at 0 h treatment. Normalized signals were indicated and plotted
   in graph on the right. D) The upregulation of TSPYL2 upon TSPYL1
   knockdown was mitigated by the inhibition of TGFβ signaling in A549 and
   HEK293FT cells. After infection, cells were treated with DMSO or 10 µm
   SB‐431542 for 72 h (A549) or 24 h (HEK293FT) and subjected to
   immunoblotting. E) The upregulation of TSPYL2 upon TSPYL1 knockdown was
   abolished by knockdown of SMAD2 but not SMAD3. A549 cells were infected
   with lentiviruses for 24 h to knockdown TSPYL1, followed by
   transfection with siRNA targeting SMAD2 or SMAD3 for 24 h. After
   transfection, cells were cultured for another 36 h and subjected to
   immunoblotting. F) Three days after transduction with pBabe or
   Flag‐TSPYL2 in A549 cells, RNA and protein samples were collected for
   measuring the transcript level of TGFBR1 (left) or immunoblotting for
   the indicated proteins (right). Data are mean ± SD, n = 3. ^** p <
   0.01, unpaired Student's t‐test. G) Seven days after transduction with
   pBabe or Flag‐TSPYL2 in A549 cells, the protein level of epithelial
   marker E‐cadherin and mesenchymal markers as indicated were determined
   by immunoblotting. H) Representative images of control (pBabe) and
   Flag‐TSPYL2 A549 cells seven days after transduction. Scale bar: 50 µm.
   I–L) Rescue of TSPYL1 knockdown‐induced EMT phenotype by co‐knockdown
   of TSPYL2. A549 cells were infected with viruses as indicated. I) 3
   days after infection, cells were treated with 5 ng mL^−1 TGFβ1 for
   40 min and subjected to immunoblotting. The protein was quantified with
   Image J to indicate reduced pSMAD2 upon double knockdown. J,K) RNA and
   protein samples were collected 5 days after infection for qPCR and
   immunoblotting. Data are shown as the mean ± SD, n = 3, ^* p < 0.05,
   ^** p < 0.01, ^*** p < 0.001, ^**** p < 0.0001 by unpaired Student's
   t‐test. L) Images under bright field were taken 6 days after infection.
   Scale bar: 50 µm.

2.5. TSPYL2 is an Effector of TGFβ Signaling

   To refine the role of TSPYL2 in TGFβ signaling, we asked if TSPYL2 is
   part of the SMAD complex following the addition of TGFβ to regulate
   transcription. We overexpressed Flag‐TSPYL2 (gift from Dr. Pier Paolo
   Pandolfi) for immunoprecipitation with anti‐Flag conjugated beads to
   allow clear detection of pSMAD2/3. Preferential binding of TSPYL2 to
   pSMAD2 was detected (Figure [142]6A). For endogenous TSPYL2,
   co‐immunoprecipitation with SMAD2 was detected only upon the addition
   of TGFβ (Figure [143]6B). To allow more sensitive detection of
   transient interaction, PLA was employed to detect the interaction
   between TSPYL2 and components of SMAD2/3/4 complex according to the
   availability of suitable antibodies. Signals were detected under the
   basal condition in both the cytoplasm and nucleus. As expected, there
   was a significant increase in nuclear signals upon the addition of TGFβ
   (Figure [144]6C,D). The abundant signals in the cytoplasm also clearly
   indicated that TSPYL2 interacted with the SMAD complex before nuclear
   import. Next, we tested whether TSPYL2 works with SMAD2/3/4 to regulate
   the transcription of SMAD7, a well‐known direct target gene of TGFβ
   signaling.^[ [145]^17 ^] Since TSPYL1 was not detected at the TGFBR1
   promoter at 48 h of TGFβ treatment (Figure [146]4H) and OE of TSPYL2
   caused increased TGFBR1 (Figure [147]5F), we also checked whether
   TSPYL2 binds to the TGFBR1 promoter. In agreement with the notion that
   TSPYL2 is part of the SMAD2/3/4 signal transducer, TSPYL2 was already
   detected at the SMAD7 promoter 0.5 h after the addition of TGFβ.
   Importantly, the binding of TSPYL2 to the TGFBR1 promoter increased
   from 0.5 to 24 h after the addition of TGFβ and showed the same binding
   kinetics with SMAD3. TSPYL1 binding was weak when the binding of TSPYL2
   prevailed (Figure [148]7A). The binding patterns show that TSPYL1 and
   TSPYL2 regulate these two TGFβ pathway component genes in the opposite
   manner by alternative binding to the promoter.

Figure 6.

   Figure 6
   [149]Open in a new tab

   TSPYL2 mediates TGFβ transcriptional responses in A549 cells. A)
   Phosphorylated SMAD2/3 complex co‐immunoprecipitated with Flag‐TSPYL2
   under TGFβ treatment. The control and Flag‐TSPYL2 transduced A549 cells
   were treated with 5 ng mL^−1 TGFβ1 for 1 h and subjected to
   immunoprecipitation (IP) with Flag beads and immunoblotted for the
   indicated protein. B) Endogenous TSPYL2 immunoprecipitated with SMAD2
   under TGFβ treatment. A549 cells were treated with or without
   5 ng mL^−1 TGFβ1 for 1 h and subjected to IP with antibodies to TSPYL2
   or IgG as indicated on the top and immunoblotted for proteins as
   indicated on the left. C) Interaction between TSPYL2 and SMAD2/3/4
   complex upon the addition of TGFβ by proximity ligation assay (PLA).
   A549 cells were treated with or without 5 ng mL^−1 TGFβ1 for 1 h. White
   punctate staining showed the interaction between TSPYL2 and components
   of SMAD complex as indicated on the top and DAPI was used to reveal the
   nucleus. Scale bar: 10 µm. D) Quantitation of punctate staining after
   TGFβ treatment relative to no TGFβ treatment group. Data are shown as
   the mean ± SD, n = 5 images, ^**** p < 0.0001 by unpaired Student's
   t‐test.

Figure 7.

   Figure 7
   [150]Open in a new tab

   TSPYL1 and TSPYL2 oppositely regulate TGFβ target genes in A549 cells.
   A) ChIP‐PCR showing the time course of binding of TSPYL1, TSPYL2 and
   SMAD3 to SMAD7 and TGFBR1 promoters upon stimulation by TGFβ. A549
   cells were treated with 5 ng mL^−1 TGFβ1 for the indicated time and
   subjected to ChIP with indicated antibodies. IgG was used as the
   negative control. B) The effects of TSPYL1 or TSPYL2 knockdown on
   transcriptional responses to TGFβ were investigated by qPCR. A549 cells
   were treated with 5 ng mL^−1 TGFβ1 for the indicated time on x‐axis
   before sample collection at 7 days post‐transduction. Data are shown as
   the mean ± SD, n = 4.

   Finally, we verified the importance of TSPYL1 and TSPYL2 in regulating
   TGFβ target genes by measuring their transcript levels in TSPYL1 or
   TSPYL2 KD cells. As expected, there was an early increase in SMAD7 and
   a delayed increase in TGFBR1 transcript levels when TGFβ was added in
   control or KD cells. Furthermore, the ground level of all four tested
   genes including also SERPINE1 and SNAI1 was elevated in TSPYL1 KD
   cells, which agreed to increased TGFβ signaling in TSPYL1 KD cells. By
   contrast, there was an impaired transcriptional response to TGFβ in
   TSPYL2 KD cells (Figure [151]7B). Together, the data support the notion
   that TSPYL1 and TSPYL2 play a counter balancing role in the
   transcription of TGFβ target genes. TSPYL1 interacts with FOXA1 to
   suppress TGFβ signaling at basal state while TSPYL2 interacts with the
   SMAD complex to promote it upon stimulation.

2.6. TSPYL1 Regulates TGFβ Signaling In Vivo

   In mice, TGFβ is important during branching morphogenesis and late lung
   development.^[ [152]^27 ^] In neonates, the respiratory surface is
   greatly increased by subdivision of alveoli by septa which reduces the
   size but increases the number of alveoli. Either under or excess TGFβ
   signaling resulted in impaired alveolar septation.^[ [153]^27 ,
   [154]^28 ^] Since most Tspyl1 KO mice die ≈15 days after birth, we
   investigated whether there was any defect in neonatal lung development.
   From histological studies, there was indeed a septation defect in
   Tspyl1 KO mice as indicated by a reduced number of alveoli which were
   of larger sizes (Figure [155]8A,B). We further examined whether there
   was perturbed expression of Tgfbr1 and Smad7 in KO lungs. We detected
   upregulated transcript levels at embryonic day 15.5 and 16.5
   (pseudoglandular stage), and significant downregulation at postnatal
   day 7 and day 14 (alveolar stage; Figure [156]8C). In addition, we
   collected wildtype and KO MEF in litters derived from Tspyl1
   heterozygous parents. Tspyl1 KO MEF consistently showed elevated
   expression of Tgfbr1 transcript and protein (Figure [157]8D,E), which
   was associated with reduced proliferation and earlier replication
   arrest at passage 5 (Figure [158]8F). The data support a critical role
   of TSPYL1 in regulating the expression of Tgfbr1 in vivo.

Figure 8.

   Figure 8
   [159]Open in a new tab

   Tspyl1 knockout mice have mis‐regulated TGFβ signaling and associated
   phenotype. A) Representative lung sections from wildtype (Tspyl1^+/+)
   and knockout (Tspyl1^−/−) mice at postnatal day 14. Lungs were inflated
   before fixation and sections were stained with hematoxylin and eosin.
   Scale bar: 100 µm. B) Morphometric analysis of sections collected as
   above. Images were digitally captured at 20X magnification for
   determining the number of alveoli and mean liner intercept (MLI). Data
   are shown as the mean ± SD, n = 4 wildtype and 3 knockout lungs. ^** p
   < 0.01, ^*** p < 0.001 by unpaired Student's t‐test. C) qPCR of lung
   samples collected at the indicated time points. Day of birth was
   designated as P0. Data are shown as the mean ± SD, n = 3–9, ^* p <
   0.05, ^** p < 0.01, ^*** p < 0.001, ns: not significant by unpaired
   Student's t‐test. D,E) Expression of TGFBR1 in wildtype and knockout
   primary mouse embryonic fibroblasts (MEF). MEF were prepared from
   wildtype and Tspyl1 knockout littermates and collected at passage 3 for
   (D) qPCR, n = 6 per genotype and (E) immunoblotting, n = 4 per
   genotype. Data are shown as the mean ± SD, ^* p < 0.05, ^** p < 0.01 by
   unpaired Student's t‐test. F) Early replication arrest in Tspyl1
   knockout MEF. Following the 3T3 protocol, MEF collected from wildtype
   and knockout littermate embryos were seeded at a density of 3 × 10^5 on
   5 cm plates (passage 1). The number of cells was counted every 3 days
   and re‐seeded at the same density. The fold change was calculated by
   dividing the cell number with 3 × 10^5 at each passage. Fold change of
   one indicates replication arrest. Data are shown as the mean ± SD, n =
   6 per genotype, ^* p < 0.05, ^*** p < 0.001, two‐way ANOVA.

3. Discussion

   This study shows that TSPYL1 is a critical negative regulator of TGFβ
   signaling and EMT in cell models. In general, TSPYL1 binds the promoter
   of TGFBR1 to refine the cellular response to TGFβ. It suppresses the
   function of the closely related TSPYL2 which can act as a signal
   transducer of TGFβ. Mechanistically, TSPYL1 partners with FOXA1 and
   EZH2 to repress TGFBR1 to maintain the cellular level of TGFβ
   receptors. In both lung carcinoma A549 and neuroblastoma BE(2)‐C cells,
   TSPYL1 depletion causes EMT. The underlying mechanism involves the
   upregulation of TGFβ signaling via elevated transcription of TGFBR1.
   Upon stimulation by TGFβ, TSPYL2 forms a complex with SMAD2/3/4 to
   regulate TGFβ target genes. The changes in gene expression upon
   activation by TGFβ follow the switch of SMAD7 and TGFBR1 promoter
   occupancy by TSPYL1 and TSPYL2. Our findings underscore the
   counterbalancing role of TSPYL1 and TSPYL2 as transcription regulators
   of TGFβ signaling. The upregulation of TSPYL2 in TSPYL1 KD cells and
   genetic rescue by co‐knockdown of TSPYL1 and TSPYL2 provide insights
   that perturbed TGFβ signaling and TSPYL2 stabilization may be the
   underlying disease mechanism in SIDDT.

   TGFβ signaling has pleiotropic effects in both normal development and
   disease processes.^[ [160]^2 ^] Depending on the cellular context, it
   controls cell proliferation, survival, and migration. In our study, we
   uncover that TSPYL1 binds the promoter of TGFBR1 in a variety of cancer
   cell lines as well as normal cells. We also observed that TSPYL1
   depletion adversely affected the proliferation of all cell lines being
   tested and in MEF. Furthermore, besides TGFBR1, SMAD7 is also a direct
   TSPYL1 target. Both genes are pivotal in determining the cellular
   sensitivity to TGFβ signaling. Upon addition of TGFβ, the intracellular
   level of TGFBR1 determines the availability of cell surface receptors;
   while SMAD7 provides the negative feedback on TGFβ signaling.^[
   [161]^29 ^] Our TSPYL1 KD and OE studies point toward the role of
   TSPYL1 in repressing the transcription of these two TGFβ signaling
   component genes in cell models. In other words, TSPYL1 is essential to
   maintain the basal levels of TGFBR1 and SMAD7. TSPYL1 is ubiquitously
   expressed. TSPYL1 might repress various TGFβ target genes through
   forming the FOXA1/EZH2 complex in cell types with FOXA1 expression.
   Candidates can be considered from the list of overlapping TSPYL1/FOXA1
   target genes for future studies (Table [162]S1, Supporting
   Information).

   Does TSPYL1 regulate TGFβ signaling in vivo? We have investigated
   whether there is perturbed TGFβ signaling in Tspyl1 KO lungs as proper
   TGFβ and BMP signaling is important in alveolar development.^[ [163]^30
   ^] We found initially up and then down‐regulation of Tgfbr1 and Smad7
   transcripts in the mutant lung. This could be due to the presence of
   heterogeneous cell types and a complex in vivo environment that
   encompasses various factors such as ligand availability, signaling
   cross‐talks, and compensatory mechanisms to modulate the cellular
   response to the absence of Tspyl1. Nevertheless, evidence that our in
   vitro model also holds in vivo was gained from the upregulated Tgfbr1
   and impaired proliferation in both Tspyl1 KO and in vitro KD MEF.

   Our data indicates that TSPYL2 forms part of the SMAD2/3/4 complex upon
   activation by TGFβ and TGFBR1 is also a TSPYL2 direct target gene. As
   part of the SMAD signaling complex, TSPYL2 may be transiently protected
   from proteasomal degradation upon stimulation by TGFβ. Since induction
   of growth arrest can reduce MDM2‐mediated ubiquitination of TSPYL2,^[
   [164]^26a ^] TSPYL2 may be further stabilized through secondary events
   of cell proliferation arrest caused by TGFβ. Intriguingly, TSPYL1 KD
   stabilizes TSPYL2 in all cell lines we have tested including A549,
   HEK293FT, BE(2)‐C, MCF7, cervical cancer HeLa and melanoma A375
   (Figure [165]5B,C; Figure [166]S8, Supporting Information). We have
   only proved that this is dependent on TGFBR1 in A549 and HEK293FT. Our
   future work is to explore whether TGFβ related or other mechanisms
   exist to stabilize TSPYL2 upon TSPYL1 depletion in different cellular
   contexts. Knowledge of the various mechanisms to stabilize TSPYL2 will
   provide understanding on how TSPYL2 helps the cell to cope with
   stresses such as DNA damage.^[ [167]^31 ^] Independent studies have
   established that ectopic expression or induction of TSPYL2 by various
   stresses induces growth arrest or apoptosis in cancer stem cells,
   cancer cell lines, human embryonic lung fibroblasts, and human
   mesenchymal stem cells.^[ [168]^7 , [169]^26 , [170]^32 ^] It is
   possible that stabilized TSPYL2 and the negative effects on cell
   proliferation and differentiation may be the caveats in SIDDT. An
   effective peptide inhibitor for blocking TSPYL2 signaling axis in renal
   fibrosis has been developed.^[ [171]^9 ^] The efficacy of this drug in
   alleviating the disease process in Tspyl1 KO mice, due to altered TGFβ
   or other cellular pathways, should be tested in the future. Finally,
   our CUT&RUN data also revealed plenty of potential TSPYL1 target genes
   in the BMP, Hippo, Wnt, Notch, and MAP kinase pathways which are very
   important in development, homeostasis, and disease. These are all
   potential mechanisms to explore in order to decipher the biological
   importance of TSPYL1.

4. Conclusion

   This study establishes the essential roles of TSPYL1 and TSPYL2 in
   controlling the expression of key components in the TGFβ signaling
   pathway. Specifically, expression of both the upstream regulator TGFBR1
   and the negative feedback messenger SMAD7 are maintained by TSPYL1 in
   cell models. TSPYL1 deficiency augments TGFβ signaling, which is
   mediated through the stabilization of TSPYL2 as part of the signal
   transducer to regulate target gene expression. The findings provide
   mechanistic insights into the etiology of SIDDT.

5. Experimental Section

Animals

   All animal protocols were approved by the Committee on the Use of Live
   Animals in Teaching & Research at the University of Hong Kong (CULATR
   6008–22, 5281‐20). Tspyl1 KO line Δ463 on pure C57BL/6N background was
   maintained by breeding between Tspyl1 ^+/− heterozygotes.^[ [172]^13 ^]

PBMC isolation

   The protocol was approved by the Institutional Review Board (IRB Ref:
   UW 22–154). PBMC were isolated from buffy coats provided by the Hong
   Kong Red Cross using Lymphoprep (Thermo Fisher Scientific).

Materials and Reagents

   Plasmids, cloning primers, qPCR primers, ChIP‐PCR primers, and
   antibodies used in this study were listed in Tables [173]S2–S6
   (Supporting Information). TGFβ1 (PHG9214), Pierce Protease Inhibitor
   Tablets, EDTA‐free (A32965), Halt Protease and Phosphatase Inhibitor
   Single‐Use Cocktail (100X) (78442), were purchased from Thermo Fisher
   Scientific, TGFBR1 inhibitor SB‐431542 (HY‐10431) and MG132 (HY‐13259)
   from MedChemExpress, and protein synthesis inhibitor cycloheximide from
   Santa Cruz Biotechnology (sc‐3508).

Cell Culture

   BE(2)‐C, A549, MCF7, HeLa, A375, and HK‐2 cells were obtained from the
   American Type Culture Collection. HEK293FT cells were a gift from Dr.
   David Wai Chan. H9 cells were a gift from Dr. Cherie Lee and directly
   used for ChIP. BE(2)‐C, A549, MCF7, HeLa, A375, and HEK293FT were
   cultured in Dulbecco's Modified Eagle Medium (Invitrogen) supplemented
   with 10% fetal bovine serum (Hyclone), and HK‐2 cells were cultured in
   DMEM/F12 (Invitrogen) with 10% fetal bovine serum at 37 °C in 5% CO[2].
   All cell cultures were tested negative for mycoplasma.

MEF Isolation and 3T3 Assay

   MEF were isolated from embryos at 13.5 days of gestation as previously
   described,^[ [174]^33 ^] and cultured in Dulbecco's Modified Eagle
   Medium (Invitrogen) supplemented with 10% fetal bovine serum (Hyclone)
   and 1% penicillin/streptomycin (Invitrogen) at 37 °C in 5% CO[2]. For
   knockdown experiments, MEF ‐were pooled from the whole litter derived
   from wildtype parents. For experiments comparing between wildtype and
   Tspyl1 KO, litters were collected from heterozygous parents. MEF
   derived from each embryo were cultured individually and genotyped with
   primers as previously described.^[ [175]^13 ^] MEF were subcultured
   using the standard 3T3 protocol. Briefly, 3 × 10^5 cells were plated on
   a 5‐cm plate and cultured for 3 days. Cells were trypsinized for
   counting at day 3 and the same number of cells (3 × 10^5) were plated
   again for further passages. The proliferation curve was constructed by
   dividing the total number of live cells before each passage to the
   number seeded to give a fold difference.

Vector Construction

   Annealed sgRNAs targeting TSPYL1 (TSPYL1sg#1 targeting sequence:
   5’‐AGCGACCAGGACGCACACCA‐3’, and TSPYL1sg#2 targeting sequence: 5’‐
   CGGGCCGTGGCGGTACTCCC‐3’) were cloned into the BbsI digested PX459 (gift
   from Dr. Feng Zhang) for CRISPR/Cas9 mediated gene KO.^[ [176]^34 ^]
   Human TSPYL1 CDS was PCR amplified from the intronless TSPYL1 gene
   using human genomic DNA and cloned into the EcoRI and XhoI sites of a
   modified pcDNA3 with HA sequence inserted with indicated primers in
   Table [177]S2 (Supporting Information). The pcDNA3‐HA‐TSPYL1‐SIDDT
   plasmid was generated via site‐directed mutagenesis of pcDNA3‐HA‐TSPYL1
   with indicated primers in Table [178]S2 (Supporting Information). The
   truncated pcDNA3‐HA‐TSPYL1 ΔC was generated from pcDNA3‐HA‐TSPYL1 by
   digestion with PfImI and XhoI, treatment with T4 DNA polymerase, and
   religation. The truncated HA‐TSPYL1‐ΔN was amplified from
   pcDNA3‐HA‐TSPYL1 by PCR and ligated into pcDNA3. For stable expression,
   human TSPYL1 CDS with C‐terminal Flag was subcloned into pSIN (gift
   from Dr. Qi‐Long Ying) using SpeI and NsiI endonucleases with indicated
   primers in Table [179]S2 (Supporting Information),^[ [180]^35 ^]
   leading to the formation of pSIN‐TSPYL1‐Flag. The shRNA against human
   TSPYL2 (targeting sequence: 5’‐CATGGTGATTGTCAAGGAG‐3’) was subcloned
   into the lentiviral vector pLKO.1‐puro with endonucleases AgeI and
   EcoRI to generate pLKO.1‐TSPYL2sh#2.

Generation of Knockout Cells

   TSPYL1 KO BE(2)‐C were generated by transfection with PX459‐TSPYL1sg#2.
   TSPYL1 KO A549 cells were generated by transfection with combined
   PX459‐TSPYL1sg#1 and PX459‐TSPYL1sg#2. Cells were transfected with
   lipofectamine 2000 (Thermo Fisher Scientific, 11668019) and selected
   with 10 µg mL^−1 puromycin (Thermo Fisher Scientific). Isolated
   colonies were picked, expanded, and screened for TSPYL1 KO through
   immunoblotting. TSPYL1 frameshift mutations were confirmed by Sanger
   sequencing (Table [181]S7, Supporting Information).

siRNA Transfection and Lentivirus Transduction

   siRNA against FOXA1 (sc‐37930), SMAD2 (sc‐38374) and SMAD3 (sc‐38376)
   and control siRNA (sc‐37007) were purchased from Santa Curz Technology.
   siRNA transfections were performed using siRNA Transfection reagent
   (sc‐29528) with siRNA transfection medium (sc‐36868). Transfections
   were carried out following the manufacturer's instructions. For
   preparing the lentivirus medium, 5 × 10^6 HEK293FT cells were seeded
   onto a 10‐cm culture plate and incubated at 37 °C with 5% CO[2]
   overnight. Lentivirus transfection was conducted with the mixture of
   7.5 µg transfer plasmid, 5 µg psPAX2, 2.5 µg pVSVG, and
   60 µL 1 µg mL^−1 PEI in 1 mL OptiMEM (Invitrogen). The transfection
   complex was incubated for 20 min and added onto HEK293FT cells
   dropwise. After 24 h, the medium was refreshed. The virus was harvested
   at 48 and 72 h. For virus collection, the medium from the transfected
   HEK293FT cells was collected and filtered with a 0.45 µm PES filter.
   The viral medium was stored at –‐80 °C and thawed at 37 °C before use.
   The retrovirus medium was prepared with the same protocol except for
   using pCL‐Ampho as the packaging plasmid rather than psPAX2. For
   transduction, the viral medium with 8 µg mL^−1 polybrene (Sigma,
   TR1003) was added to the target cell lines when they reached ≈50%
   confluency. Unless otherwise stated, the target cell lines were
   infected twice with the neat virus medium for 48 h and then cultured
   with fresh medium for at least 72 h before conducting the experiments.

Scratch Assay and Transwell Assays

   For the scratch assay, cells were grown to confluency. A549 and BE(2)‐C
   cells were treated with 10 µg mL^−1 Mitomycin C (Sigma, 50‐07‐7) for 2
   and 3 h respectively to stop cell proliferation. Cells were wounded
   with a 200 µL pipette tip across the cell monolayer, washed, and
   cultured for another 16 h (A549 cells) or 24 h (BE(2)‐C cells). Phase
   contrast pictures were taken by an inverted microscope using a
   magnification of 20X. For the transwell migration assay, 3 × 10^4 A549
   were resuspended in 150 µL complete medium and seeded in the upper
   chamber of transwells cell culture inserts (Sigma, CLS3464) placed in a
   24‐well. A complete medium (600 µL) was placed in the lower chamber.
   Cells were incubated at 37 °C in 5% CO[2] for 24 h. Cells that migrated
   to the lower surface of the transwell were fixed with cold methanol for
   10 min, and then stained with 0.1% crystal violet (Sigma, C6158) for
   5 min. Non‐invading cells on the upper surface of the membrane were
   gently removed with cotton tips. The stained cells were counted from
   images taken under an inverted microscope at a magnification of 20X. In
   each experiment, the number of stained cells, in at least five
   independent image fields, were averaged. For the invasion assay, the
   above‐described procedure of transwell migration assay was conducted,
   except for the use of transwell inserts previously coated as follows:
   Matrigel (BD Bioscience, #356234) was diluted 4 times with culture
   medium. Thirty µL was coated on each transwell insert and allowed to
   solidify for 2 h in the tissue culture incubator. For BE(2)‐C cells,
   1 × 10^5 cells were resuspended in 100 µL and added to the upper
   chamber.

Immunoblotting

   Proteins were collected in RIPA buffer (50 mm Tris‐HCl pH 8,
   150 mm NaCl, 2 mm EDTA pH 8, 1% NP‐40, 0.5% Sodium Deoxycholate, 0.1%
   SDS) supplemented with complete protease inhibitors (Thermo Fisher
   Scientific, A32965). Total proteins were quantitated with Bio‐Rad DC
   Protein Assay Reagent B. Proteins were resolved by SDS‐PAGE in the
   Mini‐PROTREAN Tetra Cell system (Bio‐Rad). Proteins were transferred to
   Nitrocellulose membranes (Bio‐Rad #1620112) using a Mini Trans‐Blot
   system (Bio‐Rad) at 100 V for 2 h. Membranes were blocked with 5%
   non‐fat dried milk or 5% BSA (Sigma) in TBS with 0.1% Tween 20 (Sigma)
   for 1 h. Membranes were incubated in primary antibodies overnight and
   subsequently washed and incubated in secondary antibodies for 1 h at
   room temperature. Proteins were visualized using ECL (Thermo Fisher
   Scientific).

Immunocytochemistry

   Cells were seeded onto coverslips 24 h in advance. Cells were fixed
   with 4% PFA for 10 min, washed three times with PBS for 5 min, and then
   blocked with 2% normal goat serum (Thermo Fisher Scientific) in PBST
   (PBS + 0.1% Tween 20) for 1 h. The primary antibodies in 2% normal goat
   serum in PBST were added to the cells. After incubation overnight at
   4 °C, the cells were washed with PBST three times for 5 min each. The
   cells then were incubated with the secondary antibodies for 1 h and
   subsequently washed three times with PBS for 5 min each. To stain the
   cell nuclei, the cells were incubated with 1 µg mL^−1 DAPI diluted in
   2% normal goat serum in PBS for 5 min. The cells then were washed for
   5 min twice with PBST. Finally, the coverslips were mounted onto slides
   and examined under the confocal microscope (LSM 980 with Airyscan2).

Luciferase Assay

   A549 cells (2 × 10^5) were seeded onto a 96‐well plate to achieve ≈ 70%
   confluence on the day of transfection. The Dual‐Luciferase reporter
   assay system (Promega, E1980) was used according to the manufacturer's
   protocol. Briefly, a mixture containing Lipofectamine 2000, luciferase
   reporter plasmid DNA (pGL3‐SBE4‐luc (gift from Dr. Peter ten Diike ^[
   [182]^16 ^]) 100 ng, RLuc 10 ng per well) was added. After 24 h, cells
   were starved with serum‐free medium for 24 h, followed by another 24 h
   of treatment with 2 ng mL^−1 TGFβ1 in complete medium. Subsequently,
   cells were harvested by Passive lysis buffer, and luciferase activity
   was measured. Each assay consisted of at least three replicate wells.

RNA Extraction and Quantitative Real‐Time PCR Analysis (qPCR)

   Total RNA was extracted with Rnazol (Molecular Research Center, RN190)
   following the manufacturer's instructions. Total RNA was used for
   reverse transcription in 20 µL using oligo(dT) (Thermo Fisher
   Scientific) and SuperscriptII reverse transcriptase (Thermo Fisher
   Scientific, 18064022). qPCR was conducted with 0.5 µL of cDNA by using
   iTAQ Universal SYBR Green Supermix (Bio‐Rad, 1725122) in Roche
   LightCycler 480 II. HPRT was used as the internal control. The ΔΔCt
   method was used for calculating the relative expression.

CUT and RUN Assay and Bioinformatics Analysis

   CUT&RUN was performed according to the protocol using the
   high‐calcium/low‐salt digestion conditions with slight modifications.^[
   [183]^36 ^] All reactions were performed at 4 °C unless stated
   otherwise. Each centrifugation step was performed at 600 g for 3 min.
   Cells were washed twice with PBS and resuspended in Buffer NE1 (20 mm
   HEPES‐KOH, pH 7.9, 10 mm KCl, 0.5 mm Spermidine, 0.1% Triton X‐100, 20%
   Glycerol, protease inhibitors). After 10 min, cells were collected and
   incubated in Buffer 1 (20 mm HEPES‐KOH pH 7.9, 150 mm NaCl, 2 mm EDTA,
   0.5 mm Spermidine, 0.1% BSA, protease inhibitors) for 5 min. Cells were
   collected and washed once with Buffer 2 (20 mm HEPES, pH 7.5,
   150 mm NaCl, 0.5 mm Spermidine, 0.1% BSA, protease inhibitors) and
   resuspended in Buffer 2. Samples were incubated with TSPYL1 antibodies
   (ProteinTech, 13932‐1‐AP) or normal Rabbit IgG (Millipore, PP64B)
   overnight. Afterward, nuclei were collected and washed twice with
   Buffer 2 and then incubated with a pA‐MNase fusion protein (150 ng) for
   1 h. Nuclei were washed twice with Buffer 2 and once with low‐salt
   buffer (3.5 mm HEPES, protease inhibitors). Subsequently, samples were
   resuspended in low‐salt buffer and 10 mm CaCl[2] was added to each
   reaction. MNase digestion was stopped after 30 min with the addition of
   Buffer 2 with 10 mm EDTA, and 20 mm EGTA. The reactions were incubated
   for another 15 min at 37 °C. The DNA in the supernatant was collected
   and purified with a QIAQUICK PCR Purification Kit (QIAGEN). Libraries
   were constructed with KAPA Hyper Prep Kit (KR0961‐V1.14). The libraries
   were denatured and diluted to the optimal concentration. Illumina
   NovaSeq 6000 was used for Pair‐End 151 bp sequencing. For the data
   analysis, the raw sequencing reads were filtered for low‐quality
   sequences and adapter sequences followed by retaining only reads with
   read length over 40 bp. The filtered reads were mapped to Human Genome
   GRCh38 using Bowtie2 ([184]http://bowtie‐bio.sourceforge.net/bowtie2/).
   The peaks were called using MACS2
   ([185]https://github.com/taoliu/MACS). The peaks were annotated with
   HOMER ([186]http://homer.ucsd.edu/homer/). The KEGG pathway enrichment
   analysis was done at the Enrichr database
   ([187]https://maayanlab.cloud/Enrichr/).

Chromatin Immunoprecipitation (ChIP)

   ChIP was performed following the X‐ChIP protocol from Abcam. Briefly,
   cells were crosslinked with 0.75% formaldehyde (Sigma) in culture
   medium for 5 min at room temperature, and chromatin from lysed nuclei
   was sheared into 200–600 bp fragments using Diagenode BioruptorPico
   Sonicator. Chromatin was immunoprecipitated with antibodies of TSPYL1
   (ProteinTech, 13932‐1‐AP), FOXA1(Abcam, ab170933), TSPYL2 (ProteinTech,
   A304‐012A), SMAD3 (Abcam, ab40854), EZH2 (Cell Signaling Technology,
   #5246), H3K27me3 (Diagenode, pAb‐069‐050) or normal Rabbit IgG
   (Millipore, PP64B) overnight at 4 °C. Pre‐washed Protein A agarose
   beads (Santa Cruz Biotechnology, sc‐2001) were added and then further
   incubated for 2 h at 4 °C. The samples were washed once with low salt
   wash buffer (0.1% SDS, 1% Triton X‐100, 2 mm EDTA, 20 mm Tris‐HCl pH
   8.0, 150 mm NaCl), once with high salt wash buffer (0.1% SDS, 1% Triton
   X‐100, 2 mm EDTA, 20 mm Tris‐HCl pH 8.0, 500 mm NaCl) and once with
   LiCl washing buffer (0.25 m LiCl, 1% NP‐40, 1% Sodium Deoxycholate, 1
   mm EDTA, 10 mm Tris‐HCl pH 8.0). After washing, immunocomplexes were
   eluted in 120 µL of elution buffer (1% SDS, 100 mm NaHCO[3]) for 15 min
   at 30 °C. The immunocomplexes were reverse cross‐linked by incubating
   with RNase A and shaking at 65 °C overnight and then treated with
   proteinase K for 1 h at 60 °C. DNA was extracted with a QIAQUICK PCR
   Purification Kit (QIAGEN) for PCR.

Electrophoretic Mobility Shift Assay (EMSA)

   For collecting the nuclear extract, cells were incubated in the
   cytoplasmic buffer (10 mm Tris pH 7.5, 1.5 mm MgCl[2], 10 mm KCl, and
   0.1% Triton X‐100) for 1 min on ice. After centrifugation at 16000 g,
   4 °C for 5 min, the supernatant was discarded. The pellet was washed
   with cytoplasmic buffer once, and then incubated with nuclear buffer
   (20 mm Tris pH 7.5, 20% glycerol, 1.5 mm MgCl[2], 420 mm NaCl, 0.2 mm
   EDTA, and 0.1% Triton X‐100) on ice. The mixture was vortexed for 15 s
   every 10 min for a total of 40 min. After centrifugation for 10 min,
   the supernatant was collected. The FOXA1 binding site in the promoter
   of TGFBR1 was predicted with MoLoTool
   ([188]https://molotool.autosome.org/) and the oligonucleotide sequences
   used were FOXA1‐TGFBR1 sense 5’‐GGTACTATTATTATTTACATTTTAGAGATGAAGA‐3’
   and antisense 5’‐ TCTTCATCTCTAAAATGTAAATAATAATAGTACC‐3’. The
   biotinylated single‐stranded oligonucleotides were generated following
   the Biotin 3’ END DNA Labeling Kit (Thermo Fisher Scientific, #89818).
   Then, the complementary 3’ biotinylated single‐stranded
   oligonucleotides were annealed with the annealing buffer (10 mM Tris,
   1 mM EDTA, 100 mm NaCl) in the thermocycler set to Step 1: 95 °C 5 min;
   Step 2: Step down 1 °C per cycle for 70 cycles; Step 3: hold at 4 °C.
   The same annealing protocol was used to generate the unlabeled (cold)
   competitor. The labeled probes were diluted to 20 fmol µL^−1 and the
   unlabeled probes were diluted to 4 pmol µL^−1 as the stock. EMSA was
   performed using the LightShift Chemiluminescent EMSA Kit (Thermo Fisher
   Scientific, #20148). Briefly, the labeling efficiency of biotinylated
   probes was estimated by dot blot. Binding reactions were set up in a 20
   µL volume containing homemade binding buffer (12.5 mm Tris‐HCl, 62.5 mm
   NaCl, 1.24 mm DTT, 10% glycerol, 50 µg µL^−1 poly dI‐dC, 0.05% Nonidet
   P‐40) and 10 µg of nuclear proteins with or without 1 µg FOXA1
   antibodies (Abcam, ab170933). Binding reactions were pre‐incubated for
   20 min at room temperature. Labeled probes were added to the binding
   reactions and incubated for another 20 min at room temperature.
   Subsequently, 5X Loading Buffer was added to the reaction and run on a
   6% TBE gel using 0.5X TBE buffer (44.5 mm Tris borate, 1 mm EDTA pH 8.2
   – 8.4) at 380 mA for 40 min at 4 °C at 100 V for ≈1 h. The TBE gel was
   pre‐run for 1 h. The DNA was transferred to a Hybond‐N+ Nylon Membrane
   which was pre‐soaked in 0.5X TBE for 10 min. The DNA was then
   UV‐crosslinked to the membrane. For detection of the biotin‐labeled
   DNA, the membrane was blocked for 20 min using a blocking buffer. The
   membrane was then incubated in conjugated/blocking buffer for 15 min.
   The membrane was washed four times with 1x Washing Buffer for 5 min.
   The membrane was then incubated in a Substrate Equilibration Buffer for
   5 min and then incubated in a Substrate Working Solution for another
   5 min. The biotin‐labeled DNA was visualized using the ECL.

Proximity Ligation Assay (PLA)

   PLA was conducted following the Duolink PLA Fluorescence Protocol
   (Sigma, DUO92008). Cells were first cultured in chamber slides (Thermo
   Fisher Scientific). Cells were washed once in PBS and fixed with 4% PFA
   in PBS for 10 min at room temperature. Afterward, cells were washed
   with PBS three times for 5 min. Cells were then blocked with the
   Duolink Blocking Solution for 60 min at 37 °C. Subsequently, the
   blocking buffer was removed, and the antibodies were diluted in the
   Duolink Antibody Diluent added to the cells, and incubated for 1 h at
   37 °C. For the probe incubation, the PLUS (Sigma, DUO92002) and MINUS
   (Sigma, DUO92004) PLA probes were diluted in the Duolink Antibody
   Diluent and added to cells. The cells were incubated in a pre‐heated
   humidity chamber for 1 h at 37 °C. For the ligation, the PLA probe was
   tapped off and the Ligase in the Ligation buffer was added to the
   cells. The cells were incubated in a pre‐heated humidity chamber for
   30 min at 37 °C. To amplify the signal, the Polymerase was diluted in
   the Amplification buffer and added to the cells for another 100‐min
   incubation in a pre‐heated humidity chamber. After final washing, the
   slides were mounted with DAPI (Sigma, DUO82040) and analyzed in a
   confocal microscope (LSM 980 with Airyscan2).

Immunoprecipitation (IP)

   Protein samples collected as above were suspended in 1 mL IP buffer
   (150 mm NaCl, 10 mm Tris‐Cl pH 8.0, 1 mm EDTA, 1 mm EGTA, 1% Triton
   X‐100, 0.5% Nonidet P‐40) with protease inhibitors and MG132. Protein
   samples were incubated with primary antibodies or anti‐Flag conjugated
   beads (Thermo Fisher Scientific, A36803) with rotation at 4 °C
   overnight. Afterward, the samples incubated with primary antibodies
   were incubated with Protein A agarose beads for another 2 h at 4 °C.
   The samples were washed with IP buffer with protease inhibitors once
   and then washed with IP buffer three times. After washing, proteins
   were dissolved in sample buffer (60 mm Tris‐Cl pH 6.8, 10% glycerol,
   2.5% β‐mercaptoethanol, 2% SDS, 0.01% bromophenol blue) and proceeded
   to immunoblotting.

Lung Morphometric Analysis

   Wildtype and Tspyl1 knockout mice at postnatal day 14 were
   anesthetized. The lung was collected and inflated with 4%
   paraformaldehyde for overnight fixation at 4 °C. The fixed lungs were
   dehydrated and embedded in paraffin. Sections of 5 µm thickness were
   cut and stained with hematoxylin and eosin (H&E) using standard
   procedures. The images of H&E‐stained sections were taken with Axioplan
   2 imaging system (Zeiss) at 20X objective magnification and analyzed
   with Image J with the genotype blinded. The number of the alveoli was
   directly counted. To quantify the Mean Linear Intercept (MLI), equally
   distributed horizontal lines were placed in the image with the ‘Grid’
   tool in Image J. The spacing between each line was 50 µm. The linear
   distance between intercepting alveolar epithelia was measured. All
   intercepts along the lines were measured. For each image, the mean
   value of the intercepts was calculated with the formula that the MLI of
   each image = the total length of intercepts / the number of intercepts.
   The MLI of each lung was calculated by adding the MLI of each image /
   the number of images. Eight to 12 images were taken for each mouse
   lung.

Statistical Analysis

   The statistical analysis of all data was performed using GraphPad Prism
   8 software. Data are presented as mean ± SD and n is shown in figure
   legends. Statistical significance was determined using the two‐tailed
   unpaired Student's t‐test or two‐way ANOVA as indicated in figure
   legends. P < 0.05 was considered statistically significant. All
   cellular experiments were repeated at least once.

Conflict of Interest

   The authors declare no conflict of interest.

Author Contributions

   J.S., K.M.C., and S.Y.C. conceived the research. H.T. and S.Y.C.
   designed and analyzed the research. H.T., M.X.M., R.X.L, J.S., L.P.
   X.Z., E.H.W.L., L.Z. performed the experiment. K.M.C., M.C. advised the
   research and provided reagents, H.T. and S.Y.C. wrote the manuscript.

Supporting information

   Supporting Information
   [189]ADVS-11-2306486-s001.docx^ (1.8MB, docx)

   Supporting Table 1
   [190]ADVS-11-2306486-s002.xlsx^ (94.2KB, xlsx)

Acknowledgements