Abstract

   A progressive subclass of early-stage non-muscle-invasive bladder
   cancer (NMIBC) frequently recurs and progress into invasive carcinoma,
   thus decreasing the overall survival rate of NMIBC. However,
   therapeutic development for progressive NMIBC has been challenging due
   to the lack of molecularly validated in vivo models and agents
   targeting its genetic vulnerability. We herein molecularly
   characterized an interventional model of progressive NMIBC and revealed
   the principal functions and therapeutic potential of microRNA-145
   (miR-145) in early bladder tumorigenesis.
   N-butyl-N-(4-hydroxybutyl)nitrosamine-induced premalignant lesions
   (BiPLs) in rats exhibited downregulated expression of miR-145 as well
   as highly similar mutation/expression profiles to those of the human
   progressive NMIBC subclass with the worst prognosis. The expression
   patterns of miR-145 inversely correlated with those of BC-related
   oncogenes in BiPLs. We also demonstrated that miR-145 dominantly
   regulated interferon pathways and c-Myc expression, which play a
   crucial role in the pathogenesis of progressive NMIBC. Furthermore, we
   demonstrated that miR-145 replacement with a novel miR-145-based
   intravesical agent (miR-145S1) significantly inhibited the progression
   of BiPLs in vivo. These results provide insights into the essential
   role of miR-145 as the earliest-acting oncogenic driver of bladder
   tumorigenesis as well as a validated interventional model and novel
   miR-145-based nucleic acid therapeutic agent for progressive NMIBC.

   Keywords: MT: Non-coding RNAs, early bladder cancer,
   non-muscle-invasive bladder cancer, NMIBC, microRNA, microRNA-145,
   early bladder cancer rat model, intravesical therapy

Graphical abstract

   graphic file with name fx1.jpg
   [33]Open in a new tab
     __________________________________________________________________

   Heishima and colleagues revealed that
   N-butyl-N-(4-hydroxybutyl)nitrosamine-induced premalignant lesions in
   rats have mutation/expression profiles highly similar to those of
   early-stage non-muscle-invasive bladder cancer (NMIBC) with the worst
   prognosis. Among many genetic alterations in NMIBC, microRNA-145 is
   likely the principal regulator and the earliest-acting oncogenic driver
   of bladder tumorigenesis.

Introduction

   Bladder cancer (BC) is the most common urogenital cancer, with more
   than 430,000 new cases and 186,000 deaths being reported each
   year.[34]^1 Approximately 75% of patients are diagnosed and treated at
   an early stage of non-muscle-invasive bladder cancer (NMIBC).[35]^2
   However, despite an early diagnosis and treatment, NMIBC recurs in
   50%–70% of patients, with 10%–15% showing disease progression. The
   recurrence and progression of BC are mainly responsible for a
   progressive subclass of NMIBC, which is defined by specific molecular
   signatures (UROMOL2021; class 1, 2a, 2b, or 3)[36]^3 rather than the
   TNM subtype (Ta, papillary non-invasive carcinoma; T1, tumors
   infiltrating the lamina propria; Tis/CIS, carcinoma in situ). Class 2a
   progressive NMIBC has the worst prognosis due to its high rate of
   recurrence and progression into muscle-invasive bladder cancer (MIBC).
   Once these lesions progress to metastatic MIBC, the 5-year survival
   rate markedly decreases to 5%[37]^4 despite treatment with highly
   invasive radical cystectomy and urinary diversion. Therefore, the
   control of progressive NMIBC has attracted enormous interest in
   clinical fields.

   The therapeutic option to treat progressive NMIBC is currently limited
   to the intravesical administration of Bacillus Calmette-Guérin (BCG), a
   non-specific immunogenic substance. Although intravesical immunotherapy
   with BCG has been the frontline therapeutic modality for decades, its
   non-specific targeting mechanism results in a high recurrence rate and
   toxicity. NMIBC recurs in more than 53.3% of patients receiving BCG
   within 3.6 years[38]^5 and more than 60% of patients experience various
   degrees of adverse effects, including chemical cystitis, irritative
   voiding symptoms, malaise, severe sepsis, and allergic reactions (skin
   rash, arthralgia, and reactive arthritis).[39]^6 Other treatment
   modalities, such as transurethral surgery and systemic chemotherapy,
   are less effective for progressive NMIBC, particularly CIS, because
   these lesions are challenging to identify and reside only in the
   mucosal layer, which is poorly vascularized. These properties indicate
   the vital necessity of developing a novel intravesical agent that can
   directly reach the urothelial mucosa and specifically target the
   genetic vulnerabilities of progressive NMIBC.

   Major genetic alterations in BC include the dysregulation of numerous
   microRNAs (miRNAs), which are small non-coding RNAs that regulate the
   expression of multiple genes through RNA interference (RNAi).[40]^7
   Among these alterations, the downregulation of microRNA-145-5p
   (miR-145) is one of the genetic hallmarks of BC, being evident from
   early-stage NMIBC to late-stage MIBC.[41]^8^,[42]^9 miR-145 plays tumor
   suppressor roles in BC and functions as the central regulator to
   inhibit the expression of BC-associated oncogenes, including
   FSCN,[43]^10 c-Myc,[44]^11^,[45]^12 PTBP1,[46]^13^,[47]^14
   IGF1R,[48]^15 and MDM2.[49]^16^,[50]^17 These findings suggest a strong
   relationship between miR-145 expression and early bladder
   tumorigenesis.

   Despite the implicated relationship, the pathogenic contribution and
   therapeutic applications of miR-145 for progressive NMIBC have not been
   examined in detail, mainly due to the lack of a validated/suitable
   model. The BC-miR-145 relationship has been investigated in cell
   lines[51]^10^,[52]^13^,[53]^15 and cell-derived xenograft
   models.[54]^18 However, these models exhibit aggressive/homogeneous
   growth in an immune-deficient microenvironment,[55]^19 in contrast to
   NMIBC, which exhibits slow/heterogeneous growth in an intact immune
   environment. Patient-derived xenograft models have the same defect as
   cell-derived xenograft models, and thus are also unsuitable as NMIBC
   models.[56]^19

   Based on the characteristics of NMIBC, rat
   N-butyl-N-(4-hydroxybutyl)nitrosamine (BBN)-induced premalignant
   lesions (BiPLs) may serve as a suitable model of progressive
   NMIBC.[57]^20^,[58]^21^,[59]^22 BBN is a DNA alkylation compound
   derived from a chemical in cigarette smoke (N-nitroso-di-N-butylamine)
   and induces premalignant lesions that predominantly progress toward
   invasive carcinoma as a part of multiple carcinogenic
   steps.[60]^23^,[61]^24 Rat BiPLs share the characteristics of human
   NMIBC in terms of the pathogenic factor (exposure to the cigarette
   smoke-derived chemical) and heterogeneous/slow growth in an intact
   immune environment.[62]^23^,[63]^24 In addition, the rat BBN-induced BC
   model offers a larger body size and greater accessibility to the
   bladder than the mouse model, with resulting advantages in testing the
   efficacy of intravesical infusion therapy. However, the majority of
   studies using rat BBN-induced BC models have focused on late-stage
   lesions[64]^21^,[65]^25 and not on early lesions, i.e., BiPLs.
   Therefore, limited information is currently available on the molecular
   characteristics and suitability of rat BiPLs as a model of human
   progressive NMIBC.

   In this study, we molecularly characterized rat BiPLs in terms of
   somatic mutations, histopathology, and oncogene expression. We
   identified that rat BiPLs had high molecular similarity to the human
   progressive NMIBC subclass with the worst prognosis (UROMOL2021 class
   2a). By using this molecularly defined model, we demonstrated that
   miR-145 predominantly regulated the major phenotypes of progressive
   NMIBC, which strongly suggests the involvement of miR-145 in the
   pathogenesis of BC. Furthermore, we developed a novel miR-145-based
   intravesical therapeutic agent and demonstrated its efficacy against
   progressive premalignant lesions in vivo.

Results

Expression of miR-145 in human and rat BC cells

   The downregulation of miR-145 is a well-known genetic alteration in
   human BC. However, the association between miR-145 expression and the
   early pathogenesis of BC and its prognostic contribution remain unclear
   in human patients and BBN-induced rat models.

   To address these issues, we initially investigated the expression
   levels of miR-145 in human and rat BC by utilizing BC cell lines and
   The Cancer Genome Atlas (TCGA)-registered BC clinical sample data. The
   expression of miR-145 was significantly downregulated in both human
   (T24 and 253JB-V) and rat (NBT-T1, NBT-L2B) BC cell lines compared with
   normal diploid urothelial cells (HBlEpC for humans and RNM for rats;
   [66]Figures 1A and [67]S1A). In addition, miR-145 was downregulated in
   human BC clinical samples registered in TCGA database ([68]Figure 1B).
   The downregulation of miR-145 was observed at all clinical stages
   ([69]Figure 1B), which is consistent with previous findings showing
   that its downregulation was already evident at the early stage of human
   BC.[70]^8

Figure 1.

   [71]Figure 1
   [72]Open in a new tab

   Expression and anti-tumor properties of miR-145 in human and rat
   bladder cancer cells

   (A) Expression levels of miR-145 in human and rat cell lines derived
   from BC or normal urothelial cells. Human BC cell lines, T24 and 253
   JB-V (JBV); rat BBN-induced BC cell lines, NBT-T1 (T1) and NBT-L2B
   (L2B); normal diploid cells, HBlEpC (HBL, human) and RNM (rat, primary
   culture). (B) Expression levels of miR-145 in human clinical samples of
   different stages from TCGA. N, normal tissues; T, tumor tissues. (C)
   Double-stranded RNA sequences of miR-145 wild-type (miR-145WT) and
   miR-145S1 with full-match passenger sequences and GG dangling. (D)
   Viable cell percent among human and rat BC cells treated with miR-145WT
   (W), miR-145S1 (S), or control miRNA (C) at 20 nM for 72 h. (E)
   Pathways significantly altered in human BC cells (JBV) treated with
   10 nM of miR-145WT or control miRNA for 48 h. A pathway enrichment
   analysis with Metascape: a p value (P-val) ≤ 0.01 and a q value (Q-val)
   ≤ 0.01. Differently expressed genes (DEGs): absolute FCs ≥ 1.5, p ≤
   0.05, and two-tailed, paired Student’s t test (n = 2). (F) A cluster
   analysis of DEGs with illustrations by heatmaps. Heatmaps are annotated
   with gray bars indicating interferon-stimulated genes (ISGs). (G) A
   qRT-PCR analysis of the expression of apoptotic ISGs in human and rat
   BC cells treated with 20 nM of miRNAs (W, S, or C) for 48 h. (H) An
   immunoblot analysis of IFN signaling pathways (p-STAT1/2, STAT1/2,
   SOCS7, and STAM) and miR-145-associated oncogenes (c-Myc, cyclin D1,
   FGFR3, PTBP1, and Fascin) in BC cells treated for 72 h with 20 nM of
   miRNAs (WT, S1, or C). (I) Scheme for the potential mechanism
   underlying miR-145-mediated anti-tumor effects. Data are presented as
   means ± SD. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001;
   one-way ANOVA with Dunnett’s post hoc test.

   On the other hand, comparisons between cancer tissue samples showed
   that high miR-145 expression levels in samples were associated with
   advanced clinical stages and short survival times ([73]Figures 1B and
   [74]S1B). This result is in direct contrast to the well-known
   anti-tumor characteristics of
   miR-145.[75]^10^,[76]^13^,[77]^15^,[78]^18 This inconsistency has been
   the subject of debate[79]^8^,[80]^26^,[81]^27; however, it may be
   attributed to contamination with smooth muscle, which strongly
   expresses miR-145.[82]^28 High-grade BC involves more stromal tissues
   containing smooth muscle or vascular smooth muscle ([83]Figure S1C).
   The expression levels of miR-145 in clinical samples strongly
   correlated with those of smooth muscle (αSMA) and vascular markers
   (CD31; [84]Figure S1D). The correlation between high miR-145 expression
   and an advanced stage and short survival time were eliminated when
   miR-145 levels were normalized to αSMA expression levels
   ([85]Figures S1E and S1F). These results were also consistent with the
   observation that these contradictory associations (high miR-145
   expression-poor prognosis) were only reported in studies that employed
   whole-tissue lysate-based methods to measure miR-145.[86]^26^,[87]^27
   In contrast, a study that used the in situ hybridization (ISH)
   technique, which distinguishes expression in urothelial cells from that
   in other cell types, reported that high miR-145 expression was
   associated with a favorable prognosis.[88]^8

   Collectively, these results indicated that the downregulation of
   miR-145 was evident in both human and rat cell lines, as well as in
   human clinical samples at earlier stages. The previously reported
   association between high miR-145 expression and poor survival may have
   been due to contaminated tissues that abundantly express miR-145.
   Therefore, ISH appears to be a suitable method for evaluating miR-145
   expression in urothelial tumor tissues.

Anti-proliferative effects of miR-145 in vitro

   To clarify the functions of miR-145 in bladder tumorigenesis, we
   examined the effects of wild-type miR-145 (miR-145WT, [89]Figure 1C) on
   the proliferation and cell death of BC cells. The transfection of
   miR-145WT significantly decreased the viable cell count in both human
   (T24 and 253JB-V) and rat (NBT-T1 and NBT-L2B) cell lines
   ([90]Figure 1D). The decrease observed in viable cells was due to
   cell-cycle arrest with the features of apoptosis, which is
   characterized by an increased sub-G fraction cell count and the
   slightly increased cleavage of PARP ([91]Figures S2A–S2C). These
   results suggested that the replacement of miR-145 had anti-oncogenic
   properties in both human and rat BC cells.

   In consideration of future clinical applications, we attempted to
   improve the efficacy of miR-145 by adding modifications to nucleotide
   sequences. We synthesized miR-145 derivatives (12 derivatives, named
   miR-145S1 to S12) with different chemical or sequence modifications
   ([92]Figures S3A and S3B). Among these, miR-145S1, with a fully
   complementary passenger sequence (full-match sequence) and GG overhang
   at the 3′ terminus (GG dangling), exhibited the strongest growth
   inhibitory activity in a human BC cell line (T24; [93]Figures 1C,
   [94]S3A, and S3B). Similar to the transfection of miR-145WT, that of
   miR-145S1 decreased viable cell counts in both human and rat BC cells.
   However, the anti-tumor efficacy of miR-145S1 was significantly higher
   than that of miR-145WT ([95]Figures 1D and [96]S2A–S2C). The
   transfection of miR-145S1 induced a stronger degree of cell-cycle
   arrest and apoptotic cell death ([97]Figures S2A–S2C). Affected cells
   showed the downregulation of a proliferation marker (p-Histone H3), an
   increased sub-G fraction cell count, and the cleavage of PARP and
   caspase-3 ([98]Figures S2A–S2C). These results suggested that miR-145S1
   has more favorable characteristics as a therapeutic agent. miR-145S1
   and miR-145WT both exerted negligible effects on normal diploid cells
   (HBlEpC and RNM; [99]Figures 1D and [100]S2A–S2C), indicating the tumor
   specificity of miR-145 replacement therapy.

Comprehensive analysis to identify functional target genes and pathways of
miR-145

   Although miR-145 has multiple target genes and pathways associated with
   bladder tumorigenesis, the mechanisms underlying its anti-tumor effects
   in BC have not yet been elucidated in detail. Therefore, we performed a
   transcriptome analysis to identify the main pathways contributing to
   the anti-tumor activity of miR-145 in BC.

   We found that the dominant phenotype induced by the replacement of
   miR-145 was the upregulation of interferon (IFN)/JAK-STAT1/2 signaling
   and its associated pathways ([101]Figures 1E and 1F). The “IFN
   signaling” pathway had the highest statistical significance (log10 p or
   q value less than −20), and most of the other enriched pathways were
   also associated with IFN signaling (5 out of 6 pathways;
   [102]Figure 1E). The majority of differently expressed genes (DEGs)
   were upregulated (143 out of 148 genes, 96.6%; [103]Figure 1F), whereas
   few were downregulated (5 out of 148 genes, 3.4%). Most of the
   upregulated DEGs were STAT1/2-regulated apoptotic IFN-stimulated genes
   (ISGs), such as IFNB1 (IFNβ), AIM2, XAF1, and OAS1 ([104]Figure 1F).
   Genes regulated by other members of the STAT family, such as STAT3,
   were not significantly enriched in DEGs, suggesting that miR-145
   exerted its anti-tumor functions mainly through STAT1/2 transcriptional
   factors rather than through other members of the STAT family.
   Furthermore, a qRT-PCR analysis showed that the replacement of miR-145
   significantly upregulated apoptotic ISGs (IFNB1, AIM2, IRF7, OAS1,
   OAS2, OAS3, PML, and XAF1, as well as their rat homologs) in all of the
   human/rat cell lines evaluated ([105]Figure 1G). The expression levels
   of ISGs were up to approximately 1,000-fold higher in cells treated
   with miR-145S1 than in steady-state cells ([106]Figure 1G), and its
   efficacy was markedly stronger than that of miR-145WT.

   Consistent with this result, an immunoblot analysis showed that
   human/rat BC cells transfected with miR-145S1 strongly induced the
   activation of IFN-STAT1/2 signaling (the upregulation of p-STAT1/2 and
   STAT1/2; [107]Figure 1H), albeit with weak effects on some cells
   treated with miR-145WT. The replacement of miR-145 also induced the
   concomitant downregulation of other oncogenes related to BC, including
   cyclin D1, FGFR3∗, Fascin∗, c-Myc∗, and PTBP1∗ (∗direct targets of
   miR-145; [108]Figure 1H). These results suggested that miR-145 exerted
   its anti-tumor effects by targeting multiple oncogenes, in addition to
   its strong efficacy in activating the IFN-STAT1/2 pathways.

   To assess the functional roles of STAT1/2 in bladder tumorigenesis in
   humans, we investigated the association between overall survival and
   STAT1/2 expression in clinical samples from BC patients using TCGA
   database. The results showed that the downregulation of STAT1/2
   significantly correlated with unfavorable outcomes in human BC
   ([109]Figure S4A). The results also suggested that the miR-145-STAT1/2
   axis was closely linked to the tumorigenesis and malignant progression
   of BC and that the interventional upregulation of the miR-145-STAT1/2
   axis may exert anti-tumor effects in human BC patients.

   We then attempted to elucidate the role of miR-145 in the upregulation
   of components of the IFN-STAT1/2 pathways. A previous study reported
   that miR-145 induced the activation of IFN via the RNAi-mediated
   inhibition of SOCS7, a negative regulator of IFN-STAT1/2 signaling, and
   these effects did not occur via the non-specific IFN activation induced
   by double-stranded RNA.[110]^29 Consistent with these findings, an
   immunoblot analysis showed that the replacement of miR-145
   downregulated the expression of SOCS7 in human/rat BC cells
   ([111]Figure 1H). Furthermore, the computational analysis identified
   another IFN suppressor, STAM as an unreported potential direct target
   of miR-145. Since STAM has not been validated as a direct target of
   miR-145, we performed a luciferase reporter assay to establish whether
   miR-145 directly bound to the 3′ UTR of STAM ([112]Figure S4B). The
   transfection of miR-145WT or miR-145S1 significantly suppressed
   luciferase activity, and this suppression of luciferase activity was
   successfully canceled in cells expressing the mutated binding site
   ([113]Figure S4C). Therefore, STAM was verified as a direct target gene
   of miR-145 ([114]Figures S4B and S4C), suggesting that miR-145 directly
   inhibited multiple IFN-STAT1/2 suppressors to induce the strong
   activation of the IFN-STAT1/2 pathways. To examine the roles of SOCS7
   and STAM in the IFN-STAT1/2 pathway and c-Myc expression, we performed
   siRNA-based knockdown in BC cells highly expressing c-Myc (human, JBV;
   rat, L2B). The results showed that the siRNA-mediated knockdown of
   SOCS7 or STAM induced a significant decrease in viable cells and the
   upregulation of STAT1/2 ([115]Figures S4D and S4E). Notably, the
   knockdown of SOCS7 or STAM markedly downregulated the expression of
   c-Myc ([116]Figure S4E), suggesting that the effects of miR-145 on
   c-Myc expression were through a synergetic mechanism involving both
   RNAi-mediated direct targeting[117]^11^,[118]^12 and an
   IFN-STAT1/2-mediated indirect pathway. These results were consistent
   with previous findings showing that IFN signaling negatively regulated
   c-Myc expression.[119]^30^,[120]^31 The downregulation of c-Myc may not
   be a result of the downregulation of the PI3K/Akt pathway because p-Akt
   expression did not correlate with the downregulation of c-Myc in BC
   cells following the knockdown of SOCS7 or STAM ([121]Figure S4E).

   In most cases, the efficacy of miR-145S1 to regulate these oncogenes
   was higher than that of miR-145WT, and miR-145S1 and miR-145WT both
   induced similar molecular changes in human and rat BC cells. These
   results suggested that miR-145S1 possessed strong anti-tumor properties
   and that the modifications added to miR-145S1 did not markedly affect
   the spectrum of the target genes, albeit with some exceptions. These
   exceptions included cyclin D1, which was downregulated by the
   transfection of miR-145 under most conditions, but was upregulated only
   in rat BC cells treated with miR-145WT. In addition, FGFR3 was
   downregulated in human BC cells transfected with miR-145WT or
   miR-145S1; however, FGFR3 was not expressed in rat BC cells. This
   result was consistent with previous findings showing that the rat
   BBN-induced BC model typically lacked the expression of FGFR3.[122]^32

   The normal urothelial cells had only weak expressions of c-Myc, cyclin
   D1, and FGFR3, and both miR-145WT and miR-145S1 exerted minimal effects
   on these normal urothelial cells. As an exception, STAT1/2 signaling
   was upregulated in a human normal urothelial cell line (HBL).

   Collectively, these results suggest that the replacement of miR-145
   exerts anti-proliferative effects on BC cells through the regulation of
   the IFN-STAT1/2 pathway and also that the underlying mechanisms appear
   to be shared between humans and rats ([123]Figure 1I). Furthermore,
   miR-145S1 displayed a similar target gene spectrum, but exerted more
   potent inhibitory effects on its targets than miR-145WT.

Gross and histopathological characterization of rat BiPLs

   The above-described in vitro results suggested that miR-145 plays
   essential roles in the pathogenesis of BC and may be targeted as a
   therapeutic strategy for BC. Therefore, we conducted in vivo studies
   using a rat BBN-induced BC model to assess the potential roles of
   miR-145 in the regulation of the progressive premalignant phenotypes.

   BBN-initiated rats develop premalignant lesions, i.e., BiPLs, which
   sequentially progress toward invasive malignant tumors. Rat BiPLs show
   a papillary growth pattern that progresses from papillary/nodular (PN)
   hyperplasia to invasive carcinoma, which differs from that of human CIS
   (the major phenotype of progressive NMIBC) showing a flat growth
   pattern that evolves toward invasive carcinoma.[124]^23^,[125]^24
   However, the majority of rat BiPLs progress toward invasive
   carcinoma,[126]^23^,[127]^24 and their strong progressive
   predisposition is rather similar to human CIS than human Ta tumors,
   which have a similar morphology but less potential to progress toward
   invasive carcinoma. Despite morphological differences, the BiPL model
   may serve as a model of human progressive NMIBC, given its strong
   progressive predisposition.[128]^23^,[129]^24 However, previous studies
   using BBN-induced rat models mainly focused on advanced/late-stage
   cancers; therefore, early bladder lesions, i.e., BiPLs, have not yet
   been characterized in detail, and the molecular similarities between
   rat BiPLs and human progressive NMIBC, as well as their utility as a
   model, remained unknown.

   To establish whether the BiPL model serves as a model of human
   progressive NMIBC, we assessed its histopathological and molecular
   characteristics. A recent study classified human NMIBCs into four
   subclasses with different prognoses and recurrence rates based on their
   mutation/expression profiles (UROMOL2021; class 1, 2a, 2b, or
   3).[130]^3 The study identified that the molecular profiles were more
   predictive for the prognosis and recurrence rates of NMIBC, rather than
   the conventional morphological characteristics or stage (Ta, T1, or
   CIS).[131]^3 Therefore, we focused on investigating class-defining
   molecular markers or signatures that may be conserved between humans
   and rats.

   BiPLs were generated by the free intake of water containing 0.05% BBN
   and 0.005% Tween 80 for 16 weeks. After an additional 2 weeks without
   BBN exposure, rats at 18 weeks exhibited mild urothelial proliferation
   in the bladder ([132]Figure 2A). To perform consistent gross and
   histopathological evaluations, bladder samples were fixed in situ by a
   uniform infusion of 300 μL of 4% paraformaldehyde into the bladder and
   were then sliced in the sagittal plane. Urothelial proliferation was
   grossly observable in in-situ-fixed samples, but was difficult to
   identify in samples processed under the normal fixation protocol
   ([133]Figures 2A and [134]S5A).

Figure 2.

   [135]Figure 2
   [136]Open in a new tab

   Pathological characterization, grading, and miR-145 expression of rat
   BiPLs

   (A) Representative gross images of bladders treated with BBN (BBN) or
   not exposed to BBN (intact). Stars and arrowheads indicate intact and
   hyperplastic areas, respectively. Scale bars, 2 mm. (B) H&E images and
   the histopathological grading of lesions in the rat BiPL model. Lesions
   included papillary/nodular hyperplasia (PN hyperplasia) and papillary
   urothelial carcinoma (PUC). Lesions were classified into three grades
   as follows: N, no morphologically abnormal proliferation; grade I,
   early PN hyperplasia with a lesion length of less than 500 μm; grade
   II, late PN hyperplasia with a lesion length of greater than 500 μm and
   PUC. Scale bars, 200 μm for low-magnification (low mag) or 50 μm for
   high-magnification (high mag) images. (C) The counts and percentages of
   each histological type and grade observed in this study. (D) Definition
   of the grading system for rat BiPL in this study. (E) Representative
   images of in situ hybridization (ISH) for miR-145 in normal bladder
   tissue. Signals were visualized in blue using the ALP-NBT/BCIP system
   with red counterstaining with Nuclear Fast Red. (F) Representative
   images of miR-145 ISH in BiPLs and normal bladder tissues. A color
   lookup table (LUT) was used to illustrate NBT-BCIP signal intensities;
   high and low expression levels are colored in red and blue to black,
   respectively. Scale bars, 30 μm. A detailed version is available in
   [137]Figure S5B. (G) Relative ISH staining intensities of miR-145. N
   and PL indicate normal and BiPL tissues, respectively.

   Histopathologically, model rats developed PN hyperplasia and papillary
   urothelial carcinoma (PUC) ([138]Figure 2B). PN hyperplasia was
   predominantly observed, accounting for 96.8% of the total lesion count
   ([139]Figure 2C). A few PUC lesions (3.2% of the total) were observed;
   however, they were early-stage non-invasive PUC with limited malignant
   areas. CIS was not identified in the rat model, which is consistent
   with previous findings showing that the BBN-induced rat model rarely
   develops CIS.[140]^21 These results indicated that rat BiPLs have
   similar morphological characteristics to those of human Ta tumors, as
   described above.

   We classified lesions into three grades based on their
   histopathological features and used them for morphological evaluation
   in this study ([141]Figure 2D). The grading system used was as follows:
   N, normal; grade I, early PN hyperplasia with a lesion length of less
   than 500 μm; grade II, late PN hyperplasia with a lesion length of
   greater than 500 μm or PUC. According to the grading system, we
   identified 123 grade I and 35 grade II lesions (grade I/II ratio, 0.28)
   as the total lesion count in 9 rats ([142]Figure 2C).

Downregulation of miR-145 in rat BiPLs

   The downregulation of miR-145 is a hallmark of bladder tumorigenesis
   evident from the early stage in humans[143]^9; however, the expression
   status of miR-145 and its association with bladder tumorigenesis in
   rats remained unclear. Therefore, we examined the expression status of
   miR-145 in rat BiPLs.

   Rats initiated by BBN developed multiple premalignant lesions of
   different stage. The lesions were very near to each other and adjacent
   to normal urothelial cells. A PCR-based assessment using bulk tissue
   lysates was not suitable for a reliable analysis because the samples
   could have been contaminated with lesions of different grade, normal
   urothelial cells, or smooth muscle tissues highly expressing miR-145.
   To analyze the expression of miR-145 with good spatial resolution, we
   utilized the miRNA ISH technique to detect miR-145 molecules in
   heterogeneous samples.

   miR-145 ISH showed that normal urothelial cells had the expression of
   miR-145, which was observed as a finely stippled blue-positive stain in
   the cytoplasm or occasionally in nucleoli under NBT/BCIP staining
   ([144]Figure 2E). The expression of miR-145 was already downregulated
   in PN hyperplasia cells (“Premalignant” in [145]Figures 2F and 2G)
   compared with adjacent urothelial cells with normal morphological
   characteristics (a uniform number of urothelial cell layers with no
   nuclear/cytoplasmic atypia) and not expressing premalignant markers
   ([146]Figure S5B). The results suggested that the downregulation
   occurred in the early stage of tumorigenesis, similar to its timing in
   human BC. Moreover, miR-145 was highly expressed in arterial smooth
   muscle and inner circular and outer longitudinal smooth muscle
   ([147]Figure 2E), which is consistent with previous findings.[148]^28
   Expression levels were markedly higher in smooth muscle than in normal
   urothelial cells; therefore, contamination with smooth muscle tissues
   may obscure actual differences between BiPL cells and normal urothelial
   cells. We also performed an miRNA qRT-PCR analysis using bulk tissue
   lysates, but found no significant differences between BiPL and normal
   urothelial samples as predicted ([149]Figure S5C).

Genetic anomalies in rat BiPLs

   To further characterize rat BiPLs, we assessed single base substitution
   (SBS) mutation signatures, somatic mutations, copy number variants
   (CNVs), and structural variants (SVs) using next-generation sequencing
   (NGS). Three rat BiPL tissues (P1-3) and rat normal urothelial tissues
   (N) from a BBN-treated bladder were sampled under a stereomicroscope.
   Rat BiPL-associated genetic anomalies were identified by comparing rat
   BiPL and normal urothelial tissues. The mutation profiles of rat BiPLs
   were compared with those of human NMIBCs
   (UROMOL2016/2021).[150]^3^,[151]^33

   The SBS mutation signatures of rat BiPL samples showed significant
   similarities to those of human BC, with a cosine similarity of greater
   than 0.92 (p < 0.01; cosine similarity: highest at 1, lowest at −1;
   [152]Figure 3A). In addition, the SBS signatures of BiPLs showed
   significant similarities to the mutation signatures associated with
   cigarette smoke-related BC ([153]Figure 3A), indicating that the
   mutational signatures of BiPL mimicked those of the most prevalent
   phenotype in human NMIBC. Notably, rat BiPLs had APOBEC mutation
   signatures ([154]Figure 3A), which are class-defining markers for the
   progressive NMIBC subclass with the worst prognosis (UROMOL2021 class
   2a),[155]^3 demonstrating the similar mutational characteristics
   between rat BiPLs and human progressive NMIBC.

Figure 3.

   [156]Figure 3
   [157]Open in a new tab

   Mutation analyses of rat BiPLs

   (A) Comparisons of single base substitution (SBS) mutation signatures
   between rat BiPLs and human BC. “Sample” shows the profiles of BiPL SBS
   signatures. “Model” shows the profiles built by proportionally
   combining reported human BC signatures. Sample-model cosine
   similarities (highest at 1, lowest at −1), signature contributions, and
   the statistical significance were calculated with Signal. The
   signatures associated with cigarette smoke exposure and APOBEC are
   marked with asterisks (∗) and daggers (†), respectively. (B) Types,
   counts, and distributions of exonic mutations in each BiPL sample. (C)
   Counts and percentages of each type of exonic mutation. (D) Comparative
   mutation maps indicating the locations of the mutations identified in
   rat BiPL and human BC samples with information about the adjacent
   domains. Pathogenic mutations from human BC clinical samples were
   obtained from COSMIC. Asterisks (∗), hotspots of PIK3CA. (E) A list of
   exonic mutations shared between human BC and rat BiPLs. The mutation
   percentages in human NMIBC (Mut. % hu-NMIBC) and BC (Mut. % hu-BC) were
   obtained from the study by Hedegaard et al. and TCGA, respectively.
   SNV, single nucleotide variant. (F) Circos plot showing the functional
   overlap of mutations at early BC and miR-145-regulated genes. The
   strings connecting the circle-shaped bars show the association between
   genes mutated in early BC (from whole-genome sequencing [WGS]) and
   genes regulated by miR-145 (from a transcriptome analysis). The purple
   and light blue strings connect identical genes and genes in the same
   functional group, respectively. (G) Heatmaps showing overlaps in the
   significantly enriched pathways “mutations at early BC” and
   “miR-145-regulated genes.” A detailed version is available in
   [158]Figure S6B. (H) Scheme showing the potential mechanism underlying
   IFN-associated tumorigenesis in early BC.

   Rat BiPLs had different types and counts of somatic mutations in their
   coding sequences (exonic mutations; [159]Figures 3B and 3C). The most
   frequent mutation type was non-synonymous single-nucleotide variants
   (87.3%), and mutations leading to substantial changes in the protein
   structure (frameshift indel, startloss, stopgain, and stoploss
   mutations) accounted for 9.1% of the total ([160]Figure 3C). These
   mutations frequently occurred on chromosomes 7, 12, 15, and 20
   (mutation frequency per chromosome, densely mutated region counts per
   chromosome; [161]Figure S6A), consistent with the chromosomal
   aberrations observed in human BC.[162]^34^,[163]^35^,[164]^36^,[165]^37
   In particular, the result showing frequent mutations on chromosome 15
   in rat BiPLs was consistent with the observation that genetic
   variations on chromosome 15 correlated with the risk of cigarette
   smoke-associated BC in humans.[166]^36

   BiPLs harbored mutations in the homologs of frequently mutated genes in
   human NMIBC (Birc6, Dync1h1, Pik3ca, Mycbp2, Atm, and Crebbp;
   [167]Figures 3D and 3E). Notably, we identified mutations in Pik3ca,
   one of the most frequently mutated oncogenes in human NMIBC, with
   mutation sites located near the hotspots of activating PIK3CA mutations
   in humans ([168]Figures 3D and 3E).[169]^38 However, no mutations were
   observed in Fgfr3, which is consistent with previous findings showing
   that rat BBN-induced tumors did not harbor Fgfr3 mutations.[170]^32

   Of note, a pathway enrichment analysis using Metascape revealed that
   the functional spectrum of mutated genes highly overlapped with that of
   miR-145-regulated genes ([171]Figure 3F). We found that the overlapped
   genes were enriched in the IFN-STAT1/2-associated pathway
   ([172]Figure 3G). We also identified the mutations in crucial apoptotic
   ISGs, including DAP-5/Eif4g2, Nup98, Pom121, Trp, and Nedd4
   ([173]Figure 3G). These results suggested that mutations in the early
   stage and the downregulation of miR-145 cooperatively contributed to
   the core mechanism of IFN-STAT1/2-mediated bladder tumorigenesis and
   also that miR-145 is the key targetable gene to treat early BC
   ([174]Figure 3H).

   Several CNVs were identified in BiPLs; however, fold changes were small
   (less than 2; [175]Figures S6B–S6D). These CNVs were mainly associated
   with the metabolism of arachidonic acid ([176]Figures S6E and S6F),
   suggesting that they were also associated with changes in the
   regulation of inflammation in BiPLs. No significant SV was detected in
   BiPL samples.

Gene expression profiles of rat BiPLs

   To assess the expression profiles of rat BiPLs, we
   immunohistochemically stained BiPL tissues (18 or 22 weeks after the
   start of BBN exposure) for BC-associated oncogenic proteins. These
   proteins were selected from the class-defining markers of the human
   NMIBC classification system (UROMOL2021)[177]^3 and were limited to the
   proteins likely conserved between humans and rats. Expression patterns
   were compared between BiPL cells (labeled “Premalignant”) and
   morphologically intact urothelial cells (labeled “Normal”;
   [178]Figure 4A). The proteins examined included cell cycle markers
   (early cell cycle marker, cyclin D1; late cell cycle marker, cyclin A2;
   Ki-67), a p53 pathway component (p53), IFN-STAT1/2 pathway components
   (activated IFN markers, p-STAT1/2; IFN suppressors, SOCS7 and STAM),
   proteins involved in cell metabolism (c-Myc and PTBP1), a protein
   involved in cytoskeletal regulation (Fascin), and RTK-PI3K-RAS pathway
   components (FGFR3, PIK3CA, HRAS, KRAS, p-Akt, Akt, p-ERK1/2, and
   ERK1/2) in nine samples of each type ([179]Figures 4B, 4C, and
   [180]5A–5C).

Figure 4.

   [181]Figure 4
   [182]Open in a new tab

   Expression of bladder cancer-associated markers in rat BiPLs

   (A) Representative H&E images of the areas of BiPLs (labeled
   “Premalignant”) at 18 weeks and of morphologically normal urothelial
   tissues (labeled “Normal”). (B and C) Heatmaps (B) and representative
   immunohistochemical images (C) showing the different expression of
   BC-associated markers in BiPL (Premalignant) and normal urothelial
   tissues (Normal). Scale bars, 20 μm. ∗∗p < 0.01, ∗∗∗p < 0.001,
   ∗∗∗∗p < 0.0001; two-tailed unpaired Student’s t test (n = 8).

Figure 5.

   [183]Figure 5
   [184]Open in a new tab

   Progression-dependent changes in bladder cancer-associated marker
   expression in BiPLs

   (A) Histopathological grading-dependent changes in the expression of
   cyclin D1 and Ki-67 in BiPLs and normal tissues. Representative H&E and
   IHC images of BiPL samples of different histopathological grades and
   normal tissues are shown, along with the counts of cells positive for
   cyclin D1 and Ki-67. N, normal; G1, grade I; G2, grade II. Scale bars,
   20 μm. Data are presented as means ± SD. ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001;
   one-way ANOVA with Dunnett’s post hoc test. (B and C) Representative
   IHC images (B) and heatmaps (C) showing time course changes in disease
   marker expression in BiPL samples. N, normal; 18w, BBN-induced tumor
   development for 18 weeks; 22w, BBN-induced tumor development for
   22 weeks. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001;
   two-tailed unpaired Student’s t test. (D) A list of markers with
   positive expression in BiPL (BiPL-positive markers) and their
   corresponding UROMOL2021 subclasses. (E) Summary of the UROMOL 2021
   classification system and the markers used to distinguish the classes
   in this study. Asterisks (∗) indicate markers with positive expression
   in BiPL.

   BiPLs in rats at 18 weeks had similar expression profiles to those of
   human class 2a NMIBC (c-Myc+, cyclin A2+/−, p-STAT1−, FGFR3−) with a
   slight mixture of class 1 and class 3 phenotypes (cyclin D1+, p53−;
   [185]Figures 4B and 4C). Most BiPLs exhibited a further enhanced class
   2a phenotype at 22 weeks (increased positivity for cyclin A2, p53, and
   p-Akt with the low expression of p-STAT1; [186]Figures 5B and 5C). At
   22 weeks, BiPLs were still negative for FGFR3, HRAS, KRAS, and PIK3CA
   ([187]Figures 5B and 5C). These results indicated that most BiPLs had
   the class 2a-like phenotype ([188]Figures 5D and 5E), i.e., the gene
   expression profile of the human progressive NMIBC subclass with the
   worst prognosis. In addition, the phenotype was consistent with
   previous findings showing that genetic anomalies in Fgfr3, Hras, and
   Kras were rare in rat BBN-induced BC models.[189]^32^,[190]^39

   Notably, these results demonstrated that the expression of cyclin D1 in
   rat BiPLs may serve as the earliest disease progression marker even
   though cyclin D1 is regarded as a marker for benign-oriented subclasses
   in the human NMIBC classification system (classes 1 and 3). The
   earliest lesions in the BBN-induced model were positive for cyclin D1
   ([191]Figure 5A), and cyclin D1-positive cell counts significantly
   increased as the grade became higher ([192]Figure 5A). This result is
   consistent with previous findings showing that cyclin D1 is the
   definitive disease progression marker that distinguishes the
   premalignant lesions that eventually progress to invasive carcinoma in
   the rat BBN-induced model.[193]^39 Furthermore, comparisons between the
   expression of cyclin D1 and Ki-67 in serial sections revealed that many
   premalignant urothelial cells showed a cyclin D1+/Ki-67− profile,
   whereas the number of cyclin D1+/Ki-67+ cells was small and limited to
   the area close to the basement membrane ([194]Figure 5A). These results
   suggested that cyclin D1 was overexpressed prior to cell cycle entry
   (Ki-67 overexpression); thus, the upregulation of cyclin D1 was not
   just a consequence of cell proliferation but an essential event
   triggering early bladder tumorigenesis in the rat BiPL model.

   We also found that many direct targets of miR-145 (Fascin, SOCS7, STAM,
   c-Myc, and PTBP1) were upregulated in BiPLs, as demonstrated by the
   positive-stained area (Fascin) or staining intensity (SOCS7, STAM,
   c-Myc, and PTBP1; [195]Figures 4B and 4C). The expression patterns of
   Fascin, SOCS7, and STAM were consistent with those in human BC samples
   ([196]Figure S7). Notably, BiPLs exhibited the co-overexpression of
   c-Myc and cyclin D1, implying the contribution of the
   miR-145-c-Myc-cyclin D1 axis to the pathogenesis of premalignant
   lesions. Moreover, BiPLs showed the suppression of IFN-STAT1/2
   signaling (downregulation of p-STAT1; [197]Figures 4B and 4C), which
   was highly consistent with the downregulation of miR-145 and
   overexpression of IFN suppressor genes (SOCS7 and STAM) in BiPLs. In
   addition, since the low expression of IFN and high expression of c-Myc
   are class-defining markers for UROMOL2021 class 2a, this result further
   supported the hypothesis that the main phenotypes of rat BiPL are
   highly similar to those of human progressive NMIBC subclass with the
   worst prognosis ([198]Figures 5D and 5E).

   Collectively, these results indicated that rat BiPLs had similar
   profiles to those of UROMOL2021 class 2a human NMIBC, suggesting the
   suitability of the BiPL model as an interventional model for predicting
   therapeutic effects on human progressive NMIBC. In addition, the
   uniform upregulation of miR-145 target genes was consistent with the
   downregulation of miR-145, suggesting the essential involvement of
   miR-145 in bladder tumorigenesis.

Correlations between miR-145 and disease marker expression

   We examined the co-localization of miR-145 with disease markers in
   BiPLs using miRNA ISH-immunohistochemical (IHC) double staining. miRNA
   ISH-IHC double staining followed by confocal microscopy successfully
   demonstrated that the expression of miR-145 was inversely correlated
   with that of the well-known direct target protein, Fascin, in normal
   urothelial tissues ([199]Figure 6A). High-magnification images showed
   that miR-145 molecules were mainly localized to the cytoplasm and
   occasionally to nucleoli, but were less prevalent in the nucleoplasm
   ([200]Figure 6A).

Figure 6.

   [201]Figure 6
   [202]Open in a new tab

   Co-localization of miR-145 and bladder cancer-associated protein
   expression in rat BiPLs

   (A) Representative images of fluorescence double ISH staining for
   miR-145 and IHC staining for Fascin (a miR-145 direct target gene) in
   morphologically normal urothelial tissue samples. High-magnification
   images show the localization of miR-145 expression in urothelial cells.
   The arrow and arrowheads indicate signals in the nucleolus and
   cytoplasm, respectively. Scale bars, 10 μm. (B) Representative images
   of fluorescence double ISH staining for miR-145 and IHC staining for
   BC-associated markers. Signal intensities and correlations are shown
   along with images of ISH-IHC double staining. N, normal tissue; P,
   premalignant tissue. Scale bars, 20 μm. Data are presented as the
   means ± SD (n = 8). ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001,
   ∗∗∗∗p < 0.0001; two-tailed unpaired Student’s t test to compare the
   signal intensities, Pearson’s correlation coefficient to test
   correlations. (C) Representative images showing the intralesional
   gradation of miR-145 and BC-associated marker expression. Line profiles
   and scatterplots to test correlations between miR-145 expression and
   protein expression are shown with images of ISH-IHC double staining.
   Scale bars, 20 μm. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001;
   Pearson’s correlation coefficient to test correlations.

   In addition, BiPLs with the low expression of miR-145 exhibited the
   significant upregulation of BC-related oncogenes (cyclin D1, Fascin∗,
   SOCS7∗, STAM∗, c-Myc∗, and PTBP1∗; ∗ indicates a direct target of
   miR-145) and a decreased level of a marker of IFN pathway activation
   (p-STAT1, [203]Figures 6B and [204]S8). Notably, normal urothelial
   cells showed the opposite pattern to that of BiPLs. A correlation
   analysis revealed that the expression of miR-145 inversely correlated
   with that of all BC-related oncogenes, except for PTBP1, between BiPL
   and normal urothelial cells ([205]Figures 6B and [206]S8).

   Areas with different degrees of the downregulation of miR-145 were
   identified within BiPLs; thus, we investigated whether the
   intralesional gradation of miR-145 expression levels also correlated
   with the expression levels of disease markers. Due to the low baseline
   expression of miR-145 in BiPL, we utilized conventional fluorescence
   microscopy, which has a higher dynamic range than confocal microscopy,
   to visualize the subtle gradation of miR-145 expression. miR-145
   expression patterns inversely correlated with those of all disease
   markers, except for PTBP1, within a single lesion ([207]Figures 6C and
   [208]S9).

   These results suggested the prominent contribution of miR-145 to the
   pathogenesis of progressive premalignant lesions and the dysregulation
   of many BC-related genes in vivo, rationalizing the miR-145 replacement
   strategy for progressive NMIBC. In addition, the results indicated that
   the rat BiPL model mimicked the characteristics of the heterogeneity of
   human progressive NMIBC (UROMOL2021 class 2a), further demonstrating
   the advantage of using the rat BiPL model.

In vivo delivery of miRNA to bladder urothelial cells

   To assess the efficacy of miR-145 replacement therapy against
   progressive premalignant lesions in vivo, we performed the intravesical
   administration of miR-145 in the rat BBN-induced model.

   We selected miR-145S1 as the therapeutic agent because of its higher
   efficacy than miR-145WT and thus more favorable properties for drug
   development. For the drug delivery system, we employed a lipid
   nanoparticle (LNP) system consisting of an ionizable lipid, a
   phospholipid, cholesterol, and a PEGylated lipid ([209]Figure 7A)
   because the Lipofectamine-based lipoplex system used in in vitro
   experiments has poor nucleic acid encapsulation and thus low nuclease
   tolerability in vivo.[210]^40^,[211]^41 We examined the transfection
   potency of LNP-encapsulated miR-145S1 into BiPL cells. We
   intravesically administered FITC-labeled miR-145S1-LNPs (20 μM solution
   in 100 μL of PBS, a total of 4.5 nmol/infusion) or PBS to the rat BiPL
   model established by the free intake of water containing 0.05% BBN and
   0.005% Tween 80 for 16 weeks ([212]Figure S10A). To prevent the early
   discharge of materials from the bladder, we sutured the external
   urethral orifice after administration, as indicated in
   [213]Figures S10A and S10B. After an incubation for 3 h, the solution
   was released by the removal of the sutures. Animals were then
   sacrificed, and bladder tissue was collected, snap-frozen,
   cryosectioned, and mounted with the NucBlue nuclear stain for the
   direct detection of FITC signals. A confocal microscopic analysis
   revealed strong FITC signals in the cytoplasm and nuclei of urothelial
   cells ([214]Figure S10C). We also performed miRNA ISH to detect the
   miR-145 molecule after administration. miR-145 signals were
   significantly increased in urothelial cells ([215]Figure S10D),
   consistent with the results obtained by the direct detection of FITC.
   These results indicated that miR-145S1-LNPs were successfully delivered
   into BiPLs and that LNPs worked as a drug delivery system for
   intravesical administration.

Figure 7.

   [216]Figure 7
   [217]Open in a new tab

   Gross analysis of the efficacy of miR-145 replacement for rat BiPLs

   (A) Illustration of the structure of Lipofectamine-based lipoplexes and
   lipid nanoparticles (LNPs). (B) Schematic illustration of the methods
   to establish the BBN-induced rat premalignant lesion model and
   treatment with miR-145S1-LNPs or ctl-RNA-LNPs. The triangles under the
   orange bar indicate the timing of administration (Admn). (C) Gross
   image analysis of the bladder. The intact bladder area was
   distinguished from the hyperplastic area based on color differences;
   e.g., the intact area had a thin bladder wall that was blue-black in
   color, whereas the hyperplastic area had a thick urothelial layer that
   was white to pink. The intact area was normalized to the total area of
   the bladder (Total mask). The actual intact area was matched with the
   calculated intact area in merged images. Scale bars, 25 mm. (D) All
   gross images and calculated intact areas of bladders treated with
   miR-145S1-LNPs (n = 8) or Ctl-RNA-LNPs (n = 9). Scale bars, 50 mm. (E)
   Representative high-magnification gross images of bladders treated with
   miR-145S1-LNP or ctl-RNA-LNP. The intact area and hyperplastic area are
   indicated with asterisks and arrowheads, respectively. Note that the
   bladder treated with miR-145S1-LNPs had a larger intact area
   (asterisks) than that treated with ctl-RNA-LNP. Scale bars, 25 mm. (F)
   Waterfall and swarm plot showing the relative intact bladder area per
   total mask in rats treated with miR-145S1-LNPs (n = 8) or Ctl-RNA-LNPs
   (n = 9). Data are presented as means ± SD. ∗∗p < 0.01; two-tailed
   unpaired Student’s t test.

In vivo anti-tumor efficacy of miR-145S1-LNPs against premalignant lesions

   We examined the anti-tumor efficacy of miR-145S1-LNP in the model using
   the multiple pathological and molecular indices examined above.

   The rat BiPL model was established by the free intake of water
   containing 0.05% BBN and 0.005% Tween 80 for 16 weeks. After confirming
   the development of early lesions in the bladder in a few rats, we
   intravesically administered miR-145S1-LNPs or ctl-RNA-LNPs (20 μM of
   miRNA solution in 100 μL of PBS, total 4.5 nmol/infusion) and incubated
   the solution in situ for at least 3 h by suturing the external urethral
   orifice ([218]Figure 7B). After twice weekly treatments for 2 weeks (a
   total of four treatments), animals were sacrificed 3 days after the
   final administration.

   To assess the overall anti-tumor efficacy of the treatment, we
   evaluated the proliferative area using gross images and image analysis
   software (ImageJ). As described earlier, bladders were fixed in situ,
   sliced in the sagittal plane, and photographed on a black background
   with a macro lens-equipped camera ([219]Figure 7C). In gross images,
   bladders had a thickened proliferative area that was white to pink in
   color, whereas the intact area was dark blue to black ([220]Figure 7C).
   Therefore, we quantified the white and blue-black areas, which
   indicated affected and intact areas, respectively ([221]Figure 7C). The
   results showed that the administration of miR-145S1-LNPs significantly
   decreased the proliferative area ([222]Figures 7D–7F).

   A histopathological examination revealed that most lesions in the
   treated group were grade I, with few grade II lesions being observed
   ([223]Figures 8A and 8B). The ratio of grade II/grade I lesions was
   significantly decreased in the treatment group (ratio, 0.14; 19/132)
   compared with the control group (ratio, 0.29; 36/126; [224]Figure 8B),
   while the total number of lesions did not significantly differ between
   the groups (treated, 151; control, 162; [225]Figure 8B). Lesions in the
   miR-145S1-treated group rarely had mitotic figures, whereas the control
   group sporadically exhibited multiple areas with a relatively high
   mitotic index.

Figure 8.

   [226]Figure 8
   [227]Open in a new tab

   Effects of miR-145 replacement on histopathological and molecular
   markers

   (A) Representative loupe images of H&E-stained bladder tissues treated
   with miR-145S1-LNPs or ctl-RNA-LNPs. Note that the bladder treated with
   miR-145S1-LNPs had fewer and more minor lesions than that treated with
   Ctl-RNA-LNPs. Scale bars, 1 mm. (B) Number of lesions of each grade
   (grade I, G1; grade II, G2) in bladders treated with miR-145S1-LNPs
   (n = 8) or ctl-RNA-LNPs (n = 9). ∗p < 0.05; two-tailed unpaired
   Student’s t test for the count comparison; two-tailed Fisher’s exact
   test for the ratio comparison. Data are presented as means ± SD. (C)
   Representative images of immunohistochemical staining for
   disease-associated markers (cyclin D1, Ki-67, p-STAT1, SOCS7, STAM,
   Fascin, c-Myc, and PTBP1) in BiPLs treated with miR-145S1-LNPs (n = 8)
   or ctl-RNA-LNPs (n = 9). Scale bars, 20 μm. (D) Relative counts of
   cyclin D1- and Ki-67-positive cells in BiPLs treated with
   miR-145S1-LNPs (n = 8) or ctl-RNA-LNPs (n = 9). (E) Relative staining
   intensities of p-STAT1, SOCS7, STAM, c-Myc, Fascin, and PTBP1 in BiPLs
   treated with miR-145S1-LNPs (n = 8) or ctl-RNA-LNPs (n = 9). Data are
   presented as the means ± SD. ∗p < 0.05, ∗∗p < 0.01; ns, not
   significant; two-tailed unpaired Student’s t test unless otherwise
   indicated.

   The IHC analysis showed that the treated group had significantly fewer
   cyclin D1-positive lesions than the control group ([228]Figures 8C and
   8D). Given that cyclin D1 could serve as the earliest marker of disease
   progression, this result suggested that miR-145 replacement inhibited
   disease progression at the early stage of tumorigenesis in the rat BiPL
   model. Similar to the results obtained for cyclin D1, the treated group
   had fewer Ki-67-positive lesions ([229]Figures 8C and 8D), suggesting
   that the treatment inhibited the growth of BiPLs. In addition, the
   administration of miR-145S1 induced the downregulation of miR-145
   target genes (Fascin, c-Myc, and STAM) and an increase in the level of
   a marker of IFN pathway activation (p-STAT1, [230]Figure 8E). The
   expression of PTBP1 and SOCS7 was slightly downregulated in the
   miR-145-treated group, but this downregulation was not significant due
   to the high statistical dispersion ([231]Figure 8E). These results
   suggested that the miR-145S1-LNP treatment inhibited the growth and
   disease progression of premalignant lesions.

   The TUNEL assay showed that the number of apoptotic cells did not
   significantly differ among the groups examined ([232]Figure S11A).
   Moreover, no massive necrosis or destruction of tissues was observed
   throughout the control and treated samples. The infiltration of
   lymphocytes and macrophages was mainly observed in the lamina propria
   of the affected areas, particularly in areas close to grade II lesions;
   however, the degree of inflammation was slight and did not markedly
   differ between the treated and control groups. No signs of bacterial
   infection secondary to the administration of LNP were observed
   throughout samples.

   To assess the systemic effects of intravesically infused miR-145S1-LNP,
   we utilized miRNA qRT-PCR to evaluate the leakage of miR-145S1-LNP into
   the bloodstream. The results showed no significant increases in the
   plasma levels of miR-145 in the samples from rats receiving
   intravesical administration of miR-145S1-LNP ([233]Figure S11B),
   indicating that the administered solution was successfully retained in
   the bladder and did not leak into the bloodstream. Throughout these
   experiments, the intravesical infusion of miR-145S1-LNP did not cause
   any significant systemic adverse effects, including behavioral
   disorders, moribund conditions, or death associated with the
   intravesical administration, and no recognizable gross lesions were
   detected in major organs, such as the heart, lungs, liver, kidneys, and
   spleen, during autopsy. Therefore, these results suggested that the
   systemic effects of the intravesical infusion of miR-145S1-LNP on the
   body were minimal.

Discussion

   Despite its pivotal necessity, the development of therapies for
   progressive NMIBC has been challenging due to the lack of validated
   in vivo models and therapeutic modalities that selectively target
   genetic vulnerability. We herein validated the molecular basis of rat
   BiPLs by clarifying their detailed profiles of somatic mutation,
   histopathological, and miR-145- and BC-related oncogene expression.
   These profiles highlighted the high molecular similarities between rat
   BiPLs and UROMOL2021 class 2a human NMIBC,[234]^3 indicating the
   suitability of the rat BiPL model as a model of human progressive NMIBC
   with a poor prognosis. We found that the downregulation of miR-145
   appeared to be one of the earliest oncogenic events and demonstrated
   that miR-145 is the principal regulator of the premalignant BC
   phenotype characterized by downregulated IFN-STAT1/2 signaling and
   upregulated BC-related oncogenes, including c-Myc, Fascin, and PTBP1.
   Furthermore, we developed a novel miR-145-based therapeutic agent
   (miR-145S1) and demonstrated its efficacy against premalignant lesions
   in vivo. The present results provide insights into the essential roles
   of miR-145 in bladder tumorigenesis, as well as a validated
   interventional model and novel miRNA-based therapeutic modality
   targeting the molecular vulnerability of progressive NMIBC.

   Rat BiPLs showed molecular signatures that were highly similar to those
   of human progressive NMIBC, particularly the UROMOL2021 class 2a
   subclass, which has the worst prognosis among NMIBCs.[235]^3 The
   expression profile of BiPLs at 18 weeks was cyclin D1+, cyclin A2+, and
   p53−, with the subsequent expression of p53 and increases in cyclin A2
   after an additional 4 weeks. In addition, BiPLs harbored somatic
   mutations in genes such as Birc6, Dync1h1, Pik3ca, Mycbp2, Atm, and
   Crebbp, as well as SBS mutation signatures associated with APOBEC and
   cigarette smoke-related BC. These molecular profiles of BiPL were
   highly similar to the profiles of class 2a human NMIBC (late cell cycle
   marker, cyclin A2; p53 anomaly; low IFN; c-Myc expression; APOBEC
   mutation signatures).[236]^3 Although rat BiPLs had a slight mixture of
   the characteristics of classes 1 and 3 (the expression of the early
   cell cycle marker, cyclin D1), their overall phenotypes were distinct
   from those of the NMIBC subclasses with a favorable prognosis,
   including class 1 (represented by FGFR3 and Ras mutations), class 2b
   (represented by high IFN signaling), and class 3 (represented by FGFR3
   mutations and RNA-editing mutation signatures),[237]^3^,[238]^33
   emphasizing the similarities between rat BiPLs and human progressive
   NMIBC with a poor prognosis. Furthermore, we identified mutations in
   Pik3ca, a frequently mutated oncogene in human NMIBC, with the
   subsequent upregulation of Akt phosphorylation in BiPL tissues. The
   mutation point was close to the hotspots of activating PIK3CA mutations
   in humans,[239]^38 which is consistent with the activation of PI3K/Akt
   in human NMIBC.[240]^3 These findings emphasize the high homology
   between the malignant progression sequences of rat BiPLs and human
   class 2a progressive NMIBC.

   The primary molecular phenotype of rat BiPLs was associated with the
   suppression of IFN-STAT1/2 pathways or high c-Myc expression. Similar
   anomalies were observed in human BC, such as the frequent homozygous
   deletion/inactivation of α-IFN,[241]^42 c-Myc signatures in human NMIBC
   profiles,[242]^33 and c-Myc-mediated malignant progression in early
   human BC.[243]^43 The low IFN and high c-Myc signatures of rat BiPLs
   were consistent with those of human progressive NMIBC (class 2a, poor
   prognosis), and the low IFN profile was the key signature that
   distinguished class 2a (a poor prognosis) from other benign-oriented
   subclasses (classes 1, 2b, and 3).[244]^3 These findings further
   emphasize the similarities in molecular signatures between rat BiPL and
   human progressive NMIBC with a poor prognosis. These results also
   strongly suggest that IFN-STAT1/2- and c-Myc-associated signaling
   alterations shape the molecular basis for the progressive/premalignant
   characteristics of rat BiPLs and human progressive NMIBC.

   Notably, the IFN-STAT1/2- and c-Myc-associated premalignant phenotypes
   were dominantly regulated by miR-145. The expression levels of
   IFN-STAT1/2-associated markers and c-Myc showed strong negative
   correlations with miR-145 expression both in vitro and in vivo.
   Transcriptome and immunoblot analyses demonstrated that miR-145
   replacement induced the strong and dominant activation of IFN-STAT1/2
   signaling and significant downregulation of c-Myc expression in BC
   cells, likely by both direct and IFN-STAT1/2-mediated indirect
   mechanisms. These results were further confirmed by in vivo miR-145
   replacement therapy significantly inhibiting disease progression with
   the activation of the IFN-STAT1/2 pathway and suppression of the c-Myc
   pathway in the rat BiPL model. These findings strongly suggest that
   miR-145 is the principal regulator of class 2a-like progressive
   characteristics in early BC, thereby rationalizing the targeting of
   miR-145 downregulation as a therapeutic approach for early BC.

   miR-145S1-based intravesical infusion therapy successfully inhibited
   the progression of premalignant lesions in vivo. The advantages of
   miR-145S1-based replacement therapy are as follows. First, miR-145S1
   potently, selectively, and simultaneously targets a wide range of
   essential oncogenes for bladder tumorigenesis; thus, it is likely to
   have higher efficacy than non-specific (BCG) or single-molecule
   targeting agents (siRNAs or small-molecule inhibitors). Second,
   intravesically administered miR-145S1 directly reaches and affects
   premalignant lesions localized in the mucosal layer despite its lower
   vascularization and thus would be more effective than systemic
   therapeutic agents. Third, miR-145 is a non-bacterial agent with safer
   properties than BCG immunotherapy and no concerns of systemic
   infection, as observed with the BCG immunotherapy. In addition, this
   strategy employs local administration, which minimizes the reported
   drawbacks of miRNA-based therapy, such as the hepatic accumulation of
   administered agents[245]^44 and toxicities from systemic immunological
   adverse events.[246]^45 These features imply the high potential of
   miR-145S1 replacement as a new intravesical therapeutic modality for
   premalignant BC lesions.

   The present results revealed that miR-145 replacement therapy had a
   feature of RNAi-mediated IFN activation therapy. Based on the results
   obtained herein, the activation mechanism appears to be based on RNAi
   against the IFN signaling inhibitors SOCS7 and STAM, which is
   consistent with previous findings.[247]^29 IFN therapy has been
   evaluated as BC therapy due to its cytotoxic and anti-angiogenic
   activities.[248]^46^,[249]^47 The intravesical instillation of rhIFNα
   was unsuccessful, which may have been due to the short half-life of the
   protein.[250]^48 However, the continuous overexpression of IFN by
   adenoviral transduction (rAd-IFN/Syn-3, also called Nadofaragene
   Firadenovec) showed significant efficacy and safety in vivo[251]^49 and
   in clinical trials, including phase I (43% CR at 3 months,
   [252]NCT01162785),[253]^50 phase II (no recurrence for 12 months in 35%
   of high-grade BCG refractory or relapsed NMIBC patients,
   [254]NCT01687244),[255]^51 and phase III (CR for 3 months in 53.4% of
   patients with BCG-unresponsive CIS; CR for 12 months in 45.5% of CIS
   patients, [256]NCT02773849) trials.[257]^52 These findings highlight
   the strong potential of miR-145-mediated IFN activation to be
   therapeutically effective against premalignant lesions with or without
   BCG resistance. Furthermore, given that miR-145 utilizes a unique
   mechanism to activate IFN signaling, miR-145 replacement may be
   synergetically effective with conventional therapies that use IFN
   ligands or BCG.

   In the assessment of the in vivo delivery of miRNA-LNP, FITC or ISH
   signals indicating the distribution of miR-145 molecules were the most
   intense on the mucosal surface and gradually decreased toward the
   deeper layer of the mucosa. Fluorescent microscopy or ISH detected only
   weak or no signals in the deeper layer of the mucosa; however, the
   target proteins of miR-145 were significantly downregulated, even in
   the deep layer without FITC/ISH signals. This inconsistency appears to
   be due to the limited sensitivity and dynamic ranges of fluorescent or
   ISH-mediated detection, which cannot identify a small number of miRNA
   molecules in cells or tissues. Given that miRNA molecules are known to
   efficiently induce RNAi because of their RISC-involved catalytic
   reactions,[258]^7 a small number of miR-145 molecules undetectable by
   ISH or fluorescent microscopy may induce the downregulation of the
   target genes. Indeed, the major target proteins of miR-145, such as
   Fascin and c-Myc, had an exact opposite expression pattern from that of
   the miR-145 signal distribution in samples from rats receiving miR-145
   replacement, i.e., the degree of target gene downregulation was the
   most intense on the surface layer, and its magnitude gradually
   decreased toward the deeper layer of the mucosa. These results support
   the hypothesis that miR-145S1 reached into the deeper layer of mucosa
   with levels that were undetectable by ISH or fluorescent microscopy and
   exerted RNAi functions; however, further studies to validate these
   results are needed in the future.

   As a limitation, rat BiPLs had different morphological characteristics
   from those of CIS, a major histological subtype of human progressive
   NMIBC. The major histological subtype of rat BiPLs was PN hyperplasia
   (papillary growth), which differs from the main histological subtype of
   CIS (flat growth). However, despite these histological differences, the
   rat BiPL model still has high potential as a molecular-based model of
   human progressive NMIBC. Rat BiPLs have a strong premalignant
   predisposition with a high progression rate along with Tp53
   mutations,[259]^23 whereas the progressive characteristics of BiPLs are
   different from the morphologically equivalent subtype in humans (human
   papillary hyperplasia), which has a benign-oriented disease
   course.[260]^21^,[261]^23^,[262]^53 These findings are consistent with
   recent studies demonstrating that the progressive characteristics of
   NMIBC (recurrence rate, progression rate, and prognosis) are defined by
   molecular characteristics rather than the histological subtype.[263]^3
   Indeed, UROMOL2021 class 2a, which includes many Ta tumors along with
   CIS, is strongly associated with the worst prognosis among NMIBCs and
   has a high recurrence rate.[264]^3 In addition, BBN-induced PN
   hyperplasia that developed in rats exhibited neither somatic mutations
   nor the overexpression of FGFR3, a benign-oriented marker frequently
   mutated/overexpressed in human NMIBC.[265]^32 Given that FGFR3
   expression is absent in class 2a progressive NMIBC and
   CIS,[266]^3^,[267]^54 these results further support the hypothesis that
   the FGFR3-negative rat BiPL model has premalignant molecular features
   similar to those of progressive NMIBC despite morphological
   differences. Nevertheless, miR-145 suppressed FGFR3 as a direct target;
   thus, regardless of the FGFR3 mutation/expression status, miR-145-based
   therapy may be effective against a broad spectrum of NMIBCs, including
   different UROMOL2021 classes (classes 1, 2a/b, and 3) and TNM subtypes
   (Ta and CIS), via the combined inhibition of c-Myc/cyclin D1, FGFR3,
   and the activation of IFN signaling.

   miR-145S1 exerted slightly different effects from miR-145WT in the
   regulation of its direct targets, such as Fascin, although the target
   spectrum of miR-145S1 was largely similar to that of miR-145WT. In
   miRNA target regulation, the guide strand and its seed sequences are
   the most crucial factors to determine the target genes for each
   miRNA.[268]^7 Although miR-145S1 has identical seed sequences to those
   of miR-145WT in the guide strand, the passenger strand of miR-145S1 is
   modified to fully match its guide strand and thus differs from
   miR-145WT. These differences in the passenger strand may affect its
   affinity to the target genes and result in the minor difference between
   miR-145S1 and miR-145WT on Fascin. However, it is important to note
   that miR-145S1 could also downregulate Fascin expression more strongly
   in the in vivo context than in the in vitro context, suggesting that
   miR-145S1 also has the capability to strongly downregulate Fascin
   expression depending on the microenvironment. Nevertheless, miR-145S1
   showed a similar spectrum for the target gene regulation to miR-145WT.
   Therefore, regardless of affinity differences in different contexts,
   miR-145S1 may serve as a therapeutic modality for the replacement of
   miR-145 expression in early BC.

Materials and methods

Cell lines and culture

   T24, JBV, NBT-T1, and NBT-L2B cells were cultured in EMEM (Wako, Osaka,
   Japan) supplemented with 10% FBS (Sigma-Aldrich, St. Louis, MO). HBlEpC
   and RNM cells were cultured in HBlEpC Growth Medium (Cell Applications,
   San Diego, CA, 217K-500). T24 cells were obtained from the Japanese
   Cancer Research Resources Bank, NBT-T1 and NBT-L2b cells were obtained
   from the Riken Bioresource Research Center, 253J-BV cells were obtained
   from Osaka Medical College, and HBlEpC cells were obtained from Cell
   Applications. RNM cells were established as primary cultured cells from
   rat bladder urothelium using a previously reported method[269]^55 and
   were confirmed to be of epithelial origin by an immunoblot analysis of
   pan-CK expression. The cell lines used were confirmed to be negative
   for mycoplasma by mycoplasma testing (MycoAlert, Lonza, Basel,
   Switzerland). All cells were cultured at 37°C in a humidified incubator
   with 5% CO[2]. Cell viability was assessed using the trypan blue
   exclusion test.

miRNA synthesis and miRNA/siRNA transfection

   miR-145WT was purchased from Thermo Fisher Scientific, Waltham, CA
   (miR-145-5p mirVana miRNA mimic, MC11480). miR-145S1 and other miR-145
   derivatives were designed by the authors and synthesized by Hokkaido
   System Science (Hokkaido, Japan). siRNAs for human SOCS7/STAM and rat
   Socs7/Stam were purchased from Sigma-Aldrich. Non-targeting control
   miRNA sequences with proven suitability in previous
   studies[270]^13^,[271]^18 were synthesized by Hokkaido System Science
   and used as negative control sequences (control miRNA).

   In transfection experiments, cells were seeded on six-well plates at a
   concentration of 0.2 × 10^5 cells/mL. Cells were transfected with a
   mixture of nucleic acids (20 nM for miRNAs and 10 nM for siRNAs) and
   Lipofectamine RNAiMAX (Thermo Fisher Scientific). After an incubation
   for 48 or 72 h, effects were examined using each method.

qRT-PCR analysis

   Total RNA was extracted and purified using the NucleoSpin miRNA Kit
   (MACHEREY-NAGEL, Düren, Germany). RNA concentrations and integrity were
   examined using UV spectrophotometry and the Agilent 2100 Bioanalyzer
   (Agilent Technologies, CA), respectively. To measure miRNA, total RNA
   (0.5 μg) was reverse transcribed to cDNA using the TaqMan MicroRNA
   Reverse Transcription Kit (Thermo Fisher Scientific). cDNA was
   amplified by PCR with the THUNDERBIRD Probe qPCR Mix (TOYOBO, Osaka,
   Japan, QPS-101) and TaqMan MicroRNA Assays (Thermo Fisher Scientific)
   to evaluate the expression of miR-145 (assay ID 002278), RNU6B (assay
   ID 001093), RNU19 (assay ID 001003), and RNU48 (assay ID 001006). To
   measure mRNA, total RNA (0.5 μg) was reverse transcribed to cDNA using
   the PrimeScript RT Reagent Kit (Takara, Shiga, Japan). cDNA was
   amplified by PCR with the THUNDERBIRD SYBR qPCR Mix (TOYOBO, QPS-201)
   with gene-specific primers (see [272]Table S1). Amplified fluorescent
   signals were recorded in each cycle using the TaKaRa Thermal Cycler
   Dice Real-Time System II (Takara). The relative expression of each
   miRNA or mRNA was calculated using the ΔCt method with normalization to
   RNU6B (for miRNAs) or TBP (for mRNAs). The internal control was
   selected from the candidate genes (RNU6B, RNU19, and RNNU48 for miRNAs;
   ACTB, GAPDH, and TBP for mRNAs) using NormFinder (Aarhus University
   Hospital, Denmark). All PCRs were performed in triplicate.

Northern blotting for miRNA

   Total RNA was extracted and purified using the NucleoSpin miRNA Kit
   (MACHEREY-NAGEL). RNA concentrations and integrity were examined using
   UV spectrophotometry and the Agilent 2100 Bioanalyzer (Agilent
   Technologies), respectively. Total RNA (5 μg) was mixed with 2× RNA
   loading buffer without ethidium bromide (Wako, 182-02571) and incubated
   at 70°C for 10 min. After a pre-run at 300 V for 1 h, the RNA samples
   and an RNA ladder marker (DynaMarker DIG Labeled Blue Color Marker for
   Small RNA, BioDynamics Laboratory, Tokyo, Japan, DM270) were
   electrophoresed at 200 V for 50 min on a 15% polyacrylamide gel with
   4 M urea (SuperSepTMRNA 15%, Wako, 194-15881) and RNase-free
   Tris-borate-EDTA buffer (TBE buffer) (Wako, 318-90041). All
   experimental equipment was cleaned with RNaseZap (RNaseZap RNase
   Decontamination Wipes, Thermo Fisher Scientific, AM9786,) to remove
   RNase. The electrophoresed gel was washed for 10 min with TBE buffer,
   and RNA in the gel was transferred onto a nylon membrane
   (BrightStar-Plus Positively Charged Nylon Membrane, 15 × 15 cm, Thermo
   Fisher Scientific, AM10100) at 15 V for 40 min with a Trans-Blot SD
   Semi-Dry Transfer Cell (Bio-Rad). The transferred membrane was rinsed
   with 0.2× SSC for 1 min and baked at 80°C for 2 h in an oven. The
   membrane was pre-hybridized at 53°C for 30 min with 10 mL of
   hybridization buffer (ULTRAhyb Ultrasensitive Hybridization Buffer,
   Thermo Fisher Scientific, AM8670) and then hybridized at 53°C for 1 h
   with 0.4 nM of the digoxigenin-labeled miR-145 probe (QIAGEN, Hilden,
   Germany, no. 339456). After three washes for 15 min with 5× SCC, the
   membrane was blocked for 1 h with the PVDF Blocking Reagent for Can Get
   Signal (TOYOBO, NYPBR01). Subsequent to a wash with TBST for 15 min,
   the membrane was reacted for 1 h with an anti-digoxigenin antibody
   conjugated with peroxidase in Can Get Signal Immunoreaction Enhancer
   Solution (final concentration [1:10,000], TOYOBO, NKB-101T). After
   three washes with TBST for 5 min, membranes were incubated with
   Luminata Forte Western HRP substrate (EMD Millipore) to visualize the
   signals.

cDNA microarray analysis

   Total RNA was extracted using the Maxwell RSC simplyRNA Cells Kit
   (Promega, Madison, WI, AS1390) and a Maxwell RSC instrument (Promega,
   AS4500). BC cells (253 JB-V) were transfected and incubated for 48 h
   with miRNAs (miR-145-5p, 10 nM; control miRNA, 10 nM) and Lipofectamine
   RNAiMAX (Thermo Fisher Scientific). After two PBS washes, cells were
   lysed in homogenization solution (400 μL) with 1-thioglycerol (4 μL).
   Subsequently, the lysate (200 μL) was mixed with lysis buffer (200 μL),
   vortexed for 15 s, and transferred to a Maxwell RSC Cartridge. After an
   incubation with DNase I, the cartridge was inserted into the Promega
   Maxwell RSC instrument to purify total RNA. RNA integrity was assessed
   using an Agilent 2100 Bioanalyzer (Agilent). RNA samples were labeled
   with the Agilent Low Input Amp Labeling Kit (Agilent, cat. no.
   5190-2305) and hybridized on slides of SurePrint G3 Human GE Microarray
   8x60K version 3.0 (Agilent) in the Agilent G2545A Hybridization Oven
   (Agilent) at 65°C overnight. Slides were then scanned with the Agilent
   Microarray Scanner System (Agilent, G2565CA). Signals were normalized
   to the 75th percentile of the signal intensity by utilizing the Subio
   platform version 1.2.2 (Subio, Kagoshima, Japan). Normalized signals
   were log2 transformed and used to calculate FC values for comparisons
   of gene expression in miR-145- and control miRNA-transfected cells. p
   values were calculated by the Student’s t test (paired, two-tailed)
   from two biological replicates. Genes with an absolute FC of greater
   than 1.5 and p values of less than 0.05 were defined as DEGs. DEGs were
   then subjected to a Metascape analysis using the Kyoto Encyclopedia of
   Genes and Genomes and Reactome pathway databases.[273]^56 JMP version
   12.2 (SAS Institute, Cary, NC) was used to generate heatmaps. A circos
   plot was generated using a function of Metascape.[274]^56

Immunoblotting

   In the immunoblot analysis, cells and tissues were lysed with 1% SDS
   buffer. In brief, cultured cells were fixed for 30 min with 10% TCA
   after two washes with ice-cold PBS (Wako). Subsequent to two washes
   with ice-cold PBS, fixed cells were lysed with 1% SDS buffer (Wako).
   Tissues were snap-frozen using liquid nitrogen and then homogenized in
   1% SDS buffer (Wako). Whole-cell lysates were sonicated and centrifuged
   at 13,000 × g for 20 min. Protein concentrations in the supernatants
   were measured with a DC Protein Assay Kit (Bio-Rad, Hercules, CA).
   Protein samples (1–5 μg/lane) were subjected to 10%–12.5% SDS-PAGE and
   transferred to polyvinylidene fluoride membranes with a pore size of
   0.45 μm (Immobilon-P Membrane, EMD Millipore, Burlington, MA). After
   1 h of blocking with PVDF Blocking Reagent (TOYOBO), membranes were
   incubated with primary antibodies at 4°C overnight. The primary
   antibodies used for immunoblotting are summarized in [275]Table S2.
   After three washes with TBS-T for 5 min each, membranes were incubated
   at room temperature with horseradish peroxidase (HRP)-conjugated
   anti-mouse or anti-rabbit secondary antibodies (1:4,000 dilution, Cell
   Signaling Technology, Danvers, MA). After three TBS-T washes, membranes
   were incubated with Luminata Forte Western HRP substrate (EMD
   Millipore) to visualize immunoreactions. Sample loading was checked
   with an anti-α-tubulin antibody (MBL, Tokyo, Japan, PM054).

Animal experiments

   All animal experiments were approved by the Institutional Animal Care
   and Use Committee (IACUC) of Gifu University. All animals were housed
   with ad libitum-access to food and water under the standard light-dark
   cycle.

   The BiPL model for gross, histological, and molecular assessments was
   established in F344/NSIc rats (n = 8, female, 4 weeks old; SLC,
   Shizuoka, Japan) by the free intake of water containing 0.05% BBN
   (Tokyo Chemical Industry, Tokyo, Japan) and 0.005% Tween 80
   (Sigma-Aldrich) for 16 weeks. BBN exposure was terminated at 16 weeks,
   and rats were housed with normal distilled water until weeks 18 or 22
   and then sacrificed for subsequent analyses. To consistently evaluate
   premalignant lesions, bladder samples for gross, hematoxylin and eosin
   (H&E), and single IHC evaluations were fixed in situ by a uniform
   infusion of 300 μL of 4% paraformaldehyde into the bladder and were
   then sliced in the sagittal plane. In miRNA ISH and ISH-IHC
   double-staining analyses, bladder samples were embedded in Tissue-Tek
   OCT Compound (Sakura Finetek Japan, Tokyo, Japan) and snap-frozen using
   liquid nitrogen. In miRNA qRT-PCR and NGS analyses, tissue samples were
   directly snap-frozen in liquid nitrogen and used for analyses.

   In miR-145S1 delivery experiments, an LNP system was employed since the
   Lipofectamine-based lipoplex system used in in vitro experiments has
   poor nucleic acid encapsulation and thus low nuclease tolerability
   in vivo.[276]^41 miR-145S1 conjugated to FITC was encapsulated in LNPs
   consisting of an ionizable lipid, a phospholipid, cholesterol, and
   PEGylated lipid using the NanoAssemblr system (Precision NanoSystems,
   Vancouver, Canada).[277]^40 To test transfection potency in vivo,
   FITC-labeled miR-145S1-LNPs (20 μM solution in 100 μL of PBS, a total
   of 4.5 nmol/infusion) or PBS was intravesically administered to rats
   with BiPLs developed over 16 weeks. The dose of miRNA-LNPs was
   determined to be 4.5 nmol/infusion based on preliminary experiments
   that test the transfection potency of miRNA-LNPs at doses of 2.25, 4.5,
   and 9 nmol/infusion. The external urethral orifice was sutured after
   administration to prevent the early discharge of materials from the
   bladder. After 3 h, the solution was released by the removal of
   sutures. Animals were then sacrificed, and bladder tissue was
   collected, snap-frozen in Tissue-Tek OCT Compound (Sakura Finetek
   Japan), cryosectioned, and mounted with the NucBlue nuclear stain for
   the direct detection of FITC signals with confocal microscopy. Frozen
   sections were also used to detect miR-145 molecules by miRNA ISH.

   In the experiment testing the therapeutic efficacy of miR-145S1, the
   BiPL model was established by the free intake of water containing 0.05%
   BBN and 0.005% Tween 80 for 16 weeks. After confirming the development
   of early lesions in the bladder in a few rats, miR-145S1-LNPs or
   ctl-RNA-LNPs (20 μM miRNA solution in 100 μL of PBS, a total of
   4.5 nmol/infusion) were intravesically administered and incubated in
   situ for at least 3 h by suturing the external urethral orifice. After
   a twice weekly treatment for 2 weeks (a total of four treatments), rats
   were sacrificed 3 days after the final administration.

Gross image analysis

   In the assessment of overall anti-tumor efficacy, the areas of
   proliferative lesions were evaluated using high-resolution gross images
   and image analysis software. As described earlier, bladders were fixed
   in situ, sliced in the sagittal plane, and photographed with a macro
   lens-equipped camera (M.Zuiko Digital ED 60 mm F2.8 Macro and E-PL9
   Mirrorless Camera, Olympus, Tokyo, Japan) on a black background. In
   gross images, the area with hyperplastic lesions had a thickened
   urothelial layer that was white to pink in color, whereas the normal
   area was dark blue to black. Therefore, blue-black and white areas were
   marked and quantified using ImageJ (NIH, Bethesda, MD) as intact and
   affected areas, respectively.

Histology, IHC, and miRNA ISH

   In H&E staining, IHC staining, and TUNEL assays, tissues were fixed
   with 4% neutral-buffered paraformaldehyde and embedded in paraffin.
   Paraffin-embedded samples were sectioned at a thickness of 5 μm and
   mounted on glass slides (S2226 for H&E staining, SCRE-05 for IHC
   staining, Matsunami Glass, Osaka, Japan). Sections were deparaffinized
   with Lemosol and rehydrated in graded alcohol series. H&E staining was
   performed with Tissue-Tek Hematoxylin 3G (Sakura Finetek Japan,
   9131-4P) and Tissue-Tek Eosin (Sakura Finetek Japan, 8659). A
   histopathological diagnosis was performed by pathologists certified by
   the Japanese College of Veterinary Pathologists (JCVP) or the American
   College of Veterinary Pathologists (ACVP). In IHC staining, sections
   were then autoclaved at 120°C for 1 min in antigen retrieval buffer
   (ImmunoActive IA6500, Matsunami Glass), cooled to room temperature, and
   incubated with 0.3% hydrogen peroxide in methanol for 20 min. After
   blocking with 2.5% normal horse serum (Vector Laboratories, Newark, CA,
   S-2012-50), sections were incubated with primary antibodies at 4°C
   overnight. The primary antibodies used are listed in [278]Table S2.
   After three washes with TBS-T, sections were incubated for 20 min with
   an HRP-conjugated secondary antibody (ImmPRESS HRP Horse Anti-Rabbit
   IgG Polymer Detection Kit, Vector Laboratories, MP-7401). The sections
   were washed three times with TBS-T, developed with DAB chromogen
   solution (ImmPACT DAB Substrate, Vector Laboratories, SK-4105), and
   counterstained with hematoxylin. Regarding TUNEL, sections were
   incubated with Proteinase K antigen retrieval solution, incubated for
   20 min in 0.3% hydrogen peroxide in methanol, and labeled with TdT
   enzyme and FITC (TUNEL Assay Kit, Abcam, Cambridge, UK, ab206386).
   After blocking with normal serum, sections were incubated with the
   HRP-conjugated anti-FITC antibody. Signals were visualized with DAB
   chromogen solution, and sections were counterstained with hematoxylin.
   Histopathological images were acquired with a microscope (KEYENCE,
   Osaka, Japan, BZ-X700).

   Single miRNA ISH was performed with hsa-miR-145-5p miRCURY LNA miRNA
   ISH Optimization Kits (QIAGEN, no. 339456). All solutions used in the
   experiment were prepared under RNase-free conditions. Tissues were
   embedded in Tissue-Tek OCT Compound (Sakura Finetek Japan), snap-frozen
   in liquid nitrogen, sectioned at a thickness of 10 μm with a cryostat
   (Leica Biosystems, Wetzlar, Germany, CM1860), mounted on glass slides
   (Matsunami Glass, PRO-01), and dried at 37°C overnight. Subsequently,
   sections were fixed with 4% neutral-buffered paraformaldehyde for
   20 min, washed with PBS four times, and incubated with 50 μL of miRNA
   hybridization mix at 51°C for 50 min in a Dako hybridizer (Dako,
   Agilent Technologies). The miRNA hybridization mix included 25 μM miRNA
   probes (0.8 μL/slide) and ISH buffer (50 μL/slide, 1:625 dilution,
   final concentration [40 nM], denatured at 90°C for 4 min). Sections
   were sequentially washed (51°C, 5 min, 3 times) with 0.2× saline sodium
   citrate buffer (UltraPure SSC 20×, Thermo Fisher Scientific,
   15557-044), washed with PBS-T for 1 min, blocked for 10 min with 2.5%
   normal horse serum (Vector Laboratories, S-2012-50), and incubated at
   room temperature for 1.5 h with 100 μL of an anti-DIG-alkaline
   phosphatase (ALP) antibody (1:50, Roche, Basel, Switzerland,
   11093274910). After three PBS-T washes for 3 min each, signals were
   visualized with NBT/BCIP solution (incubation at 30°C for 2 h;
   SIGMAFAST BCIP/NBT, Sigma-Aldrich, B5655-5TAB) supplemented with
   levamisole (final concentration [0.2 mM], Sigma-Aldrich, 31742-250MG).
   After signal visualization, the ALP reaction was terminated by two
   washes with KTBT buffer (a mixture of 50 mM Tris-HCl, 150 mM NaCl, and
   10 mM KCl at the final concentration) for 5 min each. Slides were
   washed with distilled water twice for 1 min each, counterstained for
   1 min with Nuclear Fast Red (Vector Laboratories, H-3403), washed for
   10 min with distilled water, dehydrated by passages through graded
   alcohol series, cleared with Lemosol, and mounted with MOUNT-QUICK
   (Daido Sangyo, Tokyo, Japan). Histopathological images were acquired
   with a microscope (KEYENCE, BZ-X700) and analyzed with ImageJ (NIH).

   In miRNA ISH-IHC double staining, ISH steps were performed with
   hsa-miR-145-5p miRCURY LNA miRNA ISH Optimization Kits (QIAGEN, no.
   339456). All solutions used for the ISH steps were prepared under
   RNase-free conditions. Tissues were embedded in Tissue-Tek OCT Compound
   (Sakura Finetek Japan), snap-frozen in liquid nitrogen, sectioned at a
   thickness of 10 μm with a cryostat (Leica Biosystems, CM1860), mounted
   on glass slides (Matsunami Glass, PRO-01), and dried at 37°C overnight.
   Subsequently, sections were fixed with 4% neutral-buffered
   paraformaldehyde for 20 min, washed with PBS four times, and incubated
   with 3% hydrogen peroxide in methanol for 20 min. Slides were incubated
   with 50 μL of miRNA hybridization mix at 51°C for 50 min in a Dako
   hybridizer (Dako, Agilent Technologies). The miRNA hybridization mix
   included 25 μM miRNA probes (0.8 μL/slide) and ISH buffer (50 μL/slide,
   1:625 dilution, final concentration [40 nM], denatured at 90°C for
   4 min). Sections were washed with 0.2× saline sodium citrate buffer
   (51°C, 5 min, 3 times; UltraPure SSC 20×, Thermo Fisher Scientific,
   15557-044), washed with PBS-T for 1 min, blocked for 10 min with 2.5%
   normal horse serum (Vector Laboratories, S-2012-50), and incubated at
   room temperature for 1.5 h with 100 μL of an anti-DIG-peroxidase
   antibody (1:50, Roche, 11207733910). After three PBS-T washes for 3 min
   each, signals were visualized with a cyanine 5-tyramide amplification
   system (TSA Cyanine 5 System, SKU NEL705A001KT, Akoya Biosciences,
   Marlborough, MA). Sections were washed with PBS-T for 3 min, incubated
   with primary antibodies at 4°C overnight ([279]Table S2), washed with
   PBS-T three times for 3 min each, incubated for 2 h with an Alexa Fluor
   488-conjugated secondary antibody (1:100, Cell Signaling Technology),
   washed with PBS-T three times for 5 min each, and then mounted with the
   ProLong Glass Antifade Mountant with NucBlue Stain (Thermo Fisher
   Scientific, [280]P36981). Fluorescence images were acquired with a
   confocal laser scanning microscope (Carl Zeiss, Oberkochen, Germany,
   LSM-710) or a conventional fluorescence microscope (KEYENCE, BZ-X700)
   and analyzed with ImageJ (NIH).

Whole-genome sequencing

   Samples for whole-genome sequencing were obtained from the single rat
   tissues of BiPL (P1, P2, and P3) or the adjacent normal urothelium (N1)
   at 18 weeks. Genomic DNA (0.4 μg, 10 ng/μL) was extracted using the
   QIAamp DNA Mini Kit (QIAGEN), randomly fragmented by sonication,
   analyzed for quantity, integrity, and purity using the 5400 Fragment
   Analyzer System (Agilent), and used for 150 PE sequencing on the
   NovaSeq6000 platform (Illumina, San Diego, CA). Sequence data were
   aligned with a rat reference genome (mRatBN7.2/rn7 Nov. 2020) using the
   Burrows-Wheeler Aligner (version 0.7.17), SAMtools, and Picard. Somatic
   SNPs and indels were identified by GATK (gatk-4.2.5.0, Broad Institute
   of MIT, Cambridge, MA), ANNOVAR,[281]^57 and the rat SNP/indel database
   (The Rat Genome Database[282]^58). Somatic CNVs and SVs were identified
   using CNVkit[283]^59 and BreakPointer (Broad Institute of MIT),[284]^60
   respectively. Mutation signatures were analyzed using Signal.[285]^61
   BiPL-specific somatic variants were calculated by subtracting variants
   in the adjacent normal tissue (N1) from those in BiPL tissues (P1, P2,
   and P3).

Study approval

   All animal experiments were approved by the IACUC of Gifu University
   (approval no. 2020-57) and performed according to the NIH Guide for the
   Care and Use of Laboratory Animals (National Academies Press, 2011).

Statistical analysis

   GraphPad Prism 8 version 8.4.0 (GraphPad Software, San Diego, CA), JMP
   version 12.2 (SAS Institute), and CutoffFinder[286]^62 were used for
   statistical analyses and heatmap generation. Detailed information on
   error bars, p values, and statistical tests is given in the figure
   captions. Unless otherwise specified, all experiments were conducted
   with biological replicates (sample size, n = 3), and measurements were
   taken from distinct samples in each experiment. All experiments were
   performed at least twice, and data for technical and biological
   replicates were reliably reproduced.

Acknowledgments