Abstract

   MicroRNAs may act as diagnostic and prognostic biomarkers of chronic
   kidney disease and are functionally important in disease pathogenesis.
   To identify novel microRNA biomarkers, we performed small
   RNA-sequencing on plasma from individuals with type 2 diabetes, with
   and without chronic kidney disease. MiR-190a-5p abundance was
   significantly lower in the circulation of type 2 diabetic patients with
   reduced function compared to those with normal kidney function. In an
   independent cohort of patients with chronic kidney disease of diverse
   aetiology, miR-190a-5p abundance predicted disease progression in
   individuals with no or moderate albuminuria ( < 300 mg/mmol).
   miR-190a-5p expression in kidney biopsy tissue correlated with the
   level of miR-190a-5p in the circulation and with estimated glomerular
   filtration rate, tubular mass and negatively with histological
   fibrosis. Administration of a miR-190a-5p mimic in a murine
   ischaemia-reperfusion injury model in male mice reduced tubular injury
   and fibrosis and increased expression of genes associated with tubular
   health. Our analyses suggest that miR-190a-5p is a biomarker of tubular
   cell health, low circulating levels may predict chronic kidney disease
   progression independent of existing risk factors and strategies to
   preserve miR-190a-5p may be an effective treatment for restoring
   tubular cell health following kidney injury.

   Subject terms: Prognostic markers, Kidney diseases, Molecular medicine
     __________________________________________________________________

   Chronic Kidney Disease affects 1 in 10 people worldwide with prevalence
   continuing to rise, thus there is a need to identify novel biomarkers
   that can add value to existing clinical and biochemical risk
   predictors. Here the authors identify miR190a-5p as potential indicator
   of kidney health and disease progression in patients with chronic
   kidney disease.

Introduction

   Chronic Kidney Disease (CKD) affects 1 in 10 people worldwide, with
   prevalence continuing to rise in part due to the increasing incidence
   of diabetes^[58]1. The number of patients with diabetes has more than
   doubled since 1998^[59]2, with almost 90% of those diagnosed having
   type 2 diabetes mellitus (T2DM)^[60]3. In the UK, more than 30% of
   incident end-stage kidney disease (ESKD)(need for dialysis or kidney
   transplantation) is due to diabetic kidney disease (DKD)^[61]4, while
   this figure is almost 50% in the US^[62]5.

   Measurements of serum creatinine-based estimated glomerular filtration
   rate (eGFR) and urine albumin to creatinine ratio (ACR) represent the
   current best practice for evaluating CKD and assessing
   prognosis^[63]6,[64]7. However, the pathophysiology of CKD is
   multifactorial, and many patients with CKD have no or low levels of
   albuminuria, even in kidney diseases characterised by albuminuria, such
   as diabetic nephropathy^[65]8. Furthermore, progressive decline in
   kidney function may be observed even in those with
   normoalbuminuria^[66]9. Hence, it is imperative to identify novel
   biomarkers that can add value to existing clinical and biochemical risk
   predictors.

   MicroRNAs (miRs) play important regulatory roles in diverse biological
   processes including metabolism, autophagy, inflammation and
   fibrosis^[67]10–[68]12. They have been implicated in the pathogenesis
   of kidney disease^[69]13–[70]16 and may also represent therapeutic
   targets^[71]17–[72]19. Due to their stability in body fluids, miRs have
   been widely assessed as potential biomarkers^[73]20. Quantification of
   miRs in the urine and circulation^[74]21–[75]23 can predict kidney
   outcomes in patients with CKD due to immunoglobulin A nephropathy
   (IgAN)^[76]24, T2DM^[77]25,[78]26 and predict CKD onset in those with
   normal kidney function^[79]27. However, such studies typically employ
   individual miRs selected according to known biology, which may not
   represent the optimal biomarker. The development of high-throughput
   next-generation sequencing (NGS) technologies offers new opportunities
   for unbiased quantification of miRs^[80]28,[81]29 and has led to the
   identification of potential miR-based biomarkers, including for
   paediatric transplant rejection, CKD and IgAN^[82]30–[83]32.

   The aims of this study were to characterise the circulating miRNome
   using an unbiased measurement of global miR expression profiles in
   individuals with T2DM with or without kidney disease. We further sought
   to assess their potential as prognostic biomarkers in a cohort of
   patients with unselected CKD. Finally, we employed animal and in vitro
   models to characterise the pathophysiological role of the most
   promising candidate, miR-190a-5p.

Results

Identification of miR-190a-5p as a candidate biomarker of CKD

   To identify miRs that may represent novel CKD biomarkers in patients
   with diabetes, we performed small RNA-sequencing on the plasma of
   patients who had type 2 diabetes with CKD (T2DKD, n = 9) and without
   CKD (T2DNRF, n = 13) and non-diabetic controls with normal kidney
   function (NDNRF, n = 11) (Fig. [84]1A). Summary clinical
   characteristics of this biomarker discovery cohort are provided in
   Table [85]1. All participants were of white European ancestry, and
   there were no significant differences in age or sex between the three
   groups. The T2DKD group had significantly lower mean eGFR
   (30.8 ml/min/1.73 m^2) compared to the two control groups (T2DNRF:
   76.9 ml/min/1.73 m^2, NDNRF: 77.6 ml/min/1.73 m^2). Blood pressure was
   tightly controlled in the T2DKD group, with diastolic blood pressure of
   68 mmHg being significantly lower than participants in both the T2DNRF
   (78 mmHg) and the NDNRF control groups (82 mmHg). There was no
   significant difference in glycaemic control between the two groups of
   participants with diabetes (mean HbA1c 62 mmol/mol T2DKD group vs
   64 mmol/mol in T2DNRF group).

Fig. 1. Identification of miR-190a-5p as a differentially expressed
circulating biomarker in T2D patients with normal and kidney disease.

   [86]Fig. 1
   [87]Open in a new tab

   A Schematic of the experimental approach to identify differentially
   expressed circulating miRNAs in patients with Type II diabetes with and
   without kidney disease compared to age-matched controls. BioRender.
   Denby, L. (2025) [88]https://BioRender.com/z8pwjvsB Heatmap
   representation of the unsupervised clustering of miRs with an FDR < 0.1
   in any comparison (n = 45) from the unbiased small RNA sequencing of
   plasma of patients with type 2 diabetes and kidney disease (T2DKD,
   n = 9), type 2 diabetes, with normal renal function (T2DNRF, n = 13) or
   non-diabetic controls with normal renal function (NDNRF, n = 11). Each
   column represents a sample, each row a miRNA, and the colour the
   relative expression (z score). Red box highlights miR-190a-5p.

Table 1.

   Clinical and biochemical characteristics of the patients recruited for
   the sRNA-Seq miR biomarker discovery cohort for miR-190a-5p finding
   Variable Discovery Cohort
   T2DKD (n = 9) T2DNRF (n = 13) NDNRF (n = 11) p- value
   Age (years) 72.7 ± 7.0 69.1 ± 6.9 70.7 ± 7.1 0.503
   Sex (male/female) 7/2 8/5 7/4 0.705
   SBP (mmHg) 123 ± 10 133 ± 17 138 ± 27 0.244
   DBP (mmHg) 68 ± 4 78 ± 10 82 ± 17 0.049
   HbA1c (mmol/mol) 61.7 ± 14.2 64.3 ± 18.0 39.5 ± 2.9 0.031
   Serum creatinine (µmol/l) 181.1 ± 34.3 80.0 ± 16.1 80.2 ± 15.3 <0.001
   eGFR (ml/min/ 1.73 m^2) 30.8 ± 7.3 76.9 ± 15.0 77.6 ± 10.1 <0.001
   [89]Open in a new tab

   Categorical values were assessed using a Chi-square test. For
   continuous values with a normal distribution, ANOVA was used to compare
   3 groups. T2DKD – type 2 diabetic kidney disease, T2DNRF – type 2
   diabetes, normal kidney function, NDNRF – non-diabetic, normal kidney
   function.

Hierarchical clustering of sRNA-seq data to identify the circulating miRNome

   Following alignment to miRbase, a total of 923 mature human miRs were
   detected. Principal component analysis revealed that T2DKD clustered
   away from the NDNRF and T2DNRF (Supplementary Fig. [90]1A).
   Unsupervised hierarchical clustering was performed on the most
   differentially expressed mature miRs across samples
   (Fig. [91]1B)(FDR < 0.1 in any comparison). The samples divided into
   two main groups, with one enriched for patients with T2DKD. To identify
   miRs that are altered in the circulation of those with CKD, we compared
   the miR read counts in those with T2DKD with T2DNRF. Three miRs were
   found to be differentially expressed between the T2DKD and the T2DNRF
   groups, miR-190a-5p (log FC -2.47; FDR 0.034), miR-365b-5p (log FC
   4.94; FDR 0.034) and miR-7111-5p (log FC 5.21; FDR 0.048)
   (Supplementary Data [92]1). Of these miRs, miR-190a-5p was
   significantly lower in patients with reduced kidney function
   (Supplementary Fig. [93]1B), with the other two miRs being increased in
   expression (Supplementary Data [94]1). miR-190a-5p was detected in all
   samples however, the other two miRs were detected in ~ 50% of samples
   and were not taken forward (Supplementary Data [95]1).

seNSOR CKD cohort characteristics and correlation with serum miR-190a-5p

   As circulating miR-190a-5p level was reduced in patients with type 2
   diabetes who had impaired kidney function, we next sought to determine
   whether it correlated with eGFR in patients with a diverse aetiology of
   CKD and whether it could predict kidney outcomes. To address this, we
   utilised a prospective cohort of 549 patients who attended kidney
   outpatient clinics (seNSOR study). We excluded 135 participants who
   were already receiving or were close to requiring renal replacement
   therapy (RRT: eGFR of < 20 ml/min/1.73 m^2 at recruitment). In
   addition, 6 participants were excluded due to missing data (3
   participants had no baseline eGFR measurement whilst an additional 3
   participants had no follow-up data available), as were 13 participants
   with acute kidney injury (AKI) at baseline (Fig. [96]2). In 298 of the
   remaining 395 patients, serum miR-190a-5p was measured by RT-qPCR, with
   expression undetectable in 97 participants. Median eGFR was lower in
   those with undetectable miR-190a-5p (50 ml/min/1.73 m^2, IQR 31-79 vs
   61 ml/min/1.73 m^2, IQR 37-93, p-value = 0.018, see Supplementary
   Table [97]2 for the detailed clinical characteristics of this group).

Fig. 2. Flowchart of patients from the seNSOR study cohort who were included
and excluded from the analysis of miR-190a-5p as a potential biomarker.

   Fig. 2
   [98]Open in a new tab

   From the 549 participants in the seNSOR cohort, patients were excluded
   who were already receiving, or close to requiring RRT (eGFR of
   < 20 ml/min/1.73 m^2 at recruitment). Patients with missing data or
   acute kidney injury at baseline were also excluded. From the 395
   participants meeting the study criteria, miR-190a-5p was not detected
   in the serum from 97 patient samples. Analysis was performed in the
   remaining 298 participants. RRT – renal replacement therapy.

   Clinical characteristics for participants with detectable serum
   miR-190a-5p are shown in Table [99]2, stratified according to whether
   their serum miR-190a-5p level was above or below the median.
   Participants with levels below the median tended to be older and had
   lower eGFR compared to those above median levels (median eGFR
   55 ml/min/1.73 m^2 (IQR 34-89 ml/min/1.73 m^2) vs 67 ml/min/1.73 m^2
   (IQR 39-99 ml/min/1.73 m^2), p = 0.052). Serum miR-190a-5p levels
   correlated weakly with eGFR (rho = 0.12, p = 0.04) and inversely with
   age (rho = -0.12, p = 0.04) but were not significantly correlated with
   systolic or diastolic blood pressure nor ACR (Supplementary
   Fig. [100]2).

Table 2.

   Baseline clinical and biochemical characteristics of the 298 patients
   from the seNSOR cohort who were included in the validation analyses,
   segregated according to whether their serum miR-190a-5p levels were
   above or below the median
   seNSOR cohort miR-190a-5p below Median miR-190a-5p above Median p
   comparing above vs below median
   N 298 149 149 –
   Male/Female, n 156/142 76/73 80/69 0.73
   Age (years), median(IQR) 54 (41, 68) 56 (43, 68) 51 (38, 66) 0.067
   eGFR (ml/min/1.73 m^2), median(IQR) 61 (37, 93) 55 (34, 89) 67 (39, 99)
   0.052
   SBP (mmHg), median(IQR) 130 (120, 145) 130 (120, 145) 130 (120, 144)
   0.63
   DBP (mmHg), median(IQR) 78 (70, 82) 78 (68, 84) 79 (70, 82) 0.81
   ACR (mg/mmol), median(IQR) 19 (3, 140) 15 (1, 141) 21 (3, 133) 0.36
   ACR <3 mg/mmol, n (%) 67 (22) 36 (24) 31 (21) –
   ACR 3-300 mg/mmol, n (%) 156 (52) 78 (52) 78 (52) –
   ACR >300 mg/mmol, n (%) 29 (10) 13 (9) 16 (11) –
   ACR unavailable, n (%) 46 (15) 22 (15) 24 (16) –
   Diabetes prevalence, n (%) 48 (16) 28 (18.8) 20 (13.4) 0.27
   Type 1, n (%) 3 (1) 2 (1.3) 1 (0.7) –
   Type 2, n (%) 45 (15) 26 (17.4) 19 (12.8) –
   Ethnicity – – – –
   White, n (%) 261 (88) 129 (87) 132 (89) –
   Asian, n (%) 14 (5) 10 (7) 4 (3) –
   Black, n (%) 6 (2) 3 (2) 3 (2) –
   Not recorded, n (%) 17 (6) 7 (5) 10 (7) –
   Primary kidney diagnoses grouping – – – –
   1. Primary Glomerulonephritis, n (%) 85 (29) 34 (23) 51 (34) –
   2. Interstitial Nephropathies, n (%) 77 (26) 38 (26) 39 (26) –
   3. Multisystem diseases, n (%) 67 (22) 39 (26) 28 (19) –
   4. Diabetes Nephropathy, n (%) 18 (6) 8 (5) 10 (7) –
   5. Other, n (%) 14 (5) 4 (3) 10 (7) –
   6. Not known, n (%) 37 (12) 26 (17) 11 (7) –
   [101]Open in a new tab

   Categorical values were assessed using a Chi-square test. For
   continuous values with a normal distribution, a t test was used to
   compare the 2 groups. IQR interquartile range, eGFR estimated
   glomerular filtration rate, SBP systolic blood pressure, DBP diastolic
   blood pressure, ACR urinary albumin:creatinine ratio.

Low serum miR-190a-5p level predicts risk of CKD progression

   During a median follow-up of 3.9 years, 70 of the 298 participants from
   the seNSOR cohort with detectable serum miR-190a-5p met the composite
   CKD progression endpoint, with 21 reaching ESKD. In those with no
   albuminuria (ACR < 3 mg/mmol) or moderate albuminuria
   (ACR < 3-300 mg/mmol), a serum miR-190a-5p level below the median was
   associated with increased risk of CKD progression (p = 0.028 in those
   with ACR < 3 mg/mmol and p = 0.049 in those with ACR 3-300 mg/mmol,
   Fig. [102]3A). However, in participants with ACR > 300 mg/mmol, serum
   miR-190a-5p was not predictive of CKD progression.

Fig. 3. miR-190a-5p expression predicts kidney disease progression in
patients with low to moderate proteinuria and is reduced in kidney tissue in
patients with declining kidney function.

   [103]Fig. 3
   [104]Open in a new tab

   A Kaplan-Meier survival curve for progression of kidney disease.
   Patients within the seNSOR cohort were segregated into groups by
   severity of albuminuria and whether serum miR-190a-5p levels were above
   (dashed line) or below (solid line) the median value. ACR: urinary
   albumin:creatinine ratio. The Log-rank test was used to compare curves.
   B Schemata of human CKD biopsy tissue being processed for
   quantification of miR-190a-5p expression, fibrosis, tubular mass by
   CD10/PanCK staining, and linked to clinical data. Created in BioRender.
   Denby, L. (2025) [105]https://BioRender.com/ee0461g (C–F) miR-190a-5p
   in renal biopsy tissue from patients within the seNSOR cohort
   correlates with (C). miR-190a-5p transcript abundance in the
   circulation (n = 14 paired samples). For the determination of linear
   correlation between variables, correlation coefficients (r) were
   generated using Pearson’s test. Circulating miR-190a-5p expression was
   log-transformed before entering the model due to non-normal
   distribution (D). kidney function as determined by eGFR (n = 26 biopsy
   samples). Analysed by Pearson’s correlation (E), negatively with
   percentage fibrosis (n = 25 biopsy samples). Analysed by Pearson’s
   correlation (F), tubular mass as measured by CD10:PanCK staining of
   biopsy core (n = 24 biopsy samples).

   We proceeded to examine whether miR-190a-5p levels predicted CKD
   progression in univariate and multivariate Cox proportional hazards
   models. As shown in Table [106]3, in those with baseline
   ACR < 300 mg/mmol, miR-190a-5p was associated with CKD progression
   independent of age, sex, baseline systolic/diastolic blood pressure,
   eGFR and ACR (adjusted HR 0.80, 95% CI: 0.66-0.96, p = 0.015).
   Importantly, addition of miR-190a-5p to a model using existing clinical
   risk predictors resulted in an improvement in the C-statistic from
   0.719 to 0.738, with the likelihood ratio test confirming a
   statistically significant improvement in kidney outcome prediction
   (p = 0.014). Notably, when known biomarkers urinary Epidermal Growth
   Factor (EGF) and Kidney Injury Molecule-1 (KIM-1) were included in this
   model, neither predicted CKD progression, whilst miR-190a-5p retained
   statistical significance (Supplementary Table [107]3).

Table 3.

   Univariate and multivariate Cox proportional hazards models for the
   risk of reaching end-stage kidney disease (ESKD) or having a
   sustained > 30% decline in kidney function in patients with baseline
   urinary ACR < 300 mg/mmol
   Unadjusted, univariate analysis Adjusted, multivariate analysis
   HR 95% CI p HR 95% CI p
   log[2] miR-190a-5p 0.84 0.72–0.98 0.029 0.8 0.66–0.96 0.015
   Baseline eGFR per 10 ml/min/1.73m^2 0.84 0.76–0.93 <0.001 0.76
   0.65–0.88 <0.001
   ACR per 10 mg/mmol 1.04 1.01–1.08 0.008 1.04 1.01–1.08 0.016
   Age per 10 years 1.03 0.87–1.22 0.718 0.79 0.6–1.04 0.086
   SBP per 10 mmHg 1.16 1.01–1.33 0.038 1.12 0.96–1.32 0.155
   DBP per 10 mmHg 1.04 0.82–1.34 0.728 0.97 0.73–1.29 0.831
   Female (vs male) 0.74 0.43–1.27 0.274 1.11 0.56–2.17 0.77
   [108]Open in a new tab

   Cox univariate and multivariate proportional hazards survival models
   were performed to assess factors that predicted the endpoints, for
   multivariate analysis we excluded those patients with no ACR available.

   eGFR estimated glomerular filtration rate, ACR urinary
   albumin:creatinine ratio, SBP systolic blood pressure, DBP diastolic
   blood pressure.

   When participants with an ACR > 300 mg/mmol were included in the
   analysis (see Supplementary Table [109]4), miR-190a-5p was no longer
   predictive of CKD progression in univariate and multivariate models
   (univariate HR 0.89, 95% CI: 0.77–1.02, p = 0.087, multivariate HR
   0.86, 95% CI: 0.73–1.01, p = 0.064). Secondary analysis examining ESKD
   alone as the endpoint determined that serum miR-190a-5p level was
   independently associated with reaching ESKD, regardless of whether
   participants with baseline ACR > 300 mg/mmol were included in the
   analysis (adjusted HR when including all participants 0.68, 95% CI:
   0.53-0.89, p = 0.015, Supplementary Table [110]5).

Renal miR-190a-5p expression correlates with kidney function

   We next sought to determine the biological rationale for the
   correlation of serum miR-190a-5p levels with baseline eGFR and adverse
   kidney outcomes in patients with CKD. Firstly, we wished to examine the
   likely tissue source of circulating miR-190a-5p. As miR-190a-5p is
   found in the intronic region of the TALIN2 (TLN2) gene, and has been
   shown to use the host gene transcription start site^[111]33, we
   assessed tissue expression of TLN2 as a surrogate of miR-190a-5p in a
   range of human tissues using publicly available datasets
   ([112]GSE3526)^[113]34. TLN2 expression was highest in the kidney
   (Supplementary Fig. [114]3A), suggesting that it may be the primary
   source of miR-190a-5p in the circulation. To validate this, we first
   assessed the expression of miR-190a-5p in frozen biopsy tissue from a
   subset of seNSOR patients who underwent renal biopsy at the time of
   recruitment, including patients with biopsy proven diabetic nephropathy
   (representing 15% of samples) (Fig. [115]3B). Expression of miR-190a-5p
   correlated significantly with circulating miR-190a-5p levels
   (Fig. [116]3C). The expression of miR-190a-5p in kidney biopsy tissue
   also correlated with eGFR (Fig. [117]3D) and was negatively correlated
   with the percentage of fibrosis present in the biopsy (Fig. [118]3E).
   MiR-190a-5p expression in the biopsy also correlated with kidney
   tubular cell mass as assessed by immunofluorescence staining for the
   epithelial markers CD10 and PanCK (Fig. [119]3F).

   We next wished to determine whether the loss of miR-190a-5p expression
   in the kidney was observed in pre-clinical models of kidney injury.
   miR-190a-5p expression in the renal cortex was significantly reduced in
   AKI-to-CKD transition induced by uIRI injury with nephrectomy^[120]35
   (Fig. [121]4A) and in progressive kidney disease induced by the
   subtotal nephrectomy model^[122]36 (Fig. [123]4B).

Fig. 4. miR-190a-5p is lost in pre-clinical models of kidney disease and is
enriched in proximal tubular cells.

   [124]Fig. 4
   [125]Open in a new tab

   A miR-190a-5p expression was measured by RT-qPCR, normalised to U6 in
   RNA extracted from kidneys from mice culled 2 weeks after sham (n = 4,
   male mice) or uIRI surgery with contralateral nephrectomy (Nx, n = 6,
   male mice). Data are presented as mean normalised expression +/– SEM.
   Data analysed by Mann-Whitney test. **p = 0.0095 vs sham animals (B)
   miR-190a-5p expression was measured by RT-qPCR, normalised to U6 in RNA
   extracted from kidneys from mice culled 12 weeks post sham (n = 5,
   male) or subtotal nephrectomy (STNx) surgery to induce progressive
   kidney disease (n = 8, male). Data are presented as mean normalised
   expression +/− SEM. Data analysed by Mann Whitney test. * p = 0.0186 vs
   Sham animals. C Expression of miR-190a-5p determined by small RNA-seq
   performed on cell subsets isolated by FACS from the renal cortex of
   mice undergoing a reversible ureteric obstruction model. n = 16/cell
   type. Box (IQR) is the median, Q1 (25th percentile), Q3 (75th
   percentile). The whiskers represent: Minimum Q1-1.5*IQR and Maximum
   Q3-1.5*IQR. IQR = Interquartile range. D Small RNA-Seq of sorted
   cortical proximal tubular cells during the rUUO model. n = 3-4/gp. Box
   (IQR) is the median, Q1 (25th percentile), Q3 (75th percentile). The
   whiskers represent: Minimum Q1-1.5*IQR and Maximum Q3-1.5*IQR.
   IQR = Interquartile range (E). Correlation of miR-190a-5p expression
   and Tln2 in bulk cortical rUUO tissue. n = 20. Analysed by Pearson’s
   correlation. FPKM = Fragments Per Kilobase of transcript per Million
   mapped reads. CPM = Counts per Million.

   Taken together, these data suggest that the level of circulating
   miR-190a-5p may reflect the expression in the kidney, and loss of
   miR-190a-5p is a feature of reduced kidney function.

miR-190 is highly expressed in proximal tubules

   As we observed a correlation of tissue miR-190a-5p expression with
   tubular mass, we wished to identify the specific cellular source of
   miR-190a-5p within the kidney. As miR-190a-5p is conserved across
   species (Supplementary Fig. [126]3B), we employed data from a murine
   model of reversible unilateral ureteric obstruction
   (rUUO)^[127]37,[128]38. Standard gene expression analysis and sRNA-seq
   was performed on bulk cortical tissue and on four individual cell
   populations isolated by fluorescence- activated cell sorting: proximal
   tubular cells (PT, Lotus lectin^+), macrophages (F4/80^Hi), endothelial
   cells (CD31^+) and stromal cells (PDGFRB^+)^[129]37. miR-190a-5p was
   highest in PT cells (Fig. [130]4C), and was reduced following UUO
   before being partially restored during the repair phase (Fig. [131]4D),
   mirroring that of Tln2 expression (Supplementary Fig. [132]3C). Indeed,
   Tln2 expression correlated with miR-190a-5p in the rUUO model
   (r = 0.9127; p < 0.001) (Fig. [133]4E). This suggests that, in mouse
   and humans, Tln2 expression could potentially act as a surrogate for
   miR-190a-5p expression.

Restoring miR-190a-5p abundance in the kidney ameliorates tubular injury and
renal fibrosis

   We next sought to determine whether restoring miR-190a-5p abundance in
   the kidney could mitigate tubular injury and maintain tubular cell
   health. Our previous work and that of others, has demonstrated avid
   uptake of miRNA-based therapies by kidney
   tubules^[134]10,[135]17,[136]18, therefore, we administered miR-190a-5p
   mimic or mimic control subcutaneously in the uIRI model of AKI-to-CKD
   transition (Fig. [137]5A). We found that 14 days post-IRI there was a
   significant increase in miR-190a-5p expression in the kidneys of
   miR-190a-5p mimic treated animals compared to kidneys from mimic
   control-treated animals (Fig. [138]5B), which had miR-190a-5p
   expression similar to that observed in kidneys with damage induced by
   IRI + Nephrectomy or STNx (Fig. [139]4A, B). The increased renal
   miR-190a-5p abundance resulted in a significant decrease in Havcr1
   (encodes kidney injury molecule-1, KIM-1) expression and an increase in
   Egf and Ass1 expression, markers of healthy Loop of Henle and proximal
   tubules, respectively (Fig. [140]5C). These findings are indicative of
   miR-190a-5p having a potential role in maintaining tubular health. As
   miR-190a-5p expression negatively correlated with fibrosis in the human
   biopsies, we sought to determine if preventing the loss of miR-190a-5p
   would have an effect on the induction of fibrosis in the uIRI murine
   model. We found that administration of miR-190a-5p mimic reduced
   fibrosis-related gene expression (Fig. [141]5D) and collagen
   deposition, as measured by PSR staining (Fig. [142]5E).

Fig. 5. Overexpression of miR-190a-5p in the kidney reduces injury and
fibrosis induced by uIRI.

   [143]Fig. 5
   [144]Open in a new tab

   A Schema of the intervention study in which mice were randomly assigned
   to either receive subcutaneous treatment with miR-190a-5p mimic or
   control mimic (n = 7/gp) and uIRI surgery. Animals were dosed on 7
   occasions (purple arrows) and culled at Day 14 post-surgery. Created in
   BioRender. Denby, L. (2025) [145]https://BioRender.com/s67k0ep (B)
   miR-190a-5p expression in the kidney following intervention with
   miR-190a-5p or control mimic was determined by RT-qPCR from RNA
   extracted from the IRI kidneys and normalised to U6 (n = 7/group). Data
   analysed by an unpaired Student’s t test. *p = 0.0167 vs control mimic
   treated. C RNA extracted from the IRI kidneys at Day 14 was profiled
   for the expression of genes associated with tubular health and
   normalised to Ppia (n = 7/group). Data analysed by an unpaired
   Student’s t test. For Havcr-1 *p = 0.0322; Egf ***p = 0.0002; Ass1
   *p = 0.0489 vs control mimic treated. D Gene expression of
   fibrosis-related transcripts were assessed in RNA extracted from the
   IRI kidneys at Day 14 and normalised to Ppia (n = 7/group). Data
   analysed by unpaired Student’s t test. *p = 0.0388 vs control mimic
   treated. E Fibrosis was semi-quantified by picrosirus red staining of
   non-overlapping FFPE sections prepared from IRI and sham kidneys at Day
   14 (n = 7/group). Staining was quantified using ImageJ. Data analysed
   by an unpaired Student’s t test. *** p = 0.0003 vs control treated
   kidney.

Adam10 may be a miR-190a-5p target gene

   To investigate which genes are being modulated by the loss of
   miR-190a-5p in the PTs cells, we took advantage of the transcript
   sequencing data from the sorted PTs in the rUUO model and employed
   MultimiR to identify miR-190a-5p predicted target genes that were
   significantly negatively correlated with miR-190a-5p expression within
   proximal tubular cells (Fig. [146]6A and Supplementary Table [147]6).
   The 45 most significantly negatively correlated target genes were
   over-represented in multiple pathways, including wound healing,
   regulation of cytoskeleton and actin polymerisation/stabilisation
   (Fig. [148]6B). One of the genes most significantly negatively
   correlated with miR-190a-5p was Adam10 (Fig. [149]6A; p = 0.004), which
   is a sheddase and has been shown to mediate kidney injury and fibrosis
   in pre-clinical models^[150]39. Using Targetscan, we confirmed that
   ADAM10 has a 7A-mer 3’UTR binding site for miR-190a-5p and is therefore
   a logical candidate miR-190a-5p target gene for further study.
   Mediation of Adam10 repression by miR-190a-5p is supported by the
   significant negative correlation observed between Adam10 and
   miR-190a-5p expression in bulk cortical tissue (r = − 0.77, p < 0.001)
   in the rUUO model (Fig. [151]6C).

Fig. 6. Identification and validation of ADAM10 as a miR190a-5p target in
proximal tubular cells.

   [152]Fig. 6
   [153]Open in a new tab

   A Analysis of potential negatively correlated gene targets of
   miR-190a-5p ordered by correlation coefficient (r) and –log10(p-value)
   in PT cells from the rUUO model. r < -0.6, p-value < 0.05. Analysed by
   Pearson’s correlation. PT = Proximal Tubule (B). Pathway enrichment
   analysis for negatively correlated miR-190a-5p target genes. C
   Correlation of Adam10 with miR-190a-5p expression in the renal cortex
   of the reversible ureteric obstruction (rUUO) model. For the
   determination of linear correlation between variables, correlation
   coefficients (r) were generated using Pearson’s test. FPKM = fragments
   per kilobase mapped. CPM = Counts per million. D PT cells ordered in
   pseudotime trajectory from healthy to inflammatory cell state in the
   non-tumorous portion of nephrectomy tissue from patients with and
   without ureteric obstruction. Right graph. Mean ADAM10 or TLN2
   expression in each cell along the pseudotime trajectory. E ADAM10
   expression in human proximal tubular cell line (RPTECs) with
   over-expression (50 nM mimic) or knockdown (100 nM inhibitor) of
   miR-190a-5p. Data analysed by unpaired Student’s t test. **p = 0.0073
   (mimic), *p = 0.0173 (inhibitor) vs scrambled control (n = 3 biological
   replicates). PT – proximal tubule Created in BioRender. Denby, L.
   (2025) [154]https://BioRender.com/0ljj331 (F). Adam10 expression in the
   renal cortex of mice that underwent ischaemia-reperfusion injury and
   were administered miR-190a-5p mimic or scrambled control (n = 7/group).
   Data analysed by an unpaired Student’s t test. ***p = 0.0007 vs Control
   Mimic.

   Recent studies in kidney disease have revealed the presence of a subset
   of injured PT cells that express inflammatory genes and are implicated
   in the generation of the fibrotic niche^[155]35,[156]40,[157]41. We
   utilised a single-nucleus RNA-Seq (snRNA-Seq) dataset generated from
   the non-tumorous portion of tumour nephrectomy tissue, where the
   kidneys were otherwise healthy or the ureter was obstructed by tumour,
   leading to an inflammatory and fibrotic response in the kidney^[158]35.
   We ordered cells along a trajectory from healthy to inflammatory PT
   cells (Fig. [159]6D). As miR data is not captured by snRNA-Seq, we used
   TLN2 expression as a surrogate for miR-190a-5p and found it was
   decreased in inflammatory PT cells compared to healthy PT cells, with
   ADAM10 expression being conversely increased (Fig. [160]6D).

   Finally, we tested if ADAM10 expression was regulated by the
   manipulation of miR-190a-5p expression in vitro and in vivo. We
   overexpressed and knocked down miR-190a-5p in the human renal proximal
   tubule (RPTEC) cell line (Supplementary Fig. [161]4A, B). We performed
   a dose response study (Supplementary Fig. [162]4C) and determined that
   increasing miR-190 expression reduced ADAM10 expression in the RPTEC
   cells (Fig. [163]6E). Conversely miR-190 knockdown resulted in an
   increase in ADAM10 expression (Fig. [164]6E). We then tested if Adam10
   was altered in our miR-190a-5p intervention study (Fig. [165]5) as we
   would hypothesise that over-expression of miR-190a-5p should repress
   Adam10 expression induced by the uIRI. Indeed, we found that Adam10
   expression was significantly (p < 0.001) reduced in miR-190a-5p treated
   kidneys when compared to mimic control treated uIRI kidneys
   (Fig. [166]6F). Hence, we confirm that miR-190a-5p may inhibit ADAM10
   expression in tubular cells, however we cannot define if this is an
   indirect or direct effect on ADAM10.

Discussion

   While miRs have been assessed previously as non-invasive diagnostic and
   prognostic biomarkers of CKD, most studies have employed a single
   miR^[167]22,[168]42 or a focused panel based on existing knowledge of
   miR biology^[169]23,[170]24,[171]26,[172]27, with only a few assessing
   a larger number of miRs^[173]43–[174]46. Focused approaches may not
   identify the optimal miR biomarker, therefore, we elected to perform an
   unbiased screen using sRNA-seq in plasma samples, initially in patients
   with diabetes, with or without CKD. Plasma miR-190a-5p levels were
   lower in patients with diabetes who had CKD compared to those with
   normal kidney function. The presence of diabetes did not impact on
   miR-190a-5p levels, suggesting that kidney function per se was the
   biggest determinant of miR-190a-5p levels. Indeed, miR-190a-5p was
   correlated with eGFR in a second clinical cohort (seNSOR) of patients
   with CKD due to diverse aetiologies, suggesting that this was not a
   specific feature of diabetic kidney disease, but a generic phenomenon
   in CKD. Our study is the first to explore the association between
   circulating miR-190a-5p levels and current and future kidney function.

   MiRs in the circulation may be derived from multiple organs, therefore,
   we explored the biological rationale for the reduced miR-190a-5p level
   observed in the circulation of patients with CKD. TLN2, the host gene
   for miR-190a-5p, was highly expressed in the human kidney, suggesting
   that the kidney may be the primary source of miR-190a-5p in the
   circulation and indeed we found that circulating levels correlated with
   kidney expression in a subset of our cohort. Our findings that
   miR-190a-5p was robustly expressed in human kidney tissue and lost as
   function declines, are in line with a previous report that miR-190a-5p
   was downregulated in kidney biopsy tissue of patients with progressive
   CKD compared with stable CKD^[175]46. Recent studies using single-cell
   modalities to study the kidney in human kidney injury have identified
   an important inflammatory PT subcluster of the ‘adaptive’ tubular cells
   observed in the KPMP Atlas^[176]47 that are central to the pathogenesis
   of kidney injury^[177]35. These injured PT cells are enriched for
   pro-inflammatory and pro-fibrotic genes and are consistent with Vcam1^+
   Ccl2^+ PT cells which have been observed in pre-clinical
   models^[178]40,[179]41,[180]48. We found that TLN2 was significantly
   downregulated in human inflammatory PT cells suggesting that
   miR-190a-5p may also be downregulated in this cell type. Furthermore,
   within the kidney, we have now demonstrated loss of miR-190a-5p in
   kidney biopsies in patients with CKD, enrichment of miR-190a-5p within
   proximal tubular cells and that kidney injury downregulates both
   cortical and proximal tubular miR-190a-5p expression in pre-clinical
   models. Furthermore, miR-190a-5p expression correlated with eGFR,
   tubular mass and was negatively correlated with renal fibrosis. Thus,
   we speculate that the reduced circulating levels of miR-190a-5p in
   patients with CKD may be a biomarker of reduced tubular mass and could
   be used to identify patients who may benefit from miR-190a-5p mimic
   therapy.

   Importantly, low serum miR-190a-5p levels predicted adverse kidney
   outcomes in patients with CKD who had low-moderate proteinuria
   independent of traditional clinical risk factors, suggesting that the
   addition of miR-190a-5p may provide additional prognostic information.
   However, there are important caveats to its clinical use. Firstly, it
   is found at low levels in the circulation, particularly in patients
   with CKD (median 20 reads/million in those with CKD in the sRNA-seq),
   therefore, it might be challenging to quantify by RT-qPCR; indeed, we
   could not reliably detect miR-190a-5p in the serum of all the patients,
   typically those with lower eGFR. This may reflect the lack of tubular
   mass and hence more sensitive assays to detect miR-190a-5p are required
   to enable translation into the clinic. Secondly, miR-190a-5p levels did
   not provide useful prognostic information in those patients with very
   high levels of proteinuria. It is possible that this reflects altered
   binding to serum protein or increased permeability to miRs at a damaged
   glomerular filtration barrier; however, there was no correlation
   between serum miR-190a-5p levels and the severity of proteinuria. A
   more likely explanation is that those with heavy proteinuria have
   predominantly glomerular disease, and therefore a glomerular biomarker
   such as albuminuria will more accurately predict kidney outcome than a
   surrogate marker of tubular mass, such as miR-190a-5p. Importantly,
   serum miR-190a-5p levels will be most useful as a prognostic biomarker
   in those with no or low levels of proteinuria, for whom risk prediction
   is currently more challenging.

   We provide new insight into the mechanistic role of miR-190a-5p in the
   kidney. It’s predicted target genes are enriched in pathways including
   response to wounding and actin cytoskeletal organisation, and indeed
   its host gene, TLN2, is a cytoskeletal protein important for cell
   adhesion and motility^[181]49. The results from our murine studies
   suggest that miR-190a-5p is enriched in healthy tubular cells, where it
   may regulate the cytoskeleton to maintain an epithelial phenotype;
   indeed, miR-190a-5p has been shown to prevent epithelial-mesenchymal
   transition^[182]50. miR-190a-5p was also expressed at a much lower
   level in other kidney cells, including macrophages, fibroblasts and
   endothelial cells, therefore, we could not exclude that it may regulate
   gene expression in a more subtle manner in these cell types.

   We demonstrate that Adam10 expression changes reciprocally to that of
   miR-190a-5p in the pre-clinical model of uIRI, with supplemented
   miR-190a-5p expression profoundly reducing Adam10 expression in vivo
   and in vitro, suggesting that ADAM10 may be a direct or indirect target
   of miR-190a-5p. From our data, we postulate that the loss of
   miR-190a-5p in kidney disease may lead to aberrant expression of
   ADAM10, but further study would be required to provide additional
   evidence to support this. Further experiments would be required to
   demonstrate if ADAM10 is a direct target of miR-190a-5p, such as
   employing a 3’UTR binding assay. Our pre-clinical data suggests that
   miR-190a-5p may maintain kidney cell health in part by inhibiting
   ADAM10 expression, and this fits with other data which has demonstrated
   that Adam10 knockout protects against kidney injury^[183]39. However,
   this requires confirmation in additional studies, such as by
   demonstrating that the protective effect of the miR-190a-5p mimetic is
   abrogated in Adam10 knockout mice, and that conditional tubular
   knockout of miR-190a exacerbates kidney injury through elevated ADAM10
   expression. ADAM10 promotes shedding of ectodomains of cell surface
   adhesion molecules such as cadherins, which are important in
   maintaining the integrity of the epithelial cell monolayer^[184]51. In
   addition, ADAM10 activates Notch signalling, which is important in
   embryological development of the kidney, however, sustained activity
   may promote kidney fibrosis^[185]51. Furthermore, ADAM10 is
   up-regulated by TGF-β to act as the principal ephrin-B2 sheddase in
   fibroblasts, and pharmacological inhibition of ADAM10 inhibits the
   development of pulmonary fibrosis^[186]52.

   In addition to its role as a biomarker, miR-190a-5p offers promise as a
   therapy. Over-expression of miR-190a-5p in the uIRI model had a
   pronounced effect, reducing tubular injury and fibrosis, and preserving
   markers of differentiated tubular cells. Others have reported that
   miR-190a-5p inhibits epithelial-mesenchymal differentiation in cancer
   cells, reducing tumour growth and metastases in in vivo
   models^[187]50,[188]53. In addition, miR-190a-5p promotes tissue
   adaptation to hypoxia^[189]54, protects against oxidative
   stress^[190]55, and miR-190 mimics ameliorated neurological injury in
   pre-clinical models of ischaemic stroke^[191]56. Our approach to
   overexpress miR-190a-5p was untargeted, with quite wide variation in
   the level of miR-190a-5p overexpression achieved in the kidney using
   this approach. There is likely scope to improve the efficacy using a
   more refined kidney targeting methods, such as kidney-targeted lipid
   nanoparticles^[192]57, with improved payload delivery^[193]58.

   In summary, we have used sRNA-seq to identify that miR-190a-5p levels
   are reduced in patients with lower kidney function and that low
   circulating miR-190a-5p can predict adverse kidney outcomes in patients
   with low-grade proteinuria independently of standard risk factors,
   though these findings require replication in additional cohorts.
   miR-190a-5p is predominantly expressed in healthy kidney tubules and it
   may regulate ADAM10 expression and miR-190a-5p mimics should be
   explored as a potential therapy for CKD.

Methods

Study populations

   For biomarker discovery using small RNA-sequencing, participants were
   recruited and gave written consent from kidney and cardiovascular
   clinics at the Belfast City and Royal Victoria Hospitals, Belfast. At
   the time of the study visit, the presence of CKD was determined on the
   basis of ≥ 2 eGFR measurements < 60 ml/min/1.73 m^2 > 3 months apart or
   persistent albuminuria expressed as albumin-to-creatinine ratio
   (ACR) ≥ 3 mg/mmol in line with the clinical definition of CKD^[194]6.
   T2DM was diagnosed according to the American Diabetes Association 2010
   criteria^[195]59. Patients with systemic metabolic disease other than
   T2DM, or those who had experienced infection within the previous month,
   were excluded from the study. In the discovery cohort (n = 33) patients
   were categorised into 3 groups: type 2 diabetes with kidney disease
   (n = 9) (T2DKD - T2DM and eGFR < 60 ml/min/1.73 m^2, ACR in
   3–30 mg/mmol range), type 2 diabetes and normal kidney function
   (n = 13) (T2DNRF - T2DM and eGFR > 60 ml/min/1.73 m^2, ACR ≤ 3 mg/mmol)
   and no diabetes and normal kidney function (n = 11)(NDNRF).

   To assess the utility of miR-190a-5p as a prognostic biomarker in
   patients with unselected CKD, we employed 549 patients from a
   prospective cohort study – ‘Non-invasive biomarkers of kidney disease’
   (seNSOR), which recruited patients with unselected aetiology of CKD
   from outpatient clinics at the Royal Infirmary, Edinburgh, Scotland
   between March 2017 to March 2019. Participants with acute kidney injury
   (AKI) determined using the KDIGO creatinine-based criteria^[196]6 were
   excluded.

   In both centres, research nurses obtained written consent and
   participant information was collected upon enrolment, including age,
   self-reported sex, ethnicity, diabetes status, blood pressure,
   aetiology of CKD and current prescribed medications. Baseline
   laboratory data obtained included: serum creatinine, albuminuria (as
   urinary albumin to creatinine ratio, uACR) and glycated haemoglobin A1c
   (HbA1c). Glomerular filtration rate was estimated from serum creatinine
   using the 2009 Chronic Kidney Disease Epidemiology Collaboration
   (CKD-EPI) equation, excluding race^[197]60. For the seNSOR cohort,
   kidney biopsy data and kidney outcome data were captured using the NHS
   Lothian patient record systems. The primary outcome of CKD progression
   was defined as reaching ESKD (starting renal replacement therapy (RRT)
   or maintaining an eGFR < 15 ml/min/1.73 m^2 for > 90 days, or > 30%
   reduction in kidney function from eGFR at baseline maintained for > 90
   days). Reaching ESKD alone was also used as a secondary outcome.
   Ethical approval was obtained from the respective Offices for Research
   Ethics Committees (Northern Ireland REC/14/NI/1132; Scotland
   REC/15/ES/0094). For the biopsy samples, tissue use was approved by the
   steering committee of the National Research Scotland Lothian
   Bioresource (REC 20/ES/0061, study SR2175).

Small RNA-sequencing (sRNA-Seq)

   Whole blood was collected in an ethylenediaminetetraacetic acid (EDTA)
   collection tube, centrifuged at 3000 × g for 10 minutes at 4 °C, and
   the supernatant aliquoted in RNase/DNase-free tubes and frozen at
   − 80 °C. RNA was extracted from 600 μl of plasma using the NextPrep™
   MagnaZol™ cfRNA Isolation Kit (Bio Scientific Corp, Texas, USA) and
   quantified using the Qubit miRNA Assay kit (Thermo Fisher Scientific,
   San Jose, USA).

   microRNA libraries were prepared using QIAseq miRNA Library Kit
   (Qiagen, Hilden, Germany). Library concentrations were quantified using
   the Qubit dsDNA HS Assay Kit (Thermo Fisher Scientific), and quality
   control was performed using a High Sensitivity NGS Fragment Analysis
   Kit (Agilent Technologies, Santa Clara, CA, USA). Libraries were pooled
   at 4 nM molarity and sequenced using the NextSeq 500 High-Output Kit
   (Illumina, San Diego, CA, USA). Raw read quality control was performed
   with FastQC, adaptors trimmed with Trim Galore, reads collapsed with
   seqcluster and aligned with Bowtie1 against miRBase mature miRNA and
   miRBase hairpin (miRbase Ver 22, available at [198]http://mirbase.org).
   Post-alignment processing of miRBase hairpin was performed using
   SAMtools and EdgeR (3.36.0) for miRBase analysis. The FASTQ files and
   expression matrix are available in Gene Expression Omnibus (GEO)
   ([199]GSE262414). This provided a count table of miRNA reads ready for
   downstream analysis in RStudio (Ver 4.1.1). As previously
   performed^[200]37, exploratory principal component analysis was first
   performed using DESeq2 (Version 1.34.0). After count normalisation and
   removal of lowly expressed reads, differential expression analysis was
   performed using EdgeR (Version 3.36.0). A fold change > 1.5 and false
   discovery rate (FDR, Benjamini-Hochberg method) < 0.05 was considered
   significant.

RNA-Sequencing of bulk cortical tissue and fluorescence-activated cell
sorting of specific kidney cell types

   As previously described^[201]37,[202]38, male 8–12-week-old C57BL/6
   mice (Envigo) (n = 5/group) underwent laparotomy and unilateral
   ureteric obstruction (UUO) followed by ureteric reimplantation to
   reverse the obstruction for the rUUO group. Mice were culled at day 2
   after UUO (UUO-2), at day 7 after UUO (UUO-7), and in the rUUO group at
   14 days after reversal by exsanguination under terminal isoflurane
   anaesthesia. Immediately after culling, mice were perfused with 5 mL
   PBS. The kidneys were harvested, with the kidney capsule removed and
   placed in ice-cold PBS and processed as previously
   described^[203]37,[204]38. RNA integrity was checked using Agilent
   Technologies picochips. All samples utilised for sRNA-Seq had a minimum
   RNA integrity score of 7. For the bulk and single-population sRNA-Seq,
   RNA underwent sRNA-Seq by Genewiz, using the Illumina HiSeq platform,
   generating paired-end reads of 50 bp (n = 4 per group)(sRNA-Seq dataset
   - [205]GSE150035). For gene expression analysis of sorted cell
   populations, RNA integrity was ≥ 9 for all samples, amplified cDNA was
   prepared using Ovation® RNA-Seq System V2 (NuGEN) and sequenced by
   Genewiz on the Illumina HiSeq platform with 2x150bp configuration
   (n = 4 per group). Data from the gene expression from sorted cell
   populations is available at GEO ([206]GSE262799).

   Gene targets of miR-190a-5p were identified using the R package
   multiMiR (1.6.0). multiMiR allows the user to identify validated miRNAs
   from multiple databases. For predicted miR-gene interactions, we
   selected only interactions which were conserved and present in more
   than one database. We performed pairwise correlation of proximal
   tubular cell expression changes ([207]GSE262799) of predicted
   miR-190a-5p gene targets. Genes negatively correlated to miR-190a-5p
   (correlation coefficient of < − 0.6) were considered potential gene
   targets. Finally, the pathways enriched for miR-190a-5p target genes
   were explored using the package ClusterProfiler (3.12.0).

Analysis of miR-190a-5p in seNSOR samples

   Whole blood was collected into a BD Vacutainer® SST™ Tubes, centrifuged
   at 3000 × g for 10 min at 4 °C, and the supernatant was aliquoted in
   RNase/DNase-free tubes and frozen at -80 °C. Synthetic C.elegans miR-39
   (Qiagen) was added as an exogenous reference miR prior to RNA
   extraction from 200 μl of serum using the miRNeasy serum/plasma kit
   (Qiagen). Reverse transcription was performed using the microRNA
   reverse transcription kit (Thermo Fisher Scientific) with specific
   primers for hsa-miR-190a-5p and cel-miR-39. Real-time quantitative PCR
   (RT-qPCR) was performed using specific primer:probes (Thermo Fisher
   Scientific) (Supplementary Table [208]1) with Universal Master Mix II
   (Thermo Fisher Scientific) on the QuantStudio 5 (Thermo Fisher
   Scientific). Cycle threshold was obtained for miR-190a-5p and
   normalised to cel-miR-39, and relative miRNA expression calculated
   using the 2^-ΔCT method and relative standard error of the mean
   plotted.

   For biopsy samples, a subset of seNSOR patient biopsies samples, frozen
   in optimal cutting temperature (OCT) compound, were obtained from the
   NHS Lothian Biobank (REC 20/ES/0061, studies SR2175) (n = 26). RNA was
   extracted and reverse transcription performed using the microRNA
   reverse transcription kit (Thermo Fisher Scientific) with specific
   primers for hsa-miR-190a-5p and control snoRNA RNU48 (Supplementary
   Table [209]1). RT-qPCR was performed using specific primer:probes
   (Thermo Fisher Scientific)(Supplementary Table [210]1) with Universal
   Master Mix II (Thermo Fisher Scientific) on the QuantStudio 5 (Thermo
   Fisher Scientific). Cycle threshold was obtained for miR-190a-5p and
   normalised to RNU48, and relative miRNA expression calculated using the
   2^-ΔCT method and relative standard error of the mean plotted.

Histological analysis of seNSOR biopsies

   Immunofluorescence staining on 24 human kidney biopsy samples was
   performed using a BOND III automated immunostainer (Leica Biosystems).
   Pancytokeratin (PanCK, Merck, Cat C2562) at 1:6000 dilution, and CD10
   (Leica, Cat: NCL-L-CD10-270) at 1:600 were added together, and the same
   Opal 520 fluorophore used for both, so together they represent
   pantubular epithelial markers. Images of whole slides were acquired
   using the Axio Scan Z1 whole slide scanner (Zeiss, Jena, Germany).
   Images were then analysed using QUPath (version 0.5.1) and areas of
   cortex manually annotated. Areas of positive staining for PanCK / CD10
   were expressed as a percentage of the cortical tissue to calculate
   tubular mass percentage.

   Cortical slides stained for Masson’s Trichrome were available for 25
   biopsy samples from which miR-190a-5p expression had been determined. A
   single blinded operator performed semi-quantitative cortical
   interstitial fibrosis assessment on these slides.

Cell culture of kidney proximal tubular cells (RPTECs)

   Human kidney proximal tubular epithelial cells (RPTEC/TERT1) (ATCC®
   CRL-4031^TM LGC Limited, UK) were maintained in complete growth media
   consisting of DMEM/F-12 (2.5 mM L-glutamine, 15 mM HEPES, 0.5 mM sodium
   pyruvate, and 1200 mg L^−1 sodium bicarbonate) (Gibco, New York, USA),
   geneticin 2 µL mL^−1 (Gibco, New York, USA), hydrocortisone 25 ng mL^−1
   (Sigma Aldrich, Gillingham, UK), ascorbic acid 3 µg mL^−1
   (Sigma-Aldrich, Gillingham, UK), sodium selenite 6.7 ng mL^−1
   transferrin 5.5 µg mL^−1, insulin 10 µg mL^−1 (Gibco, New York, USA),
   triiodo-L-thyronine 6 pM (Sigma-Aldrich, UK), prostaglandin E1
   25 ng mL^−1 (Sigma-Aldrich, Gillingham, UK), rhEGF 10 ng mL^−1
   (Promega, Southampton, UK). Cells were grown at 37 °C, 5% CO[2], in
   T75 cm^2 flasks to 100% confluence. For passaging, cells were detached
   using trypsin, spun and resuspended in fresh media.

Manipulation of miR-190a-5p levels in RPTECs

   Dharmacon miR-190a-5p miRNA mimic (C-300639-03-0002) and inhibitor
   (IH-300639-05-0002), along with negative control mimic
   (CN-001000-01-05) and miRNA hairpin inhibitor negative control
   (IN-001005-01-05) were purchased from Horizon Discovery. RPTECs were
   transfected with 100 nM of miR-190a-5p inhibitor or 5–50 nM miR-190a-5p
   mimic or controls using siPORT™ NeoFX™ Transfection agent in Opti-MEM
   for 6 h before adding complete media. Media was replaced after 24 h and
   cells left for a further 48 h prior to being collected in Qiazol.

   RNA was extracted from cells using the miRNeasy mini kit (Qiagen)
   following the manufacturer’s instructions. cDNA was prepared using the
   reverse transcription kit with random hexamers (Thermo Fisher
   Scientific). RT-qPCR was performed using specific primer:probes for a
   disintegrin and metalloproteinase domain-containing protein 10 (ADAM10)
   and peptidylprolyl isomerase A (PPIA) was used as the reference gene
   (Supplementary Table [211]1). Reverse transcription was performed using
   the microRNA reverse transcription kit (Thermo Fisher Scientific) using
   specific primers for miR-190a-5p and snoRNA RNU48 (Supplementary
   Table [212]1) as an endogenous reference small RNA, and RT-qPCR
   performed using specific primer:probes (Thermo Fisher Scientific). All
   RT-qPCR was performed with Universal Master Mix II (Thermo Fisher
   Scientific) on the QuantStudio 5 (Thermo Fisher Scientific). Relative
   expression was calculated as a change in cycle threshold (ΔCT) for each
   sample to the housekeeper and plotted as 2^-ΔCT, and the relative
   standard error of the mean plotted.

miR-190a-5p expression and manipulation in vivo

   All procedures were approved by the Animal Welfare and Ethical Review
   Body (AWERB) at the University of Edinburgh and were conducted in
   accordance with the United Kingdom Animals Scientific Procedures Act
   1986 and the ARRIVE guidelines under Project Licences approved by the
   UK Home Office. Our human data contained male and female biological
   sex, and the effects observed were independent of biological sex.
   Therefore, for our pre-clinical models, we used only male animals as
   these have been well characterised in-house and previous publications
   with robust data were available to perform power calculations to
   determine group size.

   C57BL/6 J mice and SV129 mice were purchased from Enviago and housed in
   a pathogen-free environment at the University of Edinburgh. Mice were
   housed at 50% humidity and 22–26 °C with 12: 12 light: dark cycles
   starting at 07:00 local time. Mice were group-housed and given ad
   libitum access to water and standard chow (rat and mouse maintenance
   diet 1 (Special Diet Services)).

   Subtotal nephrectomy was performed as previously described on SV129
   mice^[213]36. Briefly, mice were randomly allocated to either receiving
   a sham (n = 5) or subtotal nephrectomy surgery (n = 8). Animals were
   recovered and left for 12 weeks to allow progressive kidney injury to
   occur. At 12 weeks animals were culled and tissue taken.

   Unilateral IRI (uIRI) surgery was performed as previously
   described^[214]35,[215]40. Briefly, a flank incision was made, and for
   sham surgery (n = 4), the left renal pedicle identified and touched
   with a moist cotton bud, for uIRI animals (n = 6), it was clipped using
   an atraumatic clamp for 18 min. During the ischemic period, body
   temperature was maintained at 37 °C using a heating blanket with
   homoeostatic control (Harvard Apparatus), measured by a rectal
   temperature probe. The clamp was then removed, the peritoneum closed
   with 5/0 suture, and the skin incision closed with clips. At 14 days
   post-surgery, a flank incision was made on the right side and a
   contralateral nephrectomy performed to produce a function model of IRI
   injury, and animals culled 7 days post-nephrectomy.

   For the intervention study, mice (n = 7/gp) were assigned to either an
   experimental group receiving subcutaneous miR-190a-5p mimic (purchased
   from Horizon Discovery) injected at 0.625 μg per kg in PBS or scrambled
   control mimic (Horizon Discovery) using randomiser software and end
   analyses were conducted blind. Animals were dosed Day -1, 1, 3, 6, 8,
   10, 13 and culled on day 14, without a contralateral nephrectomy.

   For all in vivo studies, kidneys were divided and one portion snap
   frozen for RNA extraction, and the other portion stored in 10% formalin
   for histological analysis.

Histological analysis of mouse kidney tissue

   Formalin fixed and paraffin embedded (FFPE) mouse kidney tissue was
   sectioned at 3 μM. Sections were deparaffinized twice for 5 min in
   xylene and rehydrated in sequential 100%, 90% and 70% ethanol for 2 min
   each and placed in distilled water. Picrosirius red (PSR) staining was
   performed using the Picrosirius Red Stain Kit (Abcam, ab150681)
   following manufacturer guidelines. After rehydration, tissue sections
   were incubated in PSR for 60 min. Slides were rinsed in acetic acid
   solution and dehydrated in absolute ethanol before mounting.
   Non-overlapping images of sections were obtained for each animal and
   analysed blinded by a single operator with Fiji ImageJ to determine the
   percentage of fibrosis.

Gene Expression analysis in mouse tissue

   Total RNA was extracted from snap-frozen kidney tissue using the
   miRNeasy Mini kit (Qiagen), following the manufacturer’s instructions.
   cDNA for quantitative PCR was synthesised from the extracted RNA using
   the reverse transcription kit with random hexamers (Thermo Fisher
   Scientific). RT-qPCR was performed using TaqMan Universal Master Mix II
   (Applied Biosystems) and TaqMan Gene Expression Assay-specific primers
   (Supplementary Table [216]1) and normalised to Ppia expression.

Single-cell analyses

   For correlation analysis of ADAM10 and TLN2 single nucleus multiome
   (paired gene expression and chromatin accessibility) data was
   downloaded from GEO in Seurat (v4.4.0) format ([217]GSE254185)^[218]35.
   Due to sparsity and technical dropouts in single-cell transcriptomics
   data, we opted to assess the average expression levels in partitions of
   the nearest neighbour graph instead of individual cells for
   gene-to-gene correlations.

   For this purpose, using the original cell annotations, the data was
   subset to proximal tubule (PT) cells and subclusters of the weighted
   shared nearest neighbour (wsnn) graph were constructed using the
   FindClusters function with resolution n = 8 (median: 131
   cells/cluster). Clusters were subsequently pruned by discarding
   clusters with less than 20 cells. The average log-scale expression
   level of ADAM10 and TLN2 was derived using the AverageExpression
   function, and cluster-level average expression was used for
   visualisation as a scatterplot.

Statistical analyses

   The normality of distribution for all key variables was assessed by the
   Shapiro-Wilk test. Clinical characteristics for continuous data are
   expressed as mean ± standard deviation for normally distributed data
   and median (interquartile range) when not normally distributed.
   Categorical variables were expressed as counts (percentages). For
   continuous values with normal distribution, t-tests or ANOVA were used
   to compare 2 groups or > 2 groups, respectively. For continuous values
   without normal distribution, the Mann-Whitney test was used to compare
   2 groups and a Kruskal-Wallis test was employed if > 2 groups.
   Categorical values were assessed using a Chi-square test. For the
   determination of linear correlation between variables, correlation
   coefficients (r) were estimated using Spearman’s rank tests (if the
   data was not normally distributed) or Pearson’s (if the data was
   normally distributed). Kaplan-Meier survival curves were constructed
   for the primary kidney endpoints comparing above and below median
   miR-190a-5p separately for each uACR stage ( < 3 mg/mmol,
   3–300 mg/mmol, > 300 mg/mmol), with the log-rank test used to compare
   curves. Cox univariate and multivariate proportional hazards survival
   models were performed to assess factors that predicted the endpoints,
   for multivariate analysis we excluded those patients with no ACR
   available. Circulating miR-190a-5p expression was log-transformed
   before entering the model due to non-normal distribution. For in vitro
   miR-190 mimic and inhibitor experiments, biological repeats were used
   (n = 3), with 3 technical replicates per sample and results analysed by
   Student’s unpaired t test. All tests were performed using SPSS version
   28.0 (IBM, Armonk, NY), R version 4.1.2 (R Foundation for Statistical
   Computing, Vienna, Austria) or Graphpad Prism version 10.1.2 (Graphpad
   software, Boston, Massachusetts). A p-value < 0.05 was considered
   significant.

Supplementary information

   [219]Supplementary Information^ (1.8MB, pdf)
   [220]41467_2025_64168_MOESM2_ESM.pdf^ (87.2KB, pdf)

   Description Of Additional Supplementary File
   [221]Supplementary Data 1^ (342.3KB, xlsx)
   [222]Transparent Peer Review file^ (1.1MB, pdf)

Source data

   [223]Source data^ (407.1KB, xlsx)

Acknowledgements