Transforming growth factor-β in kidney disease in dogs and cats: a review of the evidence of its role in renal fibrosis Authors: Rachel E Cianciolo, Jonathan Elliott, Mary B Nabity, Jonathan D Foster, Dylan T Ammons, Katrina Bartas, Tracey Williams, Elizabeth A VanBeelen, Lindsay Weaver Published: 2026 ABSTRACT AbstractA growing body of literature implicates the transforming growth factor (TGF)-β pathway in the progression of chronic kidney disease (CKD) in dogs and cats by promoting renal fibrosis. In rodents and humans, TGF-β is a key player in development of glomerulosclerosis and renal interstitial fibrosis. The pathway is also immunomodulatory, underscoring the complexity of its downstream effects. Recently, evidence for the role of TGF-β in small animal CKD has been presented using multiple methodologies, including serum and urinary biomarkers, as well as gene and protein expression in renal tissue and primary renal cell cultures. The aim of this review is to integrate these data and highlight consistencies and discrepancies in the literature. Some of the most convincing data supporting the role of the TGF-β pathway includes biomarker levels and pathway analysis of transcriptional profiles in spontaneous CKD in cats and in the X-linked Alport model in dogs. Demonstration of the presence of TGF-β protein is challenging due to lack of commercially available immunohistochemistry (IHC) antibodies that are sensitive and specific for active variants and isoforms TGF-β in dogs and cats. Cross-reactivity of IHC antibodies with latent TGF-β isoforms (complexed with extracellular matrix proteins) complicates the interpretation of IHC labeling. The mechanisms by which TGF-β activates fibroblasts and injures glomeruli in small animals is an area for future research. Although the relationship between TGF-β and glomerulosclerosis is mostly theoretical based on extrapolation from humans and rodents, substantial evidence strongly supports a role for TGF-β in renal interstitial scarring of small animals. Introduction Chronic kidney disease (CKD) is a common cause of morbidity in dogs and cats. The reported prevalences in small animals vary widely and often depend on geography, age, and breed. Generally, it is estimated that approximately 1% of adult dogs1 and cats2 have CKD, whereas the prevalence in cats can reach 30% in individuals over 10 years of age.3 These reported prevalences are underestimates, because early-stage disease (before onset of clinical signs) often goes undetected. There are a variety of causes in both species, and progression partially depends on the initial insult. It is widely recognized that there is variation in disease progression, with some animals remaining stable and others showing more rapid deterioration of renal function.4,5 Factors driving progressive kidney injury in dogs and cats have are not well defined. Possible factors include the ability (or inability) to cure the primary renal disease and superimposed kidney insults. In cats, superimposed insults include renal ischemia, outflow obstruction and bacterial infection.6 In dogs, anemia, low body condition score and hyperphosphatemia are identified as factors that increase risk for progressive CKD4. Other factors driving progression (which have mostly been elucidated in experimental or natural kidney diseases in humans and rodents) result from adaptive responses to nephron loss (hyperfiltration), leading to so-called “intrinsic progression”. Hyperfiltration results in podocyte hypertrophy to cover the increased surface area of the expanded glomerular basement membrane. This causes podocyte stretching and shear stress.7 There is also higher tubular workload, (specifically higher tubular reabsorption of sodium from the urinary filtrate) and activation of inflammatory and hypoxic injury pathways.7 Proteinuria, hyperphosphatemia and anemia are linked to progression of CKD in dogs and cats.4,8–10 Interstitial fibrosis and inflammation are associated with biomarkers of progression in dogs (correlated to azotemia)11 and cats (correlated to azotemia, proteinuria and hyperphosphatemia).10–13 Phosphaturia14,15 and proteinuria16 can trigger epithelial cell stress, leading to fibrosis and inflammation of the interstitial compartment. As stated above, glomerular hyperfiltration injures podocytes, which is the precursor lesion for glomerulosclerosis.17 Both interstitial fibrosis and glomerulosclerosis are irreversible lesions which can vary in severity and distribution. Interstitial fibrosis separates the tubules from their peritubular capillary blood supply in cats,18 rodents and humans,19 exacerbating the lower oxygen supply associated with anemia. This leads to tubular degeneration and atrophy. Glomerulosclerosis eventually effaces glomerular capillary lumens, impeding blood flow to the peritubular capillary bed.20 Both interstitial fibrosis and glomerulosclerosis are collagen-based lesions, and both can be initiated and perpetuated by the transforming growth factor (TGF)-β pathway. This molecular pathway is complex and multi-directional. It is a key component of mesenchymal tissue development during embryogenesis, post-natal growth, and wound healing. The TGF-β pathway is not the only mechanism that leads to collagen production in CKD; WNT, MAPK and RhoGTPase are also involved.20 Nevertheless, TGF-β-driven renal fibrosis is considered to be a major factor in the development and progression of CKD in animals. In addition to being pro-fibrotic, TGF-β also has anti-inflammatory effects, which further complicates its role in progressive CKD. This review will describe the current knowledge of the role of TGF-β in CKD, focusing on cats and dogs, with translation, where appropriate, to humans and rodent models. There are approximately 30 TGF-β pathway genes and proteins across species. The members generally share high homology, for example, the TGF-β1 amino acid sequence shares 89% or higher percent identity (calculated as the number of identical matches divided by the total number of aligned positions) between human, canine, feline, and murine proteins (Figure 1). There are 3 known isoforms of TGF-β (TGF-β1, TGF-β2 and TGF-β3) in mammals, which are encoded by 3 separate genes, but which have similar amino acid sequences.21 TGF-β4 is an isoform that has only been observed in avian species, whereas TGF-β5 is only present in amphibians.22 Cell or organ specificity has not been demonstrated for the TGF-β isoforms in various protein/transcriptional atlases. TGF-βs are synthesized by a wide variety of cell types as pro-proteins but are soon cleaved into the mature protein and a latency associated peptide (LAP). The mature portion is subsequently cleaved in half between 2 identical regions that strongly bind to one another, forming a homodimer. The homodimer then binds non-covalently to its respective LAP, forming a small latent complex (SLC) that is then secreted into the extracellular space. The SLCs then further associate with latent TGF-β binding proteins (LTBPs), forming large latent complexes (LLCs) which are enmeshed in the extracellular matrix (ECM). LLCs represent a storage supply, so that TGF-β can be released when needed. Release of latent TGF-beta is a major control point for pathway activation. Multiple proteins can release the TGF-β in its active form, including integrins, thrombospondins, reactive oxygen species (ROS) and matrix metalloproteinases (MMPs). The structures of TGF-β1 and TGF-β3 are similar, and both can be released by integrins that are made by a variety of cells. However, TGF-β2 lacks the requisite arginine-glycine-aspartic acid (RGD) motif, so it is unable to be released by integrins.23 Thrombospondins have the capability to release all 3 isoforms, and they have particular importance in renal diseases.21 TGF-β1 amino acid sequence is highly conserved between human, mice, cats and dogs. (A) Clustal omega alignment of human, mouse, feline, and canine TGF-β1 amino acid sequences. Mismatches relative to the human sequence are colored red. (B) Heatmap depicting pairwise percent identity values between species. Percent identity was calculated as the number of identical amino acid matches between two species divided by the total number of aligned positions, multiplied by 100. Once released, a TGF-β1 or TGF-β3 homodimer binds to a monomer or a dimer of transmembrane TGF-β receptor II (TGFBRII). If the cytokine binds to a monomer of TGFBRII, then it will attract another one to create a 4-protein complex. Then, a dimer of TGF-β receptor I (TGFBRI) is recruited, ultimately forming a complex of 6 proteins. The affinity between the cytokine and TGFBRII is high, and TGFBRI is only attracted to complexes of TGF-β and TGFBRII. This stepwise binding process underscores the importance of TGFBRII in the capture of TGF-β1 or 3, and the necessary binding of TGFBRI to initiate signal transduction. In contrast, TGF-β2 does not follow the same sequence of receptor binding because TGF-β2 has a low affinity for TGFBRII on its own. In vitro studies show that TGF-β2 binds TGFBRIII, which acts as a co-receptor to enhance the binding affinity of TGF-β2 to TGFBRII. Once TGF-β2 is bound to TGFBRIII, TGFBRI is recruited, and signal transduction proceeds in the same manner as the other TGF-β isoforms.24 TGF-β receptors are promiscuous and can be activated by a wide variety of molecules in the TGF-β family, including bone morphogenic proteins (BMPs), activins and inhibins.25 BMPs can have opposite results, in comparison to the TGF-β pathway (see below). There are canonical and non-canonical pathways of TGF-β signaling. The canonical pathway involves phosphorylation of the intracellular portions of the TGFBRI and TGFBRII dimers to recruit SMAD2 and SMAD3 proteins, which will also be phosphorylated. The phosphorylated SMADs then bind to SMAD4 and other co-factors as they are transported from the cell membrane to the nucleus, where they regulate gene transcription.26 There are multiple SMADs, with SMADs 2 and 3 being part of the TGF-β pathway, whereas SMADs 1, 5 and 8 are employed by the BMP pathway. Both pathways use SMAD4 to translocate to the nucleus. SMADs bind to SMAD-binding elements, which vary according to cell and tissue type. Some of the SMAD-binding elements involve epigenetic changes, and many pathways require transcriptional cofactors. Inhibitory SMADs (typically SMAD7, and to a lesser extent SMAD6) can bind to TGFBRI and block its interaction with SMADs 2 and 3, thereby creating a negative feedback loop for the canonical pathway. Competition with the BMP pathway for SMAD4 can also limit TGF-β pathway activation. There are many other cytoplasmic proteins that can phosphorylate SMADs, as well as phosphatases and proteases to inactivate the SMADs, underscoring the complexity of the canonical pathway. Non-canonical TGF-β pathways are SMAD-independent and can activate a myriad of cell functions including morphogenesis, fibrosis, immunity and metabolism.21 The canonical and non-canonical pathways both play important roles in fibrosis. Overall, the outcome of the TGF-β pathway activation is context-dependent, influenced by which cytokine activates the receptors, how the receptors interact with one another, whether the canonical or non-canonical pathway is activated, and even which individual SMADs are recruited to the receptors (Figure 2). Diagram of the TGF-β pathway which differs between TGF-β2 and TGF-β1 and 3. All 3 ultimately activate SMAD2 and SMAD3 to bind to SMAD4 which enables the complex travel to the nucleus for gene transcription. The BMP pathway can counteract the TGF-β pathway by also binding to SMAD4, limiting its availability. BMP pathways often counteract the pro-fibrotic TGF-β pathways (figure made with BioRender). In the setting of renal injury, expression of all 3 isoforms of TGF-β is increased, but TGF-β1 is the best characterized. It can stimulate the production of ECM components by interstitial fibroblasts, tubular epithelial cells, mesangial cells, podocytes, and parietal epithelial cells.20 It also leads to downregulation of MMPs, slowing the breakdown of the newly synthesized collagen. In transgenic mice that overexpress TGF-β1 in hepatocytes, greater levels are found in the plasma. This initiates and perpetuates glomerulosclerosis by inducing production of collagens and inhibiting their breakdown via upregulation of tissue inhibitors of matrix metalloproteinases (TIMPs).27 Although that study proved that circulating TGF-β can cause renal fibrosis, local TGF-β production can also cause renal scarring. In natural disease, the cytokine is detectable in serum, but the relative influence of extra-renal vs intra-renal TGF-β in natural CKD is unknown. Renal injury is associated with upregulation of the TGF-β pathways, often in a paracrine fashion.28 Injured tubules can secrete new TGF-β or they can facilitate its release from the ECM. A key feature of fibrosis in CKD is the appearance of myofibroblasts, which have the capacity to synthesize different types of ECM components compared to a typical fibroblast.28 TGF-β is a main driver behind the appearance of this population in kidneys.28 Recent evidence has demonstrated that myofibroblasts are derived from resident interstitial fibroblasts and pericytes. Some lineages of renal myofibroblasts can further increase TGF-β signaling to transform adjacent fibroblasts and pericytes into more myofibroblasts.29 Tubular epithelial cell to mesenchymal cell transformation (EMT) was once considered to be an important mechanism for myofibroblast differentiation because gene pathways activated by TGF-β overlap with those of EMT in organ development and cancer; however, evidence for true EMT in renal scarring is lacking.29,30 Scarring of the glomerulus (glomerulosclerosis) is another downstream effect of TGF-β in the kidney. Essentially, TGF-β pathways activate the mesangial cells and podocytes to synthesize increased amounts of their normal types of collagens.27 However, these cells synthesize mostly non-fibrillar collagens, as opposed to fibrillar collagen. The effect is essentially the same: irreversible increase in ECM that can compress or obliterate capillary lumens and impede glomerular perfusion. Experimental data indicate that podocytes can undergo partial transition to a mesenchymal cell phenotype under the influence of TGF-β.31 These mesenchymal cells synthesize ECM proteins (eg, Type I collagen, fibronectin, and α-smooth muscle-actin), resulting in glomerulosclerosis.31 Podocytes undergo apoptosis in response to TGF-β1 in transgenic mice.32 In contrast, BMP protects podocytes in mouse models.33 Other studies using recombinant BMP7 in experimental rodent models of kidney disease have demonstrated a role in attenuating glomerulosclerosis, renal fibrosis, and tubular atrophy.34 As glomerulosclerosis develops, it will further lower peritubular capillary perfusion, contributing to renal tubular atrophy and interstitial fibrosis. Peritubular capillary rarefaction is another irreversible effect of the TGF-β pathway.35 Increased matrix is visible in the interstitium and is easier to detect with special stains like Masson’s trichrome that highlight collagen. Glomerulosclerosis is characterized by abundant collagen that partially or globally effaces the capillary tuft (Figure 3). (A) Biologic effects of the TGF-β pathways in the kidney, based on data from rodent models and natural disease in humans (figure made with BioRender). (B-D) renal biopsy core of a dog with IRIS stage 1 CKD (proteinuria). Mild interstitial fibrosis can be difficult to detect with hematoxylin and eosin staining (B); however, it is easier to identify with the Masson’s trichrome stain (which stains collagen blue; (C) the trichrome stain also highlights excess collagen in the glomeruli, (D) with hyalinosis staining pale peach. TGF-β1 is not only pro-fibrotic, but it is also immunomodulatory, with the ability to suppress cytotoxic T cells, block natural killer cell effector functions, and induce T-regulatory cells. This illustrates the diverse roles that TGF-β1 can play, with the ability to quell inflammatory responses in the context of nephritis.36,37 Taken together, the renal TGF-β pathway involves multiple different cell lineages (fibroblast, endothelium, renal tubular epithelium, mesangial cells, podocytes, platelets, and leukocytes38; Figure 3). Studies in humans and rodents show both positive and negative effects of TGF-β in kidney disease, and it likely reflects degree and timing of pathway of activation. Overactivation or chronic upregulation of the TGF-β pathway are scenarios that might promote CKD progression. The TGF-β pathway is also implicated in acute kidney injury (AKI). Its signaling in the proximal tubular epithelium is rapidly upregulated and exerts predominantly injurious effects. Constitutive activation of TGFBRI in tubular epithelial cells alone is sufficient to cause AKI, characterized by tubular cell apoptosis and necrosis, oxidative stress (via mitochondrial-derived ROS), epithelial dedifferentiation and interstitial accumulation of macrophages, dendritic cells and T cells.39 Selective deletion of TGFBRII in proximal tubules attenuates renal injury and reduces tubular apoptosis in toxin-induced AKI models, confirming a direct detrimental role of TGF-β signaling in acute injury response.40 Conditional knockout of SMAD2 in tubular epithelial cells also has renoprotective effects by alleviating p53-mediated apoptosis, RIPK-mediated necroptosis, and NF-KB-driven inflammation.41 TGF-β pathway activation in CKD is widely reported in the literature, with higher transcript abundance of TGF-β isoform(s) being a key feature. Although gene expression can distinguish among the TGF-β isoforms, immunohistochemistry (IHC) is generally less specific and often labels more than 1 isoform depending on the epitope recognized by the antibody. Moreover, TGF-β can remain latent in the renal interstitial ECM, so its presence in the kidney does not equate to active signaling. The interaction with the LAP and other ECM components has the potential to mask epitopes, decreasing IHC sensitivity. TGF-β is ephemeral such that identification of the active form could be missed in fibrotic kidney tissue, depending on the timepoint of evaluation. Measuring TGF-β pathway activation through evaluation of downstream pro-fibrotic and immunomodulatory gene expression (through bulk transcriptomics or quantitative reverse transcriptase polymerase chain reaction, qRT-PCR) provides the most comprehensive assessment of pathway activation. As such, pathway analysis is an optimal method to investigate which downstream effects of TGF-β including interstitial fibrosis, glomerulosclerosis, inflammation are expected in various categories of CKD. Using spatially aware approaches, such as in situ hybridization, TGF-β pathway transcripts can be localized to specific regions (eg, glomerulus vs tubulointerstitium), enabling further insights to pathogenesis. Spatial and single cell transcriptomics are powerful in illustrating the role of TGF-β in human and rodent CKD.37,42 These methods demonstrate TGF-β transcript expression and downstream pathway activation in various renal disease states. With spatial and single cell transcriptomic analyses being inherently sparse datasets, investigation of downstream gene signatures can enhance sensitivity. Although caution should be used due to the non-specificity of some gene sets. Single cell transcription analysis of feline43 and canine44 renal tissue has been performed, but it has been limited to <3 individuals in each study, so more research is needed. TGF-β in cats The clinical presentation of CKD in cats is heterogeneous, in terms of stage at which clinical signs are noted and the rate of progression. Clinical guidelines for staging CKD have been provided by the International Renal Interest Society (IRIS).45 The histopathological findings of tubulointerstitial inflammation and fibrosis are consistent and correlate best with biochemical markers of kidney function.12,13 Specifically, interstitial fibrosis correlates strongly with severity of azotemia, hyperphosphatemia and anemia,13 suggesting that it is associated with CKD progression. Furthermore, interstitial fibrosis in cats with CKD is strongly associated with the activity of a collagen cross-linking enzyme, transglutaminase.46 In vitro studies show that transglutaminase is involved in binding TGF-β to LTBPs and ECM, a step required for both sequestration and eventual activation.47,48 The evidence that TGF-β is involved in naturally occurring kidney disease in cats is accumulating. TGF-β1 gene and protein expression have been assessed in the kidneys of cats with naturally occurring disease and experimentally induced kidney injury. As discussed previously, measurement and localization of TGF-β1 protein within diseased kidney tissue is difficult because of the existence of multiple latent and active isoforms of this cytokine. Comparison of results between studies needs to carefully assess the specificity of the IHC antibody used and the validation process applied. Measurement of TGF-β1 gene expression is less problematic. However, ideally one would measure both transcript(s) and protein(s), as well as downstream mediators in the TGF-β1 pathway. RNA sequencing (RNAseq) characterized gene expression changes in renal tissue 6 months after experimentally induced ischemia and reperfusion injury.49 Previous publications from this group had validated the model and demonstrated remarkable pathological similarities to naturally occurring CKD.50 The expression changes were established through comparisons to the contralateral (non-ischemic) kidney of the same cat and to healthy control feline kidneys; of note, healthy control cats were younger than the cats undergoing renal ischemia and had not been subjected to sham operation. A gene ontology analysis approach was taken, and in comparison to control cat kidneys, the previously ischemic kidneys showed upregulation of pro-inflammatory and profibrotic pathways. Genes of interest identified within the molecular function category of “positive regulation of collagen biosynthetic process” (which was upregulated 2.94-fold), included TGFB1, TGFB3, and VIM. These are the genes for TGF-β1, TGF-β3, and vimentin proteins, respectively. Fewer pathways were differentially regulated when comparisons were made between the ischemic and non-ischemic contralateral control kidney, suggesting proinflammatory and pro-fibrotic pathways were activated in both kidneys, even 6 months after injury.49 The reason for this was not clear, but similar bilateral patterns of renal injury have been reported in other experimental models of ischemia-reperfusion injury.51 A companion paper that used the same samples took a more targeted approach to examine individual gene expression by qRT-PCR, including TGFB1.52 The authors showed that TGFB1 expression was significantly increased in the ischemic kidney compared to normalized levels found in control cat kidneys. No significant difference was found between the ischemic kidney and its contralateral non-ischemic control, or between the control kidneys and the contralateral non-ischemic controls. Other targeted genes that were upregulated in response to the ischemia–reperfusion injury included MMPs. Specifically, MMP7 and MMP9 showed the same pattern of differential expression as TGFB1. Correlations between gene expression and histological severity of fibrosis were also assessed. No significant association was found between TGFB1 expression and severity of fibrosis; however, expression of HIF1A (which encodes hypoxia inducible factor-1α), MMP2, MMP7 and TIMP1 were all significantly and relatively strongly correlated with the fibrosis score. Both this study and its companion study49 included a small number of animals (6 cats with the ischemic protocol and 8 or 9 healthy controls), which is one of the limitations. A follow-up study used the same ischemia protocol, but the contralateral kidney was removed 3 months after ischemia, and the ischemic kidney was harvested at 6 months. Micro(mi)-RNA expression was evaluated in the kidneys (ischemic compared to non-ischemic) and serum (baseline compared to terminal samples). The small study (4 cats) demonstrated differential expression of 20 miRNAs between ischemic and contralateral kidneys, 4 of which (mir-21, mir-146, mir-199, and mir-235) also exhibited increased levels in terminal serum samples compared to the baseline. These miRNAs are predicted to regulate many genes associated with CKD in cats, including TGFB1, HIF1A, HGF (which encodes hepatocyte growth factor), FN1 (which encodes fibronectin), and VEGFA (which encodes vascular endothelial growth factor A).53 The role of mir-21 in regulating the TGF-β pathway has been studied in other species, including humans.54 An additional study evaluated cats with naturally occurring CKD, targeting the same genes as detailed above.55 This study included 12 client-owned cats which varied greatly in terms of age, IRIS stage of CKD (7 were IRIS Stage 4 CKD) and severity of proteinuria. The comparator group consisted of 8 healthy laboratory cats which were young (but mature), often not neutered and fed a standard diet. They differed in many respects to the CKD cats; as such, they were not an ideal control group. Even so, this study demonstrated that kidney tissue from cats with naturally occurring CKD showed higher transcription of the TGFB1 gene when compared to the control group. The magnitude of this difference was highly variable (2- to 10-fold difference) between individual cats with CKD, and there was variability between left and right kidneys from the same cat. Similar findings were shown for other profibrotic genes (MMPs 2,7 & 9, TIMP1 and HIF1A), whereas VEGFA expression was significantly lower in CKD kidneys compared to controls. GAPDH was used as the house-keeping gene. Like the ischemia-reperfusion study, expression of TGFB1 did not correlate with histological scores for inflammation, fibrosis, or tubular atrophy; however, expression of MMP2, MMP7, and TIMP1 genes was positively correlated, and VEGFA expression was negatively correlated with fibrosis. By contrast, a study by different investigators reported that postmortem kidney samples collected after euthanasia from 6 young cats (mean age 2.54-years) with naturally occurring CKD did not show a statistically significant upregulation in TGFB1 gene expression compared to kidneys from cats with no lesions (n = 6, mean age 4 months).56 This study used β-actin as the housekeeping gene. The difference in house-keeping genes, young age of the CKD cats compared with most other studies, and the young age of the control cats (mean age 4-months) might account for the disparate results from this study.56 Preliminary data from a study involving post-mortem kidney samples from 51 client-owned cats, 14 of which were geriatric controls (with normal kidney function) and 37 of which had CKD (IRIS stages 2-4) have been presented as an abstract. This study showed weak correlations between TGFB1 expression (via qRT-PCR) and kidney fibrosis score and also between mineralocorticoid receptor expression and TGFB1 expression.57 Together, these 6 studies suggest that upregulation of TGFB1 gene expression occurs in the cat in association with induced kidney injury and naturally occurring CKD. Three publications have used IHC to localize and semi-quantify the expression of TGF-β1 protein in naturally occurring feline CKD. A monoclonal mouse anti-human TGF-β1 antibody was used to label 17 postmortem kidneys (7 cats with CKD and 10 cats from their pathology archives).58 The method of validation showing that this antibody cross-reacts with feline TGF-β1 was not provided. The mean (±SD) age was lower than many cats with end-stage CKD (7.20 ± 3.10 years). The cats were not all completely IRIS-staged, but for those that were, the stages were variable. The archived tissues were not from IRIS-staged cats, and clinical data were not provided for them. An objective scoring scheme to assess IHC staining was developed, but kidneys from cats with known normal kidney function were not included in the analysis. The IHC results showed TGFβ-1 expression in many areas of the kidney, but expression was strongest in areas of interstitial inflammation and fibrosis (Supplementary Table 1). TGF-β1 protein labeling was strongly correlated with glomerulosclerosis and tubular atrophy. Another IHC-based study examined 12 kidneys selected from 41 postmortem cases of cats with antemortem diagnoses of CKD. The 12 cases were selected because they had stable CKD with no evidence of any of the following: primary glomerular disease, acute decompensation of CKD, AKI, or polycystic kidney disease. This group included IRIS stage 1 cats that had persistently low urine specific gravity measurements (<1.035). The median plasma creatinine concentration of these cats was 2.09 (range 0.89-3.1) mg/dL. The reasons for death or euthanasia were not stated. A comparator group consisted of 4 young (23-27 months) healthy laboratory cats. TGF-β1 was labeled with a polyclonal rabbit anti-human antibody. Faint labeling was found in the control kidneys, within the interstitium and the basement membrane regions. By contrast, CKD samples showed TGF-β1 labeling of tubules and glomerular mesangium. In the distal nephron, apical cell surfaces and lumens stained strongly, whereas the cytoplasm of the proximal tubule epithelial cells (PTECs) showed positive labeling. Platelets inside blood vessels and some of the interstitial leukocytes were also positive. Correlations between the intensity of labeling, plasma creatinine and urea, glomerulosclerosis, and tubulointerstitial fibrosis were assessed. No correlations were found with fibrosis, but TGF-β1 labeling in the lumen of the distal nephron correlated positively with plasma creatinine and glomerulosclerosis. Negative correlations were found between plasma creatinine and labeling of PTEC cytoplasm, and also between plasma creatinine and labeling of platelets.59 Lastly, TGF-β IHC was described in one of the transcriptional investigations discussed above.56 Kidneys from 6 young CKD cats and from 6 kittens with normal kidney function were included. No labeling was found in the healthy control kidneys, whereas labeling was present in tubulointerstitial areas and within the cytoplasm of distal tubules and collecting ducts of CKD cats. TGF-β1 protein expression was shown by western blot of cultured Crandell-Rees Feline Kidney cells treated with doxorubicin as a model of toxic injury, in the same publication.56 These 3 IHC studies illustrate the previously discussed issues of protein labeling. Ideally, an IHC study would validate the antibody’s sensitivity using a second modality (eg, western blot or ELISA). Specificity should be demonstrated by performing isotype negative controls. There are no data in the above studies to indicate that the IHC antibodies were specific for the active TGF-β1 protein vs the protein embedded in the ECM. These limitations suggest that data reported from IHC studies should be interpreted with caution, and full validation of antibodies should be encouraged for future publications. TGF-β1 is a potential biomarker of kidney disease in cats. The isoform measured (total, latent, or active) is important to understand its biological significance. One study showed that urinary TGF-β1 measured by a multispecies ELISA kit (Biosource International Inc., Camarillo, CA) normalized to urine creatinine was significantly higher in CKD cats compared to healthy controls. They also measured serum TGF-β1 and circulating concentrations did not differ between groups. A limitation of this study was that little clinical information was provided for the 36 cats, other than the diagnostic criteria for CKD.60 Similar results were reported as part of a study examining multiple urinary cytokines obtained from healthy and CKD cats.61 The animals included in this study were well characterized as stable, non-proteinuric, non-hypertensive CKD cats (n = 26; median age 13.5 years, range 6-19 years) without significant comorbidities. Lower urinary tract disorders as causes of azotemia were excluded. Control cases were healthy cats (n = 18; median age 5.8 years, range 8 months-13 years) presented for routine screening. Two ELISA kits were used to measure TGFβ-1 in urine, one from R&D Systems and one multispecies assay from Invitrogen. Only the latter assay was able to detect feline TGF-β1, which was 5 to 10-fold higher in urine from CKD cats compared to control cats when normalized to urine creatinine. In addition, a significant positive correlation was found between urinary TGF-β1 and serum creatinine in CKD cats. Neither study published a validation of the ELISA assay, nor did they report what form of the cytokine (total, latent, or active) was measured. Furthermore, because their studies did not have kidney tissue, they could not correlate urinary TGF-β1 with tissue expression or the degree of fibrosis. Active TGF-β1 was measured in cat urine with an ELISA developed for humans (Biolegend, San Diego, CA), which was partially validated for the cat in terms of specificity, precision, reproducibility, and limit of detection. Interfering substances and matrix effects were assessed by determining spike recovery. The accuracy of the assay was not assessed because recombinant feline TGF-β1 was not available at the time. This paper presented three retrospective studies to explore the utility of measuring active TGF-β1 in urine. The first looked cross-sectionally at urine from cats that had normal kidney function (and remained normal over 12 months), cats that were non-azotemic at the time of sampling but developed azotemic CKD within 12 months, and cats with azotemic CKD. No differences in urinary active TGF-β1: creatinine ratio were detected among 3 three groups. The second study had a longitudinal design and involved 3 groups of 6 cats sampled 3 times over the course of 1 year. One group consisted of normal healthy cats whose kidney function remained normal during the year. The second group was non-azotemic initially but developed azotemic CKD over the course of a year. The third group had azotemic CKD, and their kidney function deteriorated during the year. This longitudinal study showed marked variability between individual cats, but the time course indicated that cats that developed azotemic CKD during the 12 months had elevated active urinary TGF-β1: creatinine ratios 6 months before CKD diagnosis. This elevation was sustained at the 12-month time point. Generally, progressive CKD cats had high urinary active TGF-β1: creatinine ratios, but the differences were not statistically significant, possibly because group sizes were small and there was moderate inter-cat variability. The final aspect of this study examined the relationship between urinary active TGF-β1 and renal fibrosis scores in cats that had a banked urine sample <3 months before euthanasia or death. Using multivariate analysis, only the fibrosis score remained an independent predictor of urinary active TGF-β1: creatinine ratio. These data suggest that urinary active TGF-β1 can be used as a biomarker of active fibrosis but that variability between individual cats might hamper its clinical utility.62 Experimental in vitro work shows that primary cultures of PTECs respond to stimulation by human recombinant TGF-β1. Specifically, the expression of epithelial cell markers by primary cultures of feline PTECs was characterized and compared to human renal epithelial cells (HRECs).63 Exposing these cells to increasing concentrations of human recombinant TGF-β1 (0.1-10 nmol/L) resulted in changes to cell morphology and altered gene expression, suggesting a change in phenotype of the cells and epithelial plasticity. This involved loss of expression of CDH1 (gene for E-cadherin) and increased expression of CDH2 (gene for N-cadherin), COL1A1 (gene for collagen 1α1), and FN1. Furthermore, auto-induction of TGFB1 gene expression by feline PTECs was evident, as was upregulation of connective tissue growth factor (CTGF), which is a key factor mediating the pro-fibrotic effects of the TGFβ-1 pathway. Auto-induction of TGFB1 expression is likely important in amplifying and sustaining the response. These effects were all inhibited by the TGF-β1RI/ALK5 antagonist, SB431542. Phosphorylation of Smad2 occurs in the feline PTECs. TGF-β1 also inhibited feline PTEC proliferation and upregulated apoptosis in a concentration-dependent manner. This was inhibited by SB431542. Feline PTECs behaved in a similar way to HERCs in these studies, supporting a conserved role for TGFβ-1in renal fibrosis in humans and cats. The same authors published a companion paper using primary cultures of renal cortical fibroblasts from cats with CKD and from cats with normal kidney function.64 The cells were characterized as mesenchymal cells by use of a range of cell markers and characteristic cell morphology. TGF-β1 stimulated concentration-dependent expression of the following genes: COL1A1, FN1, ACTA2 (the gene for α-smooth muscle actin) and CTGF; it also stimulated auto-induction of TGFB1. All these effects were abrogated by SB431542, demonstrating that activation of the canonical TGF-β1 pathway was involved. Some evidence indicated that the cortical fibroblasts isolated from CKD cats were activated because they expressed vascular cell adhesion molecule-1 (VCAM-1), whereas those from cats with normal kidney function did not. TGF-β1 induction of fibroblasts to express α-smooth-muscle actin suggests that these cells had been transformed (or were in the process of transforming) into feline myofibroblasts. Cells also underwent hypertrophy in response to TGF-β1 stimulation, another feature of the activated phenotype. Taken together, these experiments demonstrate that TGF-β1 can induce a strong and sustainable profibrotic response in resident cortical fibroblasts and PTECs. However, these studies did not prove that upregulation of gene expression translated to production and secretion of the encoded proteins. TGF-β in dogs Similar to cats, progressive CKD in dogs is correlated with development of interstitial inflammation and fibrosis. A large proportion of CKD dogs, particularly senior and geriatric dogs, have evidence of overt renal proteinuria.65 If routine urinalyses are performed, proteinuria often precedes declining glomerular filtration rate and irreversible tubulointerstitial lesions. IRIS stage 1 dogs with proteinuria represent a cohort that could potentially benefit from therapeutic inhibition of renal fibrosis early in its development. This is in contrast to cats, which less frequently have proteinuric CKD. To be clear, urine protein is an indicator of progressive CKD in the cat at the population level, but the magnitude of the proteinuria is often lower than in dogs. Moreover, the overlap of urine protein to creatinine ratio (UPC) between progressive and stable CKD in cats is considerable. Because of the multifaceted relationship of proteinuria, CKD, and TGF-β in dogs, proteinuric IRIS stage 1 dogs might represent a promising cohort for early intervention in renal disease progression. Multiple studies support the role of the TGF-β pathway in kidney disease in dogs. Older literature of in vitro cultured Madin-Darby canine kidney (MDCK) cells reported the involvement of TGF-β in EMT.66 However, as stated earlier, this phenomenon likely does not occur in the kidney in vivo. More recently, pathway inhibition by pirfenidone was seen in cultured MDCK cells that were treated with TGF-β1. The mechanism of pirfenidone action is not fully understood, but there is evidence that it acts in part through downregulation of Smads 2 and 3, by blocking their phosphorylation. In cultured MDCK cells, it also blocked ERK1/2 phosphorylation.67 In other model systems, pirfenidone appears to inhibit release of active TGF-β from the matrix by inhibiting one of the proteolytic enzymes, furin.68 The majority of in vivo studies demonstrating the importance of TGF-β in canine CKD involve dogs with X-linked hereditary nephropathy (XLHN). These dogs have a mutation in Type IV collagen, which is an integral component of the glomerular basement membrane. A colony of these dogs underwent serial biopsies, in addition to repeated urine and blood collections, allowing assessment of tissue gene expression over time. In one study of 10 affected and 5 unaffected dogs, increased renal expression of TGF-β (specific isoform not stated), CTGF, and platelet-derived growth factor mRNAs (assessed with qRT-PCR) was observed before development of azotemia and interstitial fibrosis, supporting that these genes are involved in the early/initiating events of tubulointerstitial lesions.69 While TGF-β mRNA expression initially increased in renal biopsy tissue early in the disease, it was not significantly overexpressed compared with unaffected male littermates (controls) at late-stage disease; therefore, mRNA expression did not correlate with degree of fibrosis. Additionally, IHC using an antibody directed against all isoforms of TGF-β showed similar expression in podocytes and some tubules in both affected dogs throughout their disease progression as well as controls (Supplementary Table 1). It was hypothesized that, in early glomerular disease, podocyte TGF-β expression results in mesangial cell proliferation and matrix expansion, which are the earliest light microscopy abnormalities in affected dogs.69 The lack of altered TGF-β expression at late-stage disease compared with unaffected littermates is consistent with an earlier microarray study comparing 4 dogs with end-stage disease due to XLHN to 4 unaffected dogs.70 In that study, cortical renal gene expression of TGF-β1 was not significantly different between the 2 groups.70 In contrast, a more recent study demonstrated increased mRNA expression of TGFB1, TGFB2, and TGFBRI at end-stage disease based on qRT-PCR, performed on autopsy samples of renal cortex from 27 affected and 5 unaffected dogs.71 Additionally, expression of miR-21, a microRNA regulated by TGF-β, was first increased at the onset of azotemia when comparing serial renal biopsy samples from 11 XLHN dogs to 10 unaffected dogs. The expression of miR-21 also correlated with the severity of interstitial fibrosis.71 This miR is of interest in cat CKD and TGF-β pathways, as noted above. Two additional studies used RNA-seq or small RNA-seq to further characterize gene expression in XLHN dogs. In the RNA-seq study, 6 serial renal biopsies from XLHN dogs with either rapid (n = 3) or slow (n = 3) progression were compared, along with those from 2 unaffected dogs. Fibrosis scores were higher in dogs with rapid progression, and pathway analysis identified TGF-β1 as the primary upstream regulator when comparing expression patterns between the rapid vs slow groups. However, the TGFB1 transcript itself was not differentially expressed in any comparison (rapid vs slow; affected vs control).72 Despite this, TGFB2 expression was generally increased in affected dogs at mid- and late-stage disease compared to controls. The TGFB3 transcript was increased at mid-stage disease in the rapidly progressive group compared to controls. Additionally, TGFB inducible (a TGF-β-induced transcript) was upregulated, similar to TGFB2. In some comparisons, TGFB1|1 (which is the transcript known as TGFB1-induced transcript 1) was also increased. These data support activation of the TGF-β pathway in this disease process.72 Using small RNAseq to compare serial renal biopsies from 5 XLHN dogs and 4 unaffected dogs, miRNA-21 and -146b, both of which are associated with fibrosis, were two of the five miRNAs consistently upregulated in affected dogs at each timepoint.73 Pathway analysis identified enrichment of “signaling by TGFβ receptor complex” pathway in affected dogs at the earliest timepoint. These findings further support the role of TGF-β in early disease progression (before onset of azotemia).73 Overall, the results of the canine XHLN studies support the importance of the TGF-β pathway in CKD progression. However, discrepancies in the results from end stage samples (eg, mRNA present in one cohort but not others and similar IHC labeling patterns across disease stages and control samples) highlight the challenges of using gene expression and IHC to determine TGF-β activity, even within one disease phenotype. The role of TGF-β in renal disease of the canine pet population has also been examined. Notably, diagnosis of spontaneous canine CKD could be based on the presence of persistent proteinuria or azotemia. Alternatively, it might occur when dogs are transitioning from an AKI event to CKD. After an AKI episode, a dog might progress to develop chronic irreversible damage consistent with CKD74. One study found 23% of dogs developed IRIS stage 2 CKD after an AKI episode.75 Although this can occur in cats, it is thought to be less common. Numerous cytokines are involved in these transitions, but TGF-β is a key player in these phenomena.76,77 TGFB mRNA from whole blood samples was higher in uremic dogs compared to controls (using qRT-PCR). This study included 46 dogs with AKI of various etiologies, 8 dogs with CKD, and 10 healthy control dogs. Interestingly, whole blood TGFB expression was significantly higher in AKI dogs compared to controls, but there was no detectable difference between CKD and control dogs.78 These findings might have diagnostic implications for whole blood TGFB expression levels as a screening tool for AKI episodes. However, modulation of the cytokine occurs with multiple systemic inflammatory conditions, and that could limit the specificity of a diagnostic test. Further investigations of whole blood TGFB expression levels with renal fibrosis and renal gene expression would be a valuable next step in developing this type of diagnostic assay. Leptospirosis is a common cause of AKI in dogs.79–81 Asymptomatic shedding of leptospires is found in most dogs with CKD living in leptospirosis endemic regions, suggesting a possible role in the development of CKD in these dogs.82 Of the numerous cytokines involved in leptospirosis in humans, TGF-β is considered one of the most important mediators of post-infection renal fibrosis.83 The relationship in dogs is not well-studied but might be an interesting avenue for future investigation. In endemic regions, leishmaniosis is a common cause of glomerular disease in dogs, most commonly glomerulonephritis.84,85 Glomerular and tubulointerstitial fibrosis was present in nearly half of dogs with leishmania-associated glomerular disease.86 An IHC study was performed in naturally and experimentally infected dogs (using 2 different parasite strains). One aim of the study was to determine if either of the 2 strains more closely mimicked renal lesions seen in natural leishmaniosis. TGF-β labeling was greatest in glomeruli and tubular basement membranes in all examined dogs. One of the 2 strains used for experimental infections and natural infections resulted in significantly larger regions of positive labeling compared to the other experimental strain and uninfected control dogs. Again, this underscores the variability of IHC labeling within a single disease entity. Interestingly, IHC also showed that vimentin and α-smooth muscle actin were increased by infections with the same experimental strain and in naturally infected dogs compared to controls.87 The immunomodulatory functions of TGF-β could also be important in both leptospirosis and leishmaniosis, because both infectious diseases incite significant nephritis. Obesity is a major risk factor for the development of kidney disease in people, but less is known about its relevance in dogs. Experimentally-induced obesity is associated with renal hyperperfusion in dogs, which can be reversed with weight loss.88 This hyperperfusion might result in glomerular hypertension and subsequent renal injury. Experimentally induced obesity also alters intrarenal cytokine expression in dogs; a 74% increase in TGFB1 was observed in the medulla of obese dogs.89 This upregulation of the TGF-β pathway might increase the risk of renal fibrosis and the loss of renal function. One study identified improvement in functional kidney variables after weight loss in obese dogs.90 This suggests that early intervention in obese dogs might help prevent associated renal injury, through normalization of renal perfusion and perhaps normalization of renal TGF-β1 levels, although the association between obesity and risk of CKD in dogs remains to be determined in any large-scale study where other co-variates can be taken into account. The data regarding the role of TGF-β in canine kidney disease has focused on AKI from a wide variety of causes and two types of glomerular-driven CKD. Many common causes of proteinuric CKD in dogs have not been evaluated for TGF-β mRNA or proteins, or pathway activation, including amyloidosis and focal segmental glomerulosclerosis. The latter disease would be of special interest, given that TGF-β is known to be involved in human and rodent glomerulosclerosis. In both cats and dogs, more research is needed to characterize its role in other types of CKD, such as obstructive nephropathy and pyelonephritis. Conclusions Progressive interstitial fibrosis with tubular atrophy and inflammation are the hallmarks of progressive kidney diseases in dogs and cats. Understanding the molecular pathogenesis is required if novel therapies of slowing progressive kidney disease are to be identified. The evidence that the TGF-β pathway links the stressors affecting PTECs to interstitial fibrosis is building for companion animal species. Links between TGF-β and glomerulosclerosis are also suspected (because they have been shown experimentally or in natural kidney disease in humans). However, this relationship is not as well characterized in small animals. The combination of in vitro and clinical data presented in this review suggests the TGF-β pathway is a logical target for future therapeutic interventions. Specific anti-TGF-β therapies for renal disease in humans were recently reviewed, using a Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) approach. Six publications on clinical trials and 14 clinical trials were identified, only 4 of which had published results (some were still in recruitment phase at the time of the review). Conflicting data from published clinical trial results indicated that there are beneficial anti-inflammatory roles of the TGF-β pathways in addition to its harmful pro-fibrotic mechanisms.91 Certainly, expectations for a successful renal therapy differ between humans and small animals; preserving renal function for one to two years would be considered beneficial in small animals, whereas it might not be in humans, because a transplanted kidney can last for decades. Given the lack of treatment modalities to lower fibrosis in scarred kidneys, prevention or mitigation is a therapeutic goal in cats and dogs. Supplementary Material