Targeting the CDA1/CDA1BP1 Axis Retards Renal Fibrosis in Experimental Diabetic Nephropathy

Tissue scarring or fibrosis is a pathological hallmark of the renal injury seen in the commonest cause of end-stage renal failure worldwide, diabetic nephropathy (DN). Our group has previously reported that cell division autoantigen 1 (CDA1) is linked to the actions of the key profibrotic growth factor transforming growth factor-β (TGF-β) (1–4) whereby CDA1 synergistically enhances TGF-β signaling in renal and vascular cells (1,2). Indeed, renal CDA1 expression levels are elevated in both animal models of DN and in human renal biopsy samples from subjects with diabetes with nephropathy (2,3). Furthermore, CDA1 knockout (KO) mice develop less renal injury in several models of DN (3). These findings complement in vitro studies demonstrating that upregulation or deletion of CDA1 leads to increased or reduced expression of TGF-β–dependent profibrotic genes, respectively. TGF-β is known to play a culprit role in DN (5–7), but unfortunately it is biologically too difficult to directly target since the deletion of TGF-β1 is lethal in mice (8,9). In contrast, CDA1 KO mice grow and breed normally without any abnormal phenotype. Thus, we consider that targeting CDA1 is not only an effective but also a safer alternative and potentially superior approach to retard diabetic nephropathy.

As part of the strategy to develop a pharmacological approach to target CDA1, we identified and characterized a novel interacting protein for CDA1, which we named CDA1 binding protein 1 (CDA1BP1). We then used both genetic and pharmacological approaches to target the CDA1/CDA1BP1 axis, providing evidence of efficacy and feasibility to reduce renal fibrosis in experimental DN.

Rabbit antibodies to CDA1 have been described previously (1,10). Antibodies to phospho-Smad3 (Ser^423/425) and Smad3 were from Abcam (Cambridge, MA); antibodies to GAPDH and β-actin were from Santa Cruz Biotechnology (Santa Cruz, CA); and recombinant human TGF-β1 was from R&D Systems, Inc. (Minneapolis, MN). Short peptides were synthesized by Mimotopes (Mulgrave, Victoria, Australia) or GL Biochem (Shanghai, People’s Republic of China). Anti-CDA1BP1 polyclonal antibodies were raised in rabbits using N- and COOH-terminal peptides of CDA1BP1 and affinity purified. The specificity and sensitivity of the antibodies were characterized by ELISA on the peptide antigens and by Western blotting using recombinant proteins.

Yeast two-hybrid (Y2H) screening was performed using full-length CDA1 as bait on a BD Matchmaker Pretransformed human testes cDNA library (BD Biosciences, San Jose, CA) from over 1 × 10⁶ independent clones using a high-stringency selection medium lacking adenine, histidine, leucine, and tryptophan. The prey fragments of the positive clones were amplified by PCR and sequenced to identify the encoded proteins. The candidate clones were recloned into the pACT2 vector at SalI and XhoI sites and fully sequenced. These constructs were further characterized for their interaction with CDA1 using a small-scale yeast Y2H interaction assay and for identification of their binding domains using mutagenesis approaches such as testing a series of sequential deletion mutant constructs for their activities to interact specifically with CDA1.

The mouse model of streptozotocin (STZ)-induced diabetes has been described previously (1–3). Briefly, 6-week-old male ApoE KO, wild-type (WT), and CDA1BP1 KO mice, all on C57BL/6 background, were rendered diabetic by five consecutive daily intraperitoneal injections of STZ at a dose of 55 mg/kg or were given citrate buffer only to serve as nondiabetic controls. Animals were kept in a metabolic cage for 24 h to collect urine samples in the last week of the experiment, and a blood sample was also collected for the measurement of relevant metabolic parameters. Animals were killed after 10 or 20 weeks of diabetes. The animal studies were approved according to the principles consistent with international guidelines including the 1985 “Principles of Laboratory Animal Care” (National Institutes of Health) and the 2013, 8th Edition “Australian Code of Practice for the Care and Use of Animals for Scientific Purposes” (National Health and Medical Council of Australia). Glycated hemoglobin (HbA_1c) levels were measured using the cobas b 101 POC System (Roche, Rotkreuz, Switzerland). Albumin in urine was measured using the Mouse Albumin ELISA Quantitation Set (Bethyl Laboratories Inc., Montgomery, TX). Urinary creatinine level was determined using a commercially available creatinine assay kit (catalog #ab65340; Abcam, Cambridge, U.K.).

Immunohistochemistry (IHC) staining on paraffin section was as described previously (1–3). Briefly, paraffin sections of mouse kidney were dewaxed, digested with 0.4% pepsin for 10 min at 37°C, and then incubated with the primary antibody overnight at 4°C for collagens III and IV (antibodies were purchased from SouthernBiotech, Birmingham, AL). For collagen I staining, the sections were heated in 10 mmol/L citrate buffer in a microwave oven for 12 min to retrieve the epitopes before staining using an anti–collagen I antibody (Abcam) overnight at 4°C. Extracellular matrix (ECM) proteins were stained using Masson’s Trichrome kit (AMT.K) from Australian Biostain P/L (Traralgon, Victoria, Australia). The positive staining was quantified by assessing ∼10 random views per section using Image-Pro Plus (Media Cybernetics, Bethesda, MD).

Paraffin-embedded kidney sections stained by the periodic acid Schiff (PAS) staining method were semiquantitatively assessed as described previously (3,11). Briefly, 20 glomeruli per section were scored in order to calculate the average glomerular injury index per mouse, with 0 for an intact glomerulus, 1 for <25%, 2 for 25–50%, 3 for 50–75%, and 4 for >75% of damaged area of a glomerulus. The mesangial area was quantified using Image-Pro Plus software.

Gene-specific mRNA quantitation by real-time RT-PCR was performed as described using a PCR system (7500 Fast Real-Time PCR and QuantStudio 7 Real-Time PCR; Applied Biosystems, Foster City, CA) and normalized against endogenous 18S ribosome RNA (TaqMan; Applied Biosystems). Results are shown as the fold change (arbitrary unit) relative to the control group. Probes and primers used to determine gene-specific mRNA levels are listed in Supplementary Table 1.

Data are presented as the mean ± SEM. Data were statistically analyzed by one-way ANOVA with least significant difference test as a post hoc test and normal distribution was examined by the D’Agostino-Pearson omnibus normality test. The unpaired t test was used to compare two experimental groups. A P value of <0.05 was considered to be statistically significant.

Using Y2H screening to identify proteins interacting with CDA1, we found that ∼36% of positive clones (18 of 50) isolated from a human testis cDNA library encode a novel protein, which we designated as CDA1BP1 (Supplementary Fig. 1A). The nucleic acid sequence of CDA1BP1 is identical to a region of Y box binding protein 1 (YBX1; accession #NM_004559), encoding CDA1BP1 by an alternative reading frame. The first possible in-frame translation initiation codon ATG was identified, leading to an open reading frame (ORF) encoding a deduced protein of 89 amino acid residues (Supplementary Fig. 1B). The mouse homolog of CDA1BP1 contains 132 residues with 2 additional amino acid residues at the N terminus and an extended COOH-terminal region of an additional 41 residues. The overlapped region of the 89 residues in mouse and humans is highly homologous with an amino acid identity of 88% (Supplementary Fig. 1C).

The direct interaction between recombinant CDA1 and CDA1BP1 was demonstrated by a glutathione S-transferase (GST) pull-down assay (Fig. 1A). Biotinylated CDA1 COOH-terminal protein was pulled down by a GST-CDA1BP1 fusion protein, but not by GST alone (Fig. 1A, top panel). The presence of both the intact GST-CDA1BP1 (Fig. 1A, middle panel) and GST (Fig. 1A, lower panel) were shown by Western blotting using a rabbit antibody to the N terminus of CDA1BP1 (Fig. 1A, middle panel) and an anti-GST antibody (Fig. 1A, bottom panel), respectively. Similarly, GST-CDA1 (COOH-terminal domain) or GST was incubated with HK-2 cell lysate or mouse kidney homogenates, followed by pull down using glutathione beads. Protein doublet bands at ∼30 kDa were detected in the complex with GST-CDA1, but not with GST alone (Fig. 1B, top panel). The presence of GST-CDA1 was detected in the complex using an anti-CDA1 antibody (Fig. 1B, middle panel) with the anti-GST antibody detecting both GST-CDA1 and GST proteins, as expected (Fig. 1B, bottom panel). Furthermore, GST-CDA1 fusion proteins were used as probes to determine whether CDA1 was able to bind to HeLa cellular proteins separated by SDS-PAGE and transferred to a nitrocellulose membrane. Indeed, the GST-CDA1 fusion protein with the complete acidic domain (GST-CDA1) bound to an ∼30 kDa protein, whereas a shorter CDA1 fusion protein containing the COOH-terminal half of the acidic domain of CDA1 with a higher concentration, bound to proteins at ∼15 and ∼30 kDa (Fig. 1C, lanes 1–2), which were equivalent to CDA1BP1 monomer and dimer, respectively. GST and no protein controls did not show any signal (Fig. 1C, lanes 3–4).

A series of COOH-terminal deletion mutants of the CDA1BP1 were tested for their ability to bind to CDA1 by Y2H. The results showed that a region of six amino acids, Tyrosine-Isoleucine-Isoleucine-Arginine-Phenylalanine-Serine (TIIRFS), within CDA1BP1 was required for binding to CDA1 (Fig. 1D). The yeast diploids containing CDA1 and a prey construct grew and formed colonies on an SD/-Leu/-Trp plate, as expected. On the reporter media, SD/-Ade/-His/-Leu/-Trp/X-α-Gal, yeasts could grow and form a colony in blue color only if the prey protein interacted with the bait CDA1. As seen in Fig. 1D (right panel), the WT CDA1BP1 (FC78) as well as the deletion mutants FC78-∆1 and FC78-∆2 grew and formed blue colonies (sections 1–3), whereas FC78-∆3 and FC78-∆4, both without the six residues, TIIRFS, did not grow (sections 4–5). Furthermore, the colonies of FC78-∆3 and FC78-∆4 on SD/-Leu/-Trp (Fig. 1D, left panel) were noticed to be red in color, a feature of adenine auxotrophy, because of the inactivation of the reporter ADE2 gene and the resultant accumulation of an intermediate of the defective adenine biosynthesis pathway in these cells. In contrast, the WT, FC78-∆1, and FC78-∆2 colonies were white, indicating the activation of the reporter ADE2 gene because of the interaction between the CDA1 bait and the prey CDA1BP1 molecules. These observations support the conclusion that the TIIRFS sequence of CDA1BP1 is responsible for binding to CDA1 in yeast cells.

The CDA1-CDA1BP1 interaction was further characterized by their colocalization in mammalian cells. Like CDA1 (10), which is also known as DENTT (differentially expressed nucleolar TGF-β1 target) (12), Myc-tagged CDA1BP1 was localized to the nucleoplasm and nucleolus in transfected cells (Supplementary Fig. 2). CDA1 and CDA1BP1 were colocalized in cotransfected HeLa and HEK293 cells by confocal microscopy (Fig. 1E), with appropriate controls demonstrating the specificity of the staining (Supplementary Fig. 3). Myc-CDA1BP1 and Flag-CDA1 in cotransfected cells were coimmunoprecipitated using either an anti-Flag or anti-Myc antibody (Fig. 1F), demonstrating that CDA1 and CDA1BP1 indeed formed a complex in cells.

Expression of both CDA1 and CDA1BP1 was detected in a broad range of tissues and cells, including human and mouse kidney and in human tubular HK-2 cells. Like CDA1, CDA1BP1 enhanced the activity of TGF-β in stimulating the expression of collagen I and other sclerotic genes in HK-2 cells. This effect was further enhanced by co-overexpression of CDA1 (Fig. 2A). Furthermore, CDA1BP1 siRNA knockdown resulted in decreases in gene expression and protein levels of collagens I and III in HK-2 cells with or without TGF-β treatment (Fig. 2B and C). In these CDA1BP1 siRNA knockdown cells, TGF-β–induced phosphorylation of Smad3 was significantly attenuated (Fig. 2D).

CDA1BP1 was found to be encoded by an alternative ORF of the YBX1 gene, the major ORF of which encodes Y-box binding protein 1 (YB1) (Supplementary Fig. 4A). In the mouse genome, the CDA1BP1 ORF shares exons 7 and 8 with YB1 ORF (Supplementary Fig. 4B). In order to functionally delete CDA1BP1 without affecting YB1, we mutated the four ATG codons within the CDA1BP1 ORF by T/C point mutation (AT/CG), resulting in the abolition of all the possible translation initiation ATG codons within the CDA1BP1 ORF, without changing the amino acid residues encoded by the modified YB1 ORF (Supplementary Fig. 4). A floxed mini-cDNA for CDA1BP1 was inserted in front of the mutated exon 7 (floxed CDA1BP1 mouse) in the CDA1BP1 targeting construct, which was deleted by crossing the CDA1BP1 floxed mouse with a global Cre mouse, resulting in a knockin (KI) of the mutated CDA1BP1 (KO/KI) (Supplementary Fig. 4B). The KO/KI allele could not produce CDA1BP1 due to the lack of a translation initiation site. The generated mouse strain containing this KO/KI allele is referred to as being the CDA1BP1 KO mouse strain. The point mutations in the genomic DNA of CDA1BP1 KO mice were confirmed by sequencing (Supplementary Fig. 5). As expected, YB1 protein expression was not changed in kidneys from CDA1BP1 KO mice when compared with WT mice (Supplementary Fig. 6). Animals used in the experiments were individually genotyped by PCR (Supplementary Fig. 7).

Both heterozygous and homozygous CDA1BP1 KO mice grew and bred normally. Twenty weeks after STZ-induced diabetes, both WT and CDA1BP1 KO mice showed typical metabolic changes such as a reduction in body weight as well as increases in blood glucose and HbA_1c levels, kidney/body weight ratio, urine excretion, water intake, food intake, and urine albumin-to-creatinine ratio (Supplementary Table 2). There were diabetes-associated, approximately threefold, increases in renal mRNA levels for collagens I and III, fibronectin, and tumor necrosis factor-α (TNF-α) seen in the diabetic WT mice, which were significantly attenuated in diabetic CDA1BP1 KO mice (Fig. 3A). Immunohistochemical staining for renal collagen III (Fig. 3B and C) and collagen I (Fig. 3D) was increased more than fourfold in diabetic WT mice, and attenuated by ∼50% in diabetic CDA1BP1 KO mice.

In order to explore the feasibility to pharmacologically target the CDA1/CDA1BP1 axis, we designed a series of synthetic short peptides containing the 6–amino acid binding site within CDA1BP1 (Fig. 4A). A 19mer peptide of the COOH-terminal region of CDA1BP1 was shown to block the binding of recombinant CDA1BP1 to CDA1 proteins, which were overexpressed in HK-2 cells (Fig. 4B, top panel, lanes 3–4). As an appropriate control, the 19mer mutant peptide with three amino acid residues replaced with alanines failed to inhibit the binding of CDA1BP1 to CDA1 (Fig. 4B, bottom panel, lanes 3–4). CDA1BP1 was undetectable when incubated with cellular proteins from cells without CDA1 overexpression (Fig. 4B, lanes 1–2).

A more quantitative ELISA format binding assay was used to evaluate a series of synthetic peptides to inhibit CDA1-CDA1BP1 binding. Coated CDA1 protein in a microtiter plate was coincubated with CDA1BP1 protein mixed with a test peptide. The inhibitory activities of the peptide were reflected by the decrease in the quantity of CDA1BP1 bound to the coated CDA1, which was detected by a rabbit anti-CDA1BP1 antibody and horseradish peroxidase (HRP)–conjugated secondary antibody. Peptides with the TIIRFS binding site were found to dose-dependently inhibit CDA1-CDA1BP1 binding, whereas peptides without the binding site (Fig. 4C) or with the binding site mutated did not inhibit the binding. Furthermore, the neighboring amino acid sequence at the N-terminal side of the binding site, but not at the COOH-terminal site, appeared to contribute to the binding activity. The 12mer peptides containing the binding site and its N-terminal neighboring amino acid sequence (CRLTISTIIRFS), such as CHA-045, CHA-046, and CHA-048, the hybrid peptides linked with a previously known cell-penetrating peptide (CPP) (13,14), had an inhibitory potency equivalent to that of the 19mer peptide (Fig. 4C).

We used HK-2 cells to determine the effect of the potential peptide inhibitors on the expression of the target genes of TGF-β and CDA1. The CHA-050 peptide, but not its scramble control peptide CHA-051, reduced gene expression of collagens I and III in a dose-dependent manner (Fig. 4D). Similar results were seen with regard to CHA-045, CHA-046, and CHA-048 peptides.

In order to extend these findings to in vivo studies, we translated the biological activity of CHA-045 using the retro-inverso principle into the d-amino acid CHA-061 with the N-C order reversed (see sequence in Fig. 4A). These modifications serve to impart plasma stability to the peptide while retaining equivalent biological function. Indeed, when added to HK-2 cells, CHA-061 attenuated TGF-β–stimulated Smad3 phosphorylation (Fig. 4E), an effect consistent with that of either a CDA1 siRNA knockdown in cells or the genetic deletion of CDA1 in mice (1–3).

ApoE KO mice were intraperitoneally injected with CHA-061 at 1, 5 and 10 mg/kg, which led to reduced renal expression of p21, α–smooth muscle actin, and collagen I 24 h later in a dose-dependent manner. The effect of a single injection of CHA-061 at 10 mg/kg was observed for at least 5 days. To explore the specific role of this peptide in reducing renal fibrosis in DN, STZ-diabetic male ApoE KO mice were randomized to receive vehicle [diabetic group (Dia)] or CHA-061 peptide treatment (intraperitoneal injection, 10 mg/kg, twice a week) (Dia + treatment) (see Supplementary Tables 3 and 4 for metabolic data at the end point). As expected, in vehicle-treated animals, diabetes for 10 weeks increased the renal expression of profibrotic genes such as various TGF-β isoforms, TGF-β receptors, and connective tissue growth factor (Fig. 5A, left panel), sclerotic genes such as collagens I, III, IV, fibronectin and MMP2 (Fig. 5A, middle panel) as well as proinflammatory genes such as TNF-α, CRP, MCP1, ICAM1, and VCAM1 (Fig. 5A, right panel). CHA-061 treatment from week 6 to week 10 after induction of diabetes significantly attenuated or blocked the diabetes associated upregulation of these genes (Fig. 5A). Immunohistochemical staining for the renal collagens I and IV showed that the diabetes associated increase in these collagens was significantly attenuated by CHA-061 (Fig. 5B and C). In a 20-week diabetes study, the diabetic animals received CHA-061 treatment in weeks 11–20 of diabetes. The treatment commenced after the mice were diabetic for 10 weeks, when renal pathology had already developed as shown in Fig. 5. In this study, the diabetic animals treated with CHA-061 showed a significant, >70% attenuation of diabetes associated increases in renal gene expression of collagens I and III, connective tissue growth factor, TNF-α and MCP1 (Fig. 6A) as well as a significant >80% attenuation in IHC staining of collagen III, which was increased >11-fold in the vehicle-treated diabetic group (Fig. 6B). Renal extracellular matrix accumulation, as determined by Masson’s Trichrome staining, showed a more than twofold increase in vehicle-treated diabetic mice, which was attenuated by ∼60% in the CHA-061–treated mice (Fig. 7A). The glomerular injury index, as assessed on PAS-stained kidney sections showed a >4.3-fold increase in association with diabetes, which was attenuated by >40% in the CHA-061 treatment group (Fig. 7B). This was consistent with the measurement of mesangial area showing that diabetes induced mesangial expansion was attenuated but not normalized by CHA-061 treatment (Fig. 7C).

The current study has provided convincing evidence that targeting the CDA1/CDA1BP1 axis using either genetic or more clinically relevant pharmacological approaches can attenuate the pathological hallmarks of a model of DN, the current commonest cause of end-stage renal failure worldwide. These findings strengthen the initial studies demonstrating the pathological role of CDA1 in DN (3).

The pathological role of CDA1 via enhancing TGF-β signaling in the diabetic complications has been described (1–4). The effect of CDA1 on TGF-β signaling appears to be via influencing the expression level of the TGF-β receptor (TβR) type I but not TβRII (1–3). Importantly, in cells lacking CDA1, TGF-β–stimulated Smad3 phosphorylation and TGF-β–responsive reporter activity were still evident, albeit at significantly reduced levels (1–3). These findings demonstrate that the CDA1/CDA1BP1 axis is able to influence TGF-β signaling but is not a vital component of the TGF-β signaling pathway. Indeed, the TGF-β signaling pathway is still intact after CDA1 deletion, and thus the physiological, in contrast to the pathological, functions of TGF-β are unlikely to be affected to a significant extent by approaches targeting CDA1. Therefore, targeting CDA1 was hypothesized to not only be efficacious in reducing fibrosis, but also likely to be a safer approach than targeting directly TGF-β or its receptors. Specifically, it was shown that CDA1 KO mice, unlike TGF-β1 KO mice, were fertile and grow normally (3).

TGF-β has been considered to be a potential target for treating diabetic nephropathy for several decades (15). However, directly targeting TGF-β appears to have ultimately been unsuitable for the purpose of developing a therapeutic strategy since TGF-β has multiple important biological functions, including the regulation of immune responses involved in immune tolerance and inflammation (16,17). Indeed, genetic deletion of the ligand TGF-β1 or other component molecules of the TGF-β pathway, such as its receptors, results in catastrophic phenotypes in the mouse, such as embryonic lethality or severe inflammation and death occurring shortly after birth (8,9,18–22). The difficulty of directly targeting TGF-β is further reflected by the recent failure of a clinical trial by Eli Lilly and Company with a TGF-β1–neutralizing monoclonal antibody, LY2382770, in subjects with diabetes in whom the dosage had to be reduced because of the concern of potential side effects. Indeed, no proven renal efficacy, including a lack of an effect on albuminuria, with that dosage could be detected (23,24).

In order to explore a pharmacological approach to target CDA1 in this study, we identified a protein, CDA1BP1, as a key partner protein for CDA1 to exhibit its profibrotic effect via enhancing TGF-β signaling, and demonstrated that inhibition of the CDA1-CDA1BP1 interaction is an efficacious strategy to reduce TGF-β signaling in cells. Building on initial in vitro studies, the pathological role of this axis was demonstrated in vivo in diabetic ApoE KO mice using a newly generated pharmacological agent, CHA-061. This finding complemented and bolstered the results obtained using a genetic approach in CDABP1 KO mice, which is consistent with the putative role of CDA1BP1 as the key partner protein of CDA1 in the CDA1/CDA1BP1 axis. Similar to the CDA1- and CDA1BP1-deficient mice, in the ApoE KO mice treated with the CDA1 inhibitor CHA-061, no obvious side effects were observed during the course of the study. Indeed, these experiments demonstrated the feasibility of pharmacologically targeting CDA1.

The pharmacological agent CHA-061 developed in this study to inhibit the profibrotic effect of CDA1 is a peptide in nature, incorporating a short CDA1BP1 sequence with an ability to bind to CDA1 and a CPP, a previously known peptide that assists with cellular entry (13,14). CHA-061 was synthesized using d-amino acids in reverse direction (retro-inverso peptide), a strategy that has been described to successfully retain the parental peptide activity with increased stability in vivo, as has been reported for other conditions previously (25–28).

ApoE KO mice were chosen to be used to test this agent in this study because this mouse strain has been previously shown to develop diabetic nephropathy at an accelerated rate with clear evidence of ECM accumulation (3,29,30). The early intervention study showed renal protection in mice with a duration of diabetes of 10 weeks that received 5 weeks of CHA-061 treatment commencing after 5 weeks of diabetes (Fig. 5). Based on these early intervention data, a more delayed intervention study was performed by treating these mice after 10 weeks of diabetes for a subsequent period of 10 weeks. The results of the later intervention study showed that CHA-061 treatment reversed these diabetes-associated molecular changes, which were reduced to levels similar to those seen in the nondiabetic controls (Fig. 6). Importantly, the diabetes-associated renal pathological changes, such as accumulation of extracellular matrix and glomerular injury seen in the untreated diabetic mice, were attenuated significantly by this specific peptide treatment approach (Fig. 7). Thus, these results demonstrate the efficacy of CHA-061 to prevent further progression of diabetic kidney disease or to reverse established renal pathological changes in association with diabetes.

It was noted that targeting the CDA1/CDA1BP1 axis had no significant effect on albuminuria in this study. This is not surprising considering that the axis appears to be specifically linked to the canonical TGF-β/Smad3/ECM pathway, which has been previously shown to have no effect on albuminuria in experimental DN using either anti–TGF-β neutralization antibody (5) or by a genetic approach to delete the Smad3 gene in mice (31). It is likely that podocytopathy and albuminuria are not influenced by classic canonical TGF-β pathways, as was recently reported in another DN model (32).

In the animal model we used in this study, diabetes was associated with an increase in tubulointerstitial ECM accumulation and glomerular injury, with both parameters attenuated by approaches to target the CDA1/CDA1BP1 axis (Figs. 3 and 5–7). Tubulointerstitial fibrosis is a common final pathology closely linked to overt renal injury and end-stage renal disease (33–36). Indeed, decline in renal function in DN has been shown to be more closely related to tubulointerstitial fibrosis than to glomerular injury (35). The pathological role of the CDA1/CDA1BP1 axis in vivo has been supported by in vitro experiments using the proximal tubule cells, HK-2 in this study (Figs. 2 and 4) as well as by our previous studies (2,3).

Since CDA1BP1 has only recently been identified, reagents required to assess this protein are limited. Thus, we focused on a gene deletion approach and an interventional strategy with an inhibitory peptide. The positive findings from this study with respect to the role of the CDA1-CDA1BP1 interaction provide a further impetus to generate additional reagents to further characterize this pathway not only in experimental models but also in human disease.

In summary, this study has demonstrated that pharmacological targeting of the CDA1-CDA1BP1 interaction attenuates, but does not block, TGF-β signaling, a property that is likely to lead to enhanced safety of such an approach when compared with directly targeting TGF-β or its receptors. Indeed, the identification of this CDA1/CDA1BP1 axis as an effective and safe target to reduce renal fibrosis provides a novel and unique opportunity to develop a new family of therapeutic agents with which to treat diabetic nephropathy.

Acknowledgments. The authors thank Elisha Lastavec and Megan Haillay for their assistance with the animal studies. The authors also thank Monash Micro Imaging, Monash University, Victoria, Australia, for use of facilities and scientific and technical assistance.

Funding. This study was supported by the National Health and Medical Research Council (NHMRC) of Australia. M.E.C. is an NHMRC Senior Principal Research Fellow.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. Z.C. and M.E.C. initiated and designed the study and wrote the manuscript. T.W., A.D., P.H., and S.R. performed experiments. F.K. generated the CDA1BP1 KO mouse strain. G.K. designed peptide reagents. Z.C. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.