AMPK-SP1–Guided Dynein Expression Represents a New Energy-Responsive Mechanism and Therapeutic Target for Diabetic Nephropathy

Visual Abstract


Introduction
Diabetic nephropathy (DN) is a major complication of diabetes.2][3] The pathophysiology of DN is considered to be injuries from chronic hyperglycemia and associated glomerular hyperperfusion, which provides the rationale for conventional treatments of DN, including the use of antidiabetic drugs, angiotensin blockers, and sodium-glucose cotransporter-2 inhibitors. 4These therapeutic strategies focus on modifying the progression of overt DN in patients who have already developed proteinuria with the underlying pathology of glomerulosclerosis. 1,5It is difficult to determine when the earliest kidney injury occurs in patients because most diabetic patients experience a long latency period of 5-8 years after the onset of hyperglycemia and before the onset of microalbuminuria, the earliest clinical presentation of DN. [6][7][8][9] However, during this latency period, irreversible chronic pathology develops. 9Therefore, early intervention in diabetic patients at risk of DN holds great promise for improving outcomes.
In diabetes, injuries to podocytes, the epithelial cell component of the glomerular filtration barrier (GFB), develop much earlier than the onset of microalbuminuria. 9Our recent work identified a cytoplasmic dynein-mediated trafficking pathway that disrupts proteostasis at the GFB under diabetic conditions. 10Dynein is a microtubule-based multiunit motor protein complex that mediates "retrograde trafficking" of membrane proteins from cell surface to cytosol.Dynein-driven mistrafficking of GFB protein is caused by hyperglycemia-induced upregulation of dynein subunits, known as "hyperglycemia-responsive dynein subunits."These include dynein heavy chain (Dync1h1), which functions as an ATPase to generate force for cargoes to slide over microtubules; intermediate chain (Dync1i1), which anchors dynein cargoes; light chain (Dynll1), which assembles the entire dynein complex [11][12][13][14] ; and dynactin (Dctn1), which activates dynein.In a high-throughput gene transcriptome study of human DN, 15 synchronized upregulation of these dynein subunits was found to be correlated with albuminuria.9][20] However, there is no streamlined mechanism that directly connects these complicated hyperglycemia-triggered pathways with kidney injury.In this study, we investigated an energy-responsive transcriptional pathway underlying hyperglycemia-induced transcription of dynein genes.
Bioinformatics analysis of the promoter sequences of the hyperglycemia-responsive dynein genes identified a shared sequence motif characterized by 59-GGGGCGGGGC-39 that binds specificity protein 1 (SP1), a two-cysteine, two-histidine zinc-finger family transcription factor (TF), suggesting a unique role for SP1 in mediating synchronous transcription of these dynein subunits.In comparative transcriptome studies, SP1 was identified as a shared TF in both human DN and the eNOS 2/2 db/db mouse model of DN. 21SP1 has been previously reported to play a role in the profibrotic pathway during the progression of mesangial expansion and glomerulosclerosis by promoting the transcription of TGF-b and extracellular matrix. 22However, these genes were less upregulated than the dynein genes in human DN 15 and were not significantly correlated with hyperglycemia in rodent models of DN, 21 suggesting a different mechanism underlying the early pathophysiology of DN.
In this study, we delineated the metabolic pathway responsible for the SP1-guided transcription of dynein and tested whether disrupting this early transcription mechanism can halt the pathogenesis of diabetic podocytopathy and prevent the development of overt DN.

Primers, Antibodies, and Reagents
Primers for chromatin immunoprecipitation (ChIP)quantitative PCR (qPCR), primers for reverse transcription qPCR, small interfering RNA duplex sequences, antibodies, and chemical compounds are listed in Tables 1-5.

Cell Culture and Experiments
The mouse podocyte cell line was maintained in Roswell Park Memorial Institute 1640 medium with 10% FBS, 1% insulin-transferrin-selenium supplement (Gibco, Grand Island, NY), and 50 IU/ml penicillin-streptomycin solution.Primary human podocytes were purchased from Celprogen Biotechnology Company (Celprogen 36036-08), cultured in precoated flasks with Human Podocyte

Live Cell Imaging and Kymograph Analysis
As described in our previous study, 10,14 nephrin endocytosis was induced by incubating podocytes with antinephrin antibody (Santa Cruz sc-376522), followed by an Alexa Fluor 594-conjugated anti-mouse secondary.Time lapse imaging was captured under a Leica SP8 Confocal microscope every 5 seconds for 10 minutes in three independent experiments.Kymograph was generated using the KymographClear plugin of Fiji software.The fraction and the average velocity of the retrograde, anterograde, and static trafficking events were quantified using Kymograph-Direct plugin.

ChIP-qPCR
According to the manufacturer's instructions for the SP1 ChIP kit (Figure 1B), chromatin samples were prepared from sonicated podocyte lysates following formalin-induced crosslinking, and normal rabbit IgG and anti-SP1 were used with A/G protein beads preabsorbed with herring sperm DNA to precipitate DNA/protein complexes.Soluble chromatin was used as an input.SYBR qPCR was conducted to target the promoter regions of dynein genes, and the results are represented as Fold Enrichment52 (Ct IgG2Ct SP1) . 23

Animal Experiment
Eight-week-old C57BL/6J wild-type mice (male) received a single intraperitoneal (i.p.) injection of streptozotocin (STZ) dissolved in 0.05 M sodium citrate buffer (pH54.5) at a dose of 150 mg/kg body weight. 24,25Agematched vehicle control mice received i.p. injections of sodium citrate buffer of the same volume and regimen.Mithramycin (MIT) was administered to both nondiabetic mice and diabetic mice (2 weeks after the STZ injection) at the dosing regimen of 0.25 mg/kg twice weekly3eight doses via i.p. injection. 26

Immunofluorescent Staining and Quantification
In paraffin-embedded mouse kidney sections, dynein subunits, phosphorylated SP1, and nephrin were coimmunofluorescent stained with podocyte markers (WT1, inverted formin 2).Using the Fiji ImageJ analysis tool, the media fluorescent intensity for podocyte-specific levels of nephrin, dynein subunit, and phosphorylated SP1 were quantified for comparison among mice with different treatments.Using the Colo2 plugin of the Fiji ImageJ software, nephrin-Dynll1 colocalization was quantified as the Manders overlap coefficient, as described in our previous work. 10anning Electron Microscopy Each kidney cortex was cut into small cubes, immersed, and fixed in Karnovsky's fixative composed of 1.5% glutaraldehyde and 1.6% paraformaldehyde buffered with 0.1 M sodium cacodylate buffer, pH 7.2.It was then treated with a 1% osmium tetroxide in 0.1 M sodium cacodylate buffer, dehydrated, and dried using a CO 2 critical point dryer.The specimens were then mounted on aluminum scanning electron microscopy (SEM) stubs with carbon tape, sputtercoated with gold and palladium in an Emitech Sputter Coater K550, and imaged using a Hitachi S-4800 SEM.

Statistical Analyses
Data analyses were performed using GraphPad Prism 9. Immunofluorescence signal analysis and densitometric analysis of the immunoblots were performed using the Fiji/ImageJ software.Data are expressed as mean6SEM.One-way ANOVA was used for comparisons among multiple groups, and a post hoc q test was used to compare the differences between groups.In a two-tailed test, P,0.05 is considered significant.

Study Approval
The animal protocol was approved by the Institutional Animal Care and Use Committee of the University of Iowa and was in accordance with the National Institutes of Health guidelines for the use of live animals.

SP1 is the Most Likely TF for the Transcription of Hyperglycemia-Responsive Dynein Genes
Using SwissRegulon tools for regulatory genomics and predicted TF-binding sites, 27 we found that SP1 is the TF with the most common binding motif in the promoters of the hyperglycemia-responsive mouse dynein genes (Dync1h1, Dync1i1, Dynll1, and Dctn1, Figure 1A) and their human orthologs (Supplemental Figure 1A), suggesting that SP1 is a key TF that mediates synchronous transcription of dynein genes in diabetes.

HG Induces Dynein Gene Transcription via an AMPK/SP1-Mediated Mechanism
Using SP1 ChIP followed by qPCR, we demonstrated an enrichment of promoter sequences for Dync1h1, Dync1i1, Dynll1, and Dctn1 immunoprecipitated with SP1 in mouse podocytes grown in HG, compared with cells cultured under normal glucose conditions (Figure 1D).This change corresponded to increased mRNA transcription and protein expression of dynein subunits (Figure 1, F and G).These changes can be reversed by chemical or genetic inhibition of SP1 using MIT, a selective chemical inhibitor that competitively displaces SP1 from its binding sequences.HGinduced transcriptional activation of SP1 was also reflected by the increased nuclear levels of Thr 453 -phosphorylated SP1, 28 relative to the stable level of total SP1 (Figure 1E).
We examined the role of AMPK, a direct cellular energy sensor, in SP1-guided dynein expression induced by HG.First, we found that HG-induced SP1 activation and SP1directed dynein expression could be replicated in cells treated with compound C, a selective inhibitor of AMPK that mimics repressed AMPK in hyperglycemia.By contrast, HG-upregulated dynein genes can be restored by 5-aminoimidazole-4-carboxamide ribonucleoside (AICAR), an activator of AMPK.Altered AMPK activity in cells was confirmed by measuring the Thr 172 phosphorylated to total AMPK-ratio 29 (AMPK-p/AMPK, Figure 1G).Second, SP1directed dynein expression following the direct repression of AMPK by compound C could be rescued by MIT.SP1 knockdown rescued HG-induced dynein overexpression, but it had no effects on dynein expression when AMPK is activated by AICAR (Figure 1, F and G), suggesting SP1mediated transcription of hyperglycemia-responsive dynein subunits is a consequence of AMPK repression.
The AMPK-SP1-regulated transcription of hyperglycemiaresponsive dynein subunits was validated in human podocytes (Supplemental Figure 1, B and C).

Inhibition of SP1-Mediated Dynein Expression Attenuates Podocytopathy and DN in STZ Mice
In a mouse model of STZ-induced type 1 diabetes, we demonstrated SP1 activation in glomerular podocytes by showing increased immunostaining of Thr 453 -phosphorylated SP1 in podocyte nucleus, which was reversed by treating the animals with i.p. injections of MIT (Figure 2A).This corroborates the synchronized upregulation of hyperglycemia-responsive dynein subunits that are colocalizes with inverted formin 2, a dynein regulating protein that is highly selectively expressed in podocyte cytosol (Figure 2D).Diabetic mice developed albuminuria at 3 months (Figure 3C), resulting from podocyte injury characterized by foot process effacement, microvillous transformation, and detachment, as shown under SEM (Figure 3B).Both the disturbed ultrastructure of podocytes and albuminuria were attenuated by administering MIT in STZ mice compared with the treatment control.Quantification of histologic features in the kidneys of STZ mice showed that MIT treatment prevented the development of mesangial expansion and interstitial fibrosis, pathologic features of overt DN (Figure 3D).Nondiabetic mice received MIT injections did not develop albuminuria or obvious kidney pathology.

Restoring the AMPK/SP1 Axis Ameliorated Dynein-Mediated Nephrin Dyshomeostasis in Diabetes
Nephrin forms the slit diaphragm, the molecular sieve of the kidney, via its extracellular domain.Therefore, accurate targeting of nephrin protein to the surface of podocytes is an important functional marker of GFB.Our previous study showed enhanced dynein-mediated mistrafficking in diabetes caused podocytopathy by diverting nephrin from surface trafficking to proteolysis pathways. 10We found that HG treatment increased dynein expression, which subsequently impaired the surface distribution of nephrin in podocytes.These changes can be restored by AICAR, MIT, or small interfering RNA-mediated knockdown of SP1 (Figure 4A).Using live cell imaging, we observed faster and prominent retrograde trafficking of nephrin (from surface to cytosol) in podocytes growing in HG (Figure 4B, tracks in warm colors exhibited by TrackMate).By quantifying the trafficking trajectories in Kymograph (Figure 4C), we demonstrated increased fraction and velocity of the retrograde trafficking events in HG, which could be reversed by MIT.The effect of MIT is similar to that of Ciliobrevin D, a direct dynein inhibitor (Figure 4D).
In vivo, compared with the linear distribution of nephrin in nondiabetic mouse podocytes, we observed a depleted nephrin with segmented distribution in diabetic mouse kidneys (Figure 2B and Supplemental Figure 2), but it is not due to podocytes loss at this early stage of DN.The nephrin dyshomeostasis in diabetic mice corresponded to the enhanced dynein-mediated trafficking of nephrin reflected by increased Dynll1 colocalizing with nephrin (Figure 2C). 10,14These changes were restored by MIT-normalized dynein expression.

Discussion
Our recent transcriptome analysis identified upregulated dynein genes in the early stages of DN, which correlates with the severity of hyperglycemia and kidney injury.The dynein genes encode subunits that are key to the integrity, motor force, cargo loading, and activity of the dynein transport complex, implicating a role for dynein in the pathogenesis of DN.Further mechanistic studies underscored a new dynein-driven mechanism of diabetic podocytopathy, in which synchronous upregulation of dynein subunits in response to hyperglycemia results in enhanced dyneinmediated mistrafficking and dyshomeostasis of nephrin, leading to podocyte injury. 10Investigating how diabetic stresses drive the upregulation of dynein will help us dissect potential therapeutic targets to remodel the pathogenesis of DN.
To understand the transcriptional signature of hyperglycemia-responsive dynein genes underlying early podocyte injury, we identified SP1 as the most likely TF because it binds to the most common motifs in the promoters of all these dynein genes.Using the ChIP assay, we demonstrated increased SP1 enrichment in dynein gene promoters in HG-treated podocytes, leading to their transcription and protein expression.In both cultured podocytes and the STZ-induced type 1 diabetes mouse model with hyperglycemia as the major pathologic factor, we demonstrated activation of SP1 transcription activity and subsequent increased expression of the hyperglycemiaresponsive dynein subunits.Furthermore, we demonstrated the essential role of SP1 in HG-induced dynein gene expression and the resultant dynein-mediated mistrafficking of nephrin because these changes can be reversed by specific gene silencing of SP1 or by MIT, a chemical inhibitor of SP1.
1][32][33] We showed that HG increased Thr 453 phosphorylation of SP1, which activates its transcriptional activity.However, the mechanism by which hyperglycemia induces the SP1-mediated transcription of dynein genes is unknown.][38] Accumulated evidence shows that AMPK, a direct cellular energy sensor, potentially connects diabetes-responsive kinases/pathways [39][40][41] with SP1. 42,43Therefore, we hypothesized that AMPK-regulated SP1 conveys the diabetic metabolic signaling that influences podocyte pathophysiology.By chemically manipulating AMPK activity in cultured podocytes, we demonstrated that direct repression of AMPK replicates the effects of HG in inducing SP1 activation and SP1dependent expression of dynein subunits.By contrast, these HG-induced changes were reversed by reactivating AMPK using AICAR.Moreover, AICAR-induced activation of AMPK rescued surface trafficking of nephrin, which was impaired by HG-upregulated dynein.
The diabetes-responsive SP1-targeting genes in different kidney cells remain to be elucidated.Previous studies in mesangial cells have emphasized the role of SP1 in the transcription of profibrotic genes for mesangial expansion and extracellular matrix production, characterizing the latestage pathology of DN. 22,44 Our study, on the other hand, identified a new set of dynein-encoding genes as key SP1targeting genes responsible for early-stage diabetic podocyte injury and delineated how hyperglycemia induces podocytopathy via the AMPK-SP1-dynein pathway.This pathway directly senses the energy changes caused by hyperglycemia and transduces metabolic signals to enhance the expression of dynein, which promotes dynein-mediated podocyte injury.While inhibitors of dynein subunits are still under pharmacokinetic and pharmacodynamic investigation for in vivo use, our findings offer alternative  molecular targets for the intervention of dynein-driven pathogenesis of diabetic podocyte injury, especially AMPK and SP1.Agonists of AMPK (AICAR and metformin) have been shown to provide protective effects in alleviating podocyte injury in mouse models of type 1 diabetes. 42,45Although there has been no SP1-targeting therapy for DN, its therapeutic potential is supported by genetic epidemiology data implicating its role in the pathogenesis of both type 1 and type 2 diabetic kidney injury.In a meta-analysis of DN gene expression datasets, SP1 was identified as one of the top TFs with differentially expressed genes enriched in DN. 46 A single nucleotide polymorphism was found to alter the sequence of the SP1 binding site in the promoter region of target genes associated with DN among type 1 diabetic patients in the Genetics of Kidneys in Diabetes Study, 47 as well as in type 2 diabetic patients of Tunisian Arab origin. 48Intriguingly, we found that HG-disturbed trafficking of nephrin can be rescued by MIT-inhibited SP1 transcription of dynein, equivalent to the effect of Ciliobrevin D, an antagonist for dynein.MIT and its analogs are pharmaceutically available and have been used to treat human malignancy and non-neoplastic animal models of ischemic neuropathy 49 and myocardiopathy, 50 providing a rationale for using MIT to interfere with SP1promoted dynein expression and activation.
We tested whether inhibiting SP1-directed transcription of dynein can rescue DN in vivo and showed that MIT treatment of STZ-induced diabetic mice indeed rescued the damaged morphology and filtering function of podocytes by preventing upregulated dynein expression, accompanied by an alleviation of albuminuria and late DN pathology, such as mesangial expansion, glomerular fibrosis, and interstitial fibrosis.In addition, the same dosing regimen of MIT did not cause obvious albuminuria or nephrotoxicity in the nondiabetic mice.MIT is an antineoplastic drug used to treat human cancers or hypercalcemia of metastatic bone disease/Paget bone disease. 51,52When used with these indications requiring high or repeated doses, MIT-related nephrotoxicity was reported, especially in patients with baseline kidney disease or predisposed to renal tubular injury. 51Pharmacokinetics study found rapid uptake of MIT by tubular cells, 53 making them more susceptible to toxicity compared with other types of kidney cells.The human equivalent dose of MIT required to counteract dynein-driven diabetic podocytopathy in our mouse model is only 20% of the doses required to treat human cancer.Therefore, fine-tuning the doses and choosing MIT analogs with less toxicity 54 to control the expression of specific genes in certain types of cells may offer a better therapeutic window for SP1-targeting therapy.SP1, a TF with broad targets, likely regulates transcription in gene-specific and cell type-specific manners.While a smaller dose of MIT suppresses the "pathogenic" upregulation of dynein expression and protects podocytes, a higher or toxic dose may impair the "essential" expression of certain SP1targeting genes to maintain the function of tubular cells (for instance, hypoxia inducible factor 1 which regulates the hypoxic adaptation 55 or sphingosine kinase 1 which maintains survival against ischemia/reperfusion injury 56 ).Conversely, the broad targets of SP1 extend the potential of MIT analogs from antineoplastic to anti-inflammatory, antireactive oxygen species and antifibrotic effects in developing new treatments for neurodegenerative disease, ischemic cerebrocardiovascular disease, 49,50,57,58 as well as epithelial-mesenchymal transition after ischemia-reperfusion injury of the kidney. 59Therefore, while exploring the protective efficacy of targeting SP-regulated gene transcription in glomerular disease, it is equally important to understand how SP1 regulates gene expression in renal tubular cells, to optimize the therapeutic window of SP1-targeting strategies.
In summary, our work delineates how the AMPKregulated SP1 pathway leads to diabetes-responsive transcription of dynein genes, provides a better understanding of the initial steps of dynein-driven pathogenesis of DN, and identifies a new molecular target for remodeling dynein-mediated pathology in vivo.In addition to DN, SP1-mediated gene transcription and dynein-mediated pathology have been reported in diabetic complications in various organ systems, 60,61 such as retinopathy 62,63 and neuropathy. 64,65The cause-effect relationship between SP1 and dynein in these organs has not been established in these tissues.By expanding our models from podocytes to other target cells of diabetes, we may provide insights into a potentially common energy-responsive AMPK-SP1dynein-driven pathogenesis of multiple diabetic complications and the value of pharmaceutically remodeling this pathway to manage them.

Figure 1 .
Figure 1.Hyperglycemia induces dynein gene expression via an AMPK/SP1-dependent mechanism.(A) Hyperglycemia-responsive dynein genes share binding motif for SP1 TF.SP1 binding motifs (red) identified in genes encoding mouse dynein subunits, using SwissRegulon tools.The promoter sequences for each individual dynein gene and their GenBank locations are highlighted in orange.ChIP primers (gray) were designed close to the SP1 binding motif and the translation starting sites (blue).(B) Workflow of crosslinking SP1 ChIP.(C) Hypothesis and research design of hyperglycemia-induced dynein gene expression via an AMPK/SP1 axis.Hyperglycemia suppresses AMPK and disinhibits SP1, which subsequently initiates the transcription of dynein subunits.(D) The fold enrichment (52 (Ct IgG2Ct SP1) ) of dynein gene locus sequences immunoprecipitated with SP1 was quantified by ChIP-qPCR.The values were normalized to that of NG.The PCR products were confirmed by running an agarose gel.(E) SP1 activity reflected by the nuclear location of phosphorylated SP1 in podocytes with different treatments, related to the unchanged total SP1 (NG; HG; CC, an AMPK inhibitor; AICAR: an AMPK agonist).Medium containing 0.3% DMSO served as a negative control for chemical intervention.(F) Relative quantification of dynein gene transcription using Gapdh as a housekeeping gene was normalized to that of NG and was compared in a heatmap.(G) Protein levels of representative dynein subunits in podocytes with different treatments were examined using Western blot.The corresponding AMPK activity was expressed as the Thr 172 phosphorylated to total AMPK ratio (AMPK-p/AMPK).The log OD values against the b-actin housekeeping protein were normalized to that of NG. n53, *P , 0.05 versus NG; ^P , 0.05 SP1 siRNA versus control siRNA.AICAR, 5-aminoimidazole-4-carboxamide ribonucleoside; AMPK, AMP-activated protein kinase; CC, compound C; ChIP, chromatin immunoprecipitation; HG, high glucose; NG, normal glucose; OD, optical density; qPCR, quantitative PCR; siRNA, small interfering RNA; SP1, specificity protein 1; TF, transcription factor.

Figure 2 .Figure 3 .
Figure 2. MIT restored SP1-mediated dynein expression and dynein-mediated nephrin homeostasis in STZ-induced diabetic mice.Type 1 diabetes was induced in 8-week-old male C57BL/6J mice by single high dose injection of STZ (150 mg/kg, i.p.).Two weeks after the STZ injection, MIT diluted in NS was given via i.p. injection at the dose of 0.25 mg/kg, twice weekly for a total of eight doses.Vehicle controls received NS injections.Nondiabetic mice that received MIT injections were included in this study to exclude nephrotoxicity of MIT at this regimen.In STZ-induced diabetic mice, MIT restored SP1 activity (reflected by immunostaining of Thr 453 phosphorylated SP1 costained with podocyte nuclear marker WT1 (A) reduced nephrin protein (B) and dynein-mediated mistrafficking of nephrin (reflected by increased Dynll1 colocalizing with nephrin, C), and the upregulated expression of dynein subunits (costained with podocyte cytosol marker INF2, D).Mean fluorescence intensity of dynein subunits per podocyte was quantified for comparison.n515 (three glomeruli or three interstitial areas per section3five mice).*P , 0.05 versus nondiabetic mice, ^P , 0.05 versus STZ-induced diabetic mice treated with NS.Scale bar: 20 mm.INF2, inverted formin 2; i.p., intraperitoneal; MIT, mithramycin; NS, normal saline; STZ, streptozotocin.

Figure 4 .
Figure 4. HG impaired nephrin proteosis via AMPK/SP1-regulated dynein expression.(A) Coimmunostaining of nephrin and Dynll1 in podocytes cultured under different conditions: NG (NG10.3%DMSO); HG (HG10.3%DMSO); HG in the presence of MIT (0.1 mM MIT); or AICAR (0.5 mM).(B-D) Nephrin expressed on podocyte surface underwent crosslink-induced endocytosis induced by fluorophore-labeled antinephrin.The postendocytic trafficking of nephrin in HG-cultured podocytes in the presence of MIT or Ciliobrevin D (50 mM) versus control cells (HG1DMSO) was visualized using live cell imaging and analyzed using the TrackMate and KymographClear plugins of Fiji/ImageJ software.With TrackMate (B), faster retrograde trafficking of nephrin (from surface membrane to cytosol) is displayed in warmer colors, and slower tracks are displayed in cooler colors.In the Kymograph (C), the retrograde, static, and anterograde (from cytosol to surface membrane) tracking events are displayed in red, blue, and green.(D) The fraction and velocities of the trafficking events were quantified using KymographDirect software.*P , 0.05 versus HG1DMSO.

Table 2 .
Primers for RT-quantitative PCR

Table 3 .
Small interfering RNA duplex sequences