Cyclosporine A inhibits MRTF-SRF signalling through Na+/K+ ATPase inhibition & Actin remodelling

Calcineurin Inhibitors (CNI) are the pillars of immunosuppression in transplantation. However, they display a potent nephrotoxicity whose mechanisms remained widely unsolved. We used an untargeted quantitative proteomic approach (iTRAQ technology) to highlight new targets of CNI in renal proximal tubular cells (RPTCs). CNI-treated RPTCs proteome displayed an over-representation of Actin-binding proteins with a CNI-specific expression profile. Cyclosporine A (CsA) induced F-Actin remodelling and depolymerisation, decreased F-Actin-stabilizing, polymerization-promoting Cofilin (CFL) oligomers and inhibited the G-Actin-regulated serum responsive factor (SRF) pathway. Inhibition of CFL canonical phosphorylation pathway reproduced CsA effects; however, Ser3, an analogue of the phosphorylation site of CFL prevented the effects of CsA which suggests that CsA acted independently from the canonical CFL regulation. CFL is known to be regulated by the Na+/K+-ATPase. Molecular docking calculations evidenced 2 inhibiting sites of CsA on Na+/K+-ATPase and a 23% decrease in Na+/K+-ATPase activity of RPTCs was observed with CsA. Ouabain, a specific inhibitor of Na+/K+-ATPase also reproduced CsA effects on Actin organization and SRF activity. Altogether, these results described a new original pathway explaining CsA nephrotoxicity.


Introduction
The development of immunosuppressive regimens based on the Calcineurin Inhibitors (CNI), namely Cyclosporine A (CsA) and Tacrolimus (Tac), has been a breakthrough in the prevention of allograft rejection in solid organ transplantation (Calne et al, 1978(Calne et al, , 1979Starzl et al, 1989). Although CNI are now widely used in clinical protocols of immunosuppression (Hart et al, 2016(Hart et al, , 2017 and have significantly improved short-term graft and patient survival, long-term exposure is associated with major limiting deleterious side effects such as nephrotoxicity, leading to end-stage renal disease (Myers et al, 1984).
These observations resulting from targeted experimental approaches only partially described CNI cell toxicity and do not allow a complete understanding of all the pathophysiological mechanisms at stake. Because the mechanisms of CNI side effects remain widely unsolved by targeted strategies, the design of untargeted experiments is of utmost importance to gain new insights and improve knowledge about the pathophysiology of CNI proximal tubule cell (PTCs) toxicity.
Omics based on the high-throughput analysis of biological systems such as Shotgun proteomics seem particularly well suited to this purpose. In the present study, we performed the quantitative proteomic analysis and dynamic mapping of CNI-exposed PTC proteome to elucidate new intracellular pathways specifically modified under CNI exposure.
Upon the significant over-representation and differential expression of Actin family cytoskeletal proteins, we focused on the deciphering of the intracellular mechanisms of CsAinduced reorganization of the Actin cytoskeleton of PTC and its downstream consequences.
We showed that CsA induced an inhibition of the Actin-dependent Myocardin-Related Transcription Factors-Serum Response Factor (MRTF-SRF) transcription activity through an original regulation of Cofilin (CFL) by the Na + /K + -ATPase.

Quantitative proteomic analysis of PTC proteome & mapping of CNI-induced perturbations
highlight the over-representation and differential expression of Actin family cytoskeletal proteins.
iTRAQ technology allowed the identification of 130 proteins in PTC lysate. The PANTHER over-representation test (Table 1)  CsA, but not Tac, elicited a strong reorganization of Actin cytoskeleton of PTCs.
In fluorescence microscopy, cell monolayers of proximal tubular LLC PK-1 showcased a puzzled pattern as phalloidin-labelled F-Actin cytoskeleton organized cell membranes into intercellular convolutions (Figure 1 A, panel a). These peculiar structuring of plasma membranes were reminiscent of the way basolateral membranes connect inside proximal tubular epithelia, in vivo.
Upon 24-hour CsA exposure, LLC PK-1 sustained a significant reorganization of cortical Actin cytoskeleton (Figure 1 A,  On the contrary, CsA effects seemed independent of Calcineurin inhibition since neither Tac, VIVIT (a specific inhibitor of NFAT dephosphorylation by Calcineurin) nor GPI 1046 (a pharmacological inhibitor of FKBP12) exhibited modifications of the Actin organization ( Figure 1A panels c, d and f, Figure 1B).
In conclusion, these observations suggested that CsA triggered the deep remodelling of proximal tubular Actin cytoskeleton and the depolymerization of F-Actin, in a Cyclophilindependent way, independently of Calcineurin inhibition. Tac and VIVIT exposure. Profiles of drug-related Actin dynamics strictly superimposed to profiles of drug-related SRF transcription activity.

Actin depolymerization led to a decrease in MRTF-SRF transcription activity in LLC
In conclusion, CsA downregulated the transcription of MRTF-SRF target genes, in a Cyclophilin-dependent way, independently of Calcineurin inhibition.

CsA regulates Actin-binding protein Cofilin through changes in Cofilin-Actin ratio and oligomerization state
Differential quantitative proteomic analysis of identified Actin family cytoskeletal proteins (PANTHER Protein Class PC00041) showed that CsA decreased the CFL: Actin ratio since Actin was overexpressed while CFL levels remained steady. On the contrary, Tac induced equivalent overexpression for both Actin and CFL ( Figure EV1).
In conclusion, CsA elicited a shift in the balance between oligomeric forms of CFL, decreasing the protein abundance of CFL dimers and tetramers.

CsA effects seemed partly independent of CFL phosphorylation/dephosphorylation
The S3-R peptide, an analogue of the CFL phosphorylation site and a non-specific inhibitor of Ser3-targeting proteins, did not affect the organization of the Actin cytoskeleton In conclusion, CsA effects involved the phosphorylation site of CFL as a target for dephosphorylation. Inhibitors of the RhoGTPases pathway triggered CsA-like effects.
CsA inhibited Na + /K + -ATPase activity in PTCs and Ouabain-induced inhibition of Na + /K + -ATPase mimicked Actin reorganization and inhibition of MRTF-SRF transcription activity observed upon CsA exposure A significant decrease of the Na + /K + -ATPase activity (-23.1 % of control, p<0.01**) was observed in LLC PK-1 exposed for 24 h to CsA (Figure 6 A) whereas Tac treatment had no effect.
LLC PK-1 exposed for 24 h to 100 nM Ouabain featured CsA-like Actin reorganization In conclusion, CsA targeted Na + /K + -ATPase and inhibited its activity. Na + /K + -ATPase inhibitor Ouabain elicited Actin reorganization the same way CsA did. Ouabain-related Actin dynamics induced the partial inhibition of MRTF-SRF transcription activity. The association of Ouabain with CsA tended to potentiate CsA effects on tubular proximal morphology and Actin dynamics.

CsA molecular docking into Na + /K + -ATPase
CsA was docked into the closed (E1) conformation of Na + /K + -ATPase with and without ATP and into the open (E2) conformation without any ligands ( Figure EV3). All the obtained binding poses had similar binding affinity, ranging from -7.7 to -6.6 kcal.mol-1. Assuming the known membrane partitioning of CsA (Haynes et al, 1985;Wang et al, 2018), only the binding poses close to the membrane boundary were considered.
On the open conformation of Na + /K + -ATPase (state E2, Figure  In the closed conformation of Na + /K + -ATPase (state E1, Figure EV3 B and C), the most prominent binding sites are at the lipid-binding sites as previously defined (Cornelius et al, 2015). Most of the binding poses are located at the so-called site C -a position of bound cholesterol and phospholipid, involving aromatic and aliphatic residues such as W32 and Y39 or V838. This site has been previously implicated in the Na + /K + -ATPase inhibition.

Discussion
In this work, we demonstrated that CsA-induced remodelling of the Actin cytoskeleton of PTC was associated with a modification of the F-Actin: G-Actin ratio through an original regulation of CFL: Indeed, CsA exposure decreased the total CFL: Actin ratio as well as the levels of CFL dimers and tetramers, independently of the phosphorylation by LIM kinases but most likely related to the CsA-induced inhibition of Na + /K + -ATPase.
In the kidney, the study of podocytes exposed to CsA revealed that the dynamic of the Actin cytoskeleton was sensitive to CsA. Indeed, CsA exerted the stabilization of the Actin cytoskeleton of podocyte foot processes, explaining the anti-proteinuric effect of CsA which is observed in nephrosis. This mechanism was demonstrated to be Calcineurin-dependent with the inhibition of the dephosphorylation and consequent proteolysis of Synaptopodin, a podocyte specific Actin-binding protein (ABP) whose activity is essential for podocyte structure-function (Faul et al, 2008). Besides, CsA induced the overexpression of CFL, a ubiquitary ABP, and its regulation in a non-phosphorylated state, independently from the effects on Synaptopodin (Li et al, 2014).
In the light of early observations of CsA effects on the Actin cytoskeleton of podocytes, the study of the Actin dynamic in PTC appeared of utmost interest in the elucidation of pathophysiological mechanisms of tubular atrophy upon CNI exposure. In PTC, we demonstrated that the Actin cytoskeleton could be modified by CsA (Descazeaud et al, 2012). Indeed, independently from the inhibition of NFAT, CsA triggered a reversible disorganization of Actin scaffolding at the periphery of the cell. Whether a mechanism involving the CsA-related regulation of CFL was responsible for the effects of CsA on PTC remained to be addressed.
CFL was initially described as an Actin-depolymerizing factor involved in Actin treadmilling. CFL modifies the physical-chemical properties of the microfilaments, promotes Pi release and ADP-Actin-G dissociation from older segments to replenish the pool of G-Actin and to supply enough material for microfilament renewal. Since then, the scope of CFL activity was expanded, in the light of the multiple mechanisms of CFL regulation. Up to date, CFL activity is known to be sensitive to F-Actin saturation, phosphorylationdephosphorylation of its Ser3 residue, PIP2 interaction, intracellular pH and oxidative stress (Vantroys et al, 2008).
Since direct interaction with F-Actin microfilaments is necessary for the Actin-related activity CFL, the degree of saturation of the microfilament (or the density of CFL near the microfilament) play a major role in the regulation of CFL. Saturation is correlated to global CFL concentration, CFL: Actin ratio and CFL oligomerization. At high saturation and CFL density, F-actin filaments are severed and prone to the nucleation and polymerization of new branches. Conversely, at low saturation and CFL density, F-actin filaments are severed and completely depolymerized into G-Actin. Low CFL concentrations are related to low saturation, and so, the higher the concentration, the higher the saturation of microfilaments (Andrianantoandro & Pollard, 2006;Yeoh et al, 2002). Besides, in vivo and in vitro, CFL may exist as monomers, dimers and tetramers with higher-order oligomers related to higher CFL concentration and F-Actin saturation (Pfannstiel et al, 2001;Goyal et al, 2013).  2003). In that way, the Actin dynamic was described as a potent intracellular fulcrum for cell adaptation to its environment, connecting upstream RhoGTPases to downstream MRTF-SRF pathway (Hill et al, 1995;Sotiropoulos et al, 1999;Maekawa, 1999).
As mentioned above, studies on podocytes reported that CsA had anti-proteinuric effects via the stabilization of the Actin cytoskeleton structuring the foot processes. This stabilization relied on the overexpression and dephosphorylation of Cofilin. Even though the study did not focus on CFL oligomerization, the CFL: Actin ratio was in favour of CFL and the CFL concentration was higher, therefore the activity of CFL was regulated to promote Actin polymerization and stabilize microfilaments. What we observed in PTC was a destabilization Outside Actin-related functions, CFL is known to interact with Na + /K + -ATPase to provide energetic fuel to the pump motion. CFL binds triose phosphate isomerase and interacts with the alpha subunit of Na + /K + -ATPase to locally provide ATP. When Na + /K + -ATPase is inhibited (by potassium depletion or the action of a pharmacological inhibitor), a feedback mechanism disrupts energetic supply by CFL dephosphorylation (Jung et al, 2002(Jung et al, , 2006a(Jung et al, , 2006b. CsA effects on TCP were reminiscent of OUA effects on HeLa cells (Jung et al, 2006a).
OUA exerted a similar reorganization of the Actin cytoskeleton by dephosphorylating CFL via a mechanism where the inhibition of Na + /K + -ATPase activates the Ras/Raf/MEK cascade, in a Src-EGFR-dependent way, leading to the inhibition of LIMK downstream the small RhoGTPases pathway. OUA effects were reproduced by Y27632 just like CsA effects were mimicked by the RhoGTPases inhibitors.
Earlier findings about the mechanism of Na + /K + -ATPase energetic supply by pCFL-TPI complexes (Jung et al, 2002) reported that the overexpression of constitutively active CFL (Ser3 was replaced by Ala) was sufficient to cut endogenous CFL out of phosphorylation/dephosphorylation-based regulation which made TPI localization insensitive to the activation/inhibition of the RhoGTPases pathway. Likewise, in our study, the addition of S3-R was enough to partially exclude endogenous CFL from CsA-induced regulation.
Dephosphorylation of CFL by Slingshot phosphatases seemed unlikely since Slingshot phosphatases are activated by Calcineurin. CsA effects on CFL were unlikely to be related to Calcineurin since Tac and VIVIT did not elicit CsA-like effects on the Actin cytoskeleton or the MRTF-SRF transcription activity.
Numerous studies have already reported the inhibition of Na + /K + -ATPase by CsA, in vitro and in vivo. However, the observations were limited to the consequences in ion transport (Ihara et al, 1990;Tumlin & Sands, 1993;Lea et al, 1994;Ferrer-Martínez et al, 1996;Deppe et al, 1997;Younes-Ibrahim et al, 2003;Marakhova et al, 1998Marakhova et al, , 1999. Although this work is not the first one to study the relationship between the Na + /K + -ATPase pump and the Actin cytoskeleton, the partial inhibition (about 25%) of proximal tubular Na + /K + -ATPase by CsA has been the first observation of a direct correlation between variations in the pump activity and cytoskeletal remodelling so far. Besides, the potentiation of the CsA-induced depolymerization of F-Actin by OUA is in favour of a Na + /K + -ATPase -related mechanism.
How CsA inhibits Na + /K + -ATPase has yet to be elucidated. CsA is known to induce overexpression of endothelin, which is a well-known inhibitor of Na + /K + -ATPase activity (Zeidel et al, 1989;Nakahama, 1990). Furthermore, CsA can downregulate Na + /K + -ATPase activity through the inhibition of Cyclophilin B. The PPIase was described as a crucial partner for the structure and activity of Na + /K + -ATPase catalytic subunit ( To what extent the Cyps are implicated has yet to be elucidated. However, our work suggests another mechanism leading to Na + /K + -ATPase inhibition. Indeed, our molecular modelling of CsA docking into Na + /K + -ATPase was performed to visualise the potential interactions. CsA is mostly made of hydrophobic amino acids favouring its membrane partitioning (Haynes et al, 1985). Therefore, the preferential binding site to Na + /K + -ATPase is likely to be in the transmembrane site C, known to bind hydrophobic compounds such as lipids and cholesterol. Such binding is mostly expected to hinder proper motions of the transmembrane helices by impairing specific protein-lipid interactions. This may explain the experimentally observed inhibition of Na + /K + -ATPase by CsA given that the Trypsin-EDTA. LLC PK-1 cells were cultured between passage 7 and passage 30.

Actin cytoskeleton characterization
Immunocytochemistry Serum-starved, hormonally-defined, treated LLC PK-1 cells seeded on glass cover slips in 6-well plates were washed twice for 3 min in PBS. Then, cells were fixed in 4 % paraformaldehyde-PBS for 10 min, at room temperature. Cells were washed once again in PBS for 3 min then permeabilised with 0.1 % Triton/X-100-PBS for 5 min, at room temperature. Permeabilised cells were washed three times for 3 min in PBS before incubation in Phalloidin-TRITC-PBS for 30 min, at 37 °C, in the dark. Eventually, cover slips were washed three times for 3 min in PBS then mounted on glass microscope slides using ProLong® Gold Antifade Reagent as a mounting medium and a DAPI nuclei staining.

Image analysis
For each exposure condition, ten photographs (*.tiff) were randomly taken to report from an independent experiment. Image analysis was conducted using the ImageJ software (v.1.48). Early image processing consisted in a restriction to the green channel of the RGB picture. The implemented Auto Local Threshold tool further selected the fluorescencepositive pixels (Niblack thresholding method, radius = 15). Thus, for each photograph, a semi-quantitative ratio was calculated as a percentage of the total image area.
All above-mentioned steps were automatized thanks to the macro editing tool of the ImageJ software.
Statistical analysis of fluorescence area data was performed using the one-way analysis of variance (1-way ANOVA) test followed by the Bonferroni's Multiple Comparison post-test with a significance threshold at p<0.05, as implemented in GraphPad Prism (v. 5.04).
Transfected LLC PK-1 cells (LLC PK-1 SRE) were incubated in routine medium for 72 h before antibiotic selection to allow proper plasmid integration. Then, cells were re-suspended in 0.05 % Trypsin-EDTA, 10 min at 37 °C. Cell suspensions were diluted 10 times and reseeded in routine medium completed with 1 % Hygromycin B (50 mg.mL-1) refreshed every 48 h until colonies appear. Colonies were harvested thanks to trypsin-soaked paper disks (#Z374431, Sigma-Aldrich) and re-seeded to start routine culture in antibiotic-supplemented growth medium.
Bioluminescence signals were detected by the Enspire Multimode Reader (Perkin-Elmer).

Oligomer cross-linking and Western blot
Serum-starved, hormonally-defined, treated LLC PK-1 cells in 60 mm Petri dishes were incubated in hormonally-defined medium -

PANTHER Overrepresentation test
The protein list analysis was performed by submitting Swiss-Prot accession IDs to the online tool powered by the PANTHER Classification system. The PANTHER Overrepresentation test (release date 20170413) parsed the PANTHER database (version 12.0, released on 2017-07-10) using the Sus scrofa reference list and the PANTHER Protein Class annotation data set. Only p<0.05 items were retained and considered significantly over-represented.

Visualisation of protein networks
The visualisation of protein networks was performed by submitting Swiss-Prot accession IDs to the online tool STRING.

Biological significance criteria
Frequency distribution of the iTRAQ peak intensities ratios could be approximated as a Gaussian distribution: iTRAQ peak intensities ratios frequency distribution significantly fitted implemented Gaussian non-linear regression (n = 370, R² = 0.9946, mean = 1.02 ± 0.10).
One of the Gaussian distribution properties is: I) 66% of the values lie in a 1 S.D. range around the mean, II) 95 % of the values lie in a 2 S.D. range around the mean, III) 99 % of the values lie in a 3 S.D. range around the mean. Applying the "mean ± n S.D." property to the set of iTRAQ peak intensity ratios, the upper thresholds for significant upregulations upon drug exposure would be "1.02 -0.10 = 0.92", "1.02 -2 x 0.10 = 0.82" and "1.02 -3 x 0.10 = 0.72" while the lower thresholds for significant downregulations upon drug exposure would be "1.02 + 0.10 = 1.12", "1.02 + 2 x 0.10 = 1.22" and "1.02 + 3 x 0.10 = 1.32". Thus, proteins were classified thanks to their distance from the unity: proteins with median ratios between 0.92 and 1.12 were annotated as non-impacted proteins; up-regulated proteins with quantitative ratios beyond 1.12 were ranked according to three tiers of significant fold

Hierarchical clustering & heat-map representation
Heat map representation and hierarchical clustering were generated using the Euclidian distance calculation as provided by the GENE-E software (version 3.0.204).

Data availability
The MS proteomics data have been deposited to the ProteomeXchange Consortium reviewer72095@ebi.ac.uk, password: TPSGICw9).

Setup of cyclosporine docking into Na + /K + -ATPase
A cyclosporine structure was docked into a human-sequence homology model of Na + /K + - T h e L a n c e t 3 1 2 : T h e L a n c e t 3 H a n S W , A m . J . T r a n s p l a n t .
T r a n s p l a n t . T r a n s p l a n t . P r o c . T r a n s p l a n t a t i o n 6 0 : 7 9 4 -7 9 8 J  e  n  n  i  n  g  s  P  ,  K  o  p  p  e  l  s  t  a  e  t  t  e  r  C  ,  A  y  d  i  n  S  ,  A  b  b  e  r  g  e  r  T  ,  W  o  l  f  A  M  ,  M  a  y  e  r  G  &  P  f  a  l  l  e  r  W  (  2  0  0  7  )  C  y  c  l  o  s  p  o  r  i  n  e  A  i  n  d  u  c  e  s  s  e  n  e  s  c  e  n  c  e  i  n  r  e  n  a  l  t  u  b  u  l  a  r  e  p  i  t  h  e  l  i  a  l  c  e  l  l  s  .   A  J  P  R  e  n  .  P  h  y  s  i  o  l  .   2  9  3  :  F  8  3  1  -F  8  3  8   J  u  n  g  J  ,  K  i  m  M  ,  C  h  o  i  S  ,  K  i  m  M  -J  ,  S  u  h  J  ,  C  h  o  i  E  C  &  L  e  e  K  (  2  0  0  6  a  )  M  o  l  e  c  u  l  a  r  m  e  c  h  a  n  i  s  m  o  f  c  o  f  i  l  i  n  d  e  p  h  o  s  p  h  o  r  y  l  a  t  i  o  n  b  y  o  u  a  b  a  i  n  . C e l l . S i g n a l .  o  r  r  o  w  T  ,  G  a  f  f  n  e  y  M  M  ,  S  l  a  t  t  e  r  y  C  ,  C  a  m  p  b  e  l  l  E  &  R  y  a  n  M  P  (  2  0  0  5  )  C  y  c  l  o  s  p  o  r  i  n  e  A  i  n  d  u  c  e  d  e  p  i  t  h  e  l  i  a  l  -m  e  s  e  n  c  h  y  m  a  l  t  r  a  n  s  i  t  i  o  n  i  n  h  u  m  a  n  r  e  n  a  l  p  r  o  x  i  m  a  l  t  u  b  u  l  a  r  e  p  i  t  h  e  l  i  a  l  c  e  l  l  s  .   N  e  p  h  r  o  l  .  D  i  a  l  . T r a n s p l a n t .  i  n  o  d  a  T  ,  O  g  a  w  a  H  ,  C  o  r  n  e  l  i  u  s  F  &  T  o  y  o  s  h  i  m  a  C  (  2  0  0  9  )  C  r  y  s  t  a  l  s  t  r  u  c  t  u  r  e  o  f  t  h  e  s  o  d  i  u  m  -p  o  t  a  s  s  i  u  m  p  u  m  p  a  t  2  .  4  Å  r  e  s  o  l  u  t  i  o  n  .   N  a  t  u  r  e   4  5  9  :  4  4  6  -4  5  0   S  i  s  s  o  n  T  H  ,  A  j  a  y  i  I  O  ,  S  u  b  b  o  t  i  n  a  N  ,  D  o  d  i  A  E  ,  R  o  d  a  n  s  k  y  E  S  ,  C  h  i  b  u  c  o  s  L  N  ,  K  i  m  K  K  ,  K  e  s  h  a  m  o  u  n  i  V  G  ,  W  h  i  t  e  E  S  ,  Z  h  o  u  Y  ,  H  i  g  g  i  n  s  P  D  R  ,  L  a  r  s  e  n  S  D  ,  N  e  u  b  i  g  R  R  &  H  o  r  o  w  i  t  z  J  C  (  2  0  1  5  )  I  n  h  i  b  i  t  i  o  n  o  f  m  y  o  c  a  r  d  i  n  r  e  l  a  t  e T h e L a n c e t 3     B. Quantification of relative protein abundance of CFL oligomeric forms. Mean ± SEM.
D. SRF transcription activity was measured by luciferase gene reporter assay in LLC PK-1 SRE. One-sample t-test for versus control comparison, One-way ANOVA plus   Class annotation data set. Bonferroni correction was applied.