Ninety-one years ago insulin was discovered, which was one of the most important medical discoveries in the past century, transforming the lives of millions of diabetic patients. Initially insulin was considered only important for rapid control of blood glucose by its action on a restricted number of tissues; however, it has now become clear that this hormone controls an array of cellular processes in many different tissues. The present review will focus on the role of insulin in the kidney in health and disease.
- diabetic nephropathy
- intracellular signalling
- metabolic syndrome
OVERVIEW OF CELLULAR INSULIN SIGNALLING
The role of insulin in the human body has been an active subject of interest since the discovery of insulin in 1921 by Banting, Best, Collip and Macleod. The critical importance of this finding was recognized by the Nobel Committee in 1923 when they awarded Banting and Macleod the Nobel Prize in Physiology or Medicine  just 2 years after their discovery. In the 50 years that followed the effects of insulin were intensely studied and revealed its glucose-controlling effects focusing on the liver, muscle and adipose tissue . Insulin is a highly potent physiological anabolic hormone that promotes the synthesis and storage of lipids, carbohydrates and proteins, while also inhibiting their degradation and release back into the circulation. In mammals insulin is the main hormone controlling blood glucose; it achieves this by stimulating glucose influx and metabolism in muscles and adipocytes, and by inhibiting gluconeogenesis by the liver. These tissues have always been considered the classically insulin-sensitive organs of the body. However, insulin has the ability to modify the expression and/or activity of an assortment of enzymes and transport systems in a wide variety of cell types , as the present review will describe.
INSULIN AND IGF (INSULIN-LIKE GROWTH FACTOR) RECEPTORS
When discussing the receptors that insulin can signal through it is important to consider another closely related collection of hormones, namely the IGF family. The reason for this is that the IGF hormones, IGF-I and IGF-II, have structural similarity to insulin and their major functional receptor, IGF-IR (IGF-I receptor) [4,5], is also structurally similar to the IR (insulin receptor). The significance of this is that insulin can signal through the IGF-IR and likewise IGF-I/-II can signal via the IR, although with differing affinities. Indeed, it is even more complicated than this as hybrid receptors are formed, by combinations of the IR and IGF-IR, through which all of the hormones can signal but with differing affinities (Table 1). Insulin has the greatest affinity for the IR, so the rest of the present review will predominantly focus on this receptor.
The IR in humans is located on chromosome 19 and is encoded by a gene containing 22 exons and 21 introns spanning 120 kb [6,7]. It is a heterotetrameric receptor consisting of two α and two β subunits , which are linked by disulfide bonds in a β-α-α-β configuration (Figure 1). The α subunits are extracellular and have the insulin-binding domain, whereas the β subunits have three compartmental domains: extracellular, transmembrane and cytosolic domains. Tyrosine residues in the cytosolic domain of β subunits are involved in signal transduction and are auto-phosphorylated when insulin binds to the receptor or by exogenous tyrosine kinase activity .
There is a further level of complexity within the IR as it exists in two different isoforms, A and B, which are formed due to the inclusion or exclusion of exon 11 of the IR gene [9–11]. IR-A lacks exon 11, whereas IR-B includes it. IR-A is widely expressed throughout the body but is importantly up-regulated during pre-natal development and when cells become cancerous . IR-B is expressed largely in the classically insulin-sensitive tissues of liver, skeletal muscle and adipose tissue. Interestingly IR-B is also expressed in the kidney [12,13]. The IR isoforms dimerize and can form either ‘pure’ or ‘hybrid’ receptors with each other or the IGF-IR. The receptor make-up dictates the affinity of the cell for insulin and/or the IGF ligands, as the different receptors have differing affinities for each of these molecules (Table 1). It should also be noted that IRs are not solely located in glucose-regulating insulin target tissues, but in many other tissue types, suggesting other functional roles of insulin signalling in multiple biological systems distinct from glucose homoeostasis.
The IR and IGF-IR mediate the actions of IGF-I, IGF-II and insulin. The IGF-IR shares a high degree of homology with the IR [14,15] (Figure 1). It is therefore unsurprising that insulin is capable of activating the IGF-IR and vice versa. IGF-I has the greatest affinity for the IGF-IR, followed by IGF-II, with insulin having a 500-fold lower affinity in comparison with its primary ligands .
CELLULAR INSULIN SIGNALLING PATHWAYS
The majority of work in this field has been performed on adipocytes, liver and skeletal muscle, as these are crucial for post-prandial glucose regulation in response to insulin.
The insulin signal transduction pathway is highly conserved and responsible for the regulation of a number of aspects of cellular physiology, most notable of which is the metabolic effects of glucose uptake and its utilization within the cell. Following a meal, increased levels of insulin encourage enhanced glucose uptake, metabolism and storage within muscle and adipose cells . Insulin levels rapidly increase approximately 10-fold after a meal from a basal level of approximately 50 pmol to 600 pmol . GLUTs (glucose transporters) are energy-independent and allow glucose to enter or leave the cell, passively down a concentration gradient, when they are incorporated into the cell membrane. The classic insulin-responsive glucose transporter is GLUT4 , which translocates from a cytoplasmic vesicular pool to the plasma membrane in response to insulin. This is the signature molecule of rapidly insulin-sensitive cells that absorb glucose. However, there is also robust evidence that GLUT1  can also translocate in a similar manner from an intracellular pool to the plasma membrane and rapidly increase its plasma membrane concentration in response to insulin. GLUT1 is also a constitutional transporter in many cells . Here it sits at the plasma membrane of cells continuously and allows a constant delivery of glucose for cellular function.
The IR differs from many other receptor tyrosine kinases in that, instead of recruiting downstream effector molecules to its phosphorylated cytoplasmic domains, when activated it phosphorylates a number of scaffolding proteins which then in turn are responsible for recruiting various downstream effector proteins . A number of intracellular substrates have been discovered, including the IRS (insulin receptor substrate) family (IRS1–IRS4), IRS5/DOK4 (downstream of kinase 4), IRS/DOK5, Gab1, Cbl, APS [adaptor protein with PH (pleckstrin homology) and SH2 (Src homology 2) domains] and Shc isoforms, and SIRP (signal regulatory protein) family members [22,23]. The best characterized have been the IRS family of proteins . IRS proteins do not possess intrinsic catalytic activities, and are instead composed of multiple interaction domains and phosphorylation motifs. Four IRS proteins have been identified (IRS1–IRS4), with IRS1 and IRS2 being the most widely expressed. Each IRS protein has the distinct characteristics of an N-terminus PH domain adjacent to a PTB (phosphotyrosine-binding) domain, ending in a variable length C-terminus. The C-terminal tail of each IRS protein contains tyrosine phosphorylation sites that serve as on/off switches and recruit the downstream signalling proteins. IRS1 and IRS2 have the longest tail therefore providing them with a greater number of possible phosphorylation sites (20) in comparison with IRS3 and IRS4 . The IRS proteins form an important node of control for the regulation of insulin and IGF signal transduction in cells.
When the intrinsic tyrosine kinase activity of the receptor is triggered by insulin binding, three major signalling pathways have been described that are propagated in response: (i) CAP (Cbl-associated protein), (ii) the PI3K (phosphoinositide 3-kinase) pathway, and (iii) the MAPK (mitogen-activated protein kinase) pathway (Figure 2). The PI3K pathway is one of the best characterized downstream effectors of the IRS proteins and activates many of the metabolic functions of insulin. Association of the p85 regulatory subunit with IRS proteins leads to the activation of further downstream molecules, including Akt substrates (which form another functional node in the pathway) and mTOR (mammalian target of rapamycin) [25,26], eventually resulting in PI3K being targeted to the plasma membrane . There is evidence that mTOR is important for kidney function and this will be discussed in more detail later in the present review.
The stimulation of glucose uptake by insulin is mediated by PI3K-dependent and PI3K-independent pathways, which play a vital role in the translocation of GLUT4. This is highlighted by use of the PI3K inhibitor wortmannin, which is able to completely block the uptake of glucose into cells upon insulin stimulation . However, despite the critical role of PI3K in insulin-stimulated glucose uptake, activation of at least a second pathway distinct from PI3K is also necessary . This is evident from a number of studies that have examined different elements of the pathway during insulin signalling. By overexpressing a constitutively active membrane-bound form of Akt in 3T3L1 adipocytes, glucose transport and GLUT4 translocation increases in the absence of insulin ; conversely, the insulin-stimulated translocation of GLUT4 is inhibited by the expression of a dominant-negative Akt mutant . These results indicate that Akt is required for insulin signalling. However, if the PI3K pathway is activated by factors other than insulin, such as PDGF (platelet-derived growth factor) or IL (interleukin)-4, although these factors can robustly activate PI3K and Akt, they do not possess insulin's ability to stimulate GLUT4 translocation and glucose uptake . This suggests that other pathways also need to be activated by insulin to elicit an effect. A number of studies have suggested that a separate signalling pathway exists for the IR in microdomains within the cell, such as lipid rafts . It has been proposed that insulin can also activate the GTPase TC10, via lipid-raft localization of the CAP–Cbl–Crk complex and the guanine-nucleotide-exchange factor C3G , and initiate glucose uptake in cells. This process occurs independently of PI3K and has been shown to be crucial to insulin-stimulated GLUT4 translocation [32,33]. However, in contrast with this, findings by Mitra et al.  have reported that Cbl/CAP isoforms are in fact not required components of insulin signalling to GLUT4 transporters; therefore the precise role of this pathway remains contentious.
The final pathway through which insulin can act is the MAPK pathway. This pathway controls a range of cellular activities, including differentiation, proliferation, transformation, survival and death [35–37]. The mammalian MAPK family consists of p38, ERK (extracellular-signal-regulated kinase) and JNK (c-Jun N-terminal kinase), each of which exist in a number of isoforms: p38-α, -β, -γ and –δ, ERK1–8 and JNK1–3 [35,38]. In terms of insulin signalling, the MAPK pathway is activated following the binding of Grb2 (growth-factor-receptor-bound protein 2) and SOS (Son of Sevenless) to phosphotyrosine residues on Shc and Gab1 . Phosphorylation of certain tyrosine residues on Gab1 are required for binding to and activation of the protein tyrosine phosphatase SHP2 (SH2 domain-containing protein tyrosine phosphatase 2) [40,41], whereas phosphorylation of certain tyrosine residues on Shc allow the binding of Grb2/SOS to these sites . This binding initiates activation of the GTPase Ras, followed shortly by Raf, leading to a kinase cascade resulting in the phosphorylation and activation of the MAPK pathway . It is important to note that the p85/p110 PI3K complex also binds to Ras, thereby connecting two pathways which are often considered to be separate . With regard to insulin signalling, the MAPK pathway is primarily associated with the regulation of mitogenesis . Again, stimulation of the MAPK pathway in isolation is not able to induce glucose uptake in fat or muscle.
In summary, insulin transduces its signal through at least three different cellular pathways. There are also a number of critical nodes involved in insulin signalling, including the IR–IGF-IR complexes, IRS molecules and Akt/MAPK substrates that are able to modify the biological effects of a ligand on a particular cell type. What is currently unclear is how cells are able to control which pathways are activated in response to specific ligands. This could be of great therapeutic benefit as it may enable novel ways of overcoming cellular resistance or hyperstimulation of either the insulin or IGF pathways.
INSULIN SIGNALLING IS NOT ALL ABOUT GLUCOSE HOMOEOSTASIS
IR, IGF-IR and hybrid receptors are expressed throughout the body in most tissues and are not restricted to the classic insulin-sensitive glucose uptake tissues, such as liver, muscle and fat . It is therefore not surprising that insulin has other important biological effects on many different tissues independent of glucose uptake, as evolution rarely allows redundant systems to persist. Two good examples of important non-glucose-controlling actions of insulin are found in the brain and the cardiovascular system. In the brain, insulin signalling is crucial for controlling appetite, regulating food intake [46–48] and preventing obesity and infertility . These are all features associated with insulin resistance. Similarly, within the cardiovascular system, insulin can directly signal to both the heart dynamically regulating metabolism  and also the peripheral vasculature controlling tone. Cardiovascular tone is controlled both centrally  and through local modulation of endothelial NO release by insulin . This is potentially relevant in the cardiovascular morbidity and mortality associated with insulin-resistant states.
INSULIN AND THE KIDNEY
The kidney is the major organ that regulates fluid balance, BP (blood pressure), acid–base status, haemoglobin production, electrolyte control and waste removal in the body. It is a highly vascular structure that receives 20% of the circulating blood volume per min through the renal artery: the RBF (renal blood flow). From here, glomeruli within the kidney are perfused and their filtrate per min is defined as the GFR (glomerular filtration rate). Within each kidney are approximately 1 million nephrons. These are the functional units of the kidney. Each nephron has a primary filtering unit, the glomerulus, and a tubule, which is able to modify the primary filtrate produced from the glomeruli. There is now accumulating evidence that both the glomeruli and tubules are insulin sensitive.
INSULIN AND THE KIDNEY: IN HEALTH
The active insulin molecule, after loss of the c-peptide, exists as a monomer; it contains 51 amino acids and is approximately 6 kDa in size, allowing it to freely traverse the GFB (glomerular filtration barrier) and pass into the tubular lumen . This is important as it allows insulin rapid access to all of the cells in the glomerulus and nephron after it has been secreted into the circulation and passes through the renal artery.
The major high-affinity receptor for insulin, the IR, is located throughout the kidney in all of the cells of the glomerulus [52–55] and the entire length of the renal tubule, from the proximal tubule to the collecting ducts [56–59]. Recently, it has also been shown that kidney, as a whole, abundantly expresses IR isoform B , which is the isoform found in the classically insulin-responsive, glucose-regulating, tissues of fat, skeletal muscle and liver. It is now also clear that insulin is involved in a number of homoeostatic physiological responses throughout the kidney and are described below.
Two main processes, glycogenolysis and gluconeogenesis, result in the release of glucose, both of which involve the hydrolysis of glucose 6-phosphate by glucose-6-phosphatase. Originally, it was thought that the liver was the only organ in the body that could release glucose in times of starvation and stress. However, it is now clear that the kidneys are also capable of gluconeogenesis. This explains why patients with fulminant liver failure can maintain their circulating glucose concentrations . It transpires that both the liver and the kidney possess sufficient gluconeogenic enzyme and glucose 6-phosphate activity to facilitate this production . Furthermore, the kidney not only releases glucose in times of acidosis or after prolonged fasting as initially thought, but is also able to release significant amounts of glucose in normal post-absorptive individuals . Insulin reduces renal gluconeogenesis [64,65], so is an important factor in the control of glucose release from here. The importance of renal gluconeogenesis is discussed in depth in a review by Gerich et al.  and highlights that the kidney may be equally as important in the production and regulation of gluconeogenesis as the liver. The cellular location of gluconeogenesis in the kidney is not entirely clear; however, the proximal tubules have been shown to express glucose-6-phosphatase as have the parietal cells and podocytes . Clinically, this may be relevant as congenital loss of glucose-6-phosphatase causes glycogen storage disease type I. These children have developmental delay, episodes of hypoglycaemia and, from a renal perspective, proximal renal tubular dysfunction and glomerular damage as illustrated by FSGS (focal segmental glomerulosclerosis) on renal biopsy . Therefore, interestingly, the renal phenotype of this condition revolves around the two cell types that are known to express glucose-6-phosphatase in the kidney.
Haemodynamic control of renal organ blood flow and local glomerular blood flow: the role of insulin
There are conflicting findings on the direct role of insulin in RBF through the major vessels and local glomerular blood flow within the kidney, which appears to be species-dependent. Early work on conscious dogs suggested that insulin decreased RBF and that this was independent of the effects of insulin on the sympathetic nervous system . However, in humans, a number of studies [69–72], but not all [73,74], have shown that RBF actually increases in response to insulin. It is possible that the differences reported in these studies are due to dual actions of insulin. Insulin is able to act systemically by stimulating a catecholamine response and activating the sympathetic nervous system . However, it can also cause a local renal vasodilatory effect in the kidney.
A more consistent finding is that insulin increases the GFR in insulin-sensitive subjects. This occurs through local renal vasodilation and is mediated by a prostoglandin-dependent pathway that can be blocked with indomethacin  and regulated by eNOS (endothelial NO synthase) . Interestingly, a number of groups have shown that insulin increases the GFR in normal subjects, but this response is lost in insulin-resistant subjects [72,78,79]. However, and in contrast, a recent study has identified a human polymorphism in the IRS1 gene which is associated with an increased GFR , and the authors speculate that loss of renal insulin signalling may, in fact, be responsible for an increased GFR.
Specific cellular actions of insulin throughout the nephron
Glomeruli are composed of three different resident cell types: podocytes, GEnCs (glomerular endothelial cells) and mesangial cells. It is now clear that all of these cells respond to insulin, but in different ways. Podocytes and GEnCs are separated by the GBM (glomerular basement membrane) and constitute the GFB. Mesangial cells are specialized smooth-muscle-like cells that are able to contract and regulate blood flow to the glomerulus (Figure 3). The glomeruli filter as much as 5 million litres of primary urine across the GFB during an average human lifetime and, as the primary urine is practically protein-free, this means that more than 200000 kg of albumin has to be prevented from crossing .
Podocytes are unique cells found on the urinary side of the GFB (Figure 4). In the past 15 years, it has become clear that they are critically important in maintaining the integrity of the GFB and preventing leakage of albumin into the urine. There are now more than ten inherited human genetic mutations, all of which cause nephrotic syndrome and all of which code for proteins found predominantly in the podocyte (Table 2). Podocytes are embryonically derived from mesenchyme and are classified by most as epithelial cells, as they are polarized and sit on a basement membrane. However, they also have features of other cell types, including smooth muscle, as they are contractile and express smooth muscle markers , and neurons due to their processes, secretory capacity and, for the most part, inability to replicate when fully formed . Podocytes adhere to the GBM through a network of anchoring proteins [84–86] and have specialized modified adherens junctions, called slit diaphragms, formed between their foot processes. These contain a set of proteins that are crucial in maintaining the integrity of the GFB [81,87]. It is now clear that the podocyte depends on its actin cytoskeleton to maintain its structure and the integrity of the GFB and many of the slit diaphragm proteins are linked to this [88,89]. Although the majority of disease-causing mutations in the podocyte are related to actin-regulating functions, it is of note that gain-of-function mutations in the Ca2+ channel modulating TRPC6 [TRPC (transient receptor potential cation channel), subfamily C, member 6] has also been discovered to cause nephrotic syndrome [90,91], as well as mitochondrial proteins  and enzymatic proteins [PLCE1 (phospholipase C, ϵ1)] .
In recent years, our group [52,94] and others  have shown that the podocyte is a rapidly insulin-responsive cell. Our initial work employed a conditionally immortalized human podocyte cell line  to study insulin responses in this cell. This was helpful as the cell line contains a temperature-sensitive transgene that enables the cells to replicate and proliferate ad infinitum at 33°C but when they are thermo-switched to 37°C they exit the cell cycle and are able to differentiate and express many of the markers of maturity. This is important as the thermo-switched cells resemble mature podocytes found in the normal glomerulus. We found that differentiated podocytes rapidly respond to insulin by doubling their glucose uptake within 15 min . This is similar to the kinetics observed in muscle, which is not surprising given the muscle-like features and markers that podocytes exhibit . Importantly proliferative immature 33°C human podocytes, human proximal tubular cells and human GEnCs did not respond to insulin in respect to glucose uptake. We went on to show that this process was dependent on the actin cytoskeleton and activated translocation of the glucose transporters GLUT1 and GLUT4 from cytoplasmic vesicles to the plasma membrane of the cell .
We extended these observations to show that the podocyte protein nephrin was also important in insulin signalling in the podocyte . This was achieved by studying conditionally immortalized podocytes derived from children with the most severe form of congenital nephrotic syndrome called Finnish type congenital nephrotic syndrome. This occurs secondary to mutations in the protein nephrin. We developed a number of conditionally immortalized natural human knockout cell lines from nephrin-deficient (no nephrin protein made) or nephrin-mutant (protein made but unable to target to the plasma membrane) kidneys. We found that these cells were completely unresponsive to insulin in respect to glucose uptake, but could be rescued by genetically reconstituting nephrin back into them. Mechanistically this was due to a failure of GLUT-rich vesicles to dock and become incorporated into the plasma membrane of the nephrin-deficient/mutant podocytes. We went on to show that the C-terminus of nephrin was able to form a protein–protein association with VAMP2 (vesicle-associated membrane 2), which is important for vesicle docking with the plasma membrane through SNARE (soluble N-ethylmaleimide-sensitive fusion protein-attachment protein receptor) processes . Although nephrin is only expressed in the podocyte in the kidney it is also found in pancreatic β-cells in the body, where it has also been shown to be potentially involved in insulin release here in response to glucose .
Interestingly given the possible role of nephrin in both insulin release and its cellular action, children with nephrin mutations who receive kidney transplants do not appear to develop overt DM (diabetes mellitus). This suggests that, although nephrin may be involved in the control of insulin release, it is not critical.
Recently, we have generated a podocyte-specific IR-deficient transgenic (podIRKO) mouse, which has proven to be highly informative . podIRKO mice develop albuminuria and loss of foot process architecture by 8 weeks of age. In light of this, we have also found that insulin is able to rapidly remodel the filamentous actin cytoskeleton of podocytes. This is via the IR and through modulation of small GTPases, which act as molecular switches for actin remodelling in cells. RhoA is activated and CDC42 is inhibited . We have also shown that in the podocyte insulin stimulates the PI3K and MAPK signalling cascades, but not the CAP/Cbl pathway.
Another recent exciting advance in our understanding of the homoeostatic role of insulin on podocyte biology has come from Dryer's group. They have discovered that insulin rapidly increases the membrane expression of Ca2+ and K+ channels in the podocyte . They have demonstrated that, within minutes, insulin causes TRPC6 to increase on the surface of podocytes, but TRPC5 channels decrease. Functionally this causes an increased influx of cations, particularly Ca2+, into the cell. This elegant study went on to show that the effects of insulin on TRPC6 were mediated through the production of ROS (reactive oxygen species) via activation of NADPH oxidases . The same group have also shown that, in addition to modulating TRPC channels in podocytes, insulin also rapidly causes Ca2+-regulated K+ channels to locate to the plasma membrane of this cell . Collectively these findings suggest that, when insulin stimulates the podocyte, as occurs after a meal, it causes the podocyte to rapidly take up a readily usable energy source, glucose, remodel its actin cytoskeleton and contract, which is facilitated by ionic flux into the cell. We think this makes biological sense in order for this cell to ‘brace’ itself for the increased work this cell is required to perform at this time (Figure 5).
GEnCs line the capillaries of the glomerulus (Figure 4). They are highly fenestrated and therefore freely permeable to an array of molecules. These cells express IRs [55,102], IGF-IRs  and hybrid receptors , so have the receptor apparatus for insulin signalling. Early work has suggested that their primary role was to remove insulin from the circulation and degrade it . However, we have shown that insulin activates the PI3K pathway rapidly in human GEnCs , but it does not have the same biological actions in these cells as in podocytes and does not cause rapid glucose uptake  or actin remodelling . Using human GEnCs that have been conditionally immortalized  in the same way as podocytes, we have found that insulin rapidly induces the production of eNOS (J. Hurcombe and R.J.M. Coward, unpublished work). This complements work by Mima et al. , who have demonstrated that glomeruli isolated from rats are insulin-sensitive and insulin also rapidly induces eNOS.
Recently, it has been shown that reduced endothelial insulin signalling elsewhere in the body reduces local eNOS production and impairs the physical delivery of insulin to extra-capillary sites . As insulin is secreted into the circulation from the pancreatic β-cells, it needs to traverse the endothelial layer of blood vessels to reach extra-vascular targets. However, in the glomerulus, the GEnC is highly fenestrated, which may protect the underlying podocytes from insulin deficiency when the GEnC is insulin-resistant. Elucidating the functional importance of insulin signalling in the GEnC in the intact glomerulus is currently challenging. This is partly due to a lack of transgenic tools to be able to genetically manipulate the GEnC. Specifically, there is currently no way of targeting cre recombinase to this cell within the kidney, which is important when developing cell-specific-knockout mice models using cre-loxp technology . However, there is now promise that this may be rectified in the future, as specific GEnC genes have recently been identified within the kidney which may be beneficial in developing these mice  and genetically manipulating the GEnC in the intact glomerulus in the future.
Mesangial cells are specialized contractile cells that support the glomerular capillaries. Their contractile properties show similarities to vascular smooth muscle cells [111,112], with the release of Ca2+ from stores within the ER (endoplasmic reticulum) upon the initiation of contraction. The released Ca2+ activates Cl− channels, which then depolarize the cell membrane and activate VOCCs (voltage-operated Ca2+ channels), resulting in an increase in intracellular Ca2+ levels and the subsequent activation of BK (large-conductance Ca2+-activated K+) channels. Activation of BK channels then causes the cell membrane potential to hyperpolarize [113,114].
Both the IR and IGF-IR are expressed in mesangial cells, although at differing levels, with the IGF-IR being predominant . Despite this, insulin at high levels is able to stimulate mesangial cell proliferation through activation of either receptor . BK channels are expressed in abundance in mesangial cells, where they contribute to the relaxation of the cell and as a result an increased GFR . Similar to podocytes, insulin has been shown to increase the density of BK channels in the plasma membrane of human mesangial cells via activation of the MAPK pathway .
Unlike the podocyte, mesangial cells do not rapidly increase glucose in response to insulin and, as a result, the intracellular glucose level observed more directly reflects its plasma concentration. Excessive extracelluar glucose in the diabetic mileu can enter mesangial cells with ease via GLUT1 in an insulin-independent manner, which can result in glucotoxicity. High glucose levels have been shown to enhance GLUT1 expression in mesangial cells, which may result in progressive damage ; in addition, these glucose-induced effects can be mimicked in ‘normal’ glucose conditions by specifically overexpressing GLUT1 , resulting in excessive production of ECM (extracellular matrix) proteins .
Apoptosis is an important mechanism within a number of organs, tissues and cells, and the glomerulus is no exception. Apoptosis of glomerular cells is a closely regulated process; it can be beneficial in allowing the removal of excess cells in order to resolve glomerular injury, but can also be detrimental if excessive apoptosis is allowed to occur leading to hypocellularity . Insulin has been shown to be a pro-survival factor via activation of PI3K, which in turn allows the recruitment of Akt to the plasma membrane, where it phosphorylates a number of molecules to suppress apoptosis [122,123]. Both insulin and IGF-I protect mesangial cells from a variety of apoptotic triggers via the PI3K/Akt pathway using this mechanism ; in conjunction with these findings, insulin has also been shown to reduce ERK1/2 activation and increase levels of the cyclin-dependent kinase inhibitor p21 during apoptosis, providing an additional level of protection.
These are segregated into a number of defined regions, including the proximal tubule, loop of Henle, distal convoluted tubule and the collecting ducts. This part of the kidney is able to modulate the primary filtrate from the glomerulus by reabsorbing ions back into the blood or secreting ions into the urine. Important ions that are regulated include glucose, Na+ and HCO3− resorption in the proximal tubule, Na+ and water resorption in the loop of Henle, and K+ and H+ secretion in the distal convoluted tubule, together with water retention in the collecting ducts. Collectively, the tubules are able to control acid–base status, Na+ and water resorption and hence BP regulation in the body.
The tubule consists of a number of different specialized polarized epithelial cells that are able to transport molecules to and from the tubular lumen into and from the circulation. Numerous groups have shown that the IR, IGF-IR and hybrid receptors are expressed throughout the tubule [56,59,125,126], including the proximal tubule, loop of Henle, distal convoluted tubule and collecting ducts. More insulin binds to the tubules in comparison with the glomeruli, although with less affinity as reported by some , but not all groups . In our initial work, we studied immortalized proximal tubular cells (HK2) and found that they did not respond to insulin in respect to glucose uptake . However, it is clear that the tubules are insulin-responsive, but that insulin elicits different cellular effects in this part of the kidney. Elegant work by Mima et al.  has shown that ex vivo renal tubules from rats are insulin-responsive in respect to activation of the PI3K and MAPK pathways. Interestingly, this insulin response is not lost in established diabetes, as was found with insulin responses in the glomeruli in these studies. Furthermore, in vitro studies of isolated tubular cell types suggest that insulin can rapidly modify a variety of transporter systems throughout the tubule [128–131]. These include the NHE3 (Na+/H+ exchanger type III) [132,133], which is the major Na+ transporter in the proximal tubule and responsible for 65% of Na+ resorption here. Modulation of this channel is also able to alter the acid–base status in the body. Insulin also augments Na+ resorption through other transporters throughout the tubules, including the loop of Henle, via the butamide-sensitive Na+–K+–2Cl− channels , and ENaC (epithelial Na+ channel), Na+/K+-ATPase and recently the Na+–Cl− co-transporter  in the collecting ducts. A number of studies examining the distal tubule have demonstrated that insulin binds to the IR and activates the PI3K pathway, which then phosphorylates and stimulates SGK1 (serum- and glucocorticoid-induced protein kinase 1) phosphorylation  that inhibits the breakdown of transporters through endocytic retrieval pathways and may also directly phosphorylate transporters resulting in enhanced actions . This is interesting, as SGK1 seems to be a connection through which mineralocorticoids and insulin can modify Na+ retention in the distal part of the nephron. In addition to Na+, insulin can also modulate the resorption of other ions in the proximal tubule, including PO43− (phosphate ion) through the Na+– PO43− co-transporter type-II in the proximal tubule  and Mg2+ in the distal convoluted tubules . A comprehensive review of the role of insulin in the renal tubules was performed in 2007 and we would recommend reading this excellent article .
INSULIN SIGNALLING AND THE KIDNEY: DISEASE
Systemic insulin-resistant states
Insulin-resistant states are a major global healthcare problem in the 21st century, with an estimated 171 million diabetics present in the world. However, owing to current sedentary lifestyles, population aging and urbanization, the anticipated number of cases is predicted to more than double in the next 15 years . The incidence of the insulin-resistant metabolic syndrome is even more pronounced, with an estimated prevalence of 20% in the people over 20 years of age in the U.S.A. rising to over 40% when over 60 years of age being reported .
Approximately one-third of all new cases of ESRD (end-stage renal disease) worldwide is accounted for by DM  and, in the U.S.A., this is even higher with over 50% of new patients having DM (U.S. Renal Data System, USRDS 2011 Annual Data Report; http://www.usrds.org). Type 2 DM is due to insulin resistance of peripheral tissues, in contrast with Type 1 DM which that occurs secondary to insulin deficiency caused by destruction of the β-cells of the pancreas . DN (diabetic nephropathy) is the most common microvascular chronic complication of DM . DN is a progressive disease which takes several years to develop; it occurs in 30–40% of patients with Type 1 DM and 8–10% of patients with Type 2 DM . Its natural history is dominated by progressive albuminuria.
There is now accumulating evidence that a loss of insulin responses in the kidney may contribute to a number of the complications that occur in insulin-resistance states, including albuminuric glomerular disease and hypertension.
The glomerulus, insulin, diabetes and the metabolic syndrome
Early renal manifestations of DN are focused on the glomerulus in the kidney consisting of glomerular hyperfiltration and microalbuminuria, alongside other changes, including GBM thickening, mesangial expansion and accumulation of ECM proteins such as laminin, collagen and fibronectin . Advanced DN is characterized by increased albuminuria (macroalbuminuria), glomerulosclerosis, interstitial fibrosis and ESRD [87,144] (Figure 6). In recent years, the podocyte has become an intense focus of research into this field as loss of this cell has been found to be the best histological predictor of progression in DN . Furthermore, as progressive albuminuria dominates the natural history of DN, this also makes the podocyte an attractive target cell in DN, because of its crucial role in preventing albuminuria, as discussed above (Table 2) .
The metabolic syndrome is also associated with microalbuminuria; indeed, it is part of the diagnostic criteria in some classifications, including that of the World Health Organization.
As described above, we have shown that the human podocyte is an insulin-sensitive cell  and we have developed podIRKO mice. These mice were highly informative as they developed a number of features of DN, including albuminuria, glomerulosclerosis, matrix accumulation (including type-IV collagen), thickening of the GBM and podocyte apoptosis. However, they all had normal blood glucose control, demonstrating that none of these features were driven by hyperglycaemia. This suggests that insulin signalling to the podocyte is critically important for normal glomerular function and may also have a role in some aspects of DN. It should be noted that the podIRKO mice only exhibited some features of DN and did not have enlarged kidneys nor did they have mesangial hypercellularity or the classic nodular Kimmelstein–Wilson lesions of DN (Figure 6). A potential explanation for this is that in diabetes there is a loss of insulin sensitivity of the podocyte and this results in some of the pathological consequences associated with DN, but other aspects of DN, for example renal hypertrophy and mesangial expansion, are driven by other factors. These may include pathways driven by high glucose levels or other growth factors such as IGF-I or IGF-II acting on the cells of the glomerulus.
Our current working hypothesis is that, in both Type 1 and Type 2 DM, as well as the metabolic syndrome, in those patients who develop nephropathy, there is insulin resistance in the podocyte which contributes to the development of renal damage. Type 2 DM and the metabolic syndrome are intrinsic cellular insulin-resistant conditions, so this hypothesis would appear intuitively to be correct. Type 1 DM occurs due to a lack of insulin; however, there is compelling evidence that those patients with Type 1 DM who develop nephropathy are more likely to also be insulin-resistant. Prolonged Type 1 DM causes insulin resistance , and nephropathic compared with non-nephropathic patients with Type 1 DM require larger doses of insulin to control their diabetes , are more insulin-resistant when assessed by euglycaemic clamps  and are more likely to have a strong family history of cellular insulin resistance .
There is further experimental evidence to support this hypothesis. Tejada et al.  have examined podocyte insulin responses in the development of albuminuria in the db/db Type 2 DM mouse model and have shown that IR and PI3K signalling is lost early in the disease process. Furthermore, a recent excellent rodent study  examining Type 1 and Type 2 DM models has examined the effect of diabetes on insulin signalling in the kidney. Mima et al.  studied rats given streptozotocin inducing Type 1 DM and the Zucker obese model of Type 2 DM. They allowed the rats to develop diabetes and then examined their insulin signalling pathways in the glomerular and tubular compartments of the kidney. They found that insulin rapidly initiated PI3K and MAPK signalling in both isolated glomeruli and tubular fractions of control rats. However, when the rats had either Type 1 or Type 2 DM they both resulted in a loss of insulin signalling via the PI3K pathway specifically in the glomerular and not in the tubular compartment. They went on to show that high glucose increased ubiquitination and hence loss of IRS1 in the glomeruli, which could be a mechanistic pathway through which Type 1 DM is able to directly modulate cellular insulin signalling in the glomerulus. Finally, building on previous work [151,152], they demonstrated that inhibiting PKCβ (protein kinase Cβ) was able to reverse high-glucose-induced insulin resistance in GEnCs via eNOS. This adds weight to the assumption that GEnC as well as podocyte insulin resistance may be important in the development of glomerular complications in systemic insulin-resistant states.
A consequence of increased insulin signalling in the podocyte is increased translocation of GLUT4 and GLUT1 to the plasma membrane of this cell, allowing more glucose to passively diffuse into the cell. Previously, it has been proposed that a major reason for cellular dysfunction in the setting of diabetes is glucose toxicity of cells . In the glomerulus there is evidence that overexpression of GLUTs in the mesangial cell is detrimental to function . However, this does not seem to be the case for podocytes. An elegant study by the Brosius group  has shown that increasing the glucose transporter GLUT1 specifically in the podocyte in a model of Type 2 DM is not detrimental to glomerular function, but intriguingly seems to be protective against the development of some aspects of DN. GLUT1 is a glucose transporter that is expressed at the cell surface constitutionally in many cells and allows basal glucose uptake, but is also found in insulin-responsive translocatable glucose transporter vesicular pools similar to GLUT4. One possibility why this mouse was protected from the development of DN is that it may have been protected from episodes of glucose deficiency when stimulated by insulin .
Our research has also shown that nephrin is crucial for the insulin sensitivity of podocytes. Many groups have demonstrated that nephrin is reduced early in DN [156–158]. Potentially this could be inducing insulin resistance in these cells resulting in pathological changes. However, it is also possible that the loss of nephrin in the development of DN is also a consequence of podocyte dysfunction and not a cause.
What is currently unclear is the in vivo insulin responsiveness of the podocyte in systemic insulin-resistant states. Some groups have found that, similar to the classically metabolically insulin-responsive tissues of adipose , liver  and skeletal muscle , that the IR is down-regulated at the protein level in the kidneys of models of Type 2 DM and systemic insulin resistance. However, other groups have found in Type 2 DM that, although adipose, liver and skeletal muscle demonstrate diminished insulin binding and signalling, the kidney does not [57,162,163]. Therefore it is possible that early in the development of diabetes there is hyperstimulation of the insulin signalling axis and not loss of insulin sensitivity in the kidney, and this may also be having a detrimental effect.
There is evidence that insulin can modulate the GFB, which may be through direct insulin ligand receptor binding. Approximately 30 years ago insulin-deficient human subjects with Type 1 DM were given insulin under euglycaemic clamp conditions, i.e. maintaining the blood glucose constant, and this resulted in a transient increase in albumin excretion into the urine . Furthermore, it was shown that giving a glucose load to healthy subjects and also patients with Type 1 DM only caused an increase in urinary albumin excretion in the healthy subjects . This suggests that hyperglycaemia is not the major driver causing a loss of albumin into the urine, as this occurred in the patients with Type 1 DM who did not develop albuminuria (and could not produce insulin in response to the glucose challenge), but that in healthy subjects high glucose levels stimulated insulin release (which lowers their blood glucose levels) and caused albumin to leak into the urine. These insulin effects may explain the immediate post-prandial proteinuria that is widely described in human subjects and rodents [166,167].
If podocyte insulin resistance is a key factor in the development of nephropathy in diabetes then it follows that strategies to enhance podocyte cellular insulin sensitivity could be beneficial in treating this condition. There is evidence that some agents which have insulin-sensitizing properties, including metformin and PPARγ (peroxisome-proliferator-activated receptor γ) agonists, are beneficial in preventing kidney damage in models of DN in both Type 1  and Type 2 [169,170] DM, as well as other non-diabetic chronic kidney diseases [171,172], which partially supports this premise. Indeed, we have shown that the PPARγ agonist rosiglitazone is able to directly enhance insulin sensitivity of the podocyte in vitro .
Conversely, we have found that factors that are increased systemically in the insulin-resistant metabolic syndrome render the podocyte insulin-resistant. These include non-esterified ‘free’ fatty acids such as palmitate . Other groups have also explored some of the molecular inhibitors of insulin signalling in the podocyte. An elegant study demonstrated that SHIP2 (SH2-domain-containing inositol phosphatase) was able to inhibit insulin signalling in the podocyte and that it was up regulated in the glomeruli in a rat model of Type 2 DM . Discovering these molecules is important, as they could potentially be good therapeutic targets when inhibited.
Finally, there are findings indicating that insulin can regulate mesangial cell function in the glomerulus and when this cell is rendered resistant it potentially contributes to matrix formation , mesangial expansion  and hyperfiltration .
In summary, there is now evidence that all three cell types found in the glomerulus can respond to insulin, but they respond in different ways. Furthermore, when insulin signalling is altered in the glomerulus it results in pathology.
mTOR, insulin and glomerular disease: another important pathway
Another potentially important role for insulin in the glomerulus is modulating the mTOR pathway. The mTOR signalling cascade controls cellular protein synthesis, growth, metabolism, autophagy and survival in response to growth factors, stress, energy and nutrient stimuli.
mTOR is a protein kinase and the catalytic subunit of two functional complexes: mTORC1 (mTOR complex 1) and mTORC2 (mTOR complex 2) . mTORC1 is a rapamycin-sensitive complex in which mTOR is associated with the Raptor (regulatory associated protein of mTOR) and regulates a number of cellular processes, including protein synthesis, cell growth and proliferation [179,180]. Insulin is able to increase the activity of mTORC1 through PI3K and ERK1/2 pathways via TSC2 (tuberous sclerosis complex 2). Both pathways inhibit TSC2, which then prevents it from suppressing mTORC1 expression. This results in protein translation, ribosomal biogenesis and autophagy . Interestingly, it has also recently become evident that mTOR inhibition causes cellular insulin resistance  and that this action is through the mTORC2 complex . Therefore the mTOR pathway is both controlled by, and also controls, insulin signalling (Figure 7).
Rapamycin (Sirolimus) is used in renal transplantation because inhibiting mTOR inhibits the response of B- and T-cells to IL-2 and hence prevents organ rejection. mTOR expression is low or undetectable in the normal kidney but following ischaemia/reperfusion injury it increases significantly, presumably to enable cellular repair and regeneration to occur. In this setting, rapamycin is detrimental as it inhibits mTOR and results in delayed renal repair and recovery .
The function of mTOR in the glomerulus remains controversial; however, the findings from a number of recent elegant murine transgenic studies [185–187] have advanced the understanding of the role of mTOR in one cell type located in the kidney, namely the podocyte. Previous work has reported that systemic administration of rapamycin in mouse models of both Type 1 and Type 2 DM can prevent the progression of DN [188–190], suggesting it may be clinically beneficial to inhibit mTOR in these settings. However, it has also been shown that mTOR inhibition with rapamycin can be detrimental in non-diabetic conditions and can cause proteinuria and glomerulosclerosis in both humans and rodents [191–194]. Studies published recently by Inoki et al. , Gödel et al.  and Cina et al.  have highlighted the critical importance of the mTOR pathway in podocyte biology. By manipulating mTORC1 and mTORC2 they have shown that loss of mTORC1 is detrimental to glomerular function  and this is exacerbated further when mTORC2 is also lost [186,187]. Interestingly increasing the activity of mTORC1 can also be harmful to the glomerulus causing albuminuric renal disease resembling DN . Increased mTORC1 activity occurs in DN and can be rescued by genetic or pharmacological inhibition of mTORC1.
In summary, insulin signalling can modify mTOR activity in the glomerulus, predominantly through mTORC1 but it can also be modified and suppressed when mTORC2 is inhibited by long-term rapamycin therapy.
The tubule, insulin, diabetes and the metabolic syndrome
As discussed above, insulin is able to modulate tubular function in a number of ways. One of the most important effects seems to be its action on Na+ resorption from the tubule into the circulation. As hypertension is a prominent feature of the metabolic syndrome and DN, it is not surprising that a great deal of work has explored the potential role of tubular insulin sensitivity and its role in the control of BP.
In 2008, Tiwari et al.  developed a mouse model in which the IR was knocked out in the renal tubules by crossing a floxed IR mouse with a ksp (kidney-specific)-cadeherin promoter-linked cre recombinase-producing mouse. Ksp-cadeherin is only found in the tubular epithelial cells in the kidney. It is predominantly expressed in the distal aspect of the tubules from the thick ascending loop of Henle through to the collecting ducts . As discussed in the previous section, work based predominantly on cell culture models had suggested that a loss of tubular insulin signalling would result in a reduction of Na+ resorption from the urinary filtrate and hence a naturesis with an associated lowering of the BP. However, intriguingly, this was not the case with this model. These mice had impaired urinary Na+ excretion in response to a Na+ load and were hypertensive in comparison with controls. The group went on to show that insulin signalling here was able to activate local NO production and reduce BP in wild-type normal animals, presumably through a vasodilatory effect, but in tubular IR-knockout mice this did not occur. This suggests another role of tubular insulin signalling in NO production and BP control. It may also be clinically relevant as hypertension is commonly associated with insulin-resistant states.
Rare renal disease associated with severe insulin resistance
In addition to the very common conditions of DM and the metabolic syndrome there are a number of rare syndromic forms of cellular insulin resistance that are associated with glomerular renal disease. It is interesting that these do not always result in classic DN, but rather in a spectrum of renal pathologies as follows.
This group of disorders is caused by a failure to deposit fat in adipose tissues. This causes abnormalities in circulating adipokines and results in the deposition of fat in ectopic locations such as skeletal muscle and liver. These patients are often severely insulin-resistant at a cellular level.
The most severe form of lipodystophy is the generalized form where the patients have no adipose tissue. The majority of patients with congenital and acquired forms of this condition have glomerular disease with albuminuria, but only a small subset have the classical features of DN. The rest have a variety of renal pathologies, including FSGS and MPGN (membranoproliferative glomerulonephritis).
Another cohort of patients suffer from acquired partial lipodystrophies. These patients lose adipose tissue from their face, neck, upper extremities, thorax and upper abdomen, and have immunological abnormalities with low C3 complement levels and elevated C3 nephritic factor levels. The most common renal lesion found in these patients is MPGN type 2 (dense deposit disease). It has been hypothesized, but not categorically proven, that this is an immunological disease; however, there may be a contribution from insulin resistance. The renal phenotypes of extreme insulin resistance are extensively reviewed by Musso et al. , which we would recommend reading.
Diseases that target the IR
Auto-antibodies against the IR (type B insulin resistance)
This is associated with extreme insulin resistance and DM in the majority, although paradoxically these auto-antibodies can also cause hypoglycaemia in some situations . Patients usually have an underlying collagen vascular disease, most commonly systemic lupus erythromatosis. Again, more than 50% of patients have albuminuria but their histology is normally that of one of the forms of lupus nephritis [198,199].
Mutations of the IR (type A insulin resistance)
Two syndromes account for the majority of patients suffering from mutations of the IR: Donogue syndrome, previously known as Leprechaunism [OMIM (Online Mendelian Inheritance in Man®) 147670], and Rabson–Mendenhall syndrome (OMIM 262190). These patients have extreme insulin resistance, acanthosis nigrans, hirsuitism and are generally slender. These patients commonly, but not always, have DN , which may be secondary to the cellular insulin signalling defect they experience.
It is now clear that the kidney is an insulin-sensitive organ, but that different regions of the kidney respond to insulin in different ways. To elucidate the clinical relevance of insulin sensitivity of the kidney there are still some fundamental questions that need to be addressed and will be a focus of research in the upcoming years. These questions include the following. (i) Does insulin resistance in the GEnC and/or mesangial cell contribute to glomerular pathology? (ii) Do human renal cells in vivo become insulin-resistant in systemic insulin-resistant states? This has never been proven. (iii) Can we therapeutically manipulate the insulin signalling pathway in the kidney to prevent renal disease from developing? Ideally this would be kidney (cell)-specific to reduce off-target side effects, as have been experienced with other insulin-sensitizing drugs, such as the glitazones . (iv) Why does insulin elicit different biological responses in different tissues? This is probably due to differences in important signalling nodes in the insulin and IGF pathways, but again this is not proven.
It is now clear that the kidney is an insulin-responsive organ in a variety of different ways. Manipulating these responses may have great therapeutic potential in treating glomerular disease and hypertension associated with DM and other insulin-resistant states.
Our own work was supported by Kidney Research UK and the Medical Research Council.
We thank Dr Tibor Toth (Department of Pathology, Southmead Hospital, Bristol, U.K.) for producing Figure 6.
Abbreviations: BK channel, large-conductance Ca2+-activated K+ channel; BP, blood pressure; CAP, Cbl-associated protein; DM, diabetes mellitus; DN, diabetic nephropathy; DOK, downstream of kinase; ECM, extracellular matrix; eNOS, endothelial NO synthase; ERK, extracellular-signal-regulated kinase; ESRD, end-stage renal disease; FSGS, focal segmental glomerulosclerosis; GBM, glomerular basement membrane; GFB, glomerular filtration barrier; GFR, glomerular filtration rate; GLUT, glucose transporter; GEnC, glomerular endothelial cell; Grb2, growth-factor-receptor-bound protein 2; IGF, insulin-like growth factor; IGF-IR, IGF-I receptor; IL, interleukin; IR, insulin receptor; IRS, insulin receptor substrate; ksp, kidney-specific; JNK, c-Jun N-terminal kinase; MAPK, mitogen-activated protein kinase; MPGN, membranoproliferative glomerulonephritis; mTOR, mammalian target of rapamycin; mTORC, mTOR complex; OMIM, Online Mendelian Inheritance in Man®; PH, pleckstrin homology; PI3K, phosphoinositide 3-kinase; podIRKO mouse, podocyte-specific IR-deficient transgenic mouse; PPARγ, peroxisome-proliferator-activated receptor γ; PTB, phosphotyrosine-binding; Raptor, regulatory associated protein of mTOR; RBF, renal blood flow; SGK1, serum- and glucocorticoid-induced protein kinase 1; SH2, Src homology 2; SOS, Son of Sevenless; TRPC, transient receptor potential cation channel; TSC, tuberous sclerosis complex
- © The Authors Journal compilation © 2013 Biochemical Society