Dr Vincent WS Lee, Department of Renal Medicine, Westmead Hospital, Westmead, NSW 2145, Australia. Email: firstname.lastname@example.org
Adriamycin nephropathy (AN) is a rodent model of chronic kidney disease that has been studied extensively and has enabled a greater understanding of the processes underlying the progression of chronic proteinuric renal disease. AN is characterized by podocyte injury followed by glomerulosclerosis, tubulointerstitial inflammation and fibrosis. Genetic studies have demonstrated a number of loci that alter both risk and severity of renal injury induced by Adriamycin. Adriamycin-induced renal injury has been shown in numerous studies to be modulated by both non-immune and immune factors, and has facilitated further study of mechanisms of tubulointerstitial injury. This review will outline the pharmacological behaviour of Adriamycin, and describe in detail the model of AN, including its key structural characteristics, genetic susceptibility and pathogenesis.
Most types of chronic kidney disease (CKD) are characterized by the development of glomerulosclerosis, tubulointerstitial inflammation and fibrosis. Adriamycin® (Pfizer, Sydney, Australia) (doxorubicin) is a well-known inducer of renal injury in rodents, which mirrors that seen in human CKD due to primary focal segmental glomerulosclerosis. The first published record of anthracyclines causing renal injury was in 1970 by Sternberg.1 The first description of Adriamycin inducing renal injury was in 1976 in rats,2 and 1998 in mice.3 In 1977, Burke and colleagues4 described a case of a 78-year-old man developing renal failure after the administration of doxorubicin. Since then, Adriamycin nephropathy (AN) in rodents has been extensively studied and has enabled a greater understanding of the processes underlying the progression of renal injury.
ADVANTAGES AND DISADVANTAGES OF AN AS AN EXPERIMENTAL MODEL OF CKD
Adriamycin nephropathy has several strengths as an experimental model of kidney disease. It is a highly reproducible model of renal injury. It is also a ‘robust’ model in that the degree of tissue injury is severe while associated with acceptable mortality (<5%) and morbidity (weight loss). Because the model is characterized by the induction of renal injury within a few days of drug administration, the timing of injury is consistent and predictable. The severity and timing of renal injury means that it is a model suitable for testing interventions that either worsen or protect against renal injury. The type of structural and functional injury is very similar to that of chronic proteinuric renal disease in humans (see below). Last but not least, this model is similar in rats and mice. Rodent models are extremely useful in the study of disease. Rodents are characterized by their short reproduction period, easy (and cheap) availability of animals and reagents, and amenability to genetic manipulation.5
There are also limitations in the use of AN as an experimental model. Batch variability occurs whereby certain batches of Adriamycin induce less severe renal injury than others. Testing of a new batch prior to commencing an experiment is recommended. Considerable operator skill is required to perform the intravenous injection of the drug, usually into the tail vein under a warm lamp to induce vasodilatation, into an animal that is a ‘moving target’ if unanaesthetized or unrestrained. Adriamycin is characterized by a narrow ‘therapeutic’ index whereby doses as little as 0.5 mg/kg lower or higher than the optimum dose may lead to either lack of renal injury or toxicity leading to death, respectively. While the model is consistent and reproducible, there is still some individual variability in response, even within the same strain of rodent. There is also variability in susceptibility across strains – an observation that has been characterized at a genetic level (see below).
Adriamycin (doxorubicin) is an anthracycline, a class of anti-tumour drugs with a very wide spectrum of activity in human cancers. The first two anthracyclines daunorubicin and doxorubicin were developed in the 1960s. Doxorubicin differs from daunorubicin only by a single hydroxyl group.6 Doxorubicin is a cytotoxic anthracycline antibiotic isolated from cultures of Streptomyces peucetius var. caesius.
Detailed pharmacokinetic studies have been performed in humans and animals, demonstrating some minor differences. In humans, Adriamycin undergoes rapid plasma clearance and there is significant tissue binding. Adriamycin is metabolized predominantly by the liver. Urinary excretion of approximately 4–5% of the administered dose occurs within 5 days. Biliary excretion accounts for 40–50% of the administered dose in 7 days.7 In rats and mice, Adriamycin is rapidly cleared from the plasma after intravenous administration, deposited in tissue, and slowly excreted into urine and bile. Adriamycin is not significantly metabolized. Adriamycin accumulates mainly in the kidney (especially in comparison with daunorubicin) but is also found in liver, heart and small intestine.8 This probably accounts for the greater nephrotoxicity and wider therapeutic index of Adriamycin compared with daunorubicin.
The optimal regimen of Adriamycin administration depends on species, strain, gender, age, source and batch. Most rat species are completely sensitive to the renal effects of Adriamycin. In male Wistar rats, the dose of Adriamycin ranges between 1.5 and 7.5 mg/kg. Male BALB/c mice require 9.8–10.4 mg/kg,9 while male BALB/c SCID mice, an inbred lymphocyte-depleted strain of BALB/c mice, require only 5.3 mg/kg.10 C57BL/c mice are highly resistant to Adriamycin-induced renal injury but renal injury may be inducible at higher doses (13–25 mg/kg)11–13 than those required in BALB/c mice. While most studies use a single injection, regimens using multiple injections (e.g. 2 mg/kg × 2 in 20 days, 1 mg/kg/day × 7 days, 2.5 mg/kg × 6 in 14 days) have also been reported.
The route of administration in most studies is intravenous. The tail vein is usually used, although other veins (e.g. penile vein,14 femoral vein15 and retro-orbital plexus16) have been used. A great deal of technical expertise is required to perform this, particularly in mice (due to the small size of their veins and their predisposition not to lie still despite restraint, particularly in the case of the C57BL/6 strain). The main complication of tail vein injection is skin necrosis in the event of tissue extravazation. Due to the narrow therapeutic index of Adriamycin, a small difference in dose administered can potentially lead to a large variation in disease severity. Another route of administration is the substernal intra-cardiac (∼7 mg/kg in male Wistar rats) approach,17 which requires general anaesthesia. The intra-renal route, whereby Adriamycin is injected directly into the kidney (pre- and post-contralateral nephrectomy) is associated with induction of renal injury within 4 weeks. Direct injection of the renal artery has not been used except in pharmacodynamic studies in dogs.18 Despite their reported safety, the invasiveness of the intra-cardiac and intra-renal routes of administration has precluded widespread application. Intraperitoneal administration has been favoured for its ease of use, particularly in mice19 but due to variable absorption through the peritoneal membrane, inconsistency in induction of renal injury compared with the intravenous route has made this method less favoured.
A variety of conditions can affect the delivery of Adriamycin to the target organ. Temporary clipping of one renal artery during the intravenous administration of Adriamycin partially protects the clipped kidney from proteinuric renal injury.14,20 In addition, inhibition of renal blood flow by nitric oxide inhibition protects against glomerulosclerosis. These studies provide substantial proof that Adriamycin acts directly on the kidney to induce tissue injury.21
Male rats are more susceptible than female rats to Adriamycin-induced nephropathy. Castration renders male rats less susceptible compared with sham-operated rats, indicating that sex hormones may contribute to the pathogenesis of Adriamycin-induced renal injury.22 Because of the difference in severity of renal injury, choice of sex is a major factor in designing an experiment using AN as a model of renal injury.
STRUCTURE AND FUNCTION
In this animal model, the histological changes resemble those of human focal glomerulosclerosis, with podocyte fusion, focal segmental and global glomerular sclerosis and tubulointerstitial inflammation and fibrosis (Fig. 1).23 Adriamycin induces thinning of the glomerular endothelium and podocyte effacement associated with loss of size- and charge-specific barrier to filtration of plasma proteins.11 These changes are seen as early as 1–2 weeks after Adriamycin injection, and are severe by 4 weeks (Fig. 2). Renal function is also affected, with rodents showing elevated serum creatinine, reduced creatinine clearance, reduced serum albumin, dyslipidaemia and increased urine protein excretion consistent with the nephrotic syndrome and CKD.
Histological assessment of the kidneys of these mice shows severe tubulointerstitial inflammation, with marked infiltration by T and B lymphocytes and macrophages (Fig. 3).23 CD4+ and CD8+ cell numbers increase in cortex and medulla of Adriamycin-affected kidneys, but not in spleen, suggesting a direct role of these cells in modulating renal injury. However, studies in severe combined immunodeficient (SCID) mice (inbred BALB/c mice that lack lymphocytes) have demonstrated that structural and functional injury induced by Adriamycin does not require lymphocytes but can be modulated by the presence or absence of specific subpopulations. Renal injury develops in mice with doses of Adriamycin approximately half (5.3 mg/kg) that of wild-type BALB/c mice (9.8–10.4 mg/kg), suggesting that while lymphocytes are not essential, it is likely that a subpopulation of these cells protects against the development of renal injury. Further evidence for this comes from adoptive transfer studies of FoxP3 expressing CD4+CD25+ T cells, which protect against renal injury in AN,24 consistent with the exacerbation of renal injury by depletion of CD4+ T cells.25 The pattern of renal injury in SCID mice is similar to that in wild-type BALB/c mice. Macrophage infiltration is prominent in the tubulointerstitium but not in glomeruli (Fig. 4). Depletion and reconstitution studies suggest a pivotal role of pro- and anti-inflammatory macrophages in the pathogenesis of Adriamycin-induced kidney injury.26–28
ADRIAMYCIN AND PREGNANCY
Adriamycin induces renal injury in the fetus as well as the mother. When Adriamycin is administered intraperitoneally 4 weeks prior to pregnancy, kidneys from the fetus show increased amounts of PAS-positive mesangial matrix, glomerulosclerosis, tubular injury and dilatation.29 Pregnant rats given Adriamycin 2 weeks prior to pregnancy develop more severe proteinuria and higher blood pressure compared with non-pregnant rats, in association with an elevated ratio of thromboxane B2 (vasoconstrictor) to prostaglandin F1α (vasodilator) synthesis, changes which normalize post-pregnancy in a manner analogous to human pre-eclampsia.30,31 In contrast, repeated pregnancies after the induction of AN are associated with persistent glomerular damage post-partum.32 Adriamycin administration early in gestation (days 7 to 9 of rat pregnancy), induces anomalies in urinary tract development, the most common being bilateral megaureters with hypoplastic bladder.33
INTERVENTIONS IN AN
We and others have examined the effect of various immunologic interventions in AN, which have enabled a greater understanding of the immune mechanisms underlying chronic proteinuric renal disease associated with tubulointerstitial fibrosis. Macrophages and lymphocytes are heterogeneous populations containing cells that act to promote or reduce inflammation and fibrosis (see review by Lee et al.34). Chemokines (such as CCL2)35 and costimulatory molecules (such as programmed death 1 and CD40) also play an active role. Table 1 lists some of these interventions.
Table 1. Interventions (both immunological and non-immunological) tested in AN
Susceptibility to AN is strain-specific, with BALB/c mice being highly sensitive,23 while C57BL/6 mice are highly resistant to renal injury.11 Breeding experiments have identified a single gene locus with recessive inheritance on chromosome 16 that confers susceptibility to AN. Susceptibility alleles at this locus are associated with blunted expression of protein arginine methyltransferase on chromosome 8, a protein implicated in cellular sensitivity to chemotherapeutic agents.56 Additionally, genetic background influences severity of AN. In these same studies a locus on chromosome 8 has been identified that influences the severity and progression of nephropathy.
Lymphocyte number is a determinant of sensitivity to Adriamycin-induced renal injury. Compared with wild-type BALB/c mice, SCID BALB/c require only half the dose of Adriamycin to induce disease10 However, Adriamycin does not cause renal injury in lymphocyte-depleted recombinase activating gene-1 knockout C57BL/6 mice (V. Lee, unpubl. obs., 2010) meaning that lymphocyte number alone does not explain the resistance of C57BL/6 mice to Adriamycin-induced renal injury.
Susceptibility to Adriamycin is likely to lie in the immunological differences between species, for example, as occurs with BALB/c and C57BL/6 mice. It is convenient to use the Th1/Th2 paradigm to summarize the differences. C57BL/6 mice have immune responses that are, in general, polarized towards the Th1 axis whereas BALB/c mice possess immune responses that deviate towards the Th2 type. Therefore, the immune system of C57BL/6 mice is better equipped against and hence less susceptible to intracellular infection (e.g. Listeria57) but is more susceptible to antibody-mediated autoimmune disease such as myasthenia gravis. The immune response of C57BL/6 mice, as compared with BALB/c mice, is characterized by greater amounts of Th1 cytokines such as IL-12 and IFN-γ and less Th2 cytokines such as IL-4. The Th1 response is also characterized by upregulation of dendritic cells to a more mature phenotype. Consistent with this hypothesis, a recent study has shown that CD4+CD25− T cells isolated from C57BL/6 mice are less susceptible to suppression by CD4+CD25+ Tregs than their BALB/c counterparts, and that C57BL/6 mice possess fewer CD4+CD25+ Tregs than BALB/c mice.58 Therefore, a possible explanation for the relative resistance of C57BL/6 mice to Adriamycin-induced renal injury may be that Th1-immune responses are protective against AN, whereas Th2 responses are not. Zheng and colleagues59 have recently reviewed susceptibilities of mice to AN (Table 2) supporting the variability in response to Adriamycin across strains.
Table 2. Mouse susceptibility to Adriamycin-induced renal injury59
Adriamycin induces injury by direct toxic damage to the glomerulus with subsequent tubulointerstitial injury. Adriamycin causes changes to the glomerular filtration barrier, including to glomerular endothelial cells (including glycocalyx), glomerular basement membrane and podocytes. Glycocalyx thickness is reduced, glomerular endothelial cell pore size is increased, glomerular charge selectivity is reduced and podocyte cell foot processes are fused. These changes are associated with reductions in glomerular cell production of proteoglycans and glycosaminoglycans contained within the glycocalyx produced by the glomerular endothelial cells.11 Further evidence for a direct effect of Adriamycin on the kidney comes from a study in which clipping of the renal artery of one kidney protects it from injury.20
Additional studies have examined the molecular mechanisms for Adriamycin-induced renal injury. Increased free radical production has been proposed as a pathogenetic mechanism. This is supported by isolation perfusion studies of hagfish (Myxine glutinosa) glomeruli in which Adriamycin was found to reduce glomerular ATPase activity in association with a reduction in water permeability, an effect reversed by the sulfhydryl donor N-acetylcysteine. In addition, depleted levels of glutathione (an anti-oxidant) and elevated levels of lipid peroxide levels in liver, kidney and heart developed after Adriamycin administration.60
Evidence for the role of advanced glycation end products comes from studies of receptor for advanced glycation end product (RAGE)-null mice. These mice are protected from Adriamycin-induced podocyte damage and proteinuria. Adriamycin induced generation of RAGE ligands, an effect reversed by treatment with soluble RAGE. The mechanism for RAGE ligand-induced renal injury involved the activation of nicotinamide adenine dinucleotide phosphate-oxidase and p44/p42 MAP kinase signalling, and upregulation of pro-fibrotic growth factors.61
The changes associated with the slit diaphragm in Adriamycin-induced nephropathy have been studied by Otaki, Kawachi and colleagues.62 Early after Adriamycin administration (day 7), expression of the slit diaphragm molecules nephrin, podocin and NEPH1 (but not ZO-1- and CD-associated protein) is altered from a continuous to a discontinuous dot-like pattern consistent with podocyte injury. In particular, NEPH1 was disproportionately affected. Using immunoprecipitation and western blot studies of glomerular lysates from animals 7 days after Adriamycin injection, Kawachi's laboratory found that a large proportion of nephrin lost its affinity with NEPH1. While these data are observational in nature, they do point to slit diaphragm abnormalities as critical early events in the pathogenesis of Adriamycin-induced proteinuric renal injury.
Gene profiling using microarray chip technology has identified gene networks that are potential drivers of tubulointerstitial fibrosis in AN. Sadlier and coworkers63 have identified pathways already identified in the pathogenesis of fibrosis (transforming growth factor beta 1-connective tissue growth factor fibronectin-1 pathway) and novel genes associated with extracellular matrix production such as endoglin, clusterin and gelsolin. Several of these genes (e.g. claudin-1) have also been implicated in the pathogenesis of epithelial–mesenchymal transition of tubular epithelial cells.
Adriamycin also has effects that are not specific to the kidney and that are currently used therapeutically in the treatment of many types of cancers. Acute cellular changes include alterations in DNA structure (intercalation, cross-linking or binding), inhibition of topoisomerase 11, free radical generation causing DNA damage and lipid peroxidation, direct cell membrane effects, necrosis, apoptosis and promotion of senescence-like growth arrest. Delayed effects include reactive oxygen species generation causing mitochondrial DNA damage.5 Adriamycin also causes tubulotoxicity independent of its effects on glomeruli via tubular cell chemokine release (CCL2 & CCL5) and oxidant injury via reactive oxygen species and/or Fas/FasL interactions. These and other organ effects (myelotoxicity,64 hepatotoxicity65 and cardiomyopathy66) may potentially contribute to Adriamycin-induced nephropathy.
TIPS IN USING THE AN MODEL IN THE LABORATORY
The most important factor in successful use of this model is the dose of Adriamycin. As there are variations in batch potency and species sensitivity, dose-finding studies are usually required to ascertain the exact dose required to induce the pathological changes required to test the investigator's hypotheses. As little as 0.5 mg/kg difference in dose can mean the difference between success and failure, particularly in mice. The intravenous route of administration is preferable.
SUMMARY AND CONCLUSIONS
Adriamycin nephropathy is a well-established rodent model, which is analogous to human focal segmental glomerulosclerosis, characterized by reductions in glomerular filtration rate, proteinuria, glomerulosclerosis associated with changes in the glomerular filtration barrier, and tubulointerstitial fibrosis. The most common method of administration is intravenous via tail vein injection as it is most reproducible in inducing renal injury. Difficulties in using the model may arise due to a number of issues including batch variation and genetic variation in the rodent used. Notwithstanding these shortcomings, this model has facilitated the study of the pathophysiology and possible therapeutics of chronic proteinuric renal disease.
The authors acknowledge the National Health and Medical Research Council of Australia for their support.