Animal models of lupus nephritis: the past, present and a future outlook

Abstract

Lupus nephritis (LN) is the most severe end-organ pathology in Systemic Lupus Erythematosus (SLE). Research has enhanced our understanding of immune effectors and inflammatory pathways in LN. However, even with the best available therapy, the rate of complete remission for proliferative LN remains below 50%. A deeper understanding of the resistance or susceptibility of renal cells to injury during the progression of SLE is critical for identifying new targets and developing effective long-term therapies. The complex and heterogeneous nature of LN, combined with the limitations of clinical research, make it challenging to investigate the aetiology of this disease directly in patients. Hence, multiple murine models resembling SLE-driven nephritis are utilised to dissect LN's cellular and genetic mechanisms, identify therapeutic targets, and screen novel compounds. This review discusses commonly used spontaneous and inducible mouse models that have provided insights into pathogenic mechanisms and long-term maintenance therapies in LN.

Keywords:

• SLE
• lupus nephritis
• animal models
• renal pathology

1. Pathophysiology of SLE and LN

SLE is a prototypic systemic autoimmune disease characterised by the loss of tolerance to self-DNA, histones, and ribonucleoproteins [1,2]. Hallmark features of SLE include immune system activation and the production of autoantibodies, resulting in the formation of immune complexes (IC) that induce local inflammation and pathology in multiple organs [2,3]. The aetiology of SLE is incompletely understood, but disease initiation results from a combination of genetic susceptibility, hormonal changes, and numerous external factors (e.g. viral infection, exposure to heavy metals) [4]. The kidneys receive about 25% of the cardiac output and are highly vascular, rendering them susceptible to IC-mediated pathology.

Nephritis is a common and severe complication of SLE and can lead to end-stage kidney disease (ESKD) [5,6]. LN can present in multiple forms in the kidneys, including proliferative lesions with crescent formation, sclerosis, necrosis, and podocyte foot process effacement. The patterns of glomerular IC include mesangial deposits (classes I, II), subendothelial deposits with endocapillary hypercellularity or prominent glomerular basement membranes (GBMs) with necrotising and crescentic lesions (classes III and IV, focal and diffuse LN), or membranous forms (class V). Focal LN (class III) is present in about ∼30% of biopsied SLE patients [7]. A simplified cross-section and description of a glomerular capillary with areas of IC deposits are depicted in Figure 1.

Figure 1. A simplified cross section of the glomerulus is depicted. The glomerular assembly comprises of a fenestrated endothelium (pink), which is contiguous with the mesangium that houses the mesangial cells (green). As such, mesangial cells are constantly exposed to blood components. The fenestrated endothelium is lined by the glomerular basement membrane (cyan). The other side of the basement membrane is lined by a specialised type of epithelial cells called podocytes and their foot processes (cyan with pink blebs).

During the evolution of lupus nephritis (LN), circulating or in-situ formed immune complexes can deposit in the mesangial (orange, class I, II LN), sub-endothelial (blue, class III and IV) or sub-epithelial (red, class V) regions and induce cellular pathology and inflammation. Some of the common pathological features include (but are not restricted to) mesangial proliferation, basement membrane thickening, endocapillary proliferation, sclerosis, crescent formation with loss of podocyte foot processes.

Traditionally, LN is thought to initiate in the glomeruli. However, tubulointerstitial inflammation, fibrosis, and atrophy strongly correlate with renal dysfunction, independently of the extent of glomerular damage, and predict worse outcomes [8–11]. Genome-wide association studies (GWAS) indicated that genes driving autoimmunity and nephritis are distinct [12], suggesting that the progression to LN in SLE patients results from genetic susceptibility to systemic autoimmunity and increased end-organ sensitivity to inflammatory stimuli.

Given the high clinical heterogeneity in SLE and LN patients and the complexity of the kidney in which multiple cell types are involved, it is challenging to recapitulate human SLE and LN in cell culture models. Furthermore, more recent studies have implicated novel contributors to LN, such as immunometabolism [13], gut dysbiosis [14], and ferroptosis [15], and the fine mechanisms of these processes are incompletely understood. Hence, multiple animal models of spontaneous and induced SLE and LN that reproduce one or multiple aspects of human disease have been described and continue to evolve. These animal models are an indispensable tool for investigating the aetiology and genetics of SLE/LN, as well as identifying drug targets and validating therapeutics. Here, we discuss the principal murine models of SLE relevant to human LN, their characteristics, and significant contributions that have led to a better understanding of human disease. Table 1 summarises the discussed models and their specific advantages and disadvantages.

Table 1. Summary of the discussed models and their specific advantages and disadvantages.

Mouse model	Strain development	Kidney immune cell accumulation	Major clinical observations	Sex predominance and severity	Kidney histopathology	Mortality	Pros	Cons
MLR/lpr	Crossing strains C3H/Di, LG/J, C57BL/6, and AKR/J	CD4 − veCD8 − veB220 + ve T cells	Autoantibodies against dsDNA and Sm, IC mediated renal and skin pathology, Lymphadenopathy	Predominant in both. More severe in female	Endothelial and mesangial cell proliferation	70% mortality by 14 weeks	Rapid with high incidence and severity of disease development	Independent of Type I interferons and FcγRs
BXSB	C57BL/6 and SB/Le mice (B6 X SB/Le) F1 to SB/Le	B lymphocytes	Reactivity to ribonucleoproteins (RNP)	Male	Exudative, proliferative nephritis; Severe proteinuria, IgG and C3 deposits in the glomerular mesangium	Males- 90% mortality at eight months, Females- 50% at fifteen months	Permits studies on male LN and consequence of TLR7 signalling pathway in SLE	Unlike human lupus, lower incidence in females
NZB/NZW F1	Breeding between B/W and NZW mice	B cell activation	Splenomegaly, anti-dsDNA antibodies	Female	Proteinaceous deposits in the mesangium, proliferation of glomerular cells and tubular casts	50% mortality around eight months and >90% mortality after one year	Close phenotypic approximation of human SLE and LN	Long disease induction time and variability in disease severity
Pristane- induced	Injection of Pristane in BALB/c mice	Monocyte infiltration	Development of anti-RNP, anti- DNA and anti- histone antibodies, lung haemorrhagic, capillary damage- associated pathology, TNFα-driven arthritis and capillaritis	Female	Proliferative glomerulonephritis, moderate proteinuria, mesangial and sub-endothelial deposition of IgM and IgG with mesangial C3 deposits	Strain dependent	Valuable tool to investigate type 1 IFN-induced pathology, excellent model to demonstrate how an environmental factor induces lupus like disease	Unlike in human SLE, anti-DNA antibodies are almost entirely IgM. Long disease induction time
NTN	Injection of nephrotoxic serum antibodies against GBM	Neutrophil, mononuclear phagocytes and T cell infiltration	Renal failure	Gender independent	IgG deposits in glomeruli	Dependent on mouse strain and dose of antiserum	Independent of genetics and gender. Rapid induction of IC mediated chronic GN	Variable disease severity depending upon the source of antiserum. Limited availability of antiserum

The spontaneous murine models of SLE vary from polygenic to single-gene knockout or transgenic expression, resulting in lupus-like phenotypes [16,17]. None of the spontaneous or induced mouse models of lupus represent the entire clinical spectrum observed in SLE patients. However, these models have delineated functional pathways in SLE and identified specific genes and genetic loci critical in maintaining systemic immune tolerance and IC-glomerulonephritis and increasing susceptibility to SLE. Multiple reviews on murine models of SLE have focused on genetic links between human and mouse SLE [16,18] or the mechanisms leading to systemic autoimmunity and clinical lupus [19]. In addition, some of these models have provided insights into novel contributors to SLE, such as the microbiome [20], immunometabolism [21], or iron metabolism [22,23] as relevant to human SLE. As in humans, the kidney is the primary target organ in multiple murine models of SLE with a clear female gender bias (except for MRL/lpr and BXSB mice, discussed below). Each model represents an overlapping subset of the clinical features of the human disease and offers unique features of interest to address specific preclinical questions. In this review, we briefly summarise the most common mouse models of SLE, stressing their unique features, provide an update on the significant advances they have contributed to the field, and discuss the advantage of using one model over the others when testing fine mechanisms and specific hypotheses.

2. Spontaneous polygenic models

2.1. MRL/lpr

The Murphy-Ross Large (MRL) strain was developed by crossing several strains, including C3H/Di, LG/J, C57BL/6, and AKR/J [24,25]. One of the MRL sub-strains carries a spontaneous mutation named lpr for lymphoproliferation and develops a severe SLE-like phenotype characterised at the cellular level by the accumulation of double negative (DN: CD4 − CD8−) B220+ T cells. Like in SLE patients [26], the DN T cells are autoreactive and pathogenic [27], making this model relevant to SLE pathogenesis. Lpr corresponds to non-functional transcripts of the Fas gene, a key regulator of apoptosis in immune cells [28]. In this strain, male and female mice are affected (though females show a more pronounced phenotype) and produce autoantibodies against dsDNA and Sm, leading to large amounts of IC that induce renal and skin pathology [29]. Unlike SLE patients, these mice develop a massive lymphadenopathy. The data generated from the MRL/lpr strain has led to, among others, a better understanding of the role of TRL7 and TLR9 in lupus [30], differences in TLR activation in B cells and dendritic cells [31], and were used for the first time to report the critical role of the extrafollicular pathway for the development of autoreactive B cells [32]. Additionally, these mice have shown that the kidney infiltrating T cells in LN have an exhausted metabolic and functional phenotype [33]. Complement plays a critical role in nephritis in these mice, as a deficiency in the regulatory factor H accelerates kidney pathology [34,35]. MRL/lpr mice become azotaemic, leading to 70% mortality by 14 weeks of age. Unlike other lupus models (e.g. NZB/NZW F1 discussed below), nephritis in MRLlpr mice is independent of Type I interferons and FcγRs, which are two major drivers of SLE pathogenesis in general, including LN. This represents clear limitations to the relevance of the MRL/lpr model to human LN, in addition to the confounding massive lymphoproliferation [36]. C57BL/6-Fas^lpr mice expressing the lpr mutation on a non-autoimmune C57BL/6 (B6) background develop systemic autoimmunity without clinical pathology and reduced lymphadenopathy [37]. These B6.lpr mice have been used to investigate the involvement of IL-17 in lupus [38,39]. Finally, the MRL/lpr model has the major advantage of being the most rapid, with the highest incidence and severity of disease amongst the spontaneous models. Hence, it has been extensively used as a pre-clinical model of LN to test novel therapeutics [23,40]. For example, the MRL/lpr mice expressing human CD20 on their B cells were used to demonstrate the superiority of obinutuzumab, a type II anti-CD20 monoclonal antibody over rituximab [41]. These mice were also utilised to highlight the efficacy of idebenone, an antioxidant that bypasses complex I activity and enhances ATP synthesis [42].

2.2. The BXSB model

The BXSB strain is a recombinant inbred strain developed in the 1970s by crossing of C57BL/6 and SB/Le mice ((B6 X SB/Le) F1 to SB/Le, hence BXSB) [43,44]. The BXSB male mice develop an accelerated disease compared to females due to the Y chromosome-linked autoimmune accelerator (Yaa) [44,45], which has been identified as the translocation of a telomeric region of the X chromosome to the Y chromosome, resulting in the duplication of the toll-like receptor 7 (Tlr7) gene [46,47]. This increased TLR7 expression activates the Type I interferon cascade, which is critical in SLE pathogenesis [48]. BXSB.Yaa mice have a relatively low titre of autoantibodies reactive to DNA and histones but produce a higher titre of anti-ribonucleoproteins (RNP) IgG with typical speckled patterns [49]. Male BXSB mice develop exudative, proliferative nephritis with neutrophilic infiltrates and severe proteinuria by three months of age [29,50]. The kidneys show prominent IgG and C3 deposits in the glomerular mesangium and along the capillary walls, with 50% of the mice developing fatal nephritis starting around five months and 90% at around eight months of age [29,49]. Female mice develop a weaker disease phenotype with around 50% mortality around 15 months of age. Therefore, despite its predominant presentation in males, the BXSB/Yaa strain models the consequences of an overreactive TLR7/Type 1 IFN pathway and associated renal failure.

Hybrid (NZW × BXSB)F1 male mice carry the Yaa-associated two copies of Tlr7 and develop accelerated lupus with antiphospholipid syndrome (APS) [51] as well as proliferative glomerulonephritis [46]. Female (NZW × BXSB)F1 mice with only one active copy of Tlr7 develop late-onset glomerulonephritis but not APS [51]. In addition to inflammatory glomerulonephritis, the males develop immune-mediated thrombocytopenia and a thrombotic vasculopathy with myocardial infarcts, myocardial fibrosis, and dilated cardiomyopathy [52,53]. Given the importance of antiphospholipid antibodies in lupus [54], this model is a valuable tool to dissect the mechanisms and interventions in SLE-associated antiphospholipid syndrome [53].

2.3. The NZB/NZW F1 model

The NZB/NZW F1 hybrid (NZB/W F1) is a spontaneous model of SLE and LN [55,56] and it is phenotypically a close approximation of human SLE and LN. As in humans, female mice are more severely affected, and anti-dsDNA antibodies are detected at around three months of age [57]. Female mice develop splenomegaly and chronic polyclonal B cell activation [29,50]. The kidney pathology presents as sclerosis with proteinaceous deposits in the mesangium, diffuse proliferation of glomerular cells, and tubular casts. As in LN patients, prominent glomerular crescents are observed, and large periglomerular and interstitial inflammatory infiltrate [29,55]. A few critical contributions of the NZB/W F1 model to the field include the identification of the therapeutic effect of BAFF inhibition [58], the role of type 1 IFN [59], and the identification of LN biomarkers [60]. We have shown that early treatment with a combination of metformin and CTLA4Ig decreases the development of LN in this model without attenuation of autoantibody production [61].

The serendipitous breeding between B/W and NZW mice led to the development of a panel of inbred New Zealand Mixed (NZM) mouse strains, several of which showed a highly penetrant SLE and LN resembling disease in parental NZB/W F1mice [56,62]. Pioneering studies on identifying the SLE susceptibility loci in murine SLE were conducted with the NZM2410 strain [62,63]. Identifying three lupus susceptibility loci (Sle1, Sle2, and Sle3 on chromosomes 1, 4, and 7, respectively) in the NZM2410 strain has also led to the development of congenic strains with unique advantages. Intercrossing the three Sle1, Sle2, and Sle3 susceptibility loci was sufficient to develop a lupus phenotype on the non-autoimmune B6 genetically resistant strain [63,64]. The B6.NZM2410.Sle1.Sle2.Sle3 is unique as it shares 95% of its genome with B6 and has provided original insight into the metabolic control of the disease [65–67]. The corresponding single (Sle1) and bi-congenic (Sle1.Sle3) are also well suited to breed with B6-based gene knockouts. For instance, deletion of the plasmacytoid dendritic cells (pDC)-specific transcription factor Tcf4 in B6.Sle1.Sle3 mice provided genetic evidence that pDCs are critically involved in developing SLE [68]. Furthermore, the transfer of bone marrow cells from B6.Sle1 mice into irradiated B6 recipients was sufficient to recapitulate the autoimmune phenotypes of B6.Sle1 mice, demonstrating that they are entirely accounted for by haematopoietic cells [69]. Bone marrow transfer from these NZM2410-derived congenics into mice with specific gene knockout (e.g. in the kidney) provides a powerful tool for studying end-organ resistance or susceptibility to lupus.

A notable mention in the NZM sub-strains is NZM2328 [70,71], which develops nephritis in two stages: acute glomerulonephritis followed by more chronic nephritis [72,73]. These mice were vital in showing that autoantibodies, especially anti-dsDNA antibodies, are not required to develop renal disease in SLE [74,75].

3. Inducible models of SLE and LN

While the spontaneous models of lupus have greatly enhanced our understanding of human SLE and LN, these models exhibit delayed and variable onset of SLE-related symptoms and renal pathology, and most have complex genetic backgrounds. These factors limit the application of these models for testing therapeutics, investigating environmental factors, and the contribution of individual genes in the pathogenesis of SLE and LN. Thus, exposing wild-type non-autoimmune mice or mice with specific gene deficiencies in individual cell types to chemical agents and toxins is an alternative valuable approach to induce an SLE-like disease. These agents mimic the impact of environmental, genetic, and non-genetic factors on susceptibility or resistance to SLE and LN and in evaluating the efficacy of therapeutic drugs. Below, we discuss two commonly used approaches to induce lupus and immune-complex mediated glomerulonephritis.

3.1. Pristane-induced lupus

Initially used for priming mice to obtain ascite fluid to produce monoclonal antibodies, an intraperitoneal injection of pristane (2,6,10,14 tetramethylpentadecane, a major component of mineral oil) leads to the development of anti-RNP, anti-DNA and anti-histone antibodies in BALB/c mice [76,77]. In this inducible model, the BALB/c and SJL strains present glomerular IgG and complement deposits with cellular proliferation and develop proteinuria, whereas B6 mice develop milder (Class II like) disease [78]. In contrast to most spontaneous models of SLE, the anti-RNP antibodies class switches to IgG, while anti-DNA antibodies are almost entirely IgM [77]. These mice gradually develop nephritis six months after injection with focal to diffuse proliferative glomerulonephritis and moderate proteinuria [77]. Pristane-treated BALB/c mice also develop lung haemorrhagic, capillary damage-associated pathology, and TNFα-driven arthritis that may resemble arthralgia and capillaritis occurring in SLE patients [79–81]. In the SJL strain, pristane-induced lupus is more severe in females than males [82]. Compared to males, the kidneys of pristane-injected SJL female mice present with significantly more glomerular hypercellularity and necrosis [82]. Pristane-induced lupus is driven by type 1 IFN [83] and thus, it is a valuable tool for investigating type 1 IFN-induced pathology, a critical pathogenic pathway in human lupus. This model was used to define the protective role of TLR9 in lupus as BALB/c mice with TLR9 deficiency developed more severe lupus than their WT TLR9-sufficient mice [84]. Pristane-induced lupus is an excellent example of how an environmental factor can induce lupus-like disease in a non-autoimmune background.

3.2. Nephrotoxic serum nephritis

The nephrotoxic serum nephritis (NTN), also known as anti-glomerular basement nephritis, is an inducible IC-mediated model of glomerulonephritis that resembles human Goodpasture’s Disease [85,86]. The antigen (anti-mouse glomerular basement membrane antibodies raised commonly in rabbits or sheep (donor) after immunisation with mouse glomerular preparation) binds the mouse glomerular basement membrane (GBM), leading to IC deposition and inflammatory response. Most nephrotoxic serum preparations are crude as they contain immunoglobulins against other structural elements of the kidney. Administration of the heterologous donor anti-GBM antiserum (NTS) leads to complement-dependent, transient renal injury (day 0–4) [87], followed by the development of mouse anti–rabbit or sheep antibodies in the later, autologous stage (day 7–14) during which IC deposits are observed in the glomeruli, resulting in glomerulonephritis with renal failure [88]. A modified accelerated model of NTS has also been developed wherein mice are pre-sensitized to the heterologous anti-GBM serum by preimmunization with normal, donor heterologous serum a few days (Day −4) before the introduction of the heterologous anti-GBM serum (Day 0), thus initiating the autologous phase of disease concomitant with the introduction of the anti-GBM serum [88,89].

The binding of polyclonal antibodies to the glomerular endothelial wall and GBM antigens triggers local immune activation and the secretion of proinflammatory chemokines and cytokines. In the heterologous phase, glomerular injury depends largely on complement activation and neutrophil-dominated inflammation [90–92]. During the autologous phase, the renal dendritic cells promote nephritogenic T-cell responses [93]. At this stage, a Th1 response stimulates the macrophages to mediate renal damage in a classical type IV hypersensitivity reaction [94], though a pathogenic Th17 response has also been reported [95]. Using this model, we have recently shown the importance of iron-mediated tubulointerstitial injury in the pathogenesis of LN [96]. Since the phenotype of NTN involves changes in the renal functional parameters, glomerular and secondary tubulointerstitial injury, renal leukocyte infiltration, and adaptive cellular and humoral immune responses [96], it serves as an all-encompassing tool to tease out delicate mechanisms associated with LN-like chronic IC-induced GN. The NTN model allows investigations into the pathogenesis of IC-mediated glomerulonephritis independent of genetic background and gender and is highly synchronised.

3.3. Chronic graft-versus-host model of systemic lupus erythematosus

Graft versus host disease (cGVHD) models are generated by transferring major histocompatibility complex (MHC) mismatched immune cells into a recipient/host. Compared to the acute GVDH model [97], the chronic GVHD (cGVHD) model, where I-A^bm12B6(C)-H2-Ab1^bm12/KhEgJ (bm12) splenocytes are transferred into B6 mice produce symptoms closely resembling SLE. In this model, the adoptive transfer of immune cells from bm12 mice, a strain identical to B6 except for three amino acid substitutions on the MHC class II _I-A chain, into I-A^b B6 mice leads to alloactivation of donor CD4+ T cells by recipient APCs [98]. The activated alloreactive donor CD4+ T cells then provide help to host self-reactive B cells [99,100]. Some of the key features include the expansion of donor-derived T follicular helper (Tfh) cells, recipient-derived germinal centre (GC) B cells and plasma cells, and the production of anti-dsDNA, anti-ssDNA, anti-chromatin, and anti-RBC antibodies [98]. During the later phase, the recipient mice develop glomerulonephritis associated with IgG deposits in the kidneys’ glomerular, interstitial, and vascular segments [101]. Host CD4+ T cells do not participate in the cGVHD induction, but their presence is essential for nurturing B cells through IL-4 and/or CD40 stimulation [101,102].

Like the NTN model, there are several advantages of this model. Genetically modified B6 mice can serve as either donors or recipients. This allows dissecting the contribution of specific genes and modified cells to disease progression. Furthermore, disease development in the bm12 model requires only two weeks for ANA production, compared to several months for most spontaneous models [103]. Furthermore, unlike spontaneous models that develop disease at different time points, the disease onset and progression in this model is highly synchronised. This permits the use of appropriately sized cohorts to test novel therapeutics during different stages of SLE/LN.

3.4. Humanised models of lupus

Though mouse models provide insights on SLE/LN pathogenesis, they do not fully recapitulate the complexities of human SLE. There are major differences between the mouse and human immune system [104,105] and genetics [106,107] pose challenges in understanding the complexities of SLE and evaluation of therapeutic protocols [105]. There have been some attempts to bridge the differences between the murine and human immune systems by using humanised mice as models of SLE [108]. Immune cell-mediated lupus-like disease can be initiated by engrafting immunodeficient mice with either peripheral blood leukocytes from SLE patients or healthy haematopoietic stem cells, followed by lupus induction with pristane [108]. Proteinuria, a marker of kidney injury in LN, was observed using both approaches [109,110]. The first approach was used to evaluate the therapeutic potential of targeting the Kv1.3 channel on T cells with nanoparticles [109]. However, the injury induced by human immune cells to murine renal parenchyma may not provide a reliable model for in-depth mechanistic studies or hold clinical relevance.

4. Conclusions

Novel pathogenic mechanisms of human SLE and LN are continuously identified. Progress has been made in the understanding of genetic risks to LN. For example, polymorphisms in the human major histocompatibility complex region increase the risk of disease development regardless of ethnicity [111,112]. In contrast, the apolipoprotein L1 gene (APOL1) risk alleles G1/G2 are more prevalent in African Americans than European Americans [113]. Even though this knowledge is more than two decades old, invasive kidney biopsy is still the gold standard for diagnosing and guiding the treatment of LN. Clinical research has severe limitations when evaluating the causal role of individual genes, environmental factors, and novel interventions to tease out fine mechanisms contributing to disease outcomes.

Herein lies the strength of spontaneous and inducible animal models of SLE/LN. As such, no singularly perfect model recapitulates all aspects of human SLE and LN. An astute selection of animal models will continue to help elucidate the intricacies of the genetics, immune, and parenchymal cell contributions to the pathogenesis of human disease. In summary, animal models of SLE/LN provide an understanding of the dominant and fine mechanisms contributing to disease. As such, animal models offer the opportunity to develop novel strategies for interventions in SLE/LN, which translates into more timely and effective therapy.

Disclosure statement

The authors have declared that no conflict of interest exists.

Additional information

Funding

Supported by grants from Vifor Pharma (P0213104, P0226952) and NIH (RO1DK136011) to Yogesh Scindia and Laurence Morel.