Development of a unilateral ureteral obstruction model in cynomolgus monkeys

Abstract Background Chronic kidney disease (CKD) has a high global prevalence and large unmet need. Central to developing new CKD therapies are in vivo models in CKD. However, next‐generation antibody, protein, and gene therapies are highly specific, meaning some do not cross‐react with rodent targets. This complicates preclinical development, as established in vivo rodent models cannot be utilized unless tool therapeutics are also developed. Tool compounds can be difficult to develop and, if available, typically have different epitopes, sequences, and/or altered affinity, making it unclear how efficacious the lead therapeutic may be, or what dosing regimen to investigate. To address this, we aimed to develop a nonhuman primate model of CKD. Methods In vivo rodent unilateral ureteral obstruction (UUO) models kidney fibrosis and is commonly used due to its rapidity, consistency, and ease. We describe translation of this model to the cynomolgus monkey, specifically optimizing the model duration to allow adequate time for assessment of novel therapeutics prior to the fibrotic plateau. Results We demonstrated that disease developed more slowly in cynomolgus monkeys than in rodents post‐UUO, with advanced fibrosis developing by 6 weeks. The tubulointerstitial fibrosis in cynomolgus monkeys was more consistent with human obstructive disease than in rodents, having a more aggressive tubular basement expansion and a higher fibroblast infiltration. The fibrosis was also associated with increased transglutaminase activity, consistent with that seen in patients with CKD. Conclusion This cynomolgus monkey UUO model can be used to test potential human‐specific therapeutics in kidney fibrosis.


| INTRODUC TI ON
Chronic kidney disease (CKD) is one of the most common diseases worldwide with kidney fibrosis being a histological hallmark. 1,2 Several animal models of CKD have been developed to understand pathogenesis and verify disease targets for therapeutic treatment.
Unilateral ureteral obstruction (UUO) is a model resembling human obstructive nephropathy. Although it is not a common cause of human renal disease, it is a well-recognized and frequently used model of kidney damage with subsequent fibrosis development. [3][4][5] By ligating a ureter of one kidney, the build-up of urine elevates intratubular pressure; this results in reduction in renal blood flow and glomerular filtration rate, as well as other changes such as interstitial inflammation, tubular dilation, tubular atrophy, and ultimately, fibrosis. Rodent UUO models have been utilized to investigate mechanisms of tubulointerstitial fibrosis, with severe fibrosis being detected within 21 days. 6,7 Development of fibrosis in the rodent UUO model is associated with changes in extracellular transglutaminase (TG) 2 8 and is a hallmark of human disease pathology. 9 TG2 is a calcium-dependent enzyme that crosslinks extracellular matrix (ECM) proteins to form a stable and proteolytic resistant ε(γ-glutamyl)-lysine dipeptide crosslink. 10 This increased crosslinking in ECM accelerates the rate of ECM deposition, while making it less susceptible to degradation by ECM proteases. Increases in extracellular TG2 expression, TG activity, and ε(γ-glutamyl)-lysine dipeptide crosslinks have been demonstrated in all types of human CKD, and are highly correlated with fibrosis levels. 9 Studies performed using rodent models of kidney fibrosis have shown increased TG2 and that blocking its activity slows renal fibrosis. 8,[11][12][13] A study is currently underway to evaluate inhibition of TG2 in patients with CKD (NCT04335578). TG2 is therefore a useful mechanistic tool to evaluate if the pathological processes of CKD occurring in an in vivo model are consistent with those in humans.
Although some proteins are homologous between species, antibodies targeting functional human epitopes (i.e., therapeutic antibodies) can be highly species-specific as they often recognize conformational epitopes, whereby changes in a single amino acid in the epitope can raise affinity and lower half-maximal inhibitory concentration (IC 50 ) significantly. Therefore, antibodies optimized for human proteins may not sufficiently cross-react with other species to allow testing in commonly used rodent in vivo models.
An option to mitigate this is to use a surrogate tool antibody; however, binding at a different epitope may complicate interpretation of data, provide different pharmacokinetics and thus, less valuable translational data to predict efficacious doses in early clinical studies. It is becoming increasingly important to use models in species with high homology and close physiology to humans to evaluate next-generation therapeutic antibodies and other modalities where species homology plays a significant role such as gene, small interfering RNA (siRNA), and oligonucleotide therapies. The nonhuman primates (NHP) most commonly used in medical research are from the genus Macaca, or more specifically the cynomolgus monkey. Their consensus with the human proteome 14 allows a high degree of cross-reactivity, making them a suitable species for studies with highly specific antibodies or other new therapeutic modalities. Models of liver 15 and lung fibrosis 16 have been established in cynomolgus monkeys and there are NHP models of renal transplantation/rejection. 17 However, to date, there are no models of renal fibrosis in NHP. Therefore, we hypothesized that a cynomolgus monkey UUO model could be developed and validated. The UUO model surgery is simple (compared with other models such as 5/6th subtotal nephrectomy) and the remaining functional kidney prevents problems with organ failure and any associated welfare issues such as end-stage renal failure. Translating the UUO model from mice to rats to rabbits was relatively consistent and quick; thus, we did not anticipate issues in translating the model to cynomolgus monkeys. Given the biggest variable between species when using the UUO model is the time to develop fibrosis consistent with end-stage renal disease, we undertook a time-course evaluation of tubulointerstitial fibrosis post-UUO to understand the optimal duration of the model and evaluate expression of a human-relevant pathological process currently being tested in the clinic: TG2.

| Animal study
Fourteen male cynomolgus monkeys (Hainan Jingang Biotech Co. Ltd., China) were used in the study. All animals were tuberculosisnegative and free of viral, bacterial, or parasitic infections (including simian immunodeficiency virus, respiratory syncytial virus, simian-T-lymphotropic virus, tuberculosis, Schmallenberg virus, shigellosis, and salmonella). Adult monkeys (aged 7-8 years) were used in this study to more accurately model human disease. Common tests for kidney disease were performed to ensure that the animals had no proteinuria (urinary protein <10 mg/dl, test with visual urinary test strip) or loss of renal function (serum creatinine <100 µmol/L, blood urea nitrogen <9 mmol/L) at study initiation.  Table 1. Animals were closely monitored post-surgery; their appearance, body weight, temperature, respiratory rate, and behavior were assessed daily and their skin wound checked by an onsite vet every day until it healed. All UUO animals were provided with soft bedding after surgery until normal behavior resumed (scored and recorded during each observation). In general, the UUO surgical procedure was associated with moderate pain, and all animals recovered within a few days. At harvest, animals were euthanized by intravenous injection of 80 mg/kg pentobarbital, and kidneys dissected longitudinally and each half cut into 4 segments. One segment from each half kidney was fixed in 10% neutral formalin solution and embedded in paraffin blocks. The other segments were snap frozen and used for in situ TG activity, extracellular TG2 staining, and hydroxyproline detection.
Researchers performing sample analysis did not know the group allocation until data were generated.

| In situ TG activity
Detection of in situ TG activity was performed as described pre-

| Extracellular TG2 antigen staining
Immunodetection of extracellular TG2 was performed as previously described. 9

| Measurement of kidney fibrosis levels, in situ TG activity and extracellular TG2 protein
Images of PSR-stained sections were downloaded onto Tissue Studio (Definiens Inc., USA). Selection of the region of interest (ROI) was performed on sections, separating out the cortex for analysis ( Figure 1A). Definiens Stain Picker options and "computer training/ learning" methods were used to generate an algorithm to identify collagen staining (red and intense staining overlaid in orange) and nuclei (blue) ( Figure 1B,C). Definiens automatically calculates the areas of the ROI, and the 3 marker areas for each slide. The ratio of collagen area to cortical area was calculated to represent the fibrotic index.
To analyze images of TG activity and extracellular TG2 protein, the dedicated Definiens Fluorescence workspace was used. ROI detection was performed automatically and corrected manually to remove the medulla from analysis. Marker detection was then performed on different image layers depending on the marker. DAPI nuclear stain (blue) was projected as green and analyzed on layer 1 (green overlay), and TG activity (yellow) or extracellular TG2 (green) was projected as red and analyzed on layer 2 (red overlay, intense staining overlaid in yellow) ( Figure 1D). The area of staining was determined by the Definiens software, and levels of TG activity or extracellular TG2 were calculated using the ratio of the staining area to DAPI nuclear stain area.

| Hydroxyproline analysis
Kidney homogenates containing 5 mg of protein per sample were hydrolyzed in 6 mol/L hydrochloric acid at 110°C for 24 h. These were centrifuged at 18 000 g for 2 min and the supernatant transferred to clean tubes and freeze dried. Samples were resuspended in 500 µl of lithium loading buffer (Biochrom, UK) and 30 µl fractionated using a lithium chloride gradient on a Biochrom 30+ amino acid analyzer using the manufacturer's standard protocol and expressed as µmol/ mg protein.

| Statistical analysis
Raw data are shown with each symbol representing one animal.
Data analyses were performed using one-way analysis of variance (ANOVA) by a Fisher's least significant difference multiple comparisons test (GraphPad Prism). A significance level of 5% (p < .05) was adopted throughout.

| In situ TG activity, extracellular TG2 expression and levels of ε-(γ-glutamyl)-lysine crosslink
In the healthy (sham) kidney, low TG activity was seen in the glomeruli and tubular basement membrane/tubulointerstitial space, based  Figure 5A) in contrast to TG2 extracellular antigen, which plateaued at weeks 5 and 6 ( Figure 5B). When image analysis was performed to quantify changes in TG activity, 3-, 6-, and 8-fold increases were seen at weeks 4, 5, and 6, respectively ( Figure 5C).
In the kidneys of healthy (sham) cynomolgus monkeys, small amounts of extracellular TG2 were detected and located in the tubular basement membrane/tubulointerstitial space, periglomerular area and mesangial matrix ( Figure 5B). In the kidneys of cynomolgus monkeys  Figure 5D).

Analysis of ε-(γ-glutamyl)-lysine crosslink in kidneys with UUO
showed a progressive trend of elevated crosslink dipeptide that reached a maximum at 5 weeks post-UUO and remained high to week 6 ( Figure 5E).

| Correlations between levels of in situ TG activity versus ε-(γ-glutamyl)-lysine crosslink and hydroxyproline
Correlation between in situ TG activity and ε-(γ-glutamyl)-lysine crosslink was r = .52 ( Figure 6A). There was also a correlation between in situ TG activity and hydroxyproline, strongly linking the accumulation of collagen with TG2 in this model (r = .82, Figure 6B).

| D ISCUSS I ON
This study sought to transfer the UUO model of CKD from rodents to cynomolgus monkeys, with the aim of allowing development of human-specific therapeutics, not cross-reactive with rodents and with a novel mode of action, for which in vivo pharmacological data are essential to define dosing for phase I human studies.
Understanding the time course of the disease process was key to allowing optimal selection of study duration and relating this to changes in collagen levels (and thus fibrosis), while using TG2 as an exemplar molecule of clinical stage target in CKD. The model was developed in adult animals to assist with translation, as CKD is typically a disease of later life in which changes occur in multiple parameters associated with fibrosis, including cell senescence, 19,20 promoter methylation, 21 ECM homeostasis, 22 and hemodynamics. 23 Overall, these data demonstrate a clear development of tubulointerstitial fibrosis over the 3-to 6-week study period, although data from the different methods used to assess the extent of fibrosis in this study were not totally aligned. Manual assessment of the sections would suggest fibrosis progressed over the 6-week study period, approaching end-stage remodeling around or just after week 6.
Image analysis of PSR suggested that the level of interstitial collagens plateau from week 4 onwards, whereas total kidney hydroxyproline is progressive until week 5. In terms of TG2, the level of activity progressed over the 6 weeks, with maximum extracellular TG2 antigen reached by week 4, and the formation of ε-(γ-glutamyl)-lysine peaking at week 5. This poses the question of the optimal time to run the model when being used to test interventional efficacy. It is generally accepted that the level of fibrosis will ultimately reach a plateau in non-functional models such as the UUO. Therefore, model duration must be optimized to allow the maximum window for measurement of interventional success before the linear formation of scar tissue declines and hides any beneficial effect on slowing progression of fibrosis. Based on the composite data here, a 4-week model in aged cynomolgus monkeys would provide the optimal balance between the disease window and the slowing of histological and mechanistic changes that underpin fibrosis. However, given the heterogeneity of data due to the intentionally small animal numbers used, it is possible that the optimal duration of this model to maximize the disease window could be 5 weeks.
Although the data allow us to conclude successful model develop- In conclusion, this study has successfully transitioned the rodent UUO model of CKD to a NHP. The model progresses to completion in approximately 6 weeks compared with 3 weeks in mice and rat models, with a suggestion that an optimal study period for testing therapeutics to avoid disease plateauing would be 4 weeks. The NHP model demonstrates a much more aggressive tubular interstitial fibrosis than rodents, with a larger expansion of the tubular basement membrane and greater infiltration of interstitial cells giving a histological picture closer to obstructive disease in humans. This NHP model can be applied to test human-specific anti-fibrotic therapeutic molecules to determine efficacy, target engagement, and pharmacokinetics.

ACK N OWLED G M ENTS
The authors would like to thank Annick Cauvin for her constructive advice during her review of the manuscript. The authors acknowledge Veronica Porkess, PhD, of UCB Pharma, for publication and editorial support. The authors also acknowledge Sarah Hibbert, PhD, of Ashfield MedComms, an Ashfield Health company, for editing support that was funded by UCB Pharma in accordance with Good Publication Practice (GPP3) guidelines (http://www.ismpp.org/gpp3).

F I G U R E 6
Correlations between in situ TG activity and ε-(γglutamyl)-lysine crosslink (A), and between in situ TG activity and hydroxyproline (B). Mean of ε-(γ-glutamyl)-lysine crosslink, in situ TG activity and hydroxyproline for each animal were calculated and used for correlation analysis, which was performed using a Pearson's correlation analysis. TG, transglutaminase; UUO, unilateral ureteral obstruction

CO N FLI C T O F I NTE R E S T
L.H., T.D., and T.S.J. are employees of UCB Pharma and may hold/have access to stock options. Z.S. is an employee of Prisys Biotechnologies Co., Ltd., and J.N. was employed by Prisys Biotechnologies Co., Ltd.
at the time this work was conducted.