Insights from engraftable immunodeficient mouse models of hyperinsulinaemia

Hyperinsulinaemia, obesity and dyslipidaemia are independent and collective risk factors for many cancers. Here, the long-term effects of a 23% Western high-fat diet (HFD) in two immunodeficient mouse strains (NOD/SCID and Rag1 −/−) suitable for engraftment with human-derived tissue xenografts, and the effect of diet-induced hyperinsulinaemia on human prostate cancer cell line xenograft growth, were investigated. Rag1 −/−and NOD/SCID HFD-fed mice demonstrated diet-induced impairments in glucose tolerance at 16 and 23 weeks post weaning. Rag1 −/− mice developed significantly higher fasting insulin levels (2.16 ± 1.01 ng/ml, P = 0.01) and increased insulin resistance (6.70 ± 1.68 HOMA-IR, P = 0.01) compared to low-fat chow-fed mice (0.71 ± 0.12 ng/ml and 2.91 ± 0.42 HOMA-IR). This was not observed in the NOD/SCID strain. Hepatic steatosis was more extensive in Rag1 −/− HFD-fed mice compared to NOD/SCID mice. Intramyocellular lipid storage was increased in Rag1 −/− HFD-fed mice, but not in NOD/SCID mice. In Rag1 −/− HFD-fed mice, LNCaP xenograft tumours grew more rapidly compared to low-fat chow-fed mice. This is the first characterisation of the metabolic effects of long-term Western HFD in two mouse strains suitable for xenograft studies. We conclude that Rag1 −/− mice are an appropriate and novel xenograft model for studying the relationship between cancer and hyperinsulinaemia.

combined immunodeficiency (SCID), non-obese diabetic/severe combined immunodeficiency (NOD/SCID), and NOD/SCIDIL2Rγ (NSG) mice, may be resistant to developing HFD-induced metabolic syndrome, due to a lack of adaptive immunity and defective innate immunity 22,23 . For example, NOD/SCID mice, which are immunocompromised due to a spontaneous mutation in Prkdc 24 , develop streptozotocin-induced, but not diet-induced hyperglycaemia 25 . Streptozotocin-induced pancreatic insulitis, a well-established model of type I diabetes mellitus, destroys pancreatic β-cell function, thereby abrogating insulin secretion. This model does not recapitulate the complex interplay between hyperinsulinaemia, β-cell stress and apoptosis, and insulin resistance phenotype of T2DM, however 26 . In contrast, mice with inactivating mutations in the genes encoding the RAG1 or RAG2 proteins, Rag1 and Rag2, are susceptible to diet-induced hyperglycaemia 20,21,27 . RAG1 and RAG2 are involved in activating the recombination of T-cell receptor molecules and immunoglobulin genes, and a null mutation of either of these genes results in adaptive immunity deficiencies, with an absence of mature B and T lymphocytes 28 . Rag1 −/− mice backcrossed onto a C57BL/6 J genetic background rapidly develop insulin resistance one week after initiation of 60% HFD feeding 20 . Additionally, when fed a 42% fat diet, Rag1 −/− mice on a C57BL/6 J background gain more weight and fat mass than wild-type C57BL/6 J mice. Both Rag1 −/− and wild-type C57BL/6 J mice exhibit glucose and insulin intolerance after 10 weeks on this diet 21 . SCID, NOD/SCID and NSG mice are often employed in xenograft studies, as the rate of xenograft establishment is high for many cancer cell lines 29 . These strains are resistant to diet-induced insulin resistance 30 , and therefore, are not suitable for studying tumour biology associated with hyperinsulinaemia.
The aim of our study was to develop a diet-induced model of hyperinsulinaemia in immunocompromised mice suitable for cancer xenograft studies. In order to confirm previous studies 22,23 , and to determine if Rag1 −/− mice provide a better engraftable model for metabolic dysfunction, we compared the effect of a normal, low-fat chow and a Western, moderate-fat diet (23% fat diet, 45% digestible energy from fat) on weight gain, glucose tolerance, hormone levels and adiposity in Rag1 −/− mice (on a C57BL/6 J background) and NOD/SCID mice. This study is the first to show that a Western HFD results in more significant diet-induced metabolic dysfunction in Rag1 −/− mice compared to NOD/SCID HFD-fed mice. Furthermore, a pilot study demonstrates that LNCaP human prostate cancer cell line xenografts grow more rapidly in Rag1 −/− mice fed a Western HFD than tumours in control Rag1 −/− mice fed a low-fat diet. The Rag1 −/− mouse, therefore, provides a novel model for studying tumour biology associated with diet-induced hyperinsulinaemia.

Discussion
The present study is the first to investigate the long-term (28 week) metabolic effects of a Western, 23% high-fat diet (HFD) on two immunodeficient mouse strains suitable for human xenograft implantation. We demonstrate that the metabolic effects of 23% HFD (46% energy from fat) are more pronounced in Rag1 −/− compared to NOD/SCID mice, manifesting as higher fasting insulin levels, increased insulin resistance and steady-state β-cell function, lower insulin sensitivity, and increased adipose accumulation in skeletal muscle and liver tissue. The Rag1 −/− strain is, therefore, a suitable immunodeficient mouse model for investigating HFD-induced hyperinsulinaemia. Furthermore, our pilot LNCaP mouse xenograft study revealed that a moderate HFD promotes tumour growth. Compared to low-fat chow-fed Rag1 −/− mice, HFD-fed Rag1 −/− mice developed tumours earlier, and exhibited increased tumour growth over time and decreased survival to ethical endpoint.
Rag1 −/− mice in this study demonstrated impaired glucose tolerance within 16 weeks of HFD initiation. We surmise that this is likely to be a reflection of their predominantly C57BL/6 J genetic background (N6). It is well established that the C57BL/6 J strain is genetically predisposed to developing metabolic syndrome when fed a HFD 18,33 . Previous studies, in which Rag1 −/− and C57Bl/6 J mice were fed a Western diet (42.2% calories from milk fat and 42.8% calories from carbohydrate) for 11 weeks, demonstrated that these strains develop impaired glucose tolerance compared to low-fat diet fed mice, and no significant difference was observed between Rag1 −/− and C57BL/6 J HFD-fed mice 21 . The Rag1 −/− mouse bred onto a C57BL/6 J genetic background is, therefore, a useful model for studying the effects of insulin resistance, hyperinsulinaemia and, potentially, other components of the metabolic syndrome on cancer development and progression.
NOD/SCID and NSG mice are commonly used for xenograft studies due to their high tumour engraftment rate 34 , however, they are relatively resistant to diet-induced hyperinsulinaemia 22 as they lack fully competent  immune systems. Rag1 −/− mice mice (on a BALB/c background) have been used for cancer allograft and xenograft studies for prostate cancer 35 . Similarly, Rag2 −/− mice, which lack mature B and T lymphocytes, due to a mutation in the recombination-activating gene encoding the RAG2 protein 36 , have been employed to study endometrial cancer 37 and oral squamous cell carcinoma 38 . Classically, Rag1 −/− mice may not be considered as amenable to xenograft studies, as they retain some innate immune function, including moderate natural killer cell (NK) activity, which reduces engraftment rate and may distort the architecture of engrafted tumours 39 . However, as the development of insulin resistance is an inflammatory process 2 , the presence of some innate immunity may be advantageous for the study of HFD and cancer. Indeed, we reveal that engraftment of human prostate cancer cell line xenografts is possible in both HFD and chow-fed Rag1 −/− mice, with tumours developing to a palpable size more rapidly in HFD-fed mice. Impaired glucose and insulin tolerance have been reported to develop in Rag1 −/− mice as early as 1 week after initiation of a 60% HFD 20 , however, this high-fat content is not representative of a human diet. The diet used in our study is closer to a Western human diet, with 46% energy from fat, (14.31% of which is saturated fat), 34% carbohydrate and 20% protein. A study by Liu and colleagues 21 employed a comparable diet, with similar energy from fat (42.2% energy) but a higher carbohydrate (42.8%) and lower protein content (15%) than the diet used in our study. Studies where mice were fed 60% or 46% HFD 20, 21 revealed similar diet-induced metabolic changes, including significantly impaired glucose tolerance in HFD-fed mice compared to chow-fed controls after 10-16 weeks on the diet 21 . Our study investigated chronic effects (28 weeks) of a high-fat diet, permitting the assessment of pancreatic β-cell function over time and demonstrating that increased β-cell activity compensates for hyperglycaemia, allowing a progression to hyperinsulinaemia. Our study is novel, as it compares the metabolic effects of HFD on two commonly-used xenograft hosts, Rag1 −/− and NOD/SCID mice. Both strains lack a competent adaptive immune system, however, only the NOD/SCID mice lack competent innate immunity 22,23 . To the best of our knowledge, this is the first study of a Western HFD in Rag1 −/− mice. In these mice, fasting blood glucose measurements at 28 weeks, and glucose tolerance at 23 weeks, improved compared to 16 weeks; possibly as a result of β-cell compensation and a significant increase in blood insulin levels. HOMA revealed significantly greater insulin resistance and steady-state β-cell function, and lower insulin sensitivity compared to low-fat chow-fed mice. This is likely to reflect the diabetogenic C57BL/6 J genetic background of Rag1 −/− mice. Similarly, glucose tolerance improves with age in C57BL/6 J mice fed a normal chow diet as a result of age-related increases in islet size and pancreatic insulin content 40 . Our pilot LNCaP xenograft study (conducted in mice backcrossed with C57BL/6 mice for 10 generations) demonstrates that Rag1 −/− mice fed a Western HFD display changes in metabolic parameters, however, greater sample size is needed to determine if these changes are statistically significant.
Although glucose tolerance in HFD-fed Rag1 −/− mice improved with time, glucose tolerance in NOD/SCID mice became progressively impaired. NOD/SCID HFD-fed mice also developed symptoms of metabolic disturbance, including increased plasma insulin, insulin resistance, hepatic steatosis, and increased adipocyte size, however, these changes were more pronounced in HFD-fed Rag1 −/− mice. Although present in Rag1 −/− HFD-fed mice, intramyocellular lipid was not observed in NOD/SCID mice. Previous studies on the effect of 46% or 60% HFD in Rag1 −/− mice investigated adipose tissue accumulation in subcutaneous, epididymal, mesenteric and perirenal depots, but lipid accumulation within metabolic organs crucial for energy homeostasis, such as the liver and skeletal musculature, were not measured 20,21 . Increased hepatic steatosis and skeletal muscle lipid accumulation measured in HFD-fed Rag1 −/− mice in our study is, thus, a novel finding.
Interestingly, Rag1 −/− HFD mice were the only group observed to retain interscapular brown adipose tissue at endpoint. Brown adipose tissue has a high thermogenic capacity and plays a role in body weight control 41,42 . This observation may suggest that Rag1 −/− HFD-fed mice adapt to a high-fat diet, partly through increased brown adipose tissue mass and minimised weight gain, to maintain energy homeostasis. As the animal facility was maintained at 20-23 °C, and the murine thermoneutral range is 30-33 °C 43 , fat stores may have been depleted in low-fat chow-fed animals in an attempt to maintain body temperature 44 .
The immune system and inflammation have an integral role in the pathogenesis of obesity, type 2 diabetes mellitus (T2DM) and metabolic syndrome 19 , and the majority of metabolic studies have employed immunocompetent mice 12 . In this study we demonstrate that white adipose tissue was infiltrated with F4/80 positive macrophages in both NOD/SCID and Rag1 −/− HFD-fed mice, while animals fed normal chow lacked visible adipose deposits. The infiltration of adipose tissue by macrophages in obesity is thought to play a critical role in mediating insulin resistance and the development of T2DM, triggering β-cell apoptosis and reducing the secretion of insulin 2, 45 . Our observation correlates with previous studies demonstrating significantly increased activated macrophage-related cytokine IL-12 46 in the circulation of Rag1 −/− HFD-fed mice compared to Rag1 −/− fed a low-fat diet (16.7% energy from fat) 21 .
Our study demonstrates that a Western 23% HFD increases fat mass, reduces insulin tolerance, increases LNCaP human prostate cancer xenograft growth, and decreases survival to ethical endpoints in male Rag1 −/− mice. Further studies are required, however, to determine if the Rag1 −/− mouse is a useful model for investigating the interaction between HFD consumption and female cancers, particularly given the growing body of evidence describing gender-specific responses to HFD in C57BL/6 mice 47 . Specifically, male mice fed a HFD develop hyperinsulinaemia and low-grade systemic inflammation, whereas females do not 47 -possibly due to the anti-inflammatory effects of oestrogen and expansion of regulatory T cells in female mice fed a HFD 48,49 . Given the strong link between the development of endometrial and breast cancers and metabolic syndrome 10, 50 , the Rag1 −/− model is likely to be useful for the further investigation of this association.
In conclusion, this is the first study to show that Rag1 −/− mice fed a Western 23% HFD from weaning develop a number of symptoms associated with metabolic dysfunction, including hyperinsulinaemia, increased fasting insulin levels, insulin resistance, decreased insulin sensitivity, increased adiposity, hepatic steatosis and intramyocellular lipid accumulation. NOD/SCID mice fed the same diet developed some metabolic sequelae, however, these effects were more pronounced in Rag1 −/− mice. Although further studies are required, this study demonstrates that a Western 23% HFD in Rag1 −/− mice increases the growth rate of prostate cancer xenografts and significantly decreases survival to ethical endpoint compared to low-fat chow-fed mice. The Rag1 −/− immunodeficient mouse is a promising mouse model for exploring the interaction between metabolic disturbances and the development and progression of cancers associated with symptoms of metabolic syndrome.

Methods
Hyperinsulinaemic mouse model and pilot xenograft study. To establish hyperinsulinaemia in immunocompromised mice, male 3-week-old NOD.CB17-Prkdc scid /Arc (NOD/SCID) and recombination-activation gene deficient mice (B6.SVJ129-Rag1 tm1Bal /Arc; Rag1 −/− ) (Jackson Laboratories; supplied by Animal Resource Centre, Murdoch, WA, Australia) were weaned onto an ad libitum diet of low-fat, normal chow (4.8% fat, 20% protein, 75.2% carbohydrate, 12152, Specialty Feeds, Glen Forrest, WA, Australia, http://www.specialtyfeeds.com), or Western, high-fat diet (23% fat, 46% digestible energy from fat, 20% energy from protein, 34% energy from carbohydrate, SF04-027, Specialty Feeds) (n = 4-8 per mouse strain and diet). Rag1 −/− mice were backcrossed onto a C57BL/6 J background for six generations (N6; Animal Resources Centre). Mice were maintained on this diet for 28 weeks in total, with bodyweight monitored twice weekly. LNCaP human prostate cancer cell line xenograft studies were performed using Rag1 −/− mice backcrossed onto a C57BL/6 J background for ten generations (N10; Animal Resources Centre). Mice were initiated on HFD, or low-fat normal chow at weaning (4 weeks of age), and injected with 2 × 10 6 LNCaP cells in Dulbecco's Phosphate Buffered Saline (DPBS) (Thermo Fisher, Waltham, MA, USA) at a 1:1 ratio with growth factor reduced Matrigel (Sigma-Aldrich, St. Louis, MO, USA) in the subcutaneous tissue of the right flank at 6 weeks of age. Mice were maintained on HFD or low-fat normal chow, and body weight and tumour volume monitored weekly using calipers. For mice with LNCaP Scientific RepoRts | 7: 491 | DOI:10.1038/s41598-017-00443-x xenografts, experimental endpoint was determined by tumour volume (>1000 mm 3 ), calculated using the equation 'tumour volume = length × width 2 /2' , or if an ethical endpoint was reached (based on a combination of signs of stress including increased heart rate, inactivity, reduced interaction with cage mates, abnormal posture and/ or >20% body weight loss as per ethical approval and the Australian Code). Xenograft volume was normalised for different durations to ethical endpoint after implantation using the equation '(xenograft volume/time since implantation) × 100' . Metabolic parameters (see Supplementary Fig. 1) were normalised for time since weaning using the equation '(original measurement/time) × 100' . Mice were housed under pathogen-free conditions in individually-ventilated cages, at a room temperature of 20-23 °C, with a 12 hour light-dark cycle. All methods were conducted in accordance with ethical guidelines and regulations. Animal ethics approval was granted from the University of Queensland and Queensland University of Technology animal ethics committees, and human ethics approval for cell line (LNCaP) use was granted from Queensland University of Technology Human Research Ethics Committee.
Intraperitoneal glucose tolerance test. At 16 and 23 weeks after initiation of the diet, intraperitoneal (i.p.) glucose tolerance tests were performed (n = 4 mice per group) to determine effect of diet on glucose tolerance. Mice were fasted for 16 hours and baseline glucose levels measured in tail-tip blood with a One-touch Ultra blood glucose monitoring system and test strips (Accu-Chek Performa, Roche, Basel, Switzerland). Glucose (20% solution, 2 g/kg) was injected i.p. and blood glucose levels assessed at 15, 30, 60 and 120 minutes post injection. Fasting blood glucose was measured at the endpoint of the experiment (28 weeks post weaning). Surrogate indices of insulin resistance, insulin sensitivity and steady-state β-cell function were determined using the homeostatic model for assessment calculator (HOMA2) 51 , available from the Oxford Centre for Diabetes, Endocrinology and Metabolism 31 , using measured fasting glucose and insulin levels. HOMA analysis is an accepted surrogate for measuring insulin resistance in rodents 52 .
Blood and tissue sample preparation. Blood for biochemical measurements was collected by terminal endpoint cardiac puncture. Tissues of interest (brown fat, epididymal fat pad, liver and skeletal muscle) were excised, frozen in Tissue-Tek O.C.T. embedding compound (VWR, Radnor, PA, USA), and stored at −80 °C or fixed in 4% paraformaldehyde for histological and immunohistochemical analysis.
Hormone measurement. Fasting serum insulin was determined by ELISA (EMD Merck Millipore Group, Darmstadt, Germany). A multiplex ELISA (metabolic panel Milliplex kit, EMD Merck Millipore Group) was used to determine fasting serum insulin, glucagon, leptin and monocyte chemoattractant protein-1 (MCP-1) in mice with LNCaP xenografts. Absorbance at 450 nm and 595 nm was determined using a FLUOstar Omega plate reader and software (BMG Labtech, Offenburg, Germany), with absorbance values interpolated using linear regression.
Histological tissue analysis. Cryosections (6-10 μm thick, Leica CM1850 cryotome) were collected onto warm, charged Menzel Superfrost slides (Thermo Fisher), air dried for 1-2 hours and stored at −80 °C. Sections were fixed with ice-cold 100% acetone for 10 mins, followed by air-drying. White adipose tissue was processed and embedded in paraffin before sectioning (5 µM sections). One section from each specimen was stained with Mayer's haematoxylin and eosin (Sigma-Aldrich), and neutral lipids were stained in skeletal muscle and liver sections using oil-red-O stain (ORO; Sigma-Aldrich). Frozen sections were fixed in formalin, rinsed in 60% isopropanol, stained with ORO for 15 minutes, rinsed in 60% isopropanol, and mounted with coverslips using CC/ Mount (Sigma-Aldrich). Stained sections were observed using an Olympus BX41/702 microscope (U-CMAD3) and the area of red, ORO-stained lipid (minimum n = 3 samples per group and n = 3 fields per section) quantified using the thresholding function in the ImageJ software (Research Services Branch, National Institute of Health, Washington, Maryland, USA) 53 . Adipocyte size (mean area of white adipose cells, minimum n = 3 samples per group and n = 3 fields per section) was quantified using the freehand area selection tool in ImageJ.
Immunohistochemistry was performed to investigate the expression of the inflammatory macrophage marker F4/80. After rehydration in a series of xylene and ethanol washes, and antigen retrieval (Carezyme Trypsin, Biocare Medical), tissue sections were incubated in 3% hydrogen peroxide for 10 min to block endogenous peroxidases. Sections were washed in phosphate buffered saline (PBS) followed by PBS with 0.05% Tween 20 (PBST) and a blocking step using 10% BSA in PBST. Rat anti-mouse F4/80 primary antibody (122602 Cell Signalling Technology, Massachusetts, USA) was diluted 1:50 in PBST with 10% BSA. Tissue sections were washed in PBST, incubated with HRP-polymer conjugates (SuperPicture, Thermo Fisher), and incubated with the chromagen diaminobenzidine (DAB) (Dako, Glostrup, Denmark), as per manufacturer's specifications. Slides were counterstained with Mayer's haematoxylin, dehydrated, and mounted with coverslips using D.P.X neutral mounting medium (Sigma-Aldrich). The number of F4/80 positive cells was quantified as a percent of the total number of cells in the field (n = 3 samples per group and n = 3 fields per section) using ImageJ software 53 . Statistics. Statistical analyses were performed using GraphPad Prism v6.01 (GraphPad Software, Inc., San Diego, CA, USA). Data were tested for normality using the Shapiro-Wilk test. Non-normally distributed data was analysed using non-parametric Kruskal-Wallis and Mann-Whitney U tests. Normally distributed data was analysed using parametric two-way ANOVA and Tukey's multiple comparison tests with P ≤ 0.05 considered to be statistically significant.