Physiological and metabolic features of mice with CRISPR/Cas9-mediated loss-of-function in growth hormone-releasing hormone

Our previous study demonstrated that the loss of growth hormone releasing hormone (GHRH) results in increased lifespan and improved metabolic homeostasis in the mouse model generated by classical embryonic stem cell-based gene-targeting method. In this study, we targeted the GHRH gene using the CRISPR/Cas9 technology to avoid passenger alleles/mutations and performed in-depth physiological and metabolic characterization. In agreement with our previous observations, male and female GHRH-/- mice have significantly reduced body weight and enhanced insulin sensitivity when compared to wild type littermates. Dual-energy X-ray absorptiometry showed that there were significant decreases in lean mass, bone mineral content and density, and a dramatic increase in fat mass of GHRH-/- mice when compared to wild type littermates. Indirect calorimetry measurements showed dramatic reductions in oxygen consumption, carbon dioxide production and energy expenditure in GHRH-/- mice compared to wild type mice in both light and dark cycles. Respiratory exchange ratio was significantly lower in GHRH-/- mice during the light cycle, but not during the dark cycle, indicating a circadian related metabolic shift towards fat utilization in the growth hormone deficient mice. The novel CRISPR/Cas9 GHRH-/- mice are exhibiting the consistent and unique physiological and metabolic characteristics, which might mediate the longevity effects of growth hormone deficiency in mice.

AGING cell (ESC) based gene-targeting method to knock out growth hormone-releasing hormone (GHRH), which is a hypothalamic peptide that controls both the synthesis and secretion of GH [17,18]. These mutant mice have significantly decreased body weight and possess increased insulin sensitivity and prolonged lifespan indicating GH deficiency is primarily responsible for longevity extension [5]. This GHRH -/model was generated using 129SV agouti color mice resulting in the co-segregation of GHRH -/and agouti alleles. Agouti is an important component of several biological pathways, including body weight homeostasis, regulation of food intake and, energy expenditure [19]. In addition, its expression was associated with metabolic syndrome [20]. Our goal for this study is to delineate the direct physiological and metabolic consequences of GH deficiency. To achieve this goal, we knocked out the GHRH gene with CRISPR/Cas9 technology, preventing the agouti gene from acting as a passenger allele. We produced our experimental knockout model on mixed genetic background to avoid any phenotype resulting from strain-specific inbreeding. This is the first in-depth metabolic and physiological profiling of GH-suppressed mice generated with CRISPR/Cas9 technology on mixed genetic background.
Our novel GHRH -/mice have decreased body weight and higher insulin sensitivity despite having normal glucose tolerance. GH deficiency resulted in dramatically decreased bone mineral density (BMD), bone mineral content (BMC), and lean mass. However, GHRH -/mice have significantly increased fat mass compared to littermate controls. Indirect calorimetry allowed us to measure physiological respiratory parameters, which were used to calculate respiratory exchange ratio (RER) and energy expenditure. RER data demonstrated a significant difference in metabolism between GHRH -/and wild type (WT) mice during light cycle. GHRH -/mice had significantly lower energy expenditure during both light and dark cycles. Our GH-deficient mouse model has physiological characteristics consistent with our previous study and similar to other GH-related mutants. It was hypothesized that slowing the biological process of aging is associated with GH deficiency. Our model helps understand key physiological and metabolic characteristics that are involved in the process of aging.

RESULTS
Utilizing CRISPR/Cas9-mediated gene-editing method, we generated homozygous GHRH -/mice ( Figure 1A). In a litter of 10 G0 pups, 8 carried indels and large deletions in the GHRH locus. Our targeted allele (#28528) is a 291 base pairs deletion that eliminates the splice donor site at exon 2, intron 2-3 and a large part of Exon 3 (77 base pairs out of 102 base pairs); this allele was selected based on, successful germline transmission ( Figure 1A, 1B). Predicted translation of the resulting sequence suggests an in-frame mutation leading to a loss of 26 amino acids, including 21 amino acids required for full activity (RMQRHVDAIFTTNYRKLLSQLYARKV). The line was used in all subsequent studies. We measured body weight and food consumption of GHRH -/and WT littermates longitudinally. Male and female GHRH -/mice were significantly lighter than their littermate controls ( Figure 1C, 1D). In addition, GHRH -/mice consumed dramatically fewer calories than WT littermates ( Figure 1E, 1F).
Previously, knockout of GHRH gene was shown to result in significant reduction of GH expression in pituitary and IGF-1 expression in liver [17]. To assess loss of function of GHRH, we measured mRNA levels of GH in pituitary and IGF-1 in liver. We observed that expression of both GH and IGF-1 genes were significantly decreased in GHRH -/mice compared to WT littermates (Figure 2A, 2B, 2E, 2F). These results confirm the reduction in GH/IGF-1 signaling. However, we did not observe significant change in prolactin expression in the pituitary ( Figure 2C, 2D).
The effect of GH signaling on body weight ( Figure  3A, 3B) and composition is well documented in GH related mutant mice [21][22][23]. Therefore, we performed dual-energy X-ray absorptiometry (DXA) to study the effects of reduced GH signaling on body composition parameters in our knockout model of GHRH. Absolute BMD, BMC, and lean mass values were significantly lower in both male and female GHRH -/mice compared to WT littermates ( Figure 3C-3H). In order to account for the significant body weight differences between GHRH -/and WT mice, we used analysis of covariance (ANCOVA) method, which revealed that BMD, BMC, and lean mass were significantly reduced ( Figure  4A-4F), whereas, fat mass was significantly increased in GHRH -/mice ( Figure 4G, 4H).
To explore the effect of GH deficiency on metabolic phenotype of mice, we measured RER using indirect calorimetry. Figure 5A, 5B provides overview of RER values for each hour recorded for 6 days for female and male mice, respectively. Notably, the RER of both male and female GHRH -/mice decreased much more rapidly during the transition between dark and light cycles compared with WT littermates ( Figure 5A, 5B). RER measurements collected for 6 days were averaged into a single day ( Figure 5C, 5D). These results indicate lower RER for both male and female GHRH -/mice compared to WT littermates during the light cycle, but not the dark cycle ( Figure 5C, 5D). Comparisons of RER confirm that the significant differences in metabolism functions in a circadian manner ( Figure 5E-5H). AGING Figure 6A, 6B, 6E, 6F show mean hourly oxygen consumption (VO2) and carbon dioxide production (VCO2) of male and female mice for 6 days. All mice presented diurnal rhythm of higher VO2, VCO2 during the dark cycles compared to VO2, and VCO2 measured during the light cycles. Figure 5C, 5D, 5G, 5H show VO2 and VCO2 measurements collected for 6 days were averaged into a single day. Overall averages of absolute VO2 and VCO2 measurements were significantly lower in GHRH -/female and male mice compared to WT littermates in both light and dark cycles ( Figure 7A-7H).

AGING
ANCOVA method, which controls for differences in body weight, showed male and female GHRH -/mice have significantly lower VO2 and VCO2 compared to WT littermates, during both light and dark cycles ( Figure 8A-8H).
We calculated energy expenditure from the respiratory parameters collected by indirect calorimetry to assess the effect of GH-deficiency on metabolic rate. Figure 9A, 9B show energy expenditure of female and male GHRH -/mice compared with WT littermates over 6 days. We  AGING averaged this data into a single day ( Figure 9C, 9D). The analyses of energy expenditure representing the 6 light and the 6 dark cycles confirmed the dramatic downward shift of metabolic rate in GH-deficient mice ( Figure 10A-10D). Controlling for the effect of significant differences in body weight, ANVOCA method, showed significant reduction in metabolic rates of both male and female GHRH -/mice compared to WT littermates in light and dark cycles ( Figure 9E, 9H). We further measured voluntary physical activity of mice during our indirect calorimetry study. Overall pattern indicates both male and female GHRH -/mice have reduced activity compared to WT littermates during light and dark cycles ( Figure 11A-11D).

AGING
To assess insulin tolerance, we performed intraperitoneal insulin injections and measured blood glucose levels. Upon injection with insulin, glucose concentrations significantly decreased in GHRH -/female and male mice compared to their littermate controls ( Figure 12A,  12B). Area under the curve (AUC) data were significantly lower GHRH -/female and male mice than their littermate controls ( Figure 12C, 12D). To evaluate the glucose homeostasis in vivo, we performed intraperitoneal glucose tolerance test (IPGTT) with mice fasted overnight. We did not observe any significant differences in blood glucose levels throughout the 2hour period following glucose injection ( Figure 12E, 12F). AUC analyses did not reveal any statistical significance due to loss of GHRH ( Figure 12G, 12H). This data strongly supports the notion that GH deficiency improves insulin sensitivity, but not glucose homeostasis in vivo.

DISCUSSION
Previously, GHRH was knocked out in mice using a neomycin resistance cassette to replace parts of both exon 2 and 3 [17]. This approach introduces the possibility of passenger flanking alleles/mutations, which may be responsible for phenotypic variations between mice generated with CRISPR/Cas9-based gene-editing and classical knockout method [24,25]. Agouti gene and GHRH -/alleles co-segregate in the mouse model generated by Salvatori lab [17]. Previously, GHRH -/mice were shown to have increased adiposity and insulin sensitivity, which are the two key parameters linked to longevity in different knockout models of the growth hormone pathway. However, insulin resistance and obesity are both linked to agouti gene expression [20]. We generated a 'clean' model for GHRH loss of function mutation in mice using the CRISPR/Cas9 system, which provides precise genome-editing without leaving any exogenous DNA sequences behind, eliminating the possibility of passenger alleles/mutations influencing the phenotype resulting from the GH-deficiency [26]. It has been shown in the literature that CRISPR/Cas9 method can result in off-target mutations. We have outcrossed our mice several generations to minimize the possibility of off-target mutations in the genome. Some of the phenotypic variability observed in transgenic animals has been attributed to the genetic background of animal models [27,28]. To rule out this possibility, we used GHRH -/mice on mixed genetic background for this study.
Indirect calorimetry provides highly sensitive, accurate, and noninvasive measurements of energy expenditure and substrate utilization in live animals [29]. One important technical issue for indirect calorimetry is the duration of acclimation and data acquisition period. AGING Generally, data collection by indirect calorimetry is limited to 24 hours, which usually takes place after a 24 hours acclimation period [5,30,31]. Temperature within the cage is thought to be closer to the thermoneutral zone, where energy expenditure required for maintaining body temperature is at its lowest. Moving mice from group housing to single housing is expected to increase their expenditure [32]. For this reason, it is critical for mice to adjust to cold stress before any respiratory measurements are performed. In order to obtain accurate and reliable results from indirect calorimetry, we acclimated mice in the respiratory chambers for 7 days and subsequently, collected the measurements for 6 days.
Interpretation of metabolic and physiological parameters has been problematic due to differences in body weight in some genetic models [33]. In order to compensate for these differences, researchers have argued for different methods of analysis, including ratio-based normalization and allometric scaling [34]. Studies illustrated that these methods are improper and result in flawed conclusions [33][34][35][36]. ANCOVA method has been promoted as the suitable method of analysis for physiological parameters that are influenced by variables such as body weight [35,37,38]. We utilized this unbiased statistical approach to control for the influence of body weight on body composition and indirect calorimetry data.

AGING
The inverse relationship between lifespan and body size within species has been observed not only in mice but also, in rats, dogs, horses and humans [6,[39][40][41]. Decreased body size is one of the strongest phenotypic characteristics of growth hormone deficiency models [5,8]. Our longitudinal study of body weight showed that GHRH -/mice are significantly lighter than the WT controls. Analyses of body composition parameters with DXA revealed remarkable effects of reduced GH signaling on bone, lean, and fat tissues. Absolute BMD, AGING BMC, lean mass measurements are significantly lower in GHRH -/female and male mice compared to littermate controls. Reduced absolute lean mass and BMD have been documented in the GHR -/model [23]. Previously, Ames dwarf mice were shown to have significantly reduced absolute lean mass and BMC [42]. For BMC, BMD, and lean mass, comparison of absolute measurements and ANCOVA method using body weight as a co-variant showed the same results. ANCOVA method demonstrated that fat mass adjusted for body weight is significantly increased in mice lacking GHRH.
We aimed to utilize indirect calorimetry to investigate the metabolic effects, which are associated with reduced GH signaling and may be related to improved longevity. Analyses of absolute energy expenditure and body weight adjusted energy expenditure values lead to the same conclusion that our novel GHRH -/mice have significantly decreased metabolic rates compared to WT littermates. Previous study using Ames dwarf and GHR -/found significantly decreased absolute energy expenditures for both models [30]. Using indirect calorimetry, we examined the RER, which is a unitless ratio obtained by dividing VCO2 by VO2. RER is close to 0.7 when mice metabolize fat as an energy source and is close to 1.0 when metabolizing carbohydrates [43]. Our results demonstrate that GH-deficiency is significantly associated with lower RER during the light cycle, but not during the dark cycle in both female and male mice. This suggests that GHRH -/mice have a greater level of fat utilization, which appears in a circadian pattern. Previous studies have shown lower RER for Ames and GHR -/mice during both light and dark cycles [30]. However, these animals were only acclimated for 24 hours and data were collected in 24 hours. This technical difference is a likely cause for the different results.
Our study shows that GHRH -/mice have decreased physical activity compared to their WT littermates. Another study showed increased physical activity in GHRH deficient mice [44]. This study recorded physical activity for only 10 minutes. However, our measurements lasted 6 days. This difference in methodology might be the likely contributor to the difference in results.
In humans, insulin resistance is one of the hallmarks of aging and is associated with chronic conditions such as diabetes, cancer, and cardiovascular disease [45,46]. One of the physiological characteristics of long-lived GH-related mutant mouse models is elimination of insulin resistance [5,47,48]. In this study, our novel CRISPR/Cas9 GHRH -/female and male mice exhibited insulin sensitivity compared to their WT littermates, affirming the relationship between growth hormone deficiency and increased sensitivity to insulin. This study validates our previous findings and establishes the GHRH -/mouse as an important animal model to study mechanisms of extended longevity and slow aging in mammals. Our next goal is to assess impact of high-fat diet, which is expected to shorten lifespan and cause metabolic dysfunction, on GHRH -/mice. To achieve this goal, we are conducting a longevity study with GHRH -/and control mice fed high-fat and control diet. Preliminary results from the ongoing longevity study indicate that our GH-deficient model is protected from harmful effects of high-fat diet. Our current work provides a practical model for a CRISPR/Cas9 system, which leads to improved locus specificity and ease of multiplexed targeting in mammalian aging and longevity studies.

CRISPR/sgRNA design and synthesis
Owing to the small size of the mouse GHRH exons (17.16 kb with 5 exons), eight CRISPR targets with high scores were identified in the sequences flanking exons 2 and 3, including the intronic sequences, using Benchling (https://www.benchling.com/) with the goal of creating deletion alleles spanning exons 2 and 3. Single guide RNA (sgRNA) molecules were generated using a cloning-free method as described earlier [49]. Cas9 protein was obtained from MacroLabs at UC Berkeley.

Generation of G0 (founder) animals and germline transmission of mutant alleles
All animal procedures were performed in accordance with the recommendations in the guide for the care and use of laboratory animals published by the National Institutes of Health. The protocols used were approved and conducted according to the University of Alabama at Birmingham institutional animal care and use committee. Pronuclear injections into C57BL/6J zygotes were performed with a solution of sgRNAs (50 ng/μl each) and Cas9 protein (50 ng/μl per guide). Injected zygotes were implanted into pseudo-pregnant CD1 recipients. Genomic DNA obtained from tail biopsies of putative founder (G0) animals were assessed for the presence of mutations in the targeted genes. G0 animals were bred to WT C57BL/6J mice for germline transmission of mutant alleles.

Detecting the presence of indels
Genomic DNA from mouse-tail biopsies was obtained by digesting in lysis buffer (50 mM Tris-HCl pH 8.0, 100 mM EDTA pH 8.0, 100 mM NaCl, 1% SDS) with proteinase K (0.3 mg/ml), followed by a phenol: chloroform extraction and ethanol precipitation AGING procedure. PCRs were set up using the oligonucleotide primers MmGhrh-gen-F1: 5'-CTTGCTTCTCTCACA CTTGC-3'; MmGhrh-gen-R1: 5'-TTAAAGGGTCGG AGCAGTAG-3' with NEB Taq 2x Master Mix. The amplicons (795 base pairs) were subjected to denaturation-slow renaturation to facilitate formation of hetero-duplexes using a thermocycler. These samples were then resolved on polyacrylamide gels (6%) and the resulting mobility profiles used to infer efficiency of CRISPR-Cas9 nuclease activity. Indels or deletions were detected by heteroduplex mobility analysis (HMA) from tail genomic DNA of potential founder animals. PCR amplicons were cloned using the TOPO-TA cloning kit (ThermoFisher/Invitrogen, Carlsbad CA). Colonies were picked from each plate and grown in 1.5 ml liquid cultures to isolate plasmid DNA using an alkaline lysis procedure. Plasmid DNA was sequenced using M13 forward or reverse primers.

Breeding
Mice were housed under standard conditions (12-hours light and 12-hours dark cycles at 20-23°C) with ad libitum access to food and water and fed with NIH-31 rodent formula. A GHRH heterozygous ( +/-) G0 mouse obtained from the University of Alabama at Birmingham (UAB) genomics core was crossed with a BALB/cByJ for increased genetic diversity, increased fecundity, and reduced aggression. To obtain experimental animals, mice heterozygous for GHRH mutation were crossed with littermates to obtain homozygous GHRH mutant and homozygous GHRH wild type animals, which served as controls. Experiments were performed with age-matched mice. Females were 6 months old and males were 8 months old at the time of the experiments.

DXA
The mice were scanned using the GE Lunar PIXImus DXA with software version 1.45. The mice were anesthetized using an Isoflurane (3%) and oxygen (500ml/min) mixture, delivered by a Surgivet anesthesia machine, and then placed in a prostrate position on the DXA imaging plate and scanned. During the scan, the mice remained anesthetized. For all scans, the head was excluded from the analysis and the data obtained included BMC, BMD, lean mass and fat mass.

Indirect calorimetry
Indirect calorimetry was performed using comprehensive lab animal monitoring system (Oxymax-CLAMS; Columbus Instruments Co., Columbus, OH). This system uses zirconia and infrared sensors to monitor oxygen (O2) and carbon dioxide (CO2), respectively. We performed indirect calorimetry with 24 mice (12 wild type and 12 GHRH -/-). We divided mice into 3 groups and collected measurements on 8 animals at a time (4 WT and 4 GHRH -/-). The mice were housed in separate respiratory chambers for 7 days for acclimatization before recording the measurements. After a 7-day acclimation period, respiratory parameters of mice were recorded for 6 days with ad libitum access to standard chow and water (12-hours light and 12-hours dark cycles at 20-23°C). Respiratory samples were measured every 9 minutes per mouse, and the data were averaged for each hour. RER was calculated by dividing VCO2 by VO2. Energy expenditure was calculated by the equation as energy expenditure = (3.815 + 1.232 × VCO2/VO2) × VO2 [43]. We used infrared beam system in X, Y, and Z coordinates to record physical activity of mice. If the mouse is standing still and starts a repetitive act, such as grooming, it will continuously break the same beam indicating locomotor activity. When the currently broken beam is different from the previous one, activity is counted as ambulatory.

Glucose and insulin tolerance tests
Overnight-fasted mice underwent glucose tolerance test by intraperitoneal injection with 1 g of glucose per kg of body weight. Blood glucose levels were measured at 0, 5, 15, 30, 60, 90, and 120 minutes. 4-hours fasted mice underwent insulin tolerance test by intraperitoneal injection with 1 IU porcine insulin (Sigma-Aldrich, St. Louis, MO) per kg of body weight. Blood glucose levels were measured at 0, 5, 15, 30, 60, 90, and 120 minutes.

Real-time quantitative PCR
RNA was harvested from tissues using RNeasy plus kit (Qiagen, Hilden, Germany). Total RNA was reverse transcribed with LunaScript RT SuperMix Kit (New England Biolabs, Ipswich, MA). Real-time quantitative PCR was performed using a QuantStudio 3 with a PowerUp SYBR green master mix (ThermoFisher Scientific, Waltham, MA). Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) or beta-actin expression was used to normalize gene of interest in each sample. Real-time quantitative PCRs were set up using the oligonucleotide primers Mm GAPDH F1 5-CCTGGAG AAACCTGCCAAGTATGATG-3'; Mm GAPDH R1 5-AAGAGTGGGAGTTGCTGTTGAAGTC-3', Mm Actb

Statistical analyses
The unpaired Student's t-test with Welch's correction was used for statistical analysis. Statistical significance was established at p<0.05, two-tailed. We used (generalized linear model) GLM package with R software for analysis of indirect calorimetry and body composition data. When interaction was not found, the code was run without the interaction term. Our GLM models were validated with the MMPC's (National Mouse Metabolic Phenotyping Center, https://www.mmpc.org/shared/regression.aspx) energy expenditure analysis tool. Graphs were generated with GraphPad Prism 8 (San Diego, CA).

ACKNOWLEDGMENTS
We thank Sofia Canlas, Joseph Jablonsky, Whitney Turner, and Matthew Joyner for technical assistance. We also thank other members of the Sun lab for their helpful discussion and comments on the revision of the manuscript.

CONFLICTS OF INTEREST
All of the contributing authors declared no conflicts of interest.

FUNDING
This work was supported in part by National Institute on Aging grants AG048264, AG057734 and AG050225 (L.S.).