Genetic Influences on Susceptibility to Oxygen-Induced Retinopathy

Abstract

purpose. To investigate the inheritance of susceptibility to oxygen-induced retinopathy in the rat with the use of formal backcross analysis.

methods. Neonatal offspring of inbred albino Fischer 344 (F344) and pigmented Dark Agouti (DA) crosses and F1×F344 and F1×DA backcrosses were exposed to alternating 24-hour cycles of hyperoxia (80% oxygen in air) and normoxia (21% oxygen in air) for 14 days. Retinal avascular area was analyzed by staining with Griffonia simplicifolia isolectin B4, a marker of vascular endothelial cells. Expression of erythropoietin (EPO) mRNA in retinas was quantified by real-time reverse-transcription polymerase chain reaction.

results. Oxygen-exposed offspring of two F344×DA F1 crosses showed retinal avascular areas and ocular and coat pigmentation that were similar to those of the DA strain. Mean retinal avascular area was 73%. Offspring of two DA×F1 backcrosses were similar to F344×DA F1 pups, with pigmented eyes and coats and a mean retinal avascular area of 76%. In contrast, offspring of two F344×F1 backcrosses exhibited a range of eye and coat pigmentation. Mean retinal avascular area of pigmented offspring of the F344×F1 backcrosses was 71% (P < 0.001 compared with F344 rats). Mean avascular area of albino offspring of the F344×F1 backcrosses was 27% (P > 0.05 compared with F344 rats). The normalized expression of EPO mRNA was 3.01 ± 1.00 in retinas from pigmented F344×F1 backcross offspring compared with 1.31 ± 0.69 for albino offspring (P < 0.001).

conclusions. Segregation of the susceptibility trait to oxygen-induced retinopathy in the DA and F344 rat strains is associated with pigmentation and erythropoietin expression and can be modeled using an autosomal dominant pattern of inheritance.

Risk factors for the development and progression of retinopathy of prematurity (ROP) include demographic factors ^{1 2 3 4 5} and the need for supplemental oxygen therapy. ^{6 7 8 9} Increasing evidence ^{3 5 6 10 11 12 13 14} indicates that factors such as racial background and genetic polymorphisms ^{15 16 17 18 19} may also be important. However, of the various animal and human studies implicating genetic factors in the risk for ROP, none has provided direct, formal evidence for a heritable susceptibility trait. 

Experimental oxygen-induced retinopathy in the rat is a useful, albeit imperfect, model of human retinopathy of prematurity. ^{20 21} We have recently shown, through the induction of oxygen-induced retinopathy, robust differences in the retinal microvascular phenotype of neonates from five different strains of inbred rat. ²² Clear and consistent differences in microvessel morphology, vascular density, and vessel tortuosity were observed among strains, and the extent of retinal avascular territories varied significantly. Fischer 344 (F344), Wistar-Furth, and Lewis strains (all albino) were relatively resistant to the effects of ischemia after hyperoxic exposure, whereas the albino Sprague–Dawley strain showed intermediate susceptibility and the pigmented Dark Agouti (DA) strain was very sensitive. The latter two strains developed the florid microvascular abnormalities and neovascular tufts characteristic of neovascular retinopathies. Strain-related differences were independent of litter size, body mass, and major histocompatibility complex haplotype. Similar differences in pairs of rat strains have since been reported by others. ²³ Accordingly, the aim of this work was to investigate the heritability of the susceptibility trait by cross and backcross analysis of the offspring of matings between susceptible and resistant strains. In addition, we sought to investigate any association between ocular pigmentation and susceptibility to oxygen-induced retinopathy in the rat and correlated the expression of mRNA for erythropoietin, a prototypic marker of angiogenesis in ROP, with retinal vascular phenotype among strains and backcross progeny. 

Methods

Experimental Animals

Inbred rat strains were derived from more than 20 consecutive brother-sister matings. F344 rats (albino coat color and red eyes) were bred within the institution. DA rats (agouti coat color and dark brown eyes) were sourced from the Institute of Medical and Veterinary Science (Adelaide, SA, Australia). Lineage records and allozyme electrophoresis confirmed genetic integrity. Rats were allowed unlimited access to water and rat chow and were exposed to a 12-hour light-dark cycle. Room temperature was maintained at 24°C and ambient humidity was between 40% and 55%. All animal experiments were approved by the institutional Animal Welfare Committee and conformed to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Exposure of Neonatal Rats to Cyclic Hyperoxia

Female rats and their newborn litters were housed in a custom-built, humidified chamber as previously described. ²² Litters were placed in the chamber within 12 hours of birth and exposed to alternating 24-hour cycles of hyperoxia (80% O₂) and normoxia (21% O₂) for the first 14 days of life. An anesthetic blender (CIG Medishield-Ramsay, Melbourne, VIC, Australia) and a high-flow oxygen regulator (Anaequip, Adelaide, SA, Australia) were used to deliver oxygen to the chamber at 25 L/min. Oxygen levels within the chamber were continuously monitored using a fuel-cell oxygen monitor and were recorded with a data logger (Gemini Dataloggers Ltd., Chichester, West Sussex, UK) for subsequent analysis. An oxygen concentration of 80% ± 1% was maintained for the duration of hyperoxic cycles. 

Tissue Processing and Isolectin Histochemistry

After the 14-day period of cyclic hyperoxia, rats were killed with an inhaled overdose of halothane anesthesia and the eyes were enucleated. Eyes were fixed in 2% wt/vol paraformaldehyde in phosphate-buffered saline (PBS), pH 7.4, at 4°C for 90 minutes and dissected according to the method of Chan-Ling. ²⁴ Four equally spaced radially oriented incisions were used to flatten each retina. The microvasculature in retinal whole-mounts was stained with fluorophore-conjugated Griffonia simplicifolia isolectin B4 (GS-IB4) ²⁵ (Alexa Fluor 488 conjugate; Molecular Probes, Eugene, OR), which stains vascular endothelial cells, according to a modification of the method of Cunningham, ²⁶ as previously described. ²² In all cases, retinal dissection and histochemistry were performed within 6 hours of enucleation. 

Image Analysis of Labeled Retinas

The right retina of each animal was used for image analysis. Imaging was performed within 12 hours of lectin labeling using a fluorescence microscope (Olympus Optical Co. Ltd., Tokyo, Japan) coupled with a CCD-digital camera (Roper Scientific, Trenton, NJ) and image acquisition software (RS Image, version 1.01; Roper Scientific). Sequential, overlapping, high-resolution images of the entire retina were captured using a 4× objective. Images were merged to construct a montage image of the retina (Adobe Photoshop, version 7.0; Adobe Systems Inc., San Jose, CA) and were analyzed using image analysis software (ImageJ version 1.30; National Institutes of Health, Bethesda, MD). A masked observer manually outlined and measured avascular areas as a percentage of the total retinal area. 

Quantitative Real-Time Reverse Transcription–Polymerase Chain Reaction

Rat eyes were enucleated immediately after death into chilled diethylpyrocarbonate (DEPC)–treated normal saline, and the retinas were dissected, snap-frozen in liquid nitrogen, and stored at –80°C. Total RNA was isolated using an RNeasy mini-kit (Qiagen, Valencia, CA). Contaminating genomic DNA was removed with DNaseI (DNA-free; Ambion, Austin, TX). Samples free of visible DNA contamination on a 1% agarose gel and with a ratio of 28S:18S rRNA approximating 2:1 were quantified by spectrophotometry. One-microgram samples with a 260:280 nm absorbance ratio of ≥1.9 were reverse transcribed using a first-strand cDNA synthesis kit (SuperScript III First-Strand Synthesis System; Invitrogen, Carlsbad, CA). A reverse transcriptase-free control sample was synthesized in parallel with each cDNA sample, with substitution of DEPC-H₂O for reverse transcriptase. For purposes of normalization, a standard cDNA pool was prepared from pooled retinal RNA extracted from 10 F344, SPD, and DA rats that had been exposed to room air or cyclic hyperoxia. 

Primers for rat erythropoietin (EPO) and for the housekeeping genes acidic hypoxanthine guanine phosphoribosyl transferase (HPRT) and ribosomal phosphoprotein (ARBP) were designed to flank an intron (Primer3 software; Whitehead Institute for Biomedical Research, Cambridge, MA) ²⁷ and tested in silico for specificity against sequences for Rattus norvegicus using BLAST software (NCBI, Bethesda, MD). Primer sequences were as follows: ACCAGAGAGTCTTCAGCTTCA (EPO forward), GAGGCGACATCAATTCCTTC (EPO reverse); TTGTTGGATATGCCCTTGACT (HPRT forward), CCGCTGTCTTTTAGGCTTTG (HPRT reverse); and AAAGGGTCCTGGCTTTGTCT (ARBP forward), GCAAATGCAGATGGATCG (ARBP reverse). Primers were then synthesized (Geneworks Ltd., Thebarton, SA, Australia). 

Real-time RT-PCR was performed (RotorGene 2000 Thermal Cycler; Corbett Research, Mortlake, NSW, Australia). Each 20-microliter reaction mixture contained 10 μL SYBR Green master-mix (QuantiTect SYBR Green PCR Master Mix; Qiagen) containing hot-start Taq DNA polymerase, SYBR Green I, dNTPs, and PCR buffer (5 mM MgCl₂, Tris-Cl, KCl, (NH₄)2SO₄ pH 8.7), 2 μL each forward and reverse primer (0.5 μM final concentration), and 6 μL cDNA sample diluted 1/100 with purified water (Ultra Pure; Fisher Biotech, West Perth, WA, Australia). Reaction conditions were initial denaturation (95°C, 15 minutes) followed by 50 cycles of denaturation (94°C, 20 seconds), annealing (50°C, 20 seconds), extension (72°C, 30 seconds), and final extension (72°C, 4 minutes, followed by 25°C, 5 minutes). The standard cDNA pool was included in triplicate in each PCR run, together with a single RT-negative control for each sample and two water (no-template) controls. Melt-curve analysis was used to confirm amplicon specificity. The melt-curve of each real-time PCR product was compared with that of the corresponding sequenced product. ²⁸ Real-time RT-PCR products were separated on agarose gels, purified, sequenced, and compared with the predicted amplicon sequence to confirm identity. Purified DNA was labeled (BigDye Terminator version 3.1 Cycle Sequencing Kit; Applied Biosystems, Foster City, CA) and resolved (ABI 3100 Genetic Analyser; Applied Biosystems). 

Statistical Analyses

Before statistical analysis of retinal areas, percentages were arc sin–transformed to normalize the variances of the data. ²⁹ Analysis of variance was performed to analyze the transformed data, including repeated-measures designs where appropriate (SPSS, version 11.0.2; SPSS Inc., Chicago, IL). Comparisons between subsets of data were made with pre-planned single degree of freedom contrasts, Ryan-Einot-Gabriel-Welsch F tests (REGWF tests), or Bonferroni tests, with significance levels (alpha) set at 0.05 in each case. Summary data were expressed as means with 95% confidence intervals (95% CIs). The Mann–Whitney U test corrected for ties was used to compare categorical data. For gene expression data, expression in each sample was determined relative to the standard cDNA pool and normalized to the housekeeping genes ARBP and HPRT (GeNorm software; Ghent University Hospital, Ghent, Belgium). ³⁰ Normalized expression data were normally distributed; thus, two-way analysis of variance (ANOVA) was used to compare gene expression among different rat strains. The significance level (alpha) was set at 0.05. 

Results

Cross-Breeding Experiments

Heritability of susceptibility to oxygen-induced retinopathy was examined in a series of cross-breeding experiments. The F344 and DA rat strains were selected as relatively resistant and susceptible, respectively, to oxygen-induced retinopathy. Each genetic cross was performed twice, and different gender pairings (male F344 crossed with female DA; female F344 crossed with male DA) were used in each of the two crosses. All the offspring of each cross were exposed to cyclic hyperoxia for the first 14 days of life, before retinal vascular area measurement. 

Susceptibility of F344×DA F1 Offspring to Oxygen-Induced Retinopathy

All 16 F1 offspring of the two F344××DA crosses had ocular and coat pigmentation similar to those of the DA strain. Eye color was dark brown and coat color was agouti, with white patches on the abdomen and paws (Fig. 1A) . All neonates exposed to cyclic hyperoxia for 14 days had large avascular regions in the central and peripheral retina (Fig. 2) . Mean total retinal avascular area was 73% (95% CI, 69%–77%). The extent of the retinal avascular area of the F1 rats was similar to that previously reported ²² for oxygen-exposed pups of the DA parental strain at the same time point (n = 14; mean difference, 1%; P = 0.02) and substantially larger than for the F344 parental strain (n = 18; mean difference, 24%; P < 0.001). 

Susceptibility of Backcross Offspring to Oxygen-Induced Retinopathy

Randomly selected adult rats of additional DA×F344 F1 crosses were mated with parental strain F344 and DA rats to yield backcross offspring. Four backcrosses were performed to accommodate all possible gender pairings. Offspring of the DA×F1 backcrosses were indistinguishable from the F1 pups: all had dark brown ocular pigmentation and agouti coat color. In contrast, the offspring of the F344×F1 backcrosses exhibited a wide range of coat colors (Fig. 1B) . Rats with albino coats had red eyes, and all rats with some coat pigmentation had dark brown ocular pigmentation. 

All newborn offspring of each backcross (45 pups in total) were exposed to cyclic hyperoxia for 14 days, and retinal avascular areas were measured. Offspring of the DA×F1 backcrosses were universally susceptible to the attenuating effects of oxygen on retinal vascularization. Mean total retinal avascular area was 76% (95% CI, 73%–80%) (Fig. 3) . Greater variation was found in the extent of retinal vascularization in the offspring of the F344×F1 backcrosses (mean avascular area, 47%; 95% CI, 38%–57%; Fig. 3 ). 

An association was identified between pigmentation and retinal vascularization in the rat. Highly significant differences were found in the extent of retinal avascular area between albino and pigmented offspring of the F344×F1 backcrosses (P < 0.001; two-tailed Mann–Whitney U test). Accordingly, when the pigmented offspring of the F344×F1 backcrosses were considered together, the mean retinal avascular area was 71% (95% CI, 65%–78%), compared with 27% (95% CI, 23%–31%) for albino offspring of the same crosses. Pigmented progeny of the F344×F1 backcrosses had areas of avascular retina that were slightly larger than those of the DA strain (mean difference, 1.6%; 95% CI, 0.1%–5%; P = 0.003) and substantially larger than those of the F344 strain (mean difference, 26%; 95% CI, 18%–34%; P < 0.001). In contrast, the albino backcross progeny had retinal avascular areas that were similar in size to those of the F344 strain (P = 1.0 for difference). 

Retinal Gene Expression in Backcross Progeny after Cyclic Hyperoxia

A cohort of seven neonatal rats—four rats from one F1×F344 backcross and three from another—incorporating the complete spectrum of coat and eye color was selected prospectively for retinal gene expression studies (Table 1) . After 14 days of cyclic hyperoxia, the right retinas were used for vascular analysis, and the left retinas were processed for RNA extraction and subsequent quantification of gene expression. The validity of using the left and right eyes of each rat for different analyses was supported by a study that demonstrated significant intereye correlation in retinal vascularization in a rat model of oxygen-induced retinopathy. ³¹ Before the analysis of gene expression, retinal cDNA samples were grouped according to the retinal avascular areas of fellow eyes. Rats with avascular areas smaller than 50% of the total retinal area were albino and were deemed resistant, whereas those with areas larger than 50% were pigmented and were deemed susceptible to the cyclic hyperoxic exposure. Mean retinal avascular areas of the resistant and susceptible groups were 27% and 68%, respectively. Real-time RT-PCR was then used to quantify expression of the EPO gene in the two retinal cDNA pools. There was significantly less EPO in the cDNA pool from the resistant (albino) rats than in the cDNA pool from the sensitive (pigmented) rats. Normalized EPO expression relative to the standard cDNA pool was 1.31 ± 0.69 for the former and 3.01 ± 1.00 for the latter (P < 0.001). By comparison, normalized EPO expression was 1.43 ± 0.23 for the hyperoxia-resistant albino F344 parental strain at the same time point and under the same conditions and 2.24 ± 0.47 for the hyperoxia-sensitive pigmented DA parental strain. 

Discussion

Our findings in a formal backcross analysis substantiated the importance of heritable factors in determining the susceptibility of rats to oxygen-induced retinopathy. If such susceptibility is arbitrarily defined as a total retinal avascular area in excess of 50% and the susceptibility allele is assumed to be dominant, then all offspring of crosses between resistant F344 and susceptible DA rats would be susceptible, as would all progeny of the DA×F1 backcrosses (Fig. 4) . Furthermore, the ratio of susceptible/resistant offspring of F344×F1 backcrosses would be expected to approximate 1:1. Experimental findings closely matched these predictions (Fig. 4) . The extent of variation in avascular retinal areas of oxygen-exposed backcross rats argues against a monogenic mode of inheritance. Nonetheless, our results demonstrate that hereditary factors are central to the risk for oxygen-induced retinopathy in the rat and that susceptibility segregates in a manner that is consistent with an autosomal dominant form of inheritance, modified by other genetic influences. 

All offspring of F344×DA crosses had pigmented eyes and coats similar to those of the parental DA strain, and their retinal vasculature was uniformly susceptible to the attenuating effects of cyclic hyperoxia in the neonatal period. Pups exhibited oxygen-induced retinopathy indistinguishable from that of the DA strain. Similar findings were observed in all oxygen-exposed progeny of DA× (F344×DA) backcrosses. In contrast, two different patterns of retinopathy were apparent in the progeny of F344× (F344×DA) backcrosses after exposure to cyclic hyperoxia. Albino offspring with red eyes showed resistance to oxygen-induced retinopathy, whereas the retinal microvasculatures of those with pigmented coats and eyes were more susceptible to the same stimulus. These data provide evidence of an association between ocular pigmentation and susceptibility to oxygen-induced retinopathy in the DA rat. 

Erythropoietin is an archetypal hypoxia-induced protein. ³² Adding EPO to cultured human vascular endothelial cells triggers receptor phosphorylation, the activation of intracellular cell signaling cascades, and the induction of a proangiogenic phenotype. ³³ Further, EPO is a key factor in the development of retinal neovascularization in oxygen-induced retinopathy in mice. ³⁴ In the experiments reported herein, EPO expression in the retinas of parental strain rats and their backcross progeny after cyclic hyperoxic exposure closely paralleled the phenotypic findings; significantly higher levels of EPO were demonstrated in neonates that were sensitive to the attenuating effects of hyperoxia, providing supporting evidence for a heritable susceptibility trait. 

The association between ocular pigmentation and susceptibility may be causal or coincident. The gene encoding tyrosinase, an enzyme in the melanin biosynthetic pathway, is often mutated in albinos, and albinism in the Wistar rat was recently attributed to a missense mutation in the tyrosinase gene. ³⁵ Dihydroxyphenylalanine (DOPA), a product of tyrosinase catalysis, is involved in cell-cycle regulation, and a deficiency of the factor is thought to account for the increase in neuronal proliferation and disordered maturation found in albino retinas. ^{36 37 38} Oxygen-free radicals are generated in the process of melanin biosynthesis, and their accumulation is enhanced in hyperoxia. ³⁹ In vitro experiments have demonstrated that vascular endothelial cells are particularly sensitive to DOPA-mediated oxidative damage. ^{36 37} DOPA-mediated endothelial cell damage may thus be responsible for the susceptibility of pigmented rats to oxygen-induced retinopathy. An alternative possibility is that the production of pigment epithelium-derived factor (PEDF), a potent inhibitor of angiogenesis, ⁴⁰ is impeded by the biochemical sequelae of aberrant melanin biosynthesis in albino eyes. The retinal expression of PEDF protein was found to be increased 3.8-fold over room air-raised controls in pigmented Brown Norway neonatal rats exposed to hyperoxia. ⁴¹ No significant increase in PEDF levels was found for albino Sprague–Dawley rats at the same time point. Oxygen excess has been associated with an increase in retinal PEDF expression in vivo and in RPE cell culture. ^{42 43} Pigmented eyes may express PEDF to a greater extent under hyperoxia than do nonpigmented eyes, leading to an increase in PEDF-mediated inhibition of retinal angiogenesis during hyperoxic exposure. 

Epidemiologic studies have clearly identified an effect of ethnic background on the risk for ROP in that African American infants are at lower risk for severe disease than white American infants. ¹² Premature neonates born to white, indigenous Australian, Maori, and Pacific Islander mothers have all been reported to exhibit the same risk for retinopathy. ¹⁴ Collectively, these data suggest that hereditary factors other than those related to ocular pigmentation are of primary importance in the risk for retinopathy in humans. Our previous finding of a hierarchy of susceptibility among albino rat strains ²² lends support to this notion. 

Strain-related heterogeneity of ocular angiogenesis in rodents is not limited to the retina. Murine strain-dependent variations in the corneal response to angiogenic factors have recently been identified. ⁴⁴ Corneal stromal implantation of basic fibroblast growth factor–impregnated micropellets was associated with angiogenesis that differed by up to 10-fold among strains. Similar heterogeneity was seen in the response to VEGF. No clear association between susceptibility to corneal neovascularization and ocular pigmentation was apparent in these studies. Further, the extent of the resting limbal vasculature has been shown to differ considerably among mouse strains and is predictive of the response to basic fibroblast growth factor in the corneal neovascularization model. ⁴⁵ Together, these studies suggest that genetic factors play important roles in regulating angiogenesis in the cornea. 

In conclusion, these experiments have confirmed the heritability of the susceptibility trait to oxygen-induced retinopathy in the rat and have identified an association with ocular pigmentation in several strains. Identification of the molecular pathology underlying the strain-related differences may clarify the role of oxygen tension in the regulation of retinal angiogenesis. 

 Supported by the Ophthalmic Research Institute of Australia and the Flinders Medical Centre Research Foundation. KAW and PvW were supported by the National Health and Medical Research Council of Australia.

 Submitted for publication May 14, 2006; revised August 25 and October 30, 2006; accepted January 31, 2007.

 Disclosure: P. van Wijngaarden, None; H.M. Brereton, None; D.J. Coster, None; K.A. Williams, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Offspring of F1 and backcross matings. (A) 14-day-old F1 rat pups. Offspring of F344×DA F1 crosses, with agouti coats and dark brown eyes, resembled the DA parental strain. (B) F344× (F344×DA) F1 backcross pups and a F344×DA F1 dam. Offspring had a wide range of coat colors including white (albino), agouti, black, and hooded (either black or agouti). Albino rats had red eyes; rats with coat pigmentation had dark brown eyes.

Figure 2.

 

Representative montages of retinas from parental (F344 and DA) and F344×DA F1 oxygen-exposed rats. (A) F344 strain. (B) DA strain. (C, D) F344×DA F1. Neonatal rats were exposed to cyclic hyperoxia for 14 days, and retinal whole-mounts were stained with fluorochrome-conjugated GS IB4 to mark the microvasculature.

Figure 3.

 

Retinal avascular areas of backcross offspring from 2 matings of DA× (F344×DA) F1 rats and 2 matings of F344× (F344×DA) F1 rats. Neonatal rats were exposed to cyclic hyperoxia for 14 days before measurement of the avascular retinal area. Values for avascular area are expressed as a percentage of the total retinal area. Albino rats had unpigmented (red) eyes, and those with pigmented coats had pigmented (dark brown) eyes.

Table 1.

 

Retinal Phenotypes of F344×F1 Backcross Progeny Used to Prepare cDNA Pools for Quantification of EPO Expression in Retina after Cyclic Hyperoxic Exposure

Coat Color	Eye Color	Retinal Avascular Area (%)
Black	Dark brown	55.6
Black hooded	Dark brown	86.4
Black	Dark brown	78.3
Agouti	Dark brown	61.2
Black hooded	Dark brown	57.9
White	Red	37.5
White	Red	17.3

Figure 4.

 

Genetic modeling of susceptibility to oxygen-induced retinopathy. S, dominant susceptibility allele; s, recessive resistance allele; unfilled symbols, albino rats; filled symbols, pigmented rats. If susceptibility to oxygen-induced retinopathy is defined as total avascular retinal area greater than 50% after 14 days of hyperoxia, then autosomal dominant inheritance of a monogenic trait can be predicted as shown. All progeny of the F344×DA F1 cross are susceptible. All progeny of the DA× (F344×DA) F1 backcross carry at least one susceptibility allele and, therefore, express the susceptible phenotype; 50% of the offspring of the F344× (F344×DA) F1 backcross are susceptible and 50% are resistant to oxygen-induced retinopathy. Observed and expected numbers of cross and backcross offspring with either the susceptible or the resistant retinal phenotype are tabulated beneath the figure.

The authors thank Anne-Louise Smith for biomedical engineering expertise and Ray Yates for animal husbandry.