Unless stated, all reagents were purchased from Sigma–Aldrich (Poole, UK).

Homozygous cytochrome b₅ complete knockout mice (BCN, Cytb₅^−/−) and control (wild-type, Cytb₅^+/+) mice were generated as previously described (McLaughlin et al. ²⁰¹⁰).

All mice were maintained on a standard chow (RM1—Special Diet Services, Essex, UK) under standard animal house conditions, with free access to food and water, and 12 h light/12 h dark cycle. All animal work was carried out in accordance with the Animal Scientific Procedures Act (1986) and after local ethical review.

Skin sections were removed post mortem from the backs of 10 week old male wild-type and BCN mice (n = 3 for each genotype) and fixed flat on Whatman 3MM paper (GE Healthcare, Amersham, UK) in formal saline (Gurr^® BDH, VWR International, Lutterworth, UK) before being sectioned and stained with hematoxylin and eosin. Measurements were taken of the epidermis, dermis and hair shaft thickness using an Olympus IX50 microscope (Olympus UK Ltd., Watford, UK) with Zeiss Axio Vision version 4.5 software (Carl Zeiss Ltd., Welwyn Garden City, UK).

Thermal imaging experiments were performed in a closed room at a constant temperature of 20°C with tolerance of 1°C and lighting kept to a minimum. This arrangement ensured temperature variations that may adversely affect the performance of the thermal camera were negligible. Temperature measurements were obtained using a high-performance thermal camera (Cedip Infrared Systems, France) containing a 320 × 240 pixel focal plane array operating in the mid-infrared range (3–5 μm) with a thermal sensitivity of 0.02°C and spatial resolution of 0.2 mm, positioned 50 cm above the subject.

Body temperature measurements of 3 week old male wild-type and BCN pups, (n = 3 for each genotype) were taken at the same time of the day with a 2-day interval over a period of 6 weeks to allow temperature changes to be related to growth. In order to determine skin temperature, a small area (20 mm diameter) on the back of each mouse was shaved, immediately before temperature measurements were taken. Average and maximum temperatures of the shaved and haired areas were calculated using the specialized software associated with the camera (Song et al. ²⁰⁰⁷).

Nursing females (first litter) were milked 10 days postpartum. Briefly, mice were given an i.p. injection of oxytocin (100 μl, 5 U/ml) to simulate milk production. After 20 min, treated mice were anesthetized by inhalation of isoflurane and milk production facilitated by gentle squeezing of the nipples. The milk released was collected by pipette and samples were stored at −70°C prior to analysis.

Detailed fatty acid profile analysis was performed on milk, plasma, skin and liver by the Nutrition Analytical Service, University of Stirling, UK, according to the following method. Total lipids were extracted from skin and liver by homogenising in 20 volumes of chloroform/methanol (2:1 v/v) in an Ultra Turrax tissue disrupter (Fisher Scientific, Loughborough, UK). Total lipids were prepared from all samples according to the Folch method and non-lipid impurities were removed by washing with 0.88% (w/v) KCl (Folch et al. ¹⁹⁵⁷). The weight of lipids was determined gravimetrically after evaporation of the solvent and overnight desiccation under vacuum. Fatty acid methyl esters (FAME) were prepared by acid-catalysed transesterification of total lipids according to the method of Christie (Christie ²⁰⁰³). Extraction and purification of FAME was performed as described by Ghioni et al. (¹⁹⁹⁶). FAME were separated and quantified by gas–liquid chromatography using a Thermo Fisher Trace GC 2000 (Thermo Fisher, Hemel Hempstead, UK) equipped with a fused silica capillary column (ZB wax, 30 m × 0.32 mm i.d.; Phenomenex, Macclesfield, UK) with hydrogen as carrier gas and using on-column injection. The temperature gradient was from 50 to 150°C at 40°C/min and then to 195°C at 1.5°C/min and finally to 220°C at 2°C/min. Individual methyl esters were identified by comparison with known standards and by reference to published data. Data were collected and processed using the Chromcard for Windows (version 2.00) computer package (Thermoquest Italia S.p.A., Milan, Italy).

Data sets were analyzed for statistical significance using an Unpaired student t-test (www.graphpad.com/quickcalcs/t-test1.cfm).

To establish the in vivo role(s) of microsomal cytochrome b₅ we generated a mouse germline deletion of exons 2–5 of the gene. Mice lacking microsomal cytochrome b₅, were unexpectedly viable, born at expected Mendelian ratios and exhibited no gross anatomical abnormalities. Cytochrome b₅ complete null (BCN) mice of both sexes were fertile, indicating that cytochrome b₅ is not necessary for sexual maturation, demonstrating that cytochrome b₅ does not play an obligatory role in testosterone biosynthesis (Akhtar et al. ²⁰⁰⁵). However, analysis of testicular testosterone levels revealed that they were approximately 40% lower in BCN males and this was accompanied by the almost complete loss of the in vitro activity of Cyp17, a P450 essential for testosterone biosynthesis (McLaughlin et al. ²⁰¹⁰). In addition to its role in transferring electrons to P450s involved in the metabolism of steroids, cytochrome b₅ can also act as an electron donor to P450s involved in the metabolism of drugs and foreign compounds. In agreement with our previous findings in the conditional hepatic knockout model, BCN mice also had a marked reduction in in vivo P450 drug metabolism confirming the pivotal role this protein may play in drug metabolism in vivo (Finn et al. ²⁰⁰⁸; McLaughlin et al. ²⁰¹⁰).

Cytochrome b₅ is thought to play a key role in the reduction of methaemoglobin to haemoglobin (Jaffe ¹⁹⁸¹) and in agreement with this role, 10 week old male BCN mice had significantly higher levels of circulating methaemoglobin (tenfold) than their wild-type counterparts (Fig. S1). Methaemoglobinemia in humans is clinically defined as having methaemoglobin levels greater than 1%, with homozygous methaemoglobin reductase-deficient patients possessing levels greater than 10% (Percy et al. ²⁰⁰⁵). By this definition, BCN mice are methaemoglobinemic, however none of the other physiological phenotypes associated with methaemoglobinemia due to disruption of cytochrome b₅ or cytochrome b₅ reductase in humans were observed (Hegesh et al. ¹⁹⁸⁶; Giordano et al. ¹⁹⁹⁴; Manabe et al. ¹⁹⁹⁶). These data suggest that although cytochrome b₅ is part of the pathway of methaemoglobin reduction, the known alternative pathways such as the glucose/glucose-6-phosphate dehydrogenase are capable of partially compensating for loss of this protein and maintaining methaemoglobinemia at asymptomatic levels (Jaffe ¹⁹⁸¹). BCN mice may, however, be more susceptible to acquired methaemoglobinemia under conditions of metabolic stress or to chemicals known to induce methaemoglobinemia (Molthrop et al. ¹⁹⁹⁴; Mary and Bhupalam ²⁰⁰⁰).

Together these findings confirm in vivo the numerous functions assigned to microsomal cytochrome b₅ through in vitro studies carried over the past four decades.

Ten days after birth hair growth had commenced in wild-type mice but in contrast BCN pups remained hairless and exhibited a generalized scaling of the skin and hyperkeratosis on the tail which by 4 weeks of age resulted in development of constrictive bands, referred to as ‘ringtail’, a condition observed in rats kept in dry, low humidity environments (Fig. 1a, b). At 4 weeks, the hair coat of BCN mice had developed a “static” appearance and was noticeably thinner than in age-matched controls (Fig. 1c). By 10 weeks of age coat condition had improved further however, the difference in thickness compared to wild-type mice could still be clearly seen on the ventral surface (Fig. 1d). Microscopic examination of hair types from BCN and wild-type mice found no significant difference in the absolute numbers or proportion of each hair type between the two genotypes (data not shown). These data therefore suggest that deletion of microsomal cytochrome b₅ delays hair follicle development.

Histological examination of dorsal skin sections from 10 week old male wild-type (Fig. 1e, g) and BCN (Fig. 1f, h) mice identified pronounced keratinization of the stratified squamous epithelium in the latter and concomitant significant thickening of the epidermis—13.2 μm ± 0.6 versus 10.2 μm ± 0.3 (Fig. 1, compare e & f). Furthermore, hair shaft width was significantly finer in BCN mice than in wild-type animals (12.7 μm ± 0.09 vs. 14.7 μm ± 0.3; Fig. 1i). Similar increases in epidermal thickness and keratinization have been reported on disruption of the 12R-lipoxygenase, ALOX12B locus, in mice, an enzyme involved in an alternative pathway of arachidonic acid metabolism in the skin, and suggests that cytochrome b₅ may also contribute to this pathway of eicosanoid biosynthesis (Juanes et al. ²⁰⁰⁹). The skin phenotypes observed in BCN mice are highly similar to those observed in the human diseases classed as autosomal recessive congenital ichthyosis (ARCI) (Akiyama and Shimizu ²⁰⁰⁸).

A static hair phenotype has also been observed in stearoyl-CoA desaturase (SCD1) null mice (Miyazaki et al. ²⁰⁰¹). However, in SCD1 null mice, the effect is associated with atrophy of the sebaceous glands, as a result of lipid depletion, whereas in BCN mice the sebaceous glands were slightly enlarged (Fig. 1g, h). Also, the characteristic phenotypes associated with ichthyosis have not been reported for SCD1 null mice. These data therefore suggest a role for cytochrome b₅ in lipid and eicosanoid metabolism in skin, disruption of which leads to the development of ichthyosis. Ichthyosis was not reported in the methaemoglobinemia patient identified by Hegesh et al. who expressed a splice variant of cytochrome b₅; however this is most likely due to the presence of residual cytochrome b₅ expression/activity and not complete loss as in BCN mice (Hegesh et al. ¹⁹⁸⁶; Giordano et al. ¹⁹⁹⁴).

In light of the above data, we carried out fatty acid analysis of the skin. Significant changes in the levels of most saturated, monounsaturated and branched-chain fatty acid species, many of which are known to have emollient and moisturizing properties e.g. behenic and erucic acid, were observed in the skin of 10 week old male BCN mice (Table S1). In addition, significant changes in the levels of α-linolenic acid (decreased by 35%) and eicosadienoic acid (increased by 40%) were found (Table S1). Calculation of desaturation indices (i.e. the ratio of unsaturated to saturated fatty acids), as a measure of SCD activity (Attie et al. ²⁰⁰²), for both palmitic and stearic acid showed a 50 and 54% reduction, respectively, in BCN compared to wild-type mice (Fig. 2a). This may account for some of the similarities observed between BCN and SCD1 null mice. The skin phenotype of SCD1 null mice was ascribed to the disruption of the epidermal lipid barrier, due to loss of ω-hydroxylated very long chain fatty acids and the abolishment of lymphoid enhancer-binding factor 1 (lef1) expression, a transcription factor essential for correct development of the hair follicle and the sebaceous gland; however, in agreement with the morphological differences of the sebaceous gland between BCN and SCD1 null mice, lef1 expression levels are not significantly changed in 10 week old male BCN mice (data not shown) (Binczek et al. ²⁰⁰⁷; DasGupta and Fuchs ¹⁹⁹⁹). Members of the CYP4 family function in lipid metabolism as ω-hydroxylases, acting on long-chain fatty acids and Loughran et al. have showed that cytochrome b₅ is essential for the in vitro ω-hydroxylation of arachidonic acid (Hardwick ²⁰⁰⁸; Kalsotra and Strobel ²⁰⁰⁶; Loughran et al. ²⁰⁰¹). The skin phenotypes observed in BCN mice may therefore result from a combination of a change in unsaturated fatty acid biosynthesis and a reduction in the ω-hydroxylation of long-chain fatty acids. In support of this possibility, Lefévre et al. have identified mutations in the gene FLJ39501, a member of the CYP4F subfamily, associated with lamellar ichthyosis (Lefevre et al. ²⁰⁰⁶).

Approximately 30% of ARCI cases cannot be explained by genetic changes in known ichthyosis loci; the data presented here suggest that polymorphisms in the cytochrome b₅ pathway represent novel candidate genes for study. The striking similarities between the BCN phenotype and the human disease suggest that the BCN mouse could potentially be a valuable genetic model for developing new treatments for this disease and suggest new approaches for therapeutic intervention.

The loss of the epidermal lipid barrier in the skin of SCD1 null mice has been associated with perturbations in temperature control (Binczek et al. ²⁰⁰⁷), we therefore used thermal imaging to assess whether the changes in skin fatty acids observed in BCN mice also resulted in disruption of this barrier. A small patch of back hair was shaved on 3 week old male BCN and wild-type mice to allow skin surface temperature measurements to be recorded together with similar measurements from haired regions over a 6 week period (Fig. 2b, c). In shaved skin areas, despite significantly higher measurements at two (out of 18) time-points, no overall difference between BCN and wild-type mice in either average or maximum temperatures was detected over the 6 week period (Fig. S2). In contrast, BCN mice had consistently significantly higher average and maximum temperatures in the haired skin areas compared to wild-type mice over the entire 6 week period, however from 5 to 6 weeks of age the differences in both average and maximum temperatures between genotypes decreased and the trend was for these measurement to approach those recorded for wild-type mice (Fig. 2d, e). These data indicated that the heat losses observed in BCN mice are primarily due differences in hair coverage rather than changes in epidermal barrier permeability and confirm the visual observations that coat thickness improved with age in BCN mice.

As indicated by the thermal imaging data, the changes in coat condition, possibly associated with the changes in skin lipid composition, improved as the mice mature, however coat condition did start to deteriorate again with age and especially in nursing females. In nursing females, this led to hair loss with skin reddening, particularly under the chin, around the eyes and snout, with the birth of each subsequent litter (Fig. 3a, b). Post-weaning, the dams’ coat condition recovered and the skin inflammation subsided. Associated with these effects, pups from successive BCN litters were progressively smaller (unlike wild-type) suggesting the possibility that the dams’ milk was nutritionally deficient (Fig. 3c, d). This hypothesis was supported by the finding that the fostering of BCN pups to a wild-type mother significantly reduced the growth retardation. In contrast, when wild-type pups were nursed by a BCN female a striking reduction in growth was observed (Fig. 3e, f).

The possibility that BCN milk is deficient in essential nutrients was investigated by comparing the fat content of milk taken 10 days postpartum from nursing wild-type and BCN females (both nursing their first litter). BCN milk contained a significantly higher level of total lipid; the predominant change being an increase in total saturated fatty acid levels; consistent with the role of cytochrome b₅ in fatty acid desaturation (Fig. 4a, Table S2). However, a small increase in n-6 polyunsaturated fatty acids was also observed. Monounsaturated and n-3 polyunsaturated fatty acid content was unchanged. The increase in saturated fat was associated with an increase in lauric, myristic, palmitic and arachidic acid (increased by 72, 75, 56 and 67% respectively) and the increase in n-6 polyunsaturated fats was due specifically to a 32% (P = 0.021) increase in linoleic acid (Fig. 4b). In addition, the level of α-linolenic acid was significantly increased in BCN milk although total n-3 polyunsaturated fatty acid levels were unchanged (Fig. 4b).

Under normal circumstances the level of fat contained in milk is tightly regulated but the composition can change depending on the maternal diet and circulating levels of individual fatty acids (Innis ²⁰⁰⁷; Rudolph et al. ²⁰⁰⁷). Genetic disruption of lactogenesis or lipogenesis, for example through the inactivation of the α-lactalbumin gene or of the activity of the mammary Akt pathway, has however been shown to increase the fat content of milk resulting in increased viscosity (Stinnakre et al. ¹⁹⁹⁴; Schwertfeger et al. ²⁰⁰³). This has been suggested in both these situations to prevent pups obtaining sufficient qualities of milk to support normal growth and thus leading to growth retardation. Indeed, milk from BCN mice appeared to have increased viscosity which would be consistent with the increase in saturated fat content and could possibly lead to an inability of BCN pups to obtain sufficient nutrition to support normal growth. Cytochrome b₅ may therefore be an important determinant in regulating the consistency of milk by virtue of its role in controlling fatty acid desaturation in the mammary gland.

However, we believe that the underlying reason for the reduced growth is due to a metabolic deficiency rather than a nurturing issue as the phenotype was progressive in subsequent generations. In addition to dietary intake, polyunsaturated fatty acids such as arachidonic, docosahexaenoic and eicosapentaenoic acid can be synthesised from linoleic and α-linolenic acid via the actions of Δ5 and Δ6 desaturases in maternal tissues. These desaturases have been shown to be expressed in lactating mammary glands of rats and are active in the conversion of linoleic acid to the γ-linolenic and arachidonic acid (Rodriguez-Cruz et al. ²⁰⁰⁶). Additionally, the in vitro findings of Guillou et al. have shown that exogenously added cytochrome b₅ is capable of enhancing Δ6 desaturase activity (Guillou et al. ²⁰⁰⁴). The loss of desaturase activity could explain the increase in linoleic acid in the milk; humans genetic variations in these gene loci have been associated with altered polyunsaturated fatty acid levels in milk (Xie and Innis ²⁰⁰⁸). It is therefore possible that the changes in the levels of linoleic and α-linolenic acid in BCN milk are responsible for the growth retardation observed in BCN pups. In support of this, studies in rodents have shown that excess levels of these fatty acids result in postnatal growth restriction (Bongiovanni et al. ²⁰⁰⁷; Church et al. ²⁰⁰⁸). The animals used in this study were on a normal diet, nutritionally balanced for saturated and unsaturated fatty acids. It is therefore intriguing that in spite of this, defects in endogenous unsaturated fatty acid biosynthesis caused the above phenotypes. The progressive degradation in the health of dams, suggests that an equilibrium exists between circulating fatty acids and breast milk, but that this is not sufficient to reverse the nutritional deficiency.

In support of the role of cytochrome b₅ in overall fatty acid homeostasis, we found that in 16 week old male BCN mice the plasma level of γ-linolenic acid was significantly decreased (by 75%) and that of dihomo-γ-linolenic acid increased (by 153%) (Fig. 5a, Table S3). Additionally, in these mice, hepatic levels of these fatty acids followed a similar pattern with γ-linolenic acid decreased by 84% and dihomo-γ-linolenic acid increased by 195% (Fig. 5b, Table S4). Furthermore, hepatic levels of octadecatetraenoic acid were decreased by 80% and eicosatetraenoic acid increased from undetectable levels to 0.03 ± 0.01 mg/g tissue (Fig. 5b). These fatty acids correspond to the substrates and products of various reactions catalysed by the Δ5 and Δ6 desaturases (Napier et al. ²⁰⁰³), and the differences observed in the desaturation indices for these fatty acids provides evidence for cytochrome b₅ as a modulator of the activity of these enzymes (Fig. 5c, d). In view of the effects ascribed to dietary ω-3 and ω-6 fatty acids on human health, it is important to understand the role of the endogenous biosynthetic pathways on circulating levels of these unsaturated fatty acids.

Monounsaturated fatty acids can be obtained either from the diet or by de novo synthesis involving a cytochrome b₅-dependent pathway (Hackett and Strittmatter ¹⁹⁸⁴; Schenkman and Jansson ²⁰⁰³). We therefore determined saturated and monounsaturated fatty acid levels in plasma and liver from 10 and 16 week old male BCN mice (Tables S3, S4). At 16 weeks of age profound differences in both plasma and hepatic levels of stearic (increasing by 50 and 47%, respectively) and palmitoleic acid (decreasing by 34 and 52%, respectively) were observed (Fig. 5a, b). Plasma and hepatic levels of palmitic and oleic acid were effectively unchanged and only the hepatic level of palmitoleic acid was significantly changed at 10 weeks of age (Table S4). Determination of desaturation indices for both palmitic and stearic acid revealed decreases for both fatty acids in BCN mice; however, only the desaturation index for palmitic acid was significantly changed (Fig. 5c, d). These data demonstrate that under normal dietary conditions cytochrome b₅-dependent pathways affect circulating mono and polyunsaturated fatty acid concentrations and their activity will therefore influence the beneficial effects of these compounds. Interestingly, Zhu et al. have recently identified a cytochrome b₅/b₅ reductase fusion protein, NAD(P)H cytochrome b₅ oxidoreductase (Ncb5OR), located in the lumen of the endoplasmic reticulum, which when deleted in mice results in marked decreases in hepatic levels of palmitoleic and oleic acid, suggesting a role for this protein in electron transport to SCD1 (Zhu et al. ²⁰⁰⁴; Larade et al. ²⁰⁰⁸).

In summary, we have generated a novel genetic model that identifies the numerous and varied in vivo roles of cytochrome b₅ and now allows these pathways to be studied in detail. Using this model we have identified that cytochrome b₅ plays a key role in a wide range of metabolic processes of central importance to both nutrition, drug metabolism and a range of human diseases and further demonstrate the importance of de novo unsaturated fatty acid biosynthesis in human health (McLaughlin et al. ²⁰¹⁰). With the exception of the changes in hepatic lipid metabolism, none of the other phenotypic changes observed in BCN mice are seen in the conditional hepatic knockout model (Finn et al. ²⁰⁰⁸). This finding indicates that these phenotypic changes are most likely due to the specific absence of cytochrome b₅ in the particular tissue and not to the systematic disruption of cytochrome b₅ function, a fact that would also explain, with the exception of plasma and hepatic fatty acid changes which mimic each other, the differences in fatty acid profiles obtained from different tissues. These studies also demonstrate that certain endogenous pathways of fatty acid homeostasis cannot be substituted by dietary means. The relationship between de novo biosynthesis and dietary intake of fatty acids and how dietary intervention may influence the pathogenesis of human disease will be a fascinating area of further study.

Below is the link to the electronic supplementary material.