Mutations in Homocysteine Metabolism Genes Increase Keratin N-Homocysteinylation and Damage in Mice

Genetic or nutritional deficiencies in homocysteine (Hcy) metabolism increase Hcy-thiolactone, which causes protein damage by forming isopetide bonds with lysine residues, generating N-Hcy-protein. In the present work, we studied the prevalence and genetic determinants of keratin damage caused by homocysteinylation. We found that in mammals and birds, 35 to 98% of Hcy was bound to hair keratin via amide or isopeptide bond (Hcy-keratin), while 2 to 65% was S-Hcy-keratin. A major fraction of hair Hcy-keratin (56% to 93%), significantly higher in birds than in mammals, was sodium dodecyl sulfate-insoluble. Genetic hyperhomocysteinemia significantly increased N-Hcy-keratin levels in the mouse pelage. N-Hcy-keratin was elevated 3.5-, 6.3-, and 11.7-fold in hair from Mthfr−/−, Cse−/−, or Cbs−/− mice, respectively. The accumulation of N-Hcy in hair keratin led to a progressive reduction of N-Hcy-keratin solubility in sodium dodecyl sulfate, from 0.39 ± 0.04 in wild-type mice to 0.19 ± 0.03, 0.14 ± 0.01, and 0.07 ± 0.03 in Mthfr−/−, Cse−/−, or Cbs−/−animals, respectively. N-Hcy-keratin accelerated aggregation of unmodified keratin in Cbs−/− mouse hair. Keratin methionine, copper, and iron levels in mouse hair were not affected by hyperhomocysteinemia. These findings provide evidence that pelage keratin is N-homocysteinylated in vivo in mammals and birds, and that this process causes keratin damage, manifested by a reduced solubility.


Introduction
Homocysteine (Hcy) is an important intermediate in folate and one-carbon metabolism. The only known source of Hcy in our body is the essential dietary protein amino acid methionine (Met). Hcy levels are regulated by vitamin B 12 -and folate-dependent remethylation to Met, catalyzed by Met synthase (with methyltetrahydrofolate cofactor provided by the MTHFR enzyme) and betaine-Hcy methyltransferase, as well as by vitamin B 6 -dependent transsulfuration to cysteine, the first step of which is catalyzed by cystathionine β-synthase (CBS). Genetic or nutritional deficiencies in folate/one-carbon metabolism lead to the hyperhomocysteinemia (HHcy) and are known to cause abnormalities in many organs, including the cardiovascular system and the brain [1].
Although homocysteinylation can cause protein damage, the prevalence and genetic determinants of Hcy-damaged proteins in vivo are not well known. For this reason, the present work has been undertaken to study hair keratin from different species of mammals and birds, as well as mouse models of genetic HHcy (Cbs −/− , Cse −/− , and Mthfr −/− mice).

Animal
Hair. The hair samples have been collected from live domestic animals and from wild animals encountered in nature as a fresh road kill. Random feathers dropped by wild birds were collected in nature. Woolly mammoth hair was purchased from Educational Innovations.

Sample Preparation for
Hair S-Hcy-Keratin and Hcy-Keratin Assays. Samples were prepared by a modification of previously described procedures as described below. The interassay and intra-assay variabilities for the quantification of various forms of Hcy were 7.3 and 11.5% [17].
2.4. S-Hcy-Keratin. Hair or feathers (2 mg) were treated in a 0.5 ml Eppendorf polyethylene tube with 50 mM Na 2 HPO 4 , 20 mM NaOH, 25mM DTT, and 1% SDS (200 μl) for 1 h at 65°C. The extracts were collected by centrifugation, hair were extracted again with a fresh solution, and the extracts were combined (400 μl). The hair pellets were saved for quantification of SDS-insoluble N-Hcy-keratin.
A 40 μl portion of each extract was treated with DTT (4 μl, 0.25M) and HCl (4 μl, 12 N) at 100°C for 30 min to convert the liberated Hcy to Hcy-thiolactone, which was then quantified by HPLC [17]. Authentic samples of Hcy were similarly processed as standards for S-Hcy assays. This procedure liberates >95% Hcy present in hair as S-Hcy-keratin.
2.5. SDS-Soluble and SDS-Insoluble Hcy-Keratin. For SDSsoluble Hcy-keratin quantification, each SDS extract (300 μl) was supplemented with 34 μl 100% trichloroacetic acid and the precipitated keratin pellets were collected by centrifugation. SDS-soluble and SDS-insoluble pellets were transferred to 1 ml Wheaton Gold Band ampoules and hydrolyzed with 6N HCl and 50 mM DTT (110 μl, 120°C, 1 h) to liberate Hcy. The hydrolysates were lyophilized on Labconco CentriVap (40 min, 70°C), dissolved in 10 μl water, purified by two-dimensional TLC, and analyzed by HPLC. Horse spleen ferritin, containing 0.49 mol Hcy/mol protein, was processed in parallel as a standard. In these procedures, N-Hcy is quantitatively converted to Hcy-thiolactone, which is then assayed by HPLC.
2.6. HPLC, Detection, and Quantification. Quantification of Hcy-thiolactone generated from Hcy-keratin or S-Hcy-keratin was carried out as previously described [17] using a Beckman-Coulter System Gold Nouveau HPLC instrumentation with a manual injector (7725i Rheodyne, with 0.1 ml loop) and a Jasco 1520 fluorescence detector, controlled by a Gold Nouveau chromatography workstation software for Windows. Samples were injected onto a cation-exchange polysulfoethyl aspartamide column (35 × 2 mm, 5 μm, 300 Å) (PolyLC Inc.), eluted isocratically with 30 mM NaCl and 10 mM sodium phosphate buffer (pH 6.6) at a flow rate 0.6 ml/min. The effluent was mixed in a three-way tee with 2.5 mM OPA and 0.25 M NaOH, delivered at a flow rate 0.3 ml/min, the mixture passed through a reaction coil (Teflon tubing, 0.3 mm I.D. × 3 m), and the fluorescence at 480 nm was recorded (excitation 370 nm). Hcy-thiolactone eluted at 3 min in a 4 min run.

Total SDS-Soluble Keratin.
Total SDS-soluble keratin was prepared by extracting mouse hair (2 mg) twice with 50 mM Na 2 HPO 4 , 20 mM NaOH, 25 mM DTT, and 1% SDS (200 μl, 65°C, 1 h each). The extracts were collected by centrifugation and combined (400 μl). A 20 μl portion of each extract was analyzed by SDS-PAGE on 10% gels. Keratin bands were visualized by staining with Coomassie Blue and quantified by densitometry.

Copper and Iron Assays.
Mouse hair samples were mineralized with 65% nitric acid (Merck) in a microwave oven (Mars 5 Digestion Microwave System, CEM Corporation). Iron and copper were quantified by flame atomic absorption spectrometry using a Zeiss AAS-3 spectrometer with deuterium background correction as previously described [19]. The accuracy of the assay was 94% for iron and 102% for copper as verified by certified reference materials (Human Hair NCS DC73347a, LGC Standards).

Statistics.
The results are reported as mean ± standard deviation. Comparisons between two groups are analyzed by using two-sided Student's t-test. Relationships of Hcykeratin solubility vs. levels were fitted to logarithmic equations and analyzed by linear regression. The level of statistical significance was set to P < 0 05.

Hcy-Keratin and S-Hcy-Keratin Content in Animal Hair.
Our previous work has shown that Hcy, bound by a disulfide or an amide bond, is a component of human hair keratin [18]. To determine whether the presence of Hcy in hair keratin is a universal feature conserved across the animal world, we analyzed Hcy content in pelages of mammals and birds. We found that both Hcy-keratin and S-Hcy-keratin were present both in mammal and bird pelages at levels ranging from 56 to 733 pmol/mg hair ( Table 1). The higher content of Hcy-keratin and lower content of S-Hcy-keratin at the tip of the horse tail hair than at its base are similar to those observed previously in the long human head hair [18].
To determine whether the variation in keratin homocysteinylation across the mammalian and bird species can alter the protein structure, we measured the solubility of Hcy-keratin in sodium dodecyl sulfate (SDS) and examined how the solubility is affected by the level of homocysteinylation. We found that SDS-insoluble Hcy-keratin  varied from 78 to 93% in birds and from 56 to 92% in mammals (Table 1). Overall, birds had significantly more SDS-insoluble Hcy-keratin than mammals (0.87 ± 0.07 vs. 0.75 ± 0.12, P = 0 019; Table 2).

Hair Copper, Iron, and Met-Keratin Are Not Affected by
Hyperhomocysteinemia. In a growing human hair, Hcykeratin levels increase along the hair shaft as a result of copper and iron-dependent demethylation of methionine residues of keratin [18]. To determine whether this process could contribute to the accumulation of Hcy-keratin in HHcy, we quantified copper and iron levels in hair as well as Met content of hair keratin in Cse −/− and Cse +/+ mice. We found that there were no differences in copper (4.83 ± 0.70 vs. 5.37 ± 1.43 μg/g hair, P = 0 296) and iron (34.6 ± 10.7 vs. 40.1 ± 16.0 μg/g hair, P = 0 373) levels between Cse −/− and Cse +/+ mice (Table 5). We also found that Met-keratin levels were not affected by the inactivation of the Cse gene (41.0 ± 10.2 in Cse −/− vs. 38.8 ± 6.4 nmol/mg hair in Cse +/+ mice, P = 0 571) ( Table 5).

Discussion
The present work shows that Hcy-keratin is highly prevalent in a variety of mammal and bird species. Although animal pelages contain both S-Hcy-keratin and Hcy-keratin, mammals have significantly more S-Hcy-keratin and less of the Hcy-keratin fraction than birds. At the same time, birds have significantly more SDS-insoluble Hcy-keratin than mammals, possibly suggesting higher keratin damage in birds than in mammals. We also show that the Cbs, Cse, and Mthfr genes control the extent of keratin N-homocysteinylation and damage in mice.
In hair, two mechanisms can account for the presence of Hcy linked by an amide bond to keratin: (i) N-homocysteinylation of keratin lysine residues by Hcy-thiolactone and (ii) copper-and iron-dependent demethylation of keratin methionine residues [18]. Although we have not directly shown that Hcy is attached to a lysine residue in keratin, we have eliminated the 2 nd possibility by showing that copper, iron, and keratin methionine levels are not affected by HHcy in the Cse −/− mice (Table 5). In conjunction with previous findings that HHcy increases Hcy-thiolactone synthesis and Table 3: Hcy-keratin and S-Hcy-keratin levels and the solubility of hair Hcy-keratin in HHcy and control wild-type mice.

SDS-soluble Hcy-keratin fraction
protein N-homocysteinylation in mice and humans [2], our present data strongly suggest that the increased hair Hcy-keratin levels in HHcy mice are due to keratin Nhomocysteinylation.
The higher content of Hcy-keratin and lower content of S-Hcy-keratin at the tip of the horse tail hair than at its base are reminiscent of those observed in the long human head hair [18]. Elevated Hcy-keratin at the tip of the horse tail hair relative to its base is most likely caused by demethylation of methionine residues in hair keratin, as we have previously shown for the human hair.
Previous work shows that N-homocysteinylation alters protein structure/function and leads to the generation of insoluble protein aggregates [20] with amyloid-like properties [2,8]. The mechanism underlying these structural changes involves free Hcy thiol in N-Hcy-protein, which is prone to one-electron redox reactions that generate radicals and/or radial ions [21]. These radicals promote the formation of disulfide bonds leading to protein multimerization [20] and generate other reaction products [4]. Thiyl radicals produced from N-Hcy-protein undergo hydrogen atom transfer reactions, which generate C α -centered radicals, well-known precursors of protein carbonyls [4,21], which can account for an increased susceptibility of N-Hcy-proteins to oxidative damage [2]. These processes can also account for the findings of the present work showing that increasing N-Hcy-keratin levels in mouse hair leads to a progressive loss of its solubility.
To substantiate a conclusion that N-homocysteinylation causes keratin damage, we used mouse models of HHcy in which N-homocysteinylation is induced by the inactivation of the Cse (the present work, Table 3), Cbs, and Mthfr genes [2]. These mice have 3-to12-fold higher N-Hcy-keratin levels than the wild-type animals. Our findings that the solubility of N-Hcy-keratin in SDS is inversely correlated with the N-Hcy-keratin content in these mice suggests that the increasing extent of N-homocysteinylation causes progressive keratin damage. This damage is also reflected by significantly reduced solubility of total keratin from hair of Cbs −/− mice compared to their Cbs +/+ littermates ( Table 4).
The Cbs −/− mice have sparse hair with a significantly smaller diameter, compared to the Cbs +/+ animals [14,22]. It is likely that this morphological hair defect could be caused, at least in part, by the physicochemical impairments of keratin structure caused by greatly increased N-homocysteinylation in Cbs −/− mice, which is reflected by its reduced solubility in SDS (Tables 3 and 4).
Assuming that 1 mg hair contains about 20 nmol keratin, the 32% loss of solubility of total keratin (Table 4) corresponds to 6.4 nmol/mg hair more of SDS-insoluble keratin in Cbs −/− mice relative to their Cbs +/+ littermates. This value is 6-fold greater than the N-Hcy-keratin content in Cbs −/− mouse hair (1014 ± 86 pmol/mg hair, Table 3), suggesting that the aggregated N-Hcy-keratin (Table 3) serves as a "seed" that causes aggregation and a loss of solubility of unmodified keratin. These findings strongly suggest that the mechanism of protein aggregation induced by N-homocysteinylation demonstrated in vitro for the aggregation of albumin in the presence of a small amount of N-Hcy-albumin "seed" [8] also occurs in vivo in the mouse hair.
Protein damage can also be induced by S-thiolation, as recently shown for human serum albumin from hyperlipidemia patients, which carries Hcy and Cys bound via a disulfide bond to albumin Cys residues normally engaged in intrachain disulfide bonds [23]. It cannot be excluded that keratin damage can also be caused by the disruption of keratin disulfide bonds by S-homocysteinylation. However, this has not been studied in the present work, because S-Hcy-keratin could not be separated from Hcy-keratin.
About three dozens of individual Hcy-proteins have been identified in humans and animals, including albumin,  Figure 2: SDS-PAGE analysis of hair keratin from Cbs −/− and Cbs +/+ mice. Keratin was extracted from mouse hair (stored at 25°C for 6 months) with SDS/DTT solution [18], and the extracts were analyzed on 10% SDS-PAGE gels. Lanes 1-4, keratin from Cbs −/− mice; lanes 5-8, keratin from their Cbs +/+ littermates. Quantification by densitometry is shown in Table 4. The last lane on the right shows molecular weight markers. fibrinogen [2], collagen [7], dynein [5], actin and E-cadherin [6], and major urinary protein [17]. The present study adds mammalian and bird keratins to the list of Hcy-proteins identified in vivo.
In conclusion, our findings show that hair keratin in animals and birds is a target for homocysteinylation in vivo and that this process causes keratin damage, manifested by the reduction of its solubility.