Isolation and Characterization of a Felinine-containing Peptide from the Blood of the Domestic Cat ( Felis catus )*

Felinine is a unique sulfur-containing amino acid found in the urine of domestic cats and select members of the Felidae family. Research over the past 50 years has led to the conclusion that felinine must be synthesized in the kidney, as free felinine is not present in the blood or tissues of cats. We propose that felinine is present in the blood as (cid:1) -glutamylfelinylglycine, a glutathione conjugate. To test our hypothesis [ 35 S]cysteine was administered intraperitoneally to one entire male cat, and two radiolabeled fractions were isolated from the blood. We showed that the amounts of both fractions in serum were linked to the gender of the cat, with entire males expressing significantly higher levels compared with castrated males, entire females, or spayed females. Both fractions were characterized using amino acid analysis with one fraction (S18), containing an equimolar ratio of Cys, Glu, and Gly, while fraction S16 was found to contain Cys, plus free amino acids. Nanospray mass spectrometry confirmed the sequence of fraction S18 as being (cid:1) -glutamylfelinylglycine and conclusively proving that felinine is present in the blood of cats as part of a larger molecule, thereby questioning the current theory that felinine is synthesized in the kidney. Felinine is a sulfur-containing amino acid that has been found in the urine of certain members of Felidae. In the domestic cat ( Felis catus claim for a cat food (17). The diet provided the following nutrients (g/100 g dry matter): crude protein, 52; crude fat, 29; methionine, 1.2; cysteine, 1.8; taurine, 0.2; and calculated metabolisable energy of 461 kcal/100 g of dry matter. All cats had been vaccinated against feline rhinotrache-itis, calicivirus, and panleukopenia using a modified live vaccine (Felo- cell CVR, Norden Laboratories, Mu¨nchen, Germany). Feline leukemia performance detected Chromatography acid content isolated peptides on duplicate using a HPLC utilizing nin-hydrin derivatization

Felinine is a sulfur-containing amino acid that has been found in the urine of certain members of Felidae. In the domestic cat (Felis catus), the entire male has been found to produce the greatest amount of urinary felinine (122 mol/kg/day compared with 41, 36, and 20 mol/kg/day in castrated males, entire females, and castrated females, respectively (1)). The biological role of felinine in the cat remains unknown, but it has been postulated that it may be a precursor to a pheromone (2).
Since its discovery in 1951, many workers have attempted to detect felinine in various tissues and body fluids of the cat. Both Westall (3) and Datta and Harris (4) failed to detect the presence of free felinine in the plasma of cats. Tallan et al. (5) detected trace amounts of felinine in the plasma and in a bladder tissue extract but none in the kidney, liver, or brain. Roberts (6) found trace amounts of felinine in the kidney, liver, and skin. The general failure to detect felinine in tissues of the cat has led to the conclusion that the kidney is the likely site of felinine synthesis. Using more sensitive methods for detecting felinine, Hendriks et al. (7) reported that felinine was detecta-ble in the kidney, urine, and bladder of the domestic cat, but that felinine could not be found in the liver, skin, blood, intestines, pancreas, or the spleen.
There is convincing evidence that felinine is synthesized from the same isoprenoid pool as cholesterol (8). The formation of felinine in vivo can occur through a condensation reaction of an allylic carbonium ion and cysteine. In a previous study we have shown that radioactivity from [ 35 S]cysteine and [ 35 S]methionine is incorporated into felinine (7), while DL- [2-14 C]leucine, DL-[2-14 C]mevalonic acid, and [2-14 C]acetate have been shown to be incorporated into felinine by Avizonis and Wriston (9), Shapiro (10), and Wang (11).
A proposed metabolic pathway for the biosynthesis of felinine in the domestic cat describes felinine being formed from isopentenylpyrophosphate (IPP) 1 and cysteine, possibly in the liver (10 -13). We propose felinine arises from a conjugation reaction between the cysteine in glutathione (GSH) and IPP to form the tripeptide ␥-glutamylfelinylglycine, rather than free felinine being formed from free cysteine and IPP. This mechanism would be similar to that described for the formation of isovalthine, another unusual sulfur amino acid that has been isolated from the urine of domestic cats, lions, and humans suffering from hypothyroidism and hypercholesterolemia (14). Isovalthine has been shown to be formed in vitro as part of a glutathione-isovaleric acid conjugate (15). In the case of the ␥-glutamylfelinylglycine conjugate, following formation in tissues such as the liver the peptide could then be transported via the blood to the kidney, where free felinine could be liberated by the actions of ␥-glutamyltransferase and aminopeptidase M and then be excreted in the urine. Our hypothesis would explain the absence of free felinine in the blood of domestic cats.
The aim of this study was to determine whether a felininecontaining peptide is present in the blood of the domestic cat, and if so, to determine the structure of the peptide.

EXPERIMENTAL PROCEDURES
The studies reported here were approved by and conformed to the requirements of the Massey University Animal Ethics Committee (16).
Animals and Diet-Domestic short-haired cats (F. catus), 4 -7 years old from the Feline Unit at Massey University (Palmerston North, New Zealand) were used as experimental animals. The body weights of the cats at the start of the experiment ranged from 3.52 to 4.11 kg. Throughout the study, the cats were fed a moist canned cat food that passed a minimum feeding protocol for proving an adult maintenance claim for a cat food (17). The diet provided the following nutrients (g/100 g dry matter): crude protein, 52; crude fat, 29; methionine, 1.2; cysteine, 1.8; taurine, 0.2; and calculated metabolisable energy of 461 kcal/100 g of dry matter. All cats had been vaccinated against feline rhinotracheitis, calicivirus, and panleukopenia using a modified live vaccine (Felocell CVR, Norden Laboratories, Mü nchen, Germany). Feline leukemia * This work was supported by a Marsden fund grant from the Royal Society of New Zealand. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Preparation of Cat Serum for Reversed-phase High Performance Liquid Chromatography (HPLC) Analysis-Approximately 1-5 ml of blood was collected into a plain tube (Vacutainer, Becton Dickinson). The blood was allowed to clot for 2 h, before being centrifuged at 4000 rpm for 10 min for the collection of the serum. The serum was then ultrafiltered through a Vivaspin Concentrator (5000 molecular weight cutoff) (Vivascience Ltd., Lincoln, UK) by centrifugation at 7,500 rpm for 30 min. The ultrafiltrate was used for further analysis by HPLC and the retentate discarded.
Analysis and Separation of Peptides by HPLC-Peptides present in the ultrafiltered serum were separated by reversed-phase HPLC performed on a Phenomenex Luna 5-m column (4.6 ϫ 150 mm), using a Waters HPLC system (Waters Corporation, Milford, MA). A linear 75-min gradient was run, from 100% solvent I (0.1% trifluoroacetic acid) to 100% solvent II (0.1% trifluoroacetic acid and 90% acetonitrile), which commenced 15 min after sample injection. Peptides were detected by absorbance at 214 nm and were manually collected, as required. Where applicable, peaks were integrated using Millennium 32 ® Chromatography Manager software (Waters Corporation).
Quantitation of 35 S Radioactivity-Radioactivity present in the [ 35 S]cysteine injection solution, serum samples, and fractions collected from reversed-phase HPLC were determined as described by Hendriks et al. (7).
Amino Acid Analysis-The amino acid content of the isolated peptides was determined on duplicate samples using a Waters ion-exchange HPLC system (Waters Corporation), utilizing post-column ninhydrin derivatization and detection at 570 and 440 nm, following hydrolysis in 6 M glass-distilled HCl containing 0.1% phenol for 24 h at 110 Ϯ 2°C in evacuated sealed tubes. Cysteine, methionine, and tryptophan are partially or completely destroyed during the acid hydrolysis stage using this method.
Mass Spectral Analysis-The mass spectral analyses were performed at the Division of Immunology, Beckman Research Institute of the City of Hope (Duarte, CA), using electrospray ionization in the positive ion mode on a ThermoFinnigan LCQ Deca iontrap mass spectrometer. Tandem mass spectrometry (MS/MS) was based on collision-induced dissociation (CID). The sample, dissolved in 1:1 acetonitrile:water and 2% acetic acid, was introduced into the mass spectrometer via a nanospray needle assembly. Experiment 1-One entire male cat (3.8 kg) was administered 1.85 ml of sterile isotonic (9 g/liter) saline containing 250 Ci (9.25 MBq) of [ 35 S]cysteine (Amersham Biosciences Inc.) by intraperitoneal injection. The cat was individually housed in a metabolism cage with free access to food and fresh water during the experimental period. At times 0, 1, 2, 4, and 8 h a 1-ml blood sample was taken by jugular vein puncture, under anesthetic (isoflurane). Serum was separated from each blood sample and ultrafiltered prior to separation by HPLC as described previously. All the major peaks present in the chromatogram of each ultrafiltered serum sample, and the base-line regions between peaks, were manually collected into preweighed tubes, after which the tubes were reweighed to determine the quantity of fluid collected prior to measurement of radioactivity.
Experiment 2-Blood samples were taken from cats of different gender by jugular vein puncture. Serum was separated from each blood sample and ultrafiltered prior to separation and quantitation of peaks S16 and S18, by HPLC as described previously.
Experiment 3-Following an appropriate "cool down" period, blood from the cat used in Experiment 1 was obtained. Fractions S16 and S18 were isolated from ultrafiltered serum as described previously and used for characterization studies.

Identification of [ 35 S]Cysteine Serum Peptides-The results
from Experiment 1 showed the presence of three major 35 Scontaining peaks, which between them contained 65-99% of the radioactivity in the loaded serum sample, depending on the time the blood samples were collected after injection of the [ 35 S]cysteine. The majority of the radioactive material was in the injection peak (material that did not bind to the column), which comprised between 59 and 83% of the radioactivity in the serum samples over the first 8 h. This material was possibly sulfate. The remaining two radioactive peaks, which eluted at ϳ16.5% (S16) and 21.5% (S18) of the gradient contained between 1-9% and 2-10% of the radioactivity in the serum sample, respectively. These two 35 S-containing peaks S16 and S18 were identified as possible candidates for the felinine-containing peptide and were therefore subjected to further analysis by amino acid analysis, mass spectrometry, and amino acid sequencing. Fig. 1 shows the incorporation of [ 35 S]cysteine into fractions S16 and S18 over time. The level of incorporation of 35 S into S16 was the greatest 1 h after the [ 35 S]cysteine was administered, after which time it plateaued at ϳ62% of its maximum level. The level of incorporation of 35 S into S18 was at a maximum 2 h after administration of the [ 35 S]cysteine, dropping to ϳ35% of the maximum level 8 h after administration.

Incorporation of [ 35 S]Cysteine into Serum Peptides Over Time-
Comparison of the Peak Areas between Cats of Different Gender (Experiment 2)-Following the identification of peaks S16 FIG. 1. Incorporation of 35 S radioactivity into fractions S16 (⅜ ⅜) and S18 (q) over time. ␥-Glutamylfelinylglycine, a Peptide Isolated from Cat Blood and S18 as possible candidates for a felinine containing peptide, serum samples from cats of different gender were analyzed to quantitate the level of these peptide fractions. Since plasma felinine levels in entire male cats are ϳ3-6 times higher than in either castrated males or females, it was postulated that differences in the amounts of felinine-containing peptide present in the serum of cats of different gender should also be observed. Table I shows the peak areas of S16 and S18 in the serum and also the calculated peptide concentration for S18 in the blood. The area of peak S16 was 1.4 -2.3 times higher in entire male cats compared with castrated males, entire females, and spayed females. Meanwhile, the area of peak S18, and consequently the amount of the peptide present, was 7-22 times higher in entire male cats compared with castrated males, entire females, and spayed females. There were significant gender-specific differences in the amounts of compounds in both peaks S16 (p Ͻ 0.001) and S18 (p Ͻ 0.0001), with entire males having significantly higher levels in their blood compared with cats of other gender. The quantity of peptide S18 present in cat serum has the strongest linkage to the gender of the cat.

Characterization of [ 35 S]Cysteine Peptide Fractions (Experiment 3)-
The molar ratios of the amino acids found in fractions S16 and S18 as determined by amino acid analysis are shown in Table II. S16 showed the presence of Cys, Leu, and Tyr; however, analysis of the fraction without prior hydrolysis indicated the presence of both free Leu and Tyr in the sample. Consequently, based on the molar ratios of unhydrolyzed and hydrolyzed samples it is likely that all of the Tyr and at least some of the Leu found in the sample originated from free amino acids that are not in peptide form.
Hydrolysis of fraction S18 yielded the amino acids Glu, Gly, and Cys in approximately equimolar amounts (Table II), and no free amino acids were found to be present. This stoichiometry is the same as seen in the tripeptide glutathione, which has the amino acid sequence ␥-Glu-Cys-Gly. Evidence that this peptide was not glutathione comes from the observation that identical acid hydrolysis of glutathione undertaken at the same time as that of S16 and S18 resulted in a different stoichiometry (1:1:0.007 for Glu, Gly, and Cys, respectively (Table II)). The majority of the Cys present in glutathione was destroyed under the hydrolysis conditions used. Further evidence that S18 is not glutathione comes from the fact that HPLC analysis of both oxidized and reduced forms of glutathione showed that neither co-elute from the HPLC column in a position similar to S18.
A possible explanation for the high recovery of Cys, following acid hydrolysis of fraction S18, is that the Cys is actually derived from felinine, and Cys itself does not occur in the S18 peptide. It has been found that free felinine is totally degraded during acid hydrolysis under the conditions used in our laboratory (18). However, Westall (3) found that during acid hydrolysis felinine degraded to cysteine and cystine, while Tallan et al. (5) reported that felinine degraded to cystine and cysteic acid, thus indicating a possible degradation pathway of felinine via Cys. Similarly Cys is normally partially degraded under acid hydrolysis conditions. In the case of S18 it is possible that felinine is more stable to acid hydrolysis when it is present within a peptide, resulting in slower degradation to Cys, which in turn is more stable to hydrolysis, since it faces a shorter exposure time to acid. Therefore the level of Cys detected as a result of the destruction of the felinine-containing peptide may be far greater than if the peptide contained Cys.
Mass Spectral Analysis of S18 -Mass spectral analysis of S18 resulted in a monoisotopic mass of 394 being detected (Fig.  2). This is consistent with the presence of a protonated species of peptide containing Glu, Felinine, and Gly. Once the monoisotopic mass and charge state was determined this species (m/z 394) was selected and fragmented. This resulted in the production of CID MS/MS spectra, which displayed sequence ions at 319, 307.9, 264.9, 247.7, 178.9, and 161.9 (Fig. 3A).
The molecular ion at m/z 319 corresponds to Glu-felinine and represents the loss of a C-terminal Gly residue (M r 75); the strong molecular ion at 264.9 corresponds to felinine-Gly, indicating the loss of an N-terminal Glu residue (M r 129), and in this case most likely a ␥-Glu residue as the cleavage of the ␥-glutamyl bond, which is favored during CID (19). The m/z at 307.9 corresponds to protonated GSH, indicating the loss of the FIG. 2. Nanospray mass spectrometry spectrum of fraction S18.   ␥-Glutamylfelinylglycine, a Peptide Isolated from Cat Blood ions in the MS/MS spectra (m/z 247.8, 178.9, and 161.9) were derived from it (Fig. 3B).
The mass spectral data indicate that the sequence of the peptide S18 is ␥-glutamylfelinylglycine.

DISCUSSION
This study has shown that intraperitoneal administration of [ 35 S]cysteine to an entire male cat results in the incorporation of the 35 S moiety into material found in three serum peaks separated by reversed-phase HPLC: the injection peak, S16, and S18. The injection peak was not characterized, but this is most likely composed of sulfate. Results from an experiment investigating S16 and S18 peaks in cats of different gender showed that although the amounts of material eluting under these peaks were gender-dependent, the level of S18 was more strongly linked to the gender of the cat, with entire male cats expressing 7-22 times more S18 material than cats of other gender. Results from the amino acid analysis of S18 showed the presence of Glu, Gly, and Cys, in a stoichiometry of ϳ1:1:1. The knowledge that felinine breaks down under conditions of normal acid hydrolysis and potentially could give rise to Cys, due to the loss of the 2-methyl butanol group, along with the fact that peak S18 did not co-elute with either GSH or GSSG, suggested that this peak was not glutathione. This information taken together with the clear linkage between the level of expression of S18 and the gender of the cat strongly indicated that S18 was more likely to contain a felinine containing peptide than S16. It was also noted that S18, when reconstituted following freeze-drying, emitted a catty odor typical of stored felinine-containing samples. 2 S18, and not S16, was therefore further characterized by mass spectral analysis and amino acid sequencing.
It is more likely that felinine is produced as the result of a glutathione conjugation reaction rather than de novo synthesis of the tripeptide, as no free felinine can be detected in any tissues or blood of the cat (7). Consequently, the 35 S label present in the felinine-containing peptide must originate from 35 S-GSH. However we did not detect any 35 S-GSH in the serum of the cat. This is not altogether unexpected, since first, although the level of GSH in the blood of most species is in the order of 1.3 mM or lower (20), most is confined to the erythrocytes, with the plasma only containing between 1 and 30 M GSH (21). Second, the apparent level of GSH in most tissues and physiological fluids can alter rapidly post mortem or postcollection. In particular plasma GSH levels can drop rapidly after collection; in fact mouse plasma has a t1 ⁄2 for GSH of around 10 min (20). Therefore it is unlikely, under the conditions used here, that the GSH present in the cat serum at the time of analysis would have been detectable.
Collision-induced dissociation of MH ϩ parent ion species has been found to be a very effective means of determining the structures of glutathione conjugates. CID has been found to specifically favor the rupture of peptide bonds in glutathione conjugates, in particular the ␥-glutamyl linkage (19). In fact MS/MS has been employed as a screening process for the detection of unknown glutathione conjugates by either searching for species that differ in molecular weight by 129, the result of the loss of the ␥-Glu, or scanning for a molecular ion of 308, which indicates those MH ϩ ions that fragment to give protonated GSH (19). The fragmentation pathway of S18 seen in our work is typical of that seen with other glutathione conjugates following MS/MS (19,22). Sequencing of S18 by mass spectrometry clearly indicates that the sequence of the peptide is Glufelinine-Gly, and the high relative abundance of the m/z 265 species is indicative that the residue removed is most likely a ␥-Glu. Therefore based on the mass spectrometry data, the sequence of the felinine-containing peptide isolated from cat serum is ␥-glutamylfelinylglycine (Fig. 4).
The biological role of GSH in protecting cells against chemically reactive and cytotoxic agents is well known. This protective function is due to the presence of the nucleophilic sulfhydryl group through which glutathione is able to form conjugates with reactive compounds. The conjugates may then be excreted into the bile or eliminated via the mercapturic acid pathway into the urine. The formation of GSH conjugates has been shown to be mediated either by specific GSH transferases in the liver or to be nonenzymatic in nature (23).
In the case of the formation of ␥-glutamylfelinylglycine, both of the substrates, GSH and IPP, are naturally found in all mammals; therefore, if the reaction was nonenzymatic in nature then felinine could be expected in the urine of species other than the Felidae. It is therefore most likely that an enzyme specific to selected cat species and under hormonal control is responsible for the formation of ␥-glutamylfelinylglycine. The liver is a prime candidate for the site of ␥-glutamylfelinylglycine synthesis because of its high metabolic capacity as well as being a major site of cholesterol synthesis and also the largest supplier of glutathione in the body. The ␥-glutamylfelinylglycine would then be transported from the liver via the blood stream to the kidney where the enzymes responsible for the degradation of GSH, ␥-glutamyltransferase and aminopeptidase M, could degrade the ␥-glutamylfelinylglycine into its constituent amino acids. The Glu and Gly could be recycled, and the felinine would be excreted in the urine. As free felinine would not be present in any organs or the blood, this would explain the observations by several researchers that free felinine has only been found to any great extent in the urine of cats, and trace amounts in the kidney, where we propose felinine is liberated from the ␥-glutamylfelinylglycine peptide, and in the bladder, where contamination from the urine is clearly possible. However, this hypothesis remains to be proven.
It can be calculated from the average glomerular filtration rate (2.94 ml/min/kg), the renal blood flow (10.61 ml/min/kg), and the 24-h excretion rate of felinine by entire male cats (1,24) that the minimum concentration of the felinine-containing peptide in the blood of entire male cats, assuming 100% clearance efficiency by the kidneys, is 41.5 nmol/ml. The concentration of tripeptide in the blood of the entire male cats in the present study was determined to be 80.6 nmol/ml (Table I). The latter value is in the same order of magnitude and sufficient to account for the levels of felinine found in the urine of entire male cats.
It is unlikely that the biological reason for the formation of ␥-glutamylfelinylglycine in cats is one of cell protection by the conjugation of potentially damaging agents given that all mammals contain the substrates required for ␥-glutamylfelinylglycine synthesis. The fact that the levels of both felinine and ␥-glutamylfelinylglycine produced by cats are so clearly gender-linked, along with the fact that females or castrated males can be induced to produce felinine following testosterone injec- ␥-Glutamylfelinylglycine, a Peptide Isolated from Cat Blood tions (25), adds further weight to the suggestion that the biological role of felinine is as a precursor to a pheromone.
We have conclusively shown that contrary to all other reports, felinine can be found in the blood of cats albeit as part of a larger molecule. This discovery raises major questions regarding the validity of the current theory, which has prevailed for the past 50 years, that felinine is synthesized in the kidney and provides for the possibility that felinine may be synthesized in tissues other than the kidney.