Anal. Chem. 2002, 74, 5960-5968
Identification and Quantification of Feathers, Down, and Hair of Avian and Mammalian Origin Using Matrix-Assisted Laser Desorption/Ionization Time-of-Flight Mass Spectrometry Klaus Hollemeyer,*,† Wolfgang Altmeyer,‡ and Elmar Heinzle†
Biochemical Engineering Institute, Im Stadtwald, Saarland University, D-66123 Saarbru¨cken, Germany, and GENE-FACTS GbR, Science Park Saar, Stuhlsatzenweg 69, D-66123 Saarbru¨cken, Germany
We describe a fast method for the identification of the origin of native and chemically processed feathers, down, and hair and for distinguishing closely related species using enzyme digestion and MALDI-TOF mass spectrometry. Additionally we present two methods for the quantification of different identified bird and mammalian samples, respectively, in binary species mixtures. Without any prior cleaning or isolation of single proteins, enzymatical digests of feathers, down, and hair are performed. Fragments generated are analyzed by MALDI-TOF mass spectrometry, and peak groups of different selectivity are established for every animal species. For the identification of individual animal species, only unique species-specific peaks are accepted while for species classification of tinted or bleached hair, the group of semispecific peaks (SEMPs) is used. Samples from native animal species show a higher coincidence of digest peak masses of SEMPs than far-related species, indicating a phylogenetical relationship of the investigated structure proteins. Quantification of mixed binary avian samples is performed by identification of preweighed single feathers and down, followed by calculation of the gravimetric mass fractions. The composition of mixed binary mammalian samples is calculated from correlation of the quotient of the relative peak intensities or areas of these species and the quotient of corresponding gravimetric mass fractions. High accuracy is achieved by both quantification methods. The fast methods are well suited for industrial quality control for example. For the identification of the origin of biological samples, several methods analyzing proteins, DNA, RNA, or lipids are commonly used. In the quality control of food, where usually soluble proteins are present, electrophoretic methods are applied under denaturing or nondenaturing conditions (PAGE) 1 as well as isoelectric focusing (IEF).2,3 Electrophoretic methods use whole proteins in mixtures from native samples and separate these proteins. The isolated single proteins are digested, and the amino acid se* Corresponding author. Fax: ++49-(0)681-302-4572. Phone: ++49-(0)681302-3721. E-mail:
[email protected]. † Saarland University. ‡ GENE-FACTS GbR.
5960 Analytical Chemistry, Vol. 74, No. 23, December 1, 2002
quences are determined by comparisons with known sequences from several species. All these methods are time-consuming and limited to a relative small number of parallel samples. A further disadvantage is the limited number of species with known amino acid sequences. Enzyme-linked immunoassays (ELISA) are also often used for the determination of animal species.4 All these methods essentially depend on the presence of soluble proteins. A nonprotein method for the determination of species is capillary gas chromatography of derivatized fatty acids, normally as fatty acid methyl esters (FAME). This method is limited to samples containing a relatively high content of fat.5 The most sensitive method is the nucleic acid-based polymerase chain reaction analysis (PCR)6 that is able to detect species-specific differences in the DNA and RNA pool. It is a powerful method in differentiating species, and even single individuals can be identified. This is done by amplifying the nucleic acids present in almost unlimited (1) Bundesinstitut fu ¨ r gesundheitlichen Verbraucherschutz und Veterina¨rmedizin, Band I/3, Lebensmittel (L), Teil 2, Beuth Verlag GmbH, Berlin, Ko ¨ln, Wien, Zu ¨ rich, Nachweis der Tierart bei nativem Muskelfleisch in Polyacrylamid- Gelen mit Hilfe der Standard- Elektrophorese (PAGE), Amtliche Sammlung von Untersuchungsverfahren nach § 35 LMBG, Methode 06.00-27, Dezember 1988. (2) Bundesinstitut fu ¨ r gesundheitlichen Verbraucherschutz und Veterina¨rmedizin, Band I/3, Lebensmittel (L), Teil 1a, Beuth Verlag GmbH, Berlin, Ko ¨ln, Wien, Zu ¨ rich, Nachweis der Tierart bei Milch, Milchprodukten und Ka¨se mit Hilfe der isoelektrischen Fokussierung (PAGIF), Amtliche Sammlung von Untersuchungsverfahren nach § 35 LMBG, Methode 01.00-39, January 1995. (3) Bundesinstitut fu ¨ r gesundheitlichen Verbraucherschutz und Veterina¨rmedizin, Band I/1b, Lebensmittel (L), Teil 1b, Beuth Verlag GmbH, Berlin, Ko ¨ln, Wien, Zu ¨ rich, Nachweis von Kuhmilchkasein in Ka¨se aus SchafZiegen- oder Bu ¨ ffelmilch oder aus Gemischen von Schaf- Ziegen oder Bu ¨ ffelmilch, Referenzmethode,. Amtliche Sammlung von Untersuchungsverfahren nach § 35 LMBG, Methode 03.52-1(EG), September 1997. (4) Bundesinstitut fu ¨ r gesundheitlichen Verbraucherschutz und Veterina¨rmedizin, Band I/1c, Lebensmittel (L), Teil 1c, Beuth Verlag GmbH, Berlin, Ko ¨ln, Wien, Zu ¨ rich, Immunoenzymatischer Nachweis der Tierart bei erhitzten Fleisch- und Fleischerzeugnissen; ELISA- Verfahren im Mikrotitertestsystem, Amtliche Sammlung von Untersuchungsverfahren nach § 35 LMBG, Methode 06.00-47, November 1999. (5) Bundesinstitut fu ¨ r gesundheitlichen Verbraucherschutz und Veterina¨rmedizin, Band I/3, Lebensmittel (L), Teil 2, Beuth Verlag GmbH, Berlin, Ko ¨ln, Wien, Zu ¨ rich, Nachweis von rohem und erhitztem Rind- und Schweinefleisch in Fleisch und Fleisch-Erzeugnissen, Screening Verfahren, Amtliche Sammlung von Untersuchungsverfahren nach § 35 LMBG, Methode 01.00-39, January 1995. (6) Genetisches Analyseverfahren zur Abstammungsu ¨ berpru ¨ fung biologischer Materialien durch Verwendung artspezifischer Primer, Patent DE 0019842991A1, 1998. 10.1021/ac020347f CCC: $22.00
© 2002 American Chemical Society Published on Web 11/01/2002
copies and subsequent electrophoresis of the highly specific oligonucleotides created by enzymatic cleavage of the amplified nucleic acids. Thus, using infinitesimal traces of original samples, the method is mainly used in science and in forensic medicine. However, it essentially depends on the presence of amplifiable nucleic acid sequences. Furthermore, the quantification with PCR is still very difficult to perform. Feathers, down, hairs, nails, scales, beaks, horns, and hooves mainly consist of keratin structures. Keratins are almost insoluble, highly fibrous, cysteine-rich intracellular structure proteins of the land vertebrates. In feathers and down, 95% of the total material is composed of β-keratin forming a twisted sheet secondary structure. Two β-keratin chains are arranged antiparallel, forming a left-twisting helical superstructure cross-linked by amino acid side chains.7 Hair is mainly composed of intermediate filament proteins (IFPs), which can be divided into two protein families (type I and type II), each consisting of four proteins.8 One protein from type I and one protein from type II, respectively, form a dimeric structure and 16 dimers form an intermediate filament.9 The matrix proteins embedding these filaments and creating bicomponent polymers are classified in several groups according to their content of glycine and cysteine, respectively. High-sulfur proteins (HSP) are cysteine-rich matrix proteins with a content of sulfur less than 30%; ultrahigh-sulfur proteins (UHSP) have a higher sulfur content.10 Glycine- and tyrosine-rich keratin-associated proteins are called high-glycine tyrosine proteins (HGTPs). Some of these proteins show a high degree of sequence homology and oligopeptide repeats reaching homologies up to 96%, whereas other proteins exhibit distinctive amino acid compositions, which can be used for the identification of species. After keratinization, the producing cells die and the nuclei and the other cell organelles are reabsorbed. This absorption of nucleic acids is one of the reasons why it is difficult to perform PCR analysis. Keratins, although mostly insoluble, can be extracted from native hair using a high concentration of urea and a high content of thiols, reducing disulfide bridges.11 Separation by electrophoresis11 and isoelectric focusing was successfully used to identify species-specific keratins,12 and sequencing data from many different types of keratins are known.13,14 Although these methods can be used for identifying animal species,15 they are suitable for neither fast analysis nor quantification in mixed samples. So far, native and chemically processed feathers, down, and hair are mostly determined by several macroscopic and micro(7) Frasier, R. D. B.; MacRae, T. P.; Rogers, G. E. Keratins: Their Composition, Structure and Biosysnthesis; Charles C. Thomas Publ.: Springfield, IL, 1972. (8) Powell, B.; Crocker, L. A.; Rogers, G. E. Development 1992, 114, 417-433. (9) Wilson, B. W.; Edwards, K. J.; Sleigh, M. J.; Bryne, C. R.; Ward, K. A. Gene 1988, 73, 21-31 (10) Powell, B. C.; Rogers, G. E. In Formation and Structure of Human Hair; Jolle`s, P., Kahn, H., Ho¨cker, H., Eds.; Birkha¨user Verlage: Basel, 1977; pp 58-148. (11) Winter, H.; Hofmann, I.; Langbein, L.; Rogers, M. A.; Schweizer, J., Jr. J. Biol. Chem. 1997, 272, 32345-32252. (12) Butler, D. J.; De Forest, P. R.; Kobilinsky, L., Jr. Forensic Sci. 1990, 35 (2), 334-344. (13) Langbein, L.; Rogers, M. A.; Winter, H.; Praetzel, S.; Schweizer, J., Jr. J. Biol. Chem. 2001, 276, 35123-35132. (14) Langbein, L.; Rogers, M. A.; Winter, H.; Praetzel, S.; Beckhaus, U.; Rackwitz, H. R.; Schweizer, J., Jr. J. Biol. Chem. 1999, 274, 19874-19884. (15) Plowman, J. E.; Bryson, W. G.; Jordan, T. W. Electrophoresis 2000, 17, 16711676.
scopic visual methods. These methods need a high degree of experience and often fail because of the lack of differing morphological patterns,16 especially in juvenile animals. Our newly developed analytical method for the determination of cell-free biological material like feathers, down, and hair overcomes all these drawbacks because the samples do not require soluble proteins, fat, nucleic acids, or visible different morphological patterns. A modified enzymatic digest of a wellknown standard protocol17 with very high contents of trypsin and 2-mercaptoethanol delivered a mixture containing species-specific and species-unspecific digest peptides. The peptides were measured by MALDI-TOF mass spectrometry, and species-specific peptide masses of certified animal species were identified and used as standards for unknown samples. The method was used to identify chemically degreased duck and goose feathers, down, and native feathers from dove, chicken, and turkey as well as hair from several mammalian species such as rabbit, sheep, camel, llama, cashmere goat, mohair goat, mink, red fox, dog, yak, and human. Tinted or bleached hair samples showed a reduced number of original species-specific peptide masses. Additionally, new peaks were observed. It was even possible to recognize a phylogenetical tendency of the relationship within the determined avian and mammalian species, respectively, by comparing the percentage of coincident peptide masses. The closer the relationship, the larger the number of coincident peptide masses that could be detected. Quantification data led to reliable results for mixtures in binary animal sample systems. With high accuracy, even small amounts of contaminating feathers, down, and hair, foreign to the investigated species, could be estimated. In binary avian test systems, quantification was done by identifying each single feather or down and summing the masses of both species. In binary mammalian test systems, mass fractions of one species were determined by calculating the ratios of the relative intensities or the peak areas of two suitable peptide masses, being specific for the two selected species. The measurement of large numbers of samples can be investigated in a short analysis time making automatic runs. EXPERIMENTAL SECTION Chemicals and Analyzed Samples. Trypsin from hog pancreas with an activity of 1645 units/mg was purchased from Fluka (Deisenhofen, Germany). Trifluoroacetic acid (TFA), R-cyano-4hydroxycinnamic acid (CCA), 2-mercaptoethanol, and the calibration peptides human angiotensin II, substance P, human neurotensin, and the human adenocorticotropic hormone fragments ACTH(1-17) and ACTH(18-39) were obtained from Sigma (Deisenhofen, Germany). Ammonium bicarbonate and acetonitrile p.a. were obtained from Merck (Darmstadt, Germany). Certified feathers and down from eider and several races of duck and goose were provided from Forschungsinstitut Hohenstein (Bo¨nnigheim, Germany) and from H. Brinkhaus GmbH & Co. KG (Warendorf, Germany). Chicken and turkey feathers were self-collected from poultry farming, and dove feathers were delivered from a breeder. Hair samples from the mammalian (16) IDFB Handbook, Determination of Feather and Down Species; proposed IDFB Method 1, 1999. (17) Kellner, R.; Houthaeve, T. In Microcharacterization of Proteins, 2nd ed.; Kellner, R., Lottspeich, F., Meyer, H. E., Eds.; Wiley-VCH Verlag: Weinheim, 1999; pp 97-116.
Analytical Chemistry, Vol. 74, No. 23, December 1, 2002
5961
species llama, camel, cashmere goat, mohair goat, sheep, yak, and human, partly chemically tinted or bleached, were obtained from the group of Prof. Dr. F. J. Wortmann, Deutsches Wollforschungsinstitut an der RWTH Aachen e.V., Germany. Rabbit, merino sheep, mink, and fox samples were obtained from Pelzatelier Hildegard Pick, Dreieich, Germany. Dog hair samples were self-collected. Sample Preparation. Feathers and down from certified and unknown samples were weighed individually. Each feather and down was taken by tweezers; extraneous fibers were mechanically removed. A single sample was composed of up to 100 individual feathers and down. About 0.1-1 mg of yak, mohair, camel, sheep, and cashmere hair standards were weighed separately, and for the mixed binary samples, they were put together in various weight percentages. The weighed feathers, down, and hair were transferred into eight-well PCR strips (Biozym Diagnostik GmbH, Hessisch Oldendorf, Germany) containing 50 µL of 25 mmol/L NH4HCO3 with 5% 2-mercaptoethanol (v/v) and carefully wetted. Caps were sealed, and after transferring into a boiling water bath for 20 min, the samples were cooled on ice and 50 µL of 25 mmol/L NH4HCO3, containing 5 mg of trypsin/mL was added to each tube. Incubation took place for 2 h in a water bath at 37° C. A 5-µL aliquot of each sample was transferred into a fresh tube, containing 45 µL of saturated CCA solution in 50% acetonitrile, 1% TFA, and mixed by pipetting. One microliter of each sample was manually pipetted on a polished steel MALDI-TOF target plate and allowed to dry (Figure 1). Mass Spectrometry. Analysis was performed on a Bruker Reflex III time-of-flight mass spectrometer (Bruker-Daltonic, Bremen, Germany) equipped with the SCOUT 384 probe ion source. The system uses a pulsed nitrogen laser (337 nm, model VSL-337ND, Laser Science Inc., Boston, MA) with energy of 400 µJ/pulse. The ions were accelerated under delayed extraction conditions in the positive mode with an acceleration voltage of 20 kV and a reflector voltage of 22.5 kV. A 6.9-kV potential difference between target and the extraction lens was applied with a time delay of 1000 ns. A Lecroy 9384C, 1-GHz digital storage oscilloscope was used for data acquisition (Lecroy Corp., Chestnut Ridge, NY). The detector signals were amplified in two stages, digitized, and transferred to the XACQ program on a SUN work station (Sun Microsystems Inc. Palo Alto, CA). Autoexecute mode was chosen for automated measurement with 68-62% attenuation of the laser intensity, a resolution better than 1400, a signal-tonoise ratio (S/N) better than 4, and a noise range of 100. Spectra with one or more peaks, larger than the measurement range, were not used. For one stored spectrum, 200 successful laser shots arising from more than three spots were summed. Saved m/zintensity data were processed with the program XMASS 5.1 (Bruker Daltonics, Bremen, Germany) using the SNAP algorithm for exclusive detection of monoisotopic masses. The 99 most intensive peaks in the range of 1000-3400 Da were collected. Relative intensities and peak areas were calculated, and the data were stored as report files. Relative intensities are always related to the maximum peak intensity of the measured spectral range. Identification of Useful Mass Spectral Peaks for the Characterization of Avian and Mammalian Samples. The procedure of data treatment is depicted in Figure 2. From each report file containing one mass spectrum, collected from 200 laser 5962 Analytical Chemistry, Vol. 74, No. 23, December 1, 2002
Figure 1. Flow diagram of sample preparation for MALDI-TOF mass spectrometry.
shots, significant peaks with intensities of >2% of the maximum peak were selected and the accepted monoisotopic peak masses were rounded to integer values. For each determined species, the frequency of occurrence of peaks was determined by inspection of 8-83 spectra. The more spectra of a species taken into account, the better was the reliability of statistical predictions of occurring peaks. Eight mass spectra per species were the minimum number required, while 16 spectra per species were a useful compromise between accuracy and simplicity of analysis. From the collected spectra, only those peaks were further used that occurred at least twice. Peaks originating from trypsin were identified from blank samples containing only trypsin. These peaks were discarded as well. This led to the absolute and percentage frequencies of avian and mammalian peak occurrences for each species. In a next step, nonspecific peaks, occurring in all investigated avian species and in more than 80% of all mammalian species, respectively, were removed. We called the remaining peaks semispecific peaks (SEMPs). A part of these peaks occurred in other species as well, but they were significantly more frequently detected with one species than with others. These peaks were the more frequent peaks (MFPs). Another part was unique for one species in a set of the investigated species and was therefore called the unique species-specific peaks (USSPs). USSPs were exclusively observed in one species but were not always detectable. Both subgroups
of peaks higher than 50% was necessary. The closer the relationship is, the more common peaks were expected. (4) Quantification of Mixed Feather and Down Samples in Avian Binary Systems. The quantification of duck and goose feather and down content of an unknown mixture was achieved by first identifying each single weighed feather or down using a minimum of six USSPs, occurring in more than 50% of all investigated spectra and, second, summing up the masses of the weighed and identified feathers or down. The final results were mass weight fractions of both species. (5) Quantification of Mixed Hair Samples in Binary Mammalian Species. The binary sample pairs, yak wool contamination in cashmere wool, yak wool contamination in mohair wool, and sheep wool contamination in camel wool, were analyzed using pure samples of each species and mixed samples of known composition as standards. Each sample was treated as described and 10 spots for every sample were put onto the target plate. For the identification of the digest peaks suitable for the quantification, peaks with a frequency of occurrence of more than 80% were selected. It was not essential that these peaks were USSPs; the chosen peaks had to be unique in this present binary test system only. For quantification over a large range of ratios in mixtures, the intensities of the selected mass peaks in a 1:1 mixture should be about equal but not less than 0.3 or larger than 3. The reason for this is the limited intensity resolution caused by the 8-bit analog-digital converter of the MALDI-ToF detector. From 10 measurements, those with the largest and the smallest intensity ratios were discarded. The average of the residual eight measurements was used for quantification.
Figure 2. Flow diagram for processing raw data from report files for avian and mammalian species. The semispecific peaks (SEMPs) were divided into two subgroups, the unique species-specific (USSPs), occurring exclusively in one species, and the more frequent peaks (MFPs), occurring in more than one species, both essential for identification and distinguishing and quantification of species.
of the SEMPs were used for solving the following analytical problems as also summarized in Table 1. (1) Identification of Native Species. For the identification of a native species, only those USSPs were taken into account that occurred in more than 50% of all investigated spectra of the species. These remaining peaks were compared to the peaks of all other avian or mammalian species, respectively. For each species, a set of USSP masses was established. (2) Recognition of Tinted or Bleached Species. For the identification of a tinted or bleached species, it was necessary to identify the native species first and to estimate the SEMPs. In a second step, remaining SEMPs, not derivatized by tinting or bleaching, were searched for. The required frequency of occurrence for these peaks was more than 50% of the investigated spectra of the derivatized species. The final result was the identification of tinted or bleaches species. (3) Determination of the Relationship of Two Species. Two avian or mammalian species, respectively, were compared by estimating the fraction of coincident SEMPs, including both the MFPs and the USSPs. For both species, a frequency of occurrence
RESULTS AND DISCUSSION The use of the high concentration of 5% (v/v) 2-mercaptoethanol was essential for the effectiveness of the following enzymatic digest. Only poor digest was achieved if 2-mercaptoethanol was used with a concentration of e1% (v/v). The high trypsin/substrate ratio ranged from 50:1 (w/w) for feathers and down to 2.5:1-0.25:1 (w/w) for hair samples. Standard trypsin protocols for cleaving proteins in polyacrylamide gels or bound to a membrane requires a maximum ratio of 1:1 (w/w) to get meaningful mass spectra. Although a high concentration of trypsin was used, no visible decay or complete digestion of even small down or hair samples was observed in the digest wells. Therefore, we can assume that only the outer protein layers of the samples were digested. As expected, the high concentration of trypsin produced many autoproteolytic digest fragments in sample free negative standards. These fragments originated from trypsin itself and from possible impurities contained in the trypsin preparation. Nevertheless, these peptide fragments did not disturb the identification or quantification of individual species. The mass range of peptide detection reaching from 1000 to 3400 Da was suitable for identifying all investigated avian and mammalian species. Several matrixes were tested, and CCA acid gave best spectra. It also gave the most homogeneous sample spots, which were necessary for automated MALDI-TOF runs. Identification and Comparison of Mammalian and Avian Species. Samples of both classes of animals, avian and mammalian, showed many identical digest fragments each. We assume this is because of the genetic relationship of feathers, down, and hair, respectively. After peak processing, the number of remaining Analytical Chemistry, Vol. 74, No. 23, December 1, 2002
5963
Table 1. Use of the MFPs and the USSPs for Identification of Species and Quantitative Determination of Mass Fractions in Binary Avian and Mammalian Systems, Respectively purpose identification of a native species recognition of a tinted or bleached species determination of relationship of two species quantitative determination of mass fractions in binary avian systems quantitative determination of mass fractions in binary mammalian systems
parameters of differentiation
FoOa
more frequent peaks (MFPs)
differences to all other tested species remaining semispecific native peaks
>50
not used
essential
>50
essential
essential
fraction of coincident peaks
>50
essential
essential
minimum six speciesspecific peaks
>50
not used
essential
two selected peaks being unique only for the tested system
>80
sufficient
not essential
(%)
species-specific peaks (USSPs)
a Frequency of occurrence of mass spectral peaks. All peaks, further used, had to be detected in more than 50% or more than 80% of the individual measured spectra.
SEMPs varied from species to species. In most cases, a sufficient number of MFPs and USSPs, occurring in more than 50% of the investigated spectra, could be detected (Table 2). For human hair, only 3 USSPs and 1 MFP could be detected, while for the fur of red fox, 54 USSPs and 39 MFPs could be established. On average, 9 of 11 remaining mammalian species showed 7 USSPs and 14 MFPs and correspondingly 21 SEMPs. For the estimation of identity of a native species, it was essential to use only the USSPs with a frequency of appearance of more than 50% of the investigated spectra. These USSPs were also the basis for the quantification of avian binary systems, described later. For comparison of two different species and for the recognition of tinted or bleached species, the additional use of the MFPs was essential. For quantification purposes of avian species, the occurrence of the USSPs had to be higher than 50% of the investigated spectra. For the quantification of mammals, two selected SEMPs occurring in more than 80% of the investigated spectra of that species were sufficient. We did not find significant differences in the spectra of feathers, down, and hair of juvenile and adult animals. Plowman et al.15 noted that the quality of wool is a function not only of species but also of nutrition, age, and environmental conditions and could discover differences in 2Dgel electrophoresis and MALDI-TOF studies. With our new method for species identification, we did not detect any significant differences of mass spectra related to nutrition, age, or environmental conditions. Figure 3 shows the fraction (% match) of common SEMPs of the investigated mammalian pairs. Very closely related animals had a high degree of peak conformity: Cashmere goat and mohair goat had 57.1% common SEMPs. Sheep and mohair goat were identical in 43.8% of the SEMPs and sheep, compared with cashmere goat, were also identical in 43.8% of the investigated peaks. Llama and camel, two species closely related too, had peak conformity of 46.7%. The comparison of several species of herbivores resulted in a different degree of peak identity reaching from 18.8 to 22.8% for cloven hoofed animals. Comparison of the cloven-hoofed sheep with rabbits, vegetarian rodents, showed only a small degree of peak identity of 6.3%. Carnivores among themselves also had peak conformity of ∼22%, whereas the 5964
Analytical Chemistry, Vol. 74, No. 23, December 1, 2002
comparison of herbivores with carnivores, including human, varied from 0 to 4.9% peak identity. For the avian species, similar results were obtained (Table 3). Waterfowl such as goose, duck, and eider had degrees of mass conformities reaching from 17.9 to 29.4%, the latter for the comparison of Chinese duck with Pyrenees duck, two biological races. In contrast, land fowls, free of down, showed a maximum degree of peak identity of 13.7% for the comparison of chicken with turkey. All other comparisons of land fowls among each other or against waterfowls resulted in smaller degrees ranging from 5 to 11.1%. Like the mammalian species, the avian species also showed higher conformity of SEMPs for closer zoological relationship. A clear tendency of relationship and coincident peaks is visible: The closer the zoological relationship, the more common peaks could be detected. The new method confirms that at least some proteins derived from keratin structures vary in their amino acid sequences depending on the genetic relationship. The fast method described here is well suited for identification of a species from a possible set. The uniqueness of the USSPs was only determined for the tested species and should not be seen as absolute values. It is possible that other birds or mammals, not investigated here, show some of the USSPs of the species studied here. The probability of finding coincident digest peaks grows with the number of investigated species. To overcome this problem and for the creation of a digest mass library it could be necessary to calculate the match of identity with clusters of SEMPs. This will be useful for the simultaneous differentiation of a large number of species. However, in practical cases of industrial quality control, there are only a few species of commercial interest and only a few species for possible adulteration. Influence of Chemical Treatment of Hair in Identifying Mammalian Species. Coloring or bleaching hair drastically reduced the number of detected SEMP identities within one species. Some SEMPs were lost and some new peaks appeared. This was expected because the chemical derivatizations of amino acid side chains dramatically changed the properties of the keratin structures and the masses of digest peptides. Tinted rabbit fur only showed 8 remaining SEMPs, from 20 original SEMPs, of the
Table 2. List of SEMPs of 11 Mammalsa
a For each determined species the list contains rounded monoisotopic m/z values, the number of investigated spectra, and the sum of the SEMPs, consisting of MFPs and the USSPs. The gray background marks the USSPs. Boldface numbers in boldface, framed boxes show the SEMPs used for quantification.
native fur (Table 4). In comparison of tinted mink fur with native mink fur, only 5 peaks of 15 were left, while in bleached human
hair 3 of original 4 SEMPs were conserved. The remaining peaks, however, were sufficient for species recognition in all three pairs Analytical Chemistry, Vol. 74, No. 23, December 1, 2002
5965
Table 4. Comparison of Semispecific Peaks (SEMPs) of Native and Tinted Rabbit Hair from Eight Individual Spectraa native rabbit
Figure 3. Match of coincident semispecific peaks (SEMPs) of compared pairs of mammals (in percent). Only native hair and fur samples were used. Closely related animals show a significantly higher match than distently related species. Above 57% of the SEMPs were common in two goat species. Sheep and goat showed ∼44% conformity similar to camel and llama with ∼47%. Table 3. Percentage of Common SEMPs of Feathers and down of Several Avian Pairs species
goose
Pyrenee duck Chinese duck eider turkey chicken dove
21.7 22.7 17.9 6.5 6.7 6.5
tinted rabbit
FOb
mass
pFOc
FOb
mass
pFOc
5 8 8 7 7 6 8 5 5 5 8 5 7 7 8 5 6 8 8 5
1097 1194 1437 1453 1480 1492 1602 1736 1816 1831 1953 1961 1975 2059 2085 2144 2245 2321 2599 2722
62.5 100 100 87.5 87.5 75 100 62.5 62.5 62.5 100 62.5 87.5 87.5 100 62.5 75 100 100 62.5
8 8 6 5 5 7 7 6 7 5 7 7 8 5
1194 1216 1370 1436 1602 1953 1962 1975 2059 2135 2245 2321 2599 2657
100 100 75 62.5 62.5 87.5 87.5 75 87.5 62.5 87.5 87.5 100 62.5
a Boldface numbers indicate SEMPs occurring in both native and tinted rabbit hair. b Absolute number peak occurrence. c Frequency of peak occurrence in percent.
Pyrenee Chinese duck duck eider turkey chicken 29.4 26.1 7.3 10.7 5
26.7 5 11.1 5.5
6.5 9.1 8.7
13.7 10.9
6
of comparison. This makes the method described suitable for the identification of tinted or bleached hair and fur, which may be of interest in the field of hair fashion, the identification of adulteration, and for forensic medicine, respectively. Quantitative Determination of Mixtures Containing Duck and Goose Feathers and Down. The quantification of mixtures of feathers and down is presently done by sampling ∼100 single feathers or down, weighing and identifying each of them.16 In the presently accepted standard method, identification is made visually using microscopy. In our method, we exactly follow the same procedure, but identification is made by MALDI-TOF measurement. Calibration is only required for the identification that is carried out using feathers and down of known origin. We found that six USSPs were sufficient for proper identification. Figure 4 shows parts of two spectra with four selected USSPs from goose and two from duck being used for differentiation. In total, the following USSPs were detected for goose: 1237, 1293, 1313, 1828, 1884, 1918, and 1994 Da. For duck, we found the following USSPs: 1527, 1543, 1727, and 1743 Da. However, not all of the possible USSPs could be detected in all samples, but it was found that the numbers of detected USSPs were sufficient for identification in all cases. Both, the existence and nonexistence of peaks were used for identification, enhancing the fidelity of identification. The quantification of binary avian systems was simply performed by adding up the weighed gravimetric masses of the identified 5966 Analytical Chemistry, Vol. 74, No. 23, December 1, 2002
Figure 4. Two tryptic digest mass spectra from 1700 to 2100 Da of feathers from goose (A) and duck (B). The tagged (/) mass peaks 1828, 1884, 1918, and 1984 were USSPs for goose while the mass peaks 1727 and 1743 were USSPs for duck.
feathers and down. The agreement with hand-sorted reference samples was better than 98%. The new method is therefore straightforwardly applicable for routine analyses. Quantitative Determination of Mixed Binary Sample Pairs of Hair. Suitable USSPs for the quantification of the three mammalian test pairs were estimated as 1432 Da for yak and 1622 Da for cashmere goat (Figure 5), 1085 Da for mohair goat, 1109 Da for camel, and 1226 Da for sheep. All of these peptide masses had larger relative intensities than 0.3 for the equally weighed mixed samples. The quotient of the relative intensity or area of the USSP of one species to the sum of the relative intensities of both tested species was compared to the quotient of the weight of that species to the total weight of both, respectively. The tested mammalian pairs showed a linear relationship between relative intensity ratios and weight ratios. Instead of the ratios of relative
Table 5. Statistical Parameters of Linear Regression Analyses of Quantitative Determinations in Binary, Mammalian Sample Pairsa
sample pairs species 1 yak yak sheep
selected peaks (integer m/z values)
species 2
species 1
species 2
cashmere goat mohair goat camel
1432 1432 1109
1622 1085 1226
max SD (n ) 8)b of individual samples intrelc areareld
R2 of linear regression intrelc areareld
slope (a) intrelc areareld
0.0342 0.0242 0.0234
0.9943 0.9904 0.9913
0.8157 1.0983 0.9446
0.0258 0.0152 0.0176
0.9913 0.9984 0.9906
0.8664 1.1393 0.9607
intercept (b) intrelc areareld -0.0259 -0.1417 -0.0248
-0.0125 -0.0841 0.0087
a The calculations showed the linear relationship of the quotients of weighed mass fractions of each sample pair and the quotients of detected relative peak intensities or detected peak areas of the SEMPs, respectively: (rel intensitypeak species1)/[(rel intensitypeak species1) + (rel intensitypeak species2)] ) a(weightspecies1)/[(weightspecies1) + (weightspecies2)] + b and (areapeak species1)/[(areapeak species1) + (areapeak species2)] ) a(weightspecies1)/ [(weightspecies1) + (weightspecies 2)] + b. b Maximum standard deviation of 8 of 10 parallel measurements for each mass fraction of each value pair. The two values deviating most from the mean were removed. c Calculation based on relative peak intensities. d Calculations based on peak areas.
Figure 5. Two tryptic digest mass spectra from 1400 to 1700 Da of hair from yak (A) and cashmere goat (B). The tagged (/) SEMPs 1432 and 1622 respectively were used for quantification of mixed binary samples.
peak intensities, the ratios of peak area could also be used, yielding almost identical results (Table 5). Correlation coefficients better than 0.99 were calculated for all of the three tested pairs. Standard deviations smaller than 4% for each of the eight parallel measurements for each measured mass fraction were calculated. It was possible to determine unknown samples with 98% accuracy. The detection limits of contaminants varied from 1 to 14.2% (w/w). Quantitative estimations of contaminants in binary mammalian systems could be measured with the necessary accuracy for all of the three compared pairs. In practical cases of commercial samples, a possible adulteration can be expected with the addition of ∼10% up to 40% of foreign hair to be of economic interest. In this expected range, our method can accurately quantify these foreign feathers, down, and hair. The only limiting conditions for the application of the new method are the knowledge of the contaminating species. Ternary systems of contaminations are not tested yet but can most likely be analyzed analogously. Further investigations showed that two almost arbitrarily chosen individual peaks fulfilling the conditions described above are sufficient for quantitative estimation of mixed mammalian samples. The only differences using other suited pairs of peaks are the different slopes and intercepts of the calculated linear regression straight lines. Usually the slope was ∼1.0 and the intercept
was close to 0, of two peaks showing very similar relative intensities or areas, respectively, in sample pairs with equal mass fractions. In our method, the selected SEMPs of one species serve as a kind of internal standard for the other species. This is a surprising fact for MALDI-TOF mass spectrometry. During the past few years, MALDI-TOF mass spectrometry was increasingly used for quantification particularly of low molecular weight molecules,18,20 but in all these cases, internal standards, preferably isotopomers, were required to get reasonable accuracy. In our newly developed method for the quantitative determination of hair in binary samples, we do not need isotopically labeled internal standards. These findings suggest that similar quantification procedures might be used for quantification of other proteins and peptides. Our new method has proven its applicability not only for qualitative identification purposes but also for quantitative use in discrimination of binary mixed samples of avian and mammalian origin. The high data accuracy and the high speed of data acquisition make this method interesting for routine analyses in a wide field of applications. Quality assurance of raw material as well as from ready-to-use products is of great interest in industrial quality management systems. The feather-processing industries producing bed linen as well as down jackets, etc., and the textile industries producing high-priced canvas such as cashmere and mohair textiles, for example, have to use quantitative methods to guarantee the product identity. This is of fundamental interest, but up to now, difficult to perform, due to the physical and biochemical properties of the original materials. Our newly developed method can help to satisfy the requested demands. Another field of application may be the possibility to recognize furs from protected animals and to help animal welfare in this way. For forensic medicine there also may be some applications to identify the origin of native and chemically processed feathers, down, and hair. Because hair and feathers are made from materials very similar to horn or scales, it should also be possible to identify animal species on the basis of these materials. Easily performed sample preparation and digest as well as simple transfer to MALDI targets make this method straightfor(18) Kang, M.-J.; Tholey, A.; Heinzle, E. Rapid Commun. Mass Spectrom. 2001, 15, 1327-1333. (19) Wittmann, C.; Heinzle, E. Biotechnol. Bioeng. 2001, 72, 642-647. (20) Kang, M.-J.; Tholey, A.; Heinzle, E. Rapid Commun. Mass Spectrom. 2000, 14, 1972-1978.
Analytical Chemistry, Vol. 74, No. 23, December 1, 2002
5967
wardly applicable. The short measuring time in the automated mode of ∼80 s/sample and the application of 384 samples/target plate upgradeable to 1536 samples/target make this method ideal for measuring large numbers of samples. ACKNOWLEDGMENT We thank Forschungsinstitut Hohenstein, Bo¨nnigheim, Germany, and Fa. Brinkhaus GmbH & Co. KG, Warendorf, Germany,
5968
Analytical Chemistry, Vol. 74, No. 23, December 1, 2002
for the certified duck, eider, and goose samples and the group of Prof. Dr. F. J. Wortmann and Pelzatelier Hildegard Pick, Dreieich, Germany, for the mammalian hair samples.
Received for review May 24, 2002. Accepted September 11, 2002. AC020347F
