Analysis of Coloring and Binding Components _AnalBioanalChem 2005

8/4/2019 Analysis of Coloring and Binding Components _AnalBioanalChem 2005

1/8

S P E C I A L I S S U E P A P E R

S. Kuckova I. Nemec R. HynekJ. Hradilova T. Grygar

Analysis of organic colouring and binding componentsin colour layer of art works

Received: 24 September 2004 / Revised: 23 December 2004 / Accepted: 16 January 2005 / Published online: 31 March 2005

Springer-Verlag 2005

Abstract Two methods of analysis of organic compo-nents of colour layers of art works have been tested: IRmicrospectroscopy of indigo, Cu-phthalocyanine, andPrussian blue, and MALDI-TOF-MS of proteinaceous

binders and a protein-containing red dye. The IR spectradistortion common for smooth outer surfaces and pol-ished cross sections of colour layer of art works is sup-pressed by reflectance measurement of microtome slices.The detection limit of the three blue pigments examinedis $0.3 wt% in reference colour layers in linseed oilbinder with calcite as extender and lead white as a dryingagent. The sensitivity has been sufficient to identifyPrussian blue in repaints on a Gothic painting. MALDI-TOF-MS has been used to identify proteinaceous bind-ers in two historical paintings, namely isinglass (fishglue) and rabbit glue. MALDI-TOF-MS has also beenproposed for identification of an insect red dye, cochi-

neal carmine, according to its specific protein compo-nent. The enzymatic cleavage with trypsin beforeMALDI-TOF-MS seems to be a very gentle and specificway of dissolution of the colour layers highly polymer-ised due to very long aging of old, e.g. medieval, samples.

Keywords Proteins Binders Blue dyes Painting MALDI

Introduction

There are several classes of organic components in col-

our layers of art works: organic pigments and dyestuffs,proteinaceous, oil, and polysaccharide binders, andnatural and synthetic resins. The analytical researchtechniques that are currently most common in identifi-cation of those organic compounds are gas and pyrolysisgas chromatography (GC-FID, GC-MS, or Py-GC-MS), high-performance liquid chromatography(HPLC), and infrared spectroscopy (FTIR). Due tolimitations in data acquisition and interpretation, thesemethods have mostly been tested in analytical researchlaboratories, but have not yet been really implementedin laboratories associated with galleries and museums inthe Czech Republic. The routinely used analyticalmethods, like optical microscopy and scanning electronmicroscopy with electron dispersive X-ray detectors(SEM-EDX), are extremely valuable for detection ofinorganic components but are almost worthless for or-ganic compounds and non-mineral blue pigments. Forexample, Prussian blue in real samples cannot beunequivocally identified according to Fe content becausethis element is quite common in almost all earthy pig-ments. Additionally, Prussian blue could be present invery low concentrations, because it is very dark in thepure form. Also copper phthalocyanine is so dark bluethat it must be substantially diluted by a white extenderand so the final concentration of Cu in a blue colourlayer is rather small; additionally Cu is common inseveral mineral blues.

Identification of blue pigments like indigo, Prussianblue, and copper phthalocyanine is required by restor-ers, because these pigments are relevant in approximatedating of art works. A combination of less commonmethods like laser induced breakdown spectroscopy andhyper-spectral imaging analysis with diffuse reflectancespectra was used to decide upon the presence of Prussianblue in an illuminated manuscript [1]. Raman micros-copy was able to detect Prussian blue and indigo in

S. Kuckova (&) I. NemecDepartment of Analytical Chemistry,Charles University, 12840 Prague 2, Czech Republic

E-mail: [email protected]. Kuckova R. HynekFaculty of Food and Biochemical Technology,Institute of Chemical Technology, Technicka 3,16628 Prague 6, Czech Republic

S. Kuckova J. HradilovaAcademy of Fine Arts in Prague,U Akademie 4, 17022 Prague 7, Czech Republic

S. Kuckova T. GrygarInstitute of Inorganic Chemistry,AS CR, 25068 Rez, Czech Republic

Anal Bioanal Chem (2005) 382: 275282DOI 10.1007/s00216-005-3108-5


2/8

colour layers of paintings [2] and iluminations [3], butRaman spectroscopy is not applicable in all samples, e.g.because of fluorescence of some organic compounds [4].

Identification of red organic pigments and dyes inheterogeneous matrices of real samples of art works isanother example of still unsolved analytical tasks. Var-ious single-purpose analytical techniques have beenproposed in individual cases, most of them requiringdissolution of target compounds. Extracts of the dyescan be analysed by reversed liquid chromatography orelectrospray mass spectrometry (ESI MSD) with higherselectivity and sensitivity than RP-HPLC [5]. Thedetection limit for ESI MSD is shown in Table 1.Standards of carminic acid, emodin, purpurin, alizarin,and laccaic acid, typical natural red dyes based onsubstituted 9,10-anthraquinone skeleton, were studiedby electrospray MS coupled to capillary electrophoresis[6]. Recently introduced matrix-assisted laser desorption(MALDI) and electrospray ionisation mass spectrome-try, both in the negative ion modes, were able to confirmthe presence of carminic acid in liquid standard mixtureswith linseed oil [7]. A drawback of the separation

methods is laborious sample pre-treatment with a risk oflosses in separation steps. A liquid sample is obtained bydissolution of a real sample in hydrochloric or sulfuricacid followed by extraction but some of the red dyes arenot stable in acidic condition [8, 9]. On the other hand,the methods of analysis of solid samples like infraredspectroscopy combined with visible spectroscopy canhelp: both methods indirectly indicated the presence ofan insect-derived anthraquinone dye in red paint of amedieval Byzantine manuscript [10] because the dye wasassociated with proteins. A much more specific waywould, however, be a direct specific identification ofinsect-derived proteins by MALDI-TOF-MS.

Organic binders are a minor part of the colour layersof easel paintings according to their total percentage, buttheir identification is of huge importance for descriptionof the painting technique and hence the binders actuallybecome the most urgently studied component of colourlayers. The analysis of oil binders [1114] and beeswax

and pine resins [15] is relatively well developed . Incontrast, the analysis of proteins has not been solved.Chromatographic methods (HPLC, GC, GC-MS, Py-GC-MS) used for identification of proteinaceous com-ponents are based on the characteristic ratios of certainamino acids [1622]. Unfortunately, polysaccharidicgums, widely used as binders and adhesives in art works,can also contain proteinaceous matter, e.g. 3% in arabicgum. Thus, a binder identification based only on thepresence of amino acids could not be correct. Furtherdisadvantages are losses of amino acids during hydro-lysis (Maillard reaction) or natural chemical reaction(oxidation, crosslinking, condensation, and dehydra-tion) [16]. In HPLC with UV or fluorescence detection,GC-FID, and GC-MS, the presence of metal ions likeCa, Cu, and Fe (from very common inorganic pigments)interferes [17]. Py-GC-MS could identify only threekinds of proteinaceous binders: animal glue, casein, andegg [16, 1820]. Bone glue, rabbit glue, skin glue as wellas isinglass (fish glue) are obviously very similar in termsof total amino acid composition or the products of theirdestruction by pyrolysis. Parallel analysis GC of the

proteinaceous (casein, egg white, albumin, pork and beefgelatine) and lipoid binders (poppy, sunflower, linseedoil) was successfully examined [22], but the method in-volved a laborious and risky pre-treatment of sampleincluding extraction separation.

A new concept in the analysis of protein bindingmedia is the use of enzymes. Meanwhile, currently en-zymes are used only for gentle but efficient cleaning inrestoring [23]. For example several different serine pro-teases (alcalase, a-chymotrypsin, elastase, esperase, sa-vinase, subtilisin A, trypsin) in immobilised form weretested for selective removal of damaged casein layers onthe surface of fragile mural paintings [24]. A great

advantage of enzymes is their specificity, i.e. they cleavethe proteins exclusively behind a definite amino acid. Afingerprint mixture of peptides is formed from a proteinthat is much more individual than amino acid ratios.The resulting mixture of peptides can then be analysedby MALDI-TOF-MS [25]. The detection is extremely

Table 1 Summary of destructive methods used for identification of paintings materials

Method Detected compounds Number of steps Amount of sample Detection limit

GC-FID AA 7 0.1 mg [17] 10100 pg [16](for derivatised orpyrolysed AA)

GC (MTBSTFA) AA 3 50100 lg [20]Py-GC-MS AA 0 0.5 mg [19], few lg [20]GC-FID FA, AA 6 0.5 mg [22]Direct chemolysis-GC-MS FA 1 0.20.5 mg [14] GC-MS Birch bark tar,

beeswax, resins2 13 mg [15]

HPLC-UV detection AA At least 2 1 ng [16]HPLC-fluorescence

detectionAA At least 2 10300 pg [16]

HPLC ESI MSD Anthraquiones 2 [4] 0.6 mg 3090 ng/ml [4]MALDI-TOF-MS Peptides 1 0.5 lg [24] 18 femtomoles [25]

FID flame ionisation detector, MTBSTFA derivatisation with N-tert-butyldimethyl-silyl-N-methyltrifluoroacetamide, Py pyrolysis, ESIMSD electrospray ionisation-mass selective detection, AA amino acids, FA fatty acids

276


3/8

sensitive: the amount as low as 18 femtomoles ofb-casein was found in positive ion MALDI-TOF-MS[26]. For comparison the detection limits and amount ofsample for analytical methods is shown in Table 1.

This work has two main aims. The first is to proposeand verify a technique of IR spectra acquisition free ofsevere deformations, e.g. Christiansens effects, andsuitable for solid samples of colour layer of art works.FTIR seems to remain one of the most perspective andflexible methods for non-mineral blues and determina-tion of binder types, and only the data acquisitionmethod must be optimised. A weak point of infraredspectroscopy is deformation of spectra due to distortionespecially severe in analysis of polished cross-sections.Identification of three non-mineral blues was chosen asan example. The reason was that data acquisition andvalidations using reasonable reference samples are not arule in artwork analysis. The second aim is to check theapplicability of enzymes to pre-treatment of colour layersamples and further develop the recently proposed [25]MALDI-TOF-MS identification of proteinaceous bind-ers of colour layers of art works.

Experimental

Materials

Reference colour layers were prepared from Prussianblue, indigo (natural indigo, product number 36000,Georg Kremer, Farbmu hle, Aichstetten/Allga u, Ger-many), or copper phthalocyanine (Sigma-Aldrich, MO,USA). Prussian blue was synthesised according to atraditional recipe by Diesbach from 1731 (taken from

[27]): by precipitation of mixed aqueous solution offerrous sulphate hydrate and potassium aluminum sul-phate hydrate (alum) with alkaline (K2CO3) solution ofpotassium cyanide followed by air oxidation on stirringfor 6 h. The phase composition of Prussian blue waschecked by powder XRD. The pigments (0.01, 0.03, 0.1,0.3, 1, 3, and 10%) were mixed with chalk (calcite,CaCO3), lead white (cerussite, PbCO3), and linseed oil(Umton, Decin, Czech Republic), the mixture wasspread over a wooden plate with white ground layer(CaCO3 with gelatin binder) and left to dry at ambientlaboratory conditions for six months.

The proteins associated with red dyes were studied by

comparing commercial red dye (Carmine, productnumber 42100, Georg Kremer, Farbmu hle, Aichstetten/Allga u, Germany) and dried specimen of Coccus cacti(product number 36040, the same supplier).

A real sample of two blue colour layers was obtainedfrom oil repaints of Our Lady the Protectress (NationalGallery in Prague, tempera on a wooden pane, originaldated before 1500 AD). The pre-restoration research wasperformed in the Academy of Fine Arts in Prague in 2001.

The protein binders were identified in two samples,one of a painter glue putty (probably nineteenth cen-

tury) found on the medieval painting Saint John theEvangelist (National Gallery in Prague), and one in asample of medieval ground of crosier of Abbess Isidoradated before year 1740 (Collections of the PragueCastle). The crosier was covered with mould because ithas laid in the wet tomb of Abbess Isidora. The pre-restoration research was done in the Academy of FineArts in Prague in 2004. For consolidation of the paintinglayer, isinglass was used by a restorer. The resultingmass spectra were compared with our own MS database[25] containing the spectra of rabbit glue, isinglass, boneglue, gelatin, hide glue, whole egg, egg yolk, and bovinecasein.

Sample preparation

All samples of painting layers were available as fewmicrograms fragments. For MALDI-TOF the fragmentswere analysed as received.

For infrared spectroscopy the reference samples werefirst measured directly in their smooth surface of the

blue layer and then the reference samples were embed-ded in polyester cast resin (Ku nstler.Farben.Fabrik C.Kreul, Hallerndorf, Germany) and polished into cross-sections. Then it was sliced into thicknesses of 2 lm byultramicrotome Pyramitom with a glass knife. The sliceswere transferred to non-absorbing silicon wafers,straightened in warm water, and finally subjected toreflectance microspectroscopy. In real samples, thepresence of blue layers was first confirmed by conven-tional light microscopy, and then SEM/EDX was usedto exclude samples containing mineral blues. The se-lected samples were embedded and polished, and thecross sections were examined by IR spectroscopy.

Light microscopy

All real samples were examined using microscopeOlympus BX 60 (magnification 100 times) to learn theirstratigraphy.

Infrared spectroscopy with Fourier transformation

IR spectra were recorded in reflection mode with aninfrared microscope Continuum with Nexus spectrom-

eter from ThermoNicolet (USA). Spectra were analysedusing Omnic Version 6. Spectra were recorded in theregion 4000650 cm1 with resolution 4 cm 1 .

Enzymatic cleavage for MALDI-TOF-MS

The enzymatic cleavage was performed with sequencinggrade modified trypsin purchased from PromegaCorporation, WI, USA. The cleavage of a few micro-grams of samples was performed by adding 6 ll of

277


4/8

trypsin solution in 50 mmol/l ammonium hydrogencarbonate containing approximately 10 lg/ml of trypsindropped directly on the sample in sealed vials (test tubes100 ll, Eppendorf AG). The cleavage was carried out atroom temperature for 2 h.

MALDI-TOF-MS

All mass spectra were acquired on a Bruker-DaltonicsBiflex IV MALDI-TOF mass spectrometer equippedwith standard nitrogen laser (337 nm) in reflector mode.Then 1.5 ll of sample was mixed with 5 ll of 2,5-di-hydroxybenzoic (DHB) acid solution (6.7 mg of DHB inacetonitrile/0.1% trifluoracetic acid (1/2; v/v)); 1.5 ll ofthe resulting mixture was spotted on the stainless steelMALDI target and dried [25]. At least 200 laser shotswere collected for each spectrum and analysed usingXMASS software (Bruker). Each experiment startedwith recalibration of the mass spectrometer with acommercial standard of peptide mixture (Pepmix,Bruker).

Results and discussion

Infrared spectroscopy

One of the most straightforward non-destructive ana-lytical techniques for identification of basic types oforganic components of colour layers is infrared spec-troscopy. Its disadvantage may be very poor quality ofthe spectra caused by scattering and reflection losseson crystals with high refractive indices relative to thematrix and the particles in the size being much greater

than the wavelength of the incident radiation (Chris-tiansens effects) [28]. Another complications met in IRspectra of solid samples are specular reflections andpeak inversions. The method of spectra acquisition ishence a critical point of the method. The IR spec-troscopy was tested using reference samples of threeblue pigments in a matrix approaching traditional oilpainting techniques. Fragments of the reference sam-ples were measured in two ways: on their smoothouter surface, and on a much more rough surface oftheir microtome slices (Fig. 1). The resulting detectionlimits are summarised in Table 2. There is no overlapbetween IR bands of individual blues listed in Table 2

and between those blues and the oil matrix of thereference samples (Table 3). In the case of Prussianblue, the detection is extremely sensitive because it waseasy to find a grain of almost pure pigment withdiameter comparable to the IR beam wavelength($50 lm).

In Fig. 2, an example is shown of spectra distortion inthe measurement of smooth outer surfaces of fragments.Similar effects caused a systematic worsening of thedetection limit in two of three series of colour layers ifsmooth outer surface is analysed (Table 2). Possible

explanation for the improvement of the spectra qualityby microtome slicing is the surface roughening (Fig. 1)and/or disintegration of particles larger than the slice

Table 2 Detection limits of selected blue pigments in referencecolour layers (see experimental for more details)

Wavenumber(cm1)

Smoothsurface(%)

Microtomeslice (%)

Prussian blue 2,094 0.3 0.3Indigo 745 1 1

755 1 0.31128 Inverse peak 0.31200 ND 11,626 1 0.3

Phthalocyanine 756 3 11,090 ND 11,122 1 1

1,165 ND 1

Smooth surface denotes measurements of an outer surface ofsamples with properties similar to polished cross-sectionsND not detected

Fig. 1 Ultramicrotome slice of cross-section of reference colourlayer with 0.3% Cu-phthalocyanine. The area from which thespectrum was acquired is indicated by a whiter rectangle

Table 3 The IR bands of the matrix of the reference samples (ourresults) and other commonest binders (taken from [ 28])

Oil binderin the matrix

Proteins Carbonates[28]

Carbohydrates[28]

723 1,458 1,4001,500 1,419968 1,562 1,622

1,165 1,698 Sulphates [28] 2,8002,9001,377 3,515 1,1201,1501,464 SiO2 [28]1,746 Chromates [28] 9942,854 830930 7808002,925 1,170

278


5/8

thickness (2 lm) as well as to the incident irradiationbeam wavelength (2.517 lm).

Because the sensitivity of IR spectroscopy of the threeblue pigments in oil painting was checked with referencesamples, microtome slices of a real sample of blue re-paints (Our Lady the Protectress, National Gallery inPrague) were analysed. The presence of Prussian blue

was unequivocally proved. Although there is only oneapplicable diagnostic IR band in the Prussian bluespectra (Table 2 [27]), it is assigned to an uncommonCN group and hence no overlaps can be expected withother substances used in medieval paintings or Baroquerepaints (Table 3).

MALDI-TOF

MALDI TOF MS is based on searching for the matchbetween peptide fragments obtained from the samplesanalysed and reference compounds. Because trypsin

cleaves the protein chain after lysine and arginine, thefragment sizes ranges from several units to about 30amino acids and their m/z hence ranges from about 5003000 Da. The matrix (dihydroxybenzoic acid) obscurespeaks with m/z lower than approximately 700 Da, andhence a window useful for the peptide identificationranges roughly from 700 to 3000 m/z. Accuracy ofdetermination of m/z of each peak is about 0.6 Da, andhence mere statistic probability of accidental appearanceof one peak is about 0.0003. In a real case, this proba-bility is raised to a power equal to the number of

Fig. 2 Infrared spectra of samples containing Prussian blue. Thespectra distortions are marked by arrows. Spectra description fromthe bottom to top: the reference sample containing 0.3% pigmentmeasured on the smooth outer surface of the painting beforeembedding in polyester resin (curve A) and on the microtome slice(curve B) and real sample of the blue repaints of Our Lady theProtectress measured on the polished cross-section (curve C).Note the inverse IR bands of carbonates at about 1500 cm1 andsubstantial spectra distortion in curve A

Table4

Comparisonofthemos

trelevantm/zvalues(M

+

H+

)ofreferencematerialsandrealsamples

Reference

material

Realsample

Reference

material

Real

sample

Reference

material

Reference

material

Reference

material

Reference

material

Reference

material

Referen

ce

material

Analysedsample

Rabbitglue

PaintingSt.Joh

n

theEvangelist

Isinglass

(fishglue)

Crosierof

AbbessIsidora

Eggwhite

Eggyolk

Gelatin

Casein

Boneglue

Driedb

odies

ofC.ca

cti

Commercial

Carmine(42,1

00)

678.0

827.5

827.5

1,0

43.4

804.4

742.4

830.5

898.4

884.5

884.2

836.5

836.2

842.5

842.6

1,0

74.6

806.5

830.6

1,1

37.6

1,1

61.5

954.5

954.2

868.5

868.2

874.5

847.5

1,3

07.7

881.5

1,3

84.6

1,2

51.7

1,4

59.5

1,2

48.8

1,2

48.3

1,0

95.6

908.5

908.6

1,4

75.7

895.4

2,1

86.1

1,3

37.7

1,7

26.7

1,2

57.8

1,2

57.4

1,1

05.6

1,1

05.4

967.5

967.5

1,9

93.7

905.8

1,3

67.7

1,9

75.7

1,3

37.8

1,3

37.4

1,2

41.6

1,2

41.5

969.5

969.6

2,7

17.7

1,0

85.6

1,3

83.8

2,0

55.9

1,5

30.0

1,5

92.4

1,4

27.7

1,4

27.6

993.5

993.5

1,1

64.6

1,7

60.0

1,4

35.7

1,4

35.5

1,1

01.5

1,1

01.7

1,4

01.7

1,8

68.8

1,4

53.6

1,2

82.5

1,2

82.7

1,4

40.4

1,9

52.0

1,4

59.7

1,4

59.7

1,3

05.5

1,3

05.8

1,4

45.8

1,9

80.2

1,4

73.6

1,4

73.5

1,3

20.5

1,3

20.7

2,1

86.3

1,5

86.7

2,2

02.4

1,6

48.8

1,6

48.3

2,2

35.4

1,7

19.8

2,3

16.3

2,1

31.1

2,1

31.0

2,3

32.4

279


6/8

diagnostic peaks. Hence the spectrum of m/z is a fin-gerprint of a given protein.

Because the proteinaceous binders are complicatedmixtures of proteins, they are not placed in public da-tabases. In this work all relevant proteinaceous bindersexpected to have been used in paintings were examinedunder the same conditions (time and temperature ofcleavage, amount of enzyme, buffer, matrix, calibration).Results are summarised in Table 4.

There is probably no method of non-destructiveidentification of individual proteinaceous components ofcolour layers. The dissolution, the first step of anydestructive technique, could be most gently done by anenzymatic cleavage of polymerised colour layers. Theenzymatic cleavage is simple and can be done directly ina single-step procedure (Experimental) applied to thesame microtome slice [25] as could be used for IRspectroscopy. Beside simplicity, a further advantage isthat there is no risk of denaturation of enzyme by heavymetal ions, because the cleavage is done in mild condi-tions of pH neutral NH4HCO3 buffer in which theactivity of heavy metals is negligible.

MALDI-TOF-MS was tested for identification of thespecific protein component of the insect-derived dye.The mass spectra of the peptide fragments of a com-mercial red dye, Carmine, and of the specimens of theinsect Coccus cacti that was traditionally used for thedye preparation were compared. The result is shown inFig. 3 and Table 4. Six peptide fragments correspondingwith precision 0.6 Da to fragments from tryptic digestof dried bodies of insect C. cacti were found in theMALDI-TOF spectrum of tryptic digest of the com-mercial red dye, Carmine 42100 (Fig. 3). The proteins of

the sample from commercial red dye were hence identi-fied as proteins from dried bodies of C. cacti.

The insect processing obviously does neither destroynor remove the proteinaceous component, and hence thecochineal carmine could theoretically be identified alsoin colour layers of art works. The MS pattern is a morereliable evidence of the dye origin than mere presence ofa protein associated to the dye according to by Langet al. [10]. This result is particularly valuable because ofthe similarity of UVVis spectral properties and struc-ture of several other 9,10-anthraquinone derived redsthat differ mainly by their provenance. Contrarily tocochineal carmine, the red dyes from plants containmore polysaccharides than proteins.

Further examples of identification of proteinaceousbinders in real samples are shown in Figs. 4 and 5. Theidentification is based on a match of the mass spectraof real samples and a representative set of typicalproteinaceous binders used by Middle Age andBaroque artists. Thirteen peptide fragments corre-sponding to fragments from tryptic digest of referencesample of rabbit glue were found in the MALDI-TOF

spectrum of tryptic digest of the real sample (Fig. 4).The protein of the sample from the painting StaintJohn the Evangelist was hence identified as proteinfrom rabbit glue. Eleven peptide fragments corre-sponding to fragments from tryptic digest of isinglasswere found in the MALDI-TOF spectrum of trypticdigest of the real sample taken from the crosier ofAbbess Isidora (Fig. 5). Isinglass was really applied inthe restoration of the crosier in the spring of 2004. Thisresult indicates that the method used is sufficientlysensitive and works in a real matrix.

Fig. 3 The MALDI-TOF massspectrum of tryptic digest ofCarmine 42,100. The indicatedmasses marked by black circlecorrespond to proteins inoriginal sample of C. cacti(Table 4)

280


7/8

Conclusion

Infrared spectroscopy is a method suitable for identifi-cation of three common non-mineral blues, indigo,Cu-phthalocyanine, and Prussian blue in oil matrix.Problems with IR spectra distortion, usual when pol-ished cross-sections are analysed, can be suppressed by

analysis of rough surfaces obtained by ultramicrotomeslicing. Infrared spectroscopy can easily confirm thepresence, but not the kind of proteins in colour layer ofart works. It could hence be used as a preliminaryscreening method before MALDI-TOF-MS.

A novel method of binder analysis, enzymatic cleav-age followed by MALDI-TOF-MS, was successfully

Fig. 5 The MALDI-TOF massspectrum of tryptic digest of theground layer real sampleobtained from the crosier ofAbbess Isidora. Theindicated masses marked byblack circle correspond toisinglass

Fig. 4 The MALDI-TOF massspectrum of tryptic digest of thepainter glue putty on theSt. John the Evangelist. Theindicated masses marked byblack circle correspond toreference sample of rabbit glue(Table 3)

281


8/8

used for identification of proteinaceous binders. Massspectra obtained by enzymatic cleavage with trypsinwere not influenced by the sample ageing, as followsfrom the very satisfactory match between spectra offresh reference specimens and real historic samples. Nointerference was observed due to the presence of CaCO3and PbCO3 in reference samples, common inorganiccomponents of ground and colour layers of easelpaintings. The enzymatic cleavage is a promising ap-proach to dissolution and analysis of colour layers of artworks.

Acknowledgements The work was supported by Grant Agency ofCzech Republic (project number 203/04/2091). We thank torestorers K. Stretti, D. Frank, and J. Hamsk (Academy of FineArts in Prague, Czech Republic) for providing samples, TatyanaBayerova (University of Applied Arts, Vienna, Austria), for aninspiring discussion about IR spectroscopy in artwork analysis, andLadislava Kratinova and Jindrich Martinek (First Faculty ofMedicine, Charles University, Prague, Czech Republic) for cuttingthin layers.

References

1. Melessanaki K, Papadakis V, Balas C, Anglos D (2001) Spec-trochim Acta B 56:23372346

2. Burgio L, Clark RJH, Theodoraki K (2003) Spectrochim ActaA 59:23712389

3. Burgio L, Clark RJH (2000) J Raman Spectrosc 31:3954014. Vandenabeele P, Wehling B, Moens L, Edwards H, De Reu M,

Van Hooydonk G (2000) Anal Chim Acta 407:2612745. Ackacha MA, Poec -Pawlak K, Jarosz M (2003) J Sep Sci

26:102810346. Puchalska M, Orlinska M, Ackacha MA, Poec -Pawlak K,

Jarosz M (2003) J Mass Spectrom 38:125212587. Maier MS, Parera SD, Seldes AM (2004) Int J Mass Spectrom

232:225229

8. Kuckova S, Grygar T, Hradil T, Hradilova J (2003) J SolidState Electrochem 7:706713

9. Novotna P, Pacakova V, Bosakova Z, Stulik K (1999)J Chromatogr A 863:235241

10. Lang PL, Orna MV, Richwine LJ, Mathews TF, Nelson RS(1992) Microchem J 46:234248

11. Van den Berg JDJ, Vermist ND, Carlyle L, Holcapek M, BoonJJ (2004) J Sep Sci 27:181199

12. Spyros A, Anglos D (2004) Anal Chem 76:4929493613. Keune K, Boon JJ (2004) Anal Chem 76:1374138514. Pitthard V, Finch P, Bayerova T (2004) J Sep Sci 27:200208

15. Regert M (2004) J Sep Sci 27:24425416. Colombini MP, Modungo F (2004) J Sep Sci 27:14716017. De la Cruz-Canizares J, Dome nech-Carbo MT, Gimeno-

Adelantado JV, Mateo-Castro R, Bosch-Reig F (2004)J Chromatogr A 1025:277285

18. Carbini M, Stevanato R, Rovea M, Traldi P, Favretto D (1996)Rapid Commun Mass Spectrom 10:12401242

19. Chiavari G, Gandini N, Russo P, Fabbri D (1998) Chroma-tographia 47:420426

20. Bonaduce I, Colombini MP (2003) Rapid Commun MassSpectrom 17:25232527

21. Challinor JM (2001) J Anal Appl Pyrol 61:33422. Mateo-Castro R, Gimeno-Adelantado JV, Bosch-Reig F,

Dome nech-Carbo A, Casas-Catala n MJ, Osete-Cortina L, Dela Cruz-Canizares J, Dome nech-Carbo MT (2001) FreseniusJ Anal Chem 369:642646

23. Makes F (1988) Enzymatic consolidation of the portrait ofRudolph II as Vertumnus by Giuseppe Arcimboldo with anew multi-enzyme preparation isolated from Antarctic krill(Euphausia superba), Acta Universitatis Gothoburgensis, Swe-den

24. Beutel S, Klein K, Knobbe G, Ko nigfeld P, Petersen K, UlberR, Scheper T (2002) Biotechnol Bioeng 80:1321

25. Hynek R, Kuckova S, Hradilova J, Kodicek M (2004) RapidCommun Mass Spectrom 18:15

26. Ma Y, Lu Y, Zeng H, Ron D, Mo W, Neubert TA (2001)Rapid Commun Mass Spectrom 15:16931700

27. Berrie HB (1997) Prussian blue. In: Fitzhugh EW (ed) Artistspigments, a handbook of their history and characteristics, chap.7, vol 3. National Gallery of Art, New York, pp 191217

28. Newman R (1979) JAIC 19:4262

282

Documents

Analysis of Coloring and Binding Components _AnalBioanalChem 2005