Biotechnology applied to cultural heritage: Biorestoration of frescoes using viable bacterial cells and enzymes

Abstract

To set up and employ, for the biorestoration of cultural heritage (altered frescoes), an advanced and innovative biotechnology method based on the sequential use of whole viable bacterial cells and specific enzymes.
The bioremediation intervention consisted of the direct application onto an artwork surface of whole bacterial cells of the Pseudomonas stutzeri A29 strain (bioaugmentation), followed by, in a final step, a purified Protease enzyme. The bioremediation was performed on a Spinello Aretino fresco that had become altered by the animal glue residues of past restoration. For the reader's interest the fresco is the 14th century Conversione di S. Efisio e battaglia (Conversion of S. Efisio and battle), size 3.5 x 7.8 m at the Pisa Camposanto Monumentale, Italy. An assessment was made of the final costs of the biological tests (whole bacterial cells, enzymes) so as to compare them with other intervention techniques.
A successful innovative biological approach to recover valuable frescoes was set up, and the best conditions for treatment efficiency were identified. Furthermore the cost of the biological cleaning using viable bacterial cells and enzymes (P. stutzeri, Protease, Collagenase, 1 : 3 : 10, ratio respectively) was much lower than that of other conventional methods, making this biotechnology not only very interesting but also very competitive.
New biotechnologies with an innovative, soft approach to the 'biocleaning' and 'biorestoration' of cultural heritage are in constant demand, and our results are clear evidence that such an approach has been achieved; the technique could be of significant importance towards developing other goals.

Full text

Biotechnology applied to cultural heritage: biorestoration of
frescoes using viable bacterial cells and enzymes
G. Ranalli
1
, G. Alfano
1
, C. Belli
1
, G. Lustrato
1
, M.P. Colombini
2
, I. Bonaduce
2
, E. Zanardini
3
,
P. Abbruscato
3
, F. Cappitelli
3
and C. Sorlini
3
1
DISTAAM, Universita
`del Molise, Campobasso, Italy,
2
Dip. Chimica e Chimica Industriale, Universita
`di Pisa, Pisa, Italy, and
3
DISTAM,
Universita
`di Milano, Milan, Italy
2003/1162: received 17 December 2003, revised 2 July 2004 and accepted 17 July 2004
ABSTRACT
G. RANALLI, G. ALFANO, C. BELLI, G. LUSTRATO, M.P. COLOMBINI, I. BONADUCE, E. ZANARDINI,
P . A B B R U S C A T O , F . C A P P I T E L L I A N D C . S O R L I N I . 2004.
Aims: To set up and employ, for the biorestoration of cultural heritage (altered frescoes), an advanced and
innovative biotechnology method based on the sequential use of whole viable bacterial cells and specific enzymes.
Methods and Results: The bioremediation intervention consisted of the direct application onto an artwork
surface of whole bacterial cells of the Pseudomonas stutzeri A29 strain (bioaugmentation), followed by, in a final step,
a purified Protease enzyme. The bioremediation was performed on a Spinello Aretino fresco that had become altered
by the animal glue residues of past restoration. For the reader’s interest the fresco is the 14th century Conversione di
S. Efisio e battaglia (Conversion of S. Efisio and battle), size 3Æ5·7Æ8 m at the Pisa Camposanto Monumentale,
Italy. An assessment was made of the final costs of the biological tests (whole bacterial cells, enzymes) so as to
compare them with other intervention techniques.
Conclusions: A successful innovative biological approach to recover valuable frescoes was set up, and the best
conditions for treatment efficiency were identified. Furthermore the cost of the biological cleaning using viable
bacterial cells and enzymes (P. stutzeri,Protease,Collagenase, 1 : 3 : 10, ratio respectively) was much lower than that
of other conventional methods, making this biotechnology not only very interesting but also very competitive.
Significance and Impact of the Study: New biotechnologies with an innovative, soft approach to the
biocleaningand biorestorationof cultural heritage are in constant demand, and our results are clear evidence that
such an approach has been achieved; the technique could be of significant importance towards developing other
goals.
Keywords: bacteria, biocleaning, biorestoration, enzymes, frescoes, gas chromatography coupled with mass
spectrometry (GC-MS) and pyrolysis/GC-MS analyses.
INTRODUCTION
Outdoor artwork, especially where lithoid materials, stones,
frescoes and paint are involved, is very susceptible to
deterioration that is brought about by changes mainly caused
by ageing and, in recent decades, to pollution. In fact, in
urban areas, damage because of the increased pollution of
today is manifested on monument surfaces by alterations
such as black crusts, nitration, sulphation and the deposition
of dust and residual hydro-carbons (Saiz-Jimenez 1995).
In addition to air pollutants, the surfaces of man-made
artistic stonework can also be altered by organic matter that
has been applied, but then not completely removed, during
restoration; in many cases this can pose serious danger even
to the preservation of the artwork itself (Ranalli et al. 1996,
2000). In fact such leftovercompounds often act as a good
growth substrate for micro-organisms and mycetes that
Correspondence to: Giancarlo Ranalli, DISTAAM, University of Molise, Via De
Sanctis 46, 86100 Campobasso, Italy (e-mail: ranalli@unimol.it).
ª2004 The Society for Applied Microbiology
Journal of Applied Microbiology 2005, 98, 73–83 doi:10.1111/j.1365-2672.2004.02429.x

destroy the surface, and allow hyphae penetration to some
depth.
Moreover, the process of detaching frescoes from walls
prior to restoration calls for notable quantities of organic
compounds (such as glue and casein) that become distributed
on both the painted surface and at the back of the fresco.
This was the condition of the Pisa frescoes (in the
Monumental Cemetery, Camposanto) that cover a surface
area of about 1500 m
2
; these frescoes date back to the XIV
century and are by famous painters such as Antonio
Bonaiuti, Antonio Veneziano, Benozzo Gozzoli, Taddeo
Gaddi, Francesco Traini, Bonamico Buffalmacco and Spi-
nello Aretino.
Most of the frescoes had been restored using traditional
chemical and physical techniques in which residual organic
substances and salts are removed by ammonium carbonate
solution and organic solvents. However even today’s
enzymes, surfactants and solubilizing agents have failed to
clear away past restoration material and restore some of the
frescoes (Makes 1988; Bellucci and Cremonesi 1994; Bonomi
1994; Wolbers 2000) that today exhibit a particular hydro-
phobic behaviour related to weathering and protein poly-
merization.
In order to choose suitable conservation practices, and
to resolve the above-mentioned problems, it is most
important to identify both the original organic materials
applied and those belonging to restoration work, and also
materials that have undergone transformation through
natural ageing and pollution. As artwork samples are
obviously unique and generally very small, and the
compounds to be determined are in low concentration,
it is fundamental to use sensitive and selective techniques
like gas chromatography coupled with mass spectrometry
(GC-MS) and pyrolysis/GC-MS (PY/GC-MS) (Colom-
bini et al. 1999, 2003).
Up until this realwork on the frescos we had only used
viable micro-organisms to degrade and remove substrate
supported organic matter under laboratory conditions
(Ranalli et al. 2003a). Thus we had to first identify the
unwanted organic matter on the frescos, and verify the
difficulty of removing such organic compounds by tradi-
tional methods. We then began experimenting the use of
viable bacterial cells to degrade the organic matter, i.e. using
them as biorestorationagents on the frescoes; as the organic
matter degraded the residues were rinsed offand removed
by enzymatic treatments.
The present research was aimed at dissolving the adhesive
(animal glue) between the painted fresco face and the
adhering gauze, and thus allow the gauze to be detached. We
took the following steps:
i the characterization of the organic adhesive matter on the
gauze used to detach the fresco from the wall: GC-MS
and PY/GC-MS analytical procedures identified the
natural organic compounds and their degradation prod-
ucts
ii the developing and improving of an innovative bioresto-
ration system, that we had already set up on a laboratory
scale, to remove the organic matter from the fresco
surface.
We also assessed the costs of the whole bacterial cell and
enzyme biotechnologies, and compared them with tradi-
tional methods.
MATERIALS AND METHODS
Micro-organisms, media and cultural methods
We tested Pseudomonas cepacia strain DSMZ 7288
(Deutsche Sammlung von mikroorganismen und Zellk-
ulturen GmbH, Braunschweig, Germany), P. testosteroni
strain DSMZ 1455, P. fluorescens strain DSMZ 304,
P. flavescens strain DSMZ 12071, P. stutzeri strain A 29
(DISTAM-DISTAAM Strains Collections, University of
Milan–University of Molise, Italy), P. stutzeri strains
DSMZ 5190, and ATCC 23856 (American Type Culture
Collection, Rockville, MD, USA). To select the best
culture to apply to the altered fresco surfaces we assessed
the biodegrading activity of the culture by adding animal
glue at 0Æ5% and 1%, supplied as the complex organic
matter and energy sources in the mineral medium M9,
under aerobic conditions, at 10, 20, 28 and 37C (Ranalli
et al. 1996, 2003a). Microbial growth was monitored by
O.D.
560
, and an assay was made of ATP content and the
viable cell count in plate count medium. The selected
strains were stored on standard plate count agar (Difco).
Preparation of the bacterial suspension for
application
Suspensions containing exponentially growing bacteria,
ca 10
8
CFU ml
)1
, were obtained by inoculating 3 ml of
an overnight broth culture into 300 ml fresh broth
medium and incubating it in a shaker (200 rev min
)1
)
for 24 h at 28C. The cells were centrifuged at 7000 gfor
10 min at 4C, washed twice with phosphate-buffered
saline, and re-suspended in sterile 0Æ8% (w/v) NaCl
solution, pH 7Æ2; the final cell concentration was above
10
8
cells ml
)1
, corresponding to an O.D.
560nm
of ca 1Æ6;
the cells were used immediately or stored at 4C during
conservation and/or transport from the laboratory to the
field application.
A fully computerized laboratory batch fermenter (20 l
useable volume) (mod. Biostat C, B. Braun; Biotech
International, Helsungen, Germany) was adopted to obtain,
when needed, a relevant amount of viable bacterial cell
biomass. The best selected bacterium for the in situ
74 G. RANALLI ET AL.
ª2004 The Society for Applied Microbiology, Journal of Applied Microbiology,98, 73–83, doi:10.1111/j.1365-2672.2004.02429.x

biorestoration was cultivated, at the same concentration as
the cultures, for 24 h on broth medium containing animal
glue as the complex organic matter. The cells were
recovered in vitroby centrifugation and re-suspended in
NaCl solution at a final abundance of 1–3 ·10
8
cells per ml.
Whole bacterial cell applications on a fresco
Viable cells of the selected strain were applied to the fresco
surface at ca 20C and 10C to evaluate the effect of
temperature on the removal of organic matter. The bacterial
culture was applied in three ways to test the application
effect: (i) spraying, (ii) brushing and (iii) a covering cotton
wool layer soaked with bacterial suspension.
In the first case a manually operated atomizer (useful
volume 0Æ5 l) was used to spray the fresco surface from a
distance of 25–30 cm, distributing the bacterial cells regu-
larly by consecutive left-right and top-bottom dispersions;
to favour distribution, large fresco areas were subdivided
into areas to form rectangular reticules (treated areas ca 200–
400 cm
2
).
For the brushing distribution we adopted a flat paintbrush
(8–12 cm wide) to deliver the cells manually, making gentle
brush strokes to deliver the suspension over all the surface in
a regular layer and taking great care to avoid detaching
altered pellicles.
In the third application type we employed the same
brushed-on cell distribution (as in application 2) and then
covered the fresco with a thin layer (0Æ5 cm depth, 10 cm
wide, 10–20 cm long) of hydrophilic sterile white cotton
wool wet with the same bacterial cell suspension.
Monitoring of viable microbial cell counts
Throughout the biorestoration of the fresco, we constantly
evaluated the total viable microbial cell numbers. For
application 3 this was carried out on small samples taken
from the cotton layer (ca 1Æ0 g), using the plate counts
technique in standard plate count agar (Difco), incubation
for 48 h at 28C expressed as CFU g
)1
. In applications 1
and 2, when the microbial count was determined directly on
the fresco surface, we adopted the replica-plating technique;
the results are expressed as CFU per unit area. Sterile velvet
was pressed onto the cleaned surfaceof the fresco, and then
onto plate count agarized medium. Finally, in order to verify
microbial viability during bacterial cell application, ATP
assays (total and free) were carried out using the sterile swab
technique (Ranalli et al. 2003b).
Enzyme treatments
Enzyme treatments were performed after the bacterial cell
applications had removed as much glue as possible.
Five pure enzymes (Collagenase Type IA and Type V
from Clostridium histolyticum, 1–3 U mg
)1
,pH7Æ5–8Æ0;
Protease Type XIV from Streptomyces griseus,4Umg
)1
,pH
7Æ5–8Æ5; Protease Type XIX from Aspergillus sojae,
0Æ4Umg
)1
,pH8Æ4; Lipase Type VII from Candida
cylindracea, 400–900 U mg
)1
,pH8Æ4–8Æ8; Sigma-Aldrich,
St Louis, MO, USA) were tested, both separately and
mixed, for the removal of the weathered glue.
Purified enzyme solutions were added delicately to the
altered fresco surface by three different techniques: (i) swab,
(ii) paper-disc and (iii) brush.
Enzymatic activity
The levels of enzymatic activity of the pure bacterial
cultures were determined using the API-ZYM
TM
system
(bioMe
´rieux Italia, Rome, Italy). The specific galleries (nm
5, 11, 12) were inoculated with 65 llof10
)1
adequate
suspension (O.D.
560 nm
of ca 1.1) of 1 ml of an overnight
broth culture (at 12th hour) to 9 ml of sterile solution (NaCl
9Æ0gl
)1
). After incubation for 4 h at 20, 28 and 37C, the
galleries were activated by adding 30 ll of Reagent ZYM A
and Reagent ZYM B (bioMe
´rieux) and after 5 min we
assigned values ranging from 0 to 5 in relation to the colour
developed in each enzymatic reaction, using the colour chart
provided by the manufacturer (Principi et al. 2003). For the
purposes of this study, the results are reported as reactions
of low intensity (1), moderate intensity (2–3) and high
intensity (4–5) (Tiquia 2002).
The grade of enzymatic activity on native animal glue and
on animal glue cross-linked was determined using the Feller
test (Feller et al. 1985; Cremonesi 1999).
ATP determination
Total ATP assays to monitor both culture growth on the
laboratory scale, and the viability of the bacterial cell
suspensions during the in situ biorestoration processes,
were performed using a specific enzymatic kit (NRM/
Lumit-QM, code 9332-l; Lumac B.V., Landgraaf, the
Netherlands). A Biocounter 1500 P luminometer (Lumac
B.V.) equipped with a photomultiplier tube set at
7200 RLU with 200 pg ATP in 100 ll of Lumit buffer
and Lumit-QM reagent was used (Ranalli et al. 1998,
2003b).
Microscope observations
Microbial growth and cell survival were determined by
optical microscope (Axiophot, Zeiss; LEO Elektronen-
mikroskopie GmbH, Oberkochen, Germany) and scanning
electron microscopy (SEM) observations throughout the
application of the biocleaning treatment to the altered
BIOTECHNOLOGY APPLIED TO CULTURAL HERITAGE 75
ª2004 The Society for Applied Microbiology, Journal of Applied Microbiology,98, 73–83, doi:10.1111/j.1365-2672.2004.02429.x

surface of the fresco. The samples for SEM observation
were treated overnight in a solution of 2% glutaraldehyde
(0Æ01 mol l
)1
phosphate buffer), and then immersed in 1%
osmium tetroxide. A microscope operating at 10 kV was
used (Zeiss DSM 940A; LEO Elektronenmikroskopie
GmbH).
Chemical analyses
Reagents. All the solvents were Baker HPLC grade.
Hexadecane and tridecanoic acid were used as internal
standards (IS). Hexamethyldisilazane (HMDS) and
N,O,bis(trimethylsilyl)trifluoroacetamide (BSTFA) con-
taining 1% trimethylchlorosilane were purchased from
Sigma (Milan, Italy), N-tert-butyldimethylsilyl-N-methyl-
trifluoroacetamide (MTBSTFA) with 1% trimethylchlo-
rosilane was from Fluka (Milwaukee, WI, USA); all were
used without any further purification. Standard solutions of
amino acids in HCl 0Æ1
NN
, norleucine (IS2) and hexadecane
(IS1) as injection internal standard were purchased from
Sigma-Aldrich.
Analytical procedure based on selective extraction and
hydrolysis assisted by microwave and GC-MS analy-
sis. The glue sample (50–100 lg) was extracted with
ammonia solution, subjected to microwave-assisted acidic
hydrolysis (microwave oven model MLS-1200 MEGA
Milestone; FKV, Sorisole, Italy), and the resultant hydro-
lysed solution derivatized with MTBSTFA and GC-MS
analysis (6890N GC System Gas Chromatograph coupled
with a 5973 Mass Selective Detector; Agilent Technologies,
Palo Alto, CA, USA) to determine the protein content. The
residual of the ammonia extraction was subjected to
saponification, extraction with n-hexane, acidification and
extraction with diethyl ether to obtain, respectively, the
neutral and acidic fractions, subsequently analysed by GC-
MS after derivatization with BSTFA to determine the
presence of natural resins, lipids and waxes. The detailed
procedure is reported in the literature (Colombini et al.
1999, 2003; Bonaduce and Colombini 2003).
Analytical procedure based on PY/GC-MS. A few lgof
samples together with 5 ll of a HMDS solution (20% in
acetone) were pyrolysed at 600C in a pyrolyser operating at
constant temperature mode Pyrojector II (SGE, Austin,
TX, USA). Experimental conditions are reported in detail
in previous work (Colombini et al. 2003).
Mass spectra assignment. Mass spectra assignment was
generally based on direct matching with spectra from the
Wiley 275 library: spectra were retained when the correla-
tion match index was >95%. We compared pure compounds
whenever possible.
Physical analyses. The thickness of the glue layers was
measured at different points of the fresco using a laser
sensor micrometer, Thrubeam LX2-V model (Keyence
Corp., Osaka, Japan).
Spinello Aretino fresco
During World War II the frescoes belonging to the complex
of the Monumental Cemetery of Pisa were ruined by a bomb
(1944) but, after detachment from the walls, were restored at
a later date. From among them we chose, for the present
study, the XIV century fresco, Conversione di S. Efisio e
battaglia, painted by Spinello Aretino; the surface area is ca
50 m
2
(26 m
2
treated) and the contained inorganic pigments
are known from relative documentation.
Unfortunately the restoration techniques and conservation
conditions were inappropriate, and ourfresco had soon
shown alteration phenomena such as swelling and detachment
of the painting pellicle. In the 1980s, the fresco was again
detached from the walls at Camposanto using tear off
techniques that involve, on site, the application of a wide gauz e
directly onto the fresco surface, using animal glue as the
consolidating agent; thus the gauze and fresco become one
layer that is then detached from the wall. The back of the
fresco is reinforced by a canvas cloth, applied using casein,
which is, in turn, supported on asbestos, again with casein.
One of the greatest difficulties during the restoration was
the detachment of gauze still adhering, from the previous
restoration, to the paint layer on the front surface; this was
despite soaking with solvents. Over the years the glue had
altered greatly, and had become very hard and resistant to
commonly used solvents. Our task was to dissolve the animal
glue on the paintwork itself to release the gauze residues
without removing the casein supporting the back of the fresco.
The conservation state was also worsened by the
contraction of the glue and the swelling of the casein at
the back; furthermore different synthetic organic substances
had been used in the past restoration and these had favoured
the polymerization of the compounds, making their removal
an onerous task.
RESULTS
Biological treatment
Selection of the strain for application in the bioresto-
ration. Table 1 shows the results of the laboratory scale trials
of bacterial culture growth rate at 28C on broth medium
containing animal glue as the complex organic matter.
There were marked growth rate differences among the
tested Pseudomonas species and the P. stutzeri strains;
P. stutzeri A 29 showed the highest cell density (1Æ8 O.D.
540
and >8Æ5 log CFU ml
)1
), a relevant ATP content, higher
76 G. RANALLI ET AL.
ª2004 The Society for Applied Microbiology, Journal of Applied Microbiology,98, 73–83, doi:10.1111/j.1365-2672.2004.02429.x

enzymatic protease activity and, from optical microscope
observations, the greatest number of bacterial cells per field.
The influence of temperature on the growth of the
selected bacterial strains was investigated by cultural,
enzymatic and microscopic tests, the response of P. stutzeri
strain A29 growing cells being noted at 10, 20, 28 and 37C.
Cell density increased slightly when the incubation tem-
perature was maintained under 10C for 36 h, and improved
a little at 20C (data not shown). However at constant
temperatures of 28C and 37C there was a very marked
increase in cell growth at 24 and 36 h, with abundant
biomass production. Moreover, notable increases in enzy-
matic activity were observed at the tested temperatures: in
fact, the response of the protease reaction in P. stutzeri strain
A29 on overnight broth culture at 12th hour, showed low
intensity (value of 1) at 10C, moderate intensity (value of
2–4) at 20C, and high intensity (value of 5) at both 28 and
37C (Table 1).
Figure 1 shows the SEM observations on the selected
P. stutzeri strain A29 cells.
Chemical analyses
The GC-MS and PY/GC-MS results from several fresco
samples (1–5 mg) evidenced both animal glue, used to glue the
gauze onto the front of the fresco, and casein which came from
the back of the painting. The PY/GC-MS identified, by
specific markers, the proteins indole for casein and pyrrole and
alkylpyrroles for the animal glue. The GC-MS amino acidic
profiles showed that all the proteins had undergone degrada-
tion. This was particularly evident in the samples taken from
the front of the fresco (Fig. 2) where there was a typical
collagen protein pattern; the samples from the back of the
fresco showed a high degree of degradation in the casein amino
acidic pattern, because of natural ageing and micro-organism
attack (Fig. 3 in GC-MS). In fact, this phosphoprotein had a
low content of amino acids like methionine, phenylalanine and
lysine. These results concord with those of samples from other
analysed frescoes (Colombini et al. 1995, Colombini et al.
1999). Moreover, as protein recovery was ca 20% w/w of the
sample weight, it would appear that cross-linking reactions
had occurred.
This hypothesis was also confirmed by the presence of
formalin, which had been added to the glue mixture to avoid
Table 1 Growth yields in minimal medium with added glue, assessed by O.D.
560
and CFU at 36th hour, at 28C; ATP content, protease activity at
10, 20, 28 and 37C and microscope observations at 12th hour
Bacterial culture
O.D.
560
(±
S.DS.D
.)
(36 h)
Log CFU ml
)1
(36 h)
Total ATP
(RLU)* (12 h)
Protease activity
(trypsin)(C) Microscope
observation(12 h)
10 20 28 37
P. cepacia DSMZ 7288 1Æ5(0Æ15) <8 ++ 1334++
P. testosteroni DSMZ 1455 0Æ1(0Æ1) <3 +/)0000+/)
P. fluorescens DSMZ 304 1Æ45 (0Æ2) <8 ++ 0233++
P. flavescens DSMZ 12071 1Æ4(0Æ15) <8 ++ 0234++
P. stutzeri A29
(DISTAM-DISTAAM, Italy)
1Æ8(0Æ2) >8Æ5 ++++ 2455+++
P. stutzeri DSMZ 5190 1Æ65 (0Æ15) 8 +++ 1344+++
P. stutzeri ATCC 23856 1Æ7(0Æ2) 8 +++ 1345+++
RLU, relative luminose unit.
*+/), very low; ++, low; +++, high; ++++, very high.
Overnight broth-culture at 12th hour: reaction of low intensity (value of 1: 5 nm of hydrolysed substrate); moderate intensity (value of 2–4: 10–
30 nm of hydrolysed substrate); high intensity (value of 5: ‡40 nm of hydrolysed substrate).
Presence of bacterial cells per field (+/), very rare; ++, low; +++, high; ++++, very high).
Fig. 1 Scanning electron microscopy of Pseudomonas stutzeri strain
A29 cells in 0Æ8% NaCl solution
BIOTECHNOLOGY APPLIED TO CULTURAL HERITAGE 77
ª2004 The Society for Applied Microbiology, Journal of Applied Microbiology,98, 73–83, doi:10.1111/j.1365-2672.2004.02429.x

fermentation in alkaline conditions; furthermore, formalin in
the presence of casein during drying gives rise to insoluble
compounds that are difficult to remove.
The relevant amounts of sulphates, phosphates and
oxalates were because of the eternit, that together with the
lime putty contributed to the sulphation of the frescoes, to
both the phospholipids of the casein and the final products
of the oxidative degradation of the animal glues and casein,
and to metabolic micro-organism degradation.
These salts were responsible for the whitish patina
observed on the surface of almost all the frescoes.
Our results have revealed that the painted surface of
Spinello Aretino’s fresco is embedded in a hard net of
polymerized and cross-linked proteins which, in turn, are
glued to the canvas.
Physical analyses
The thickness of the glue layer was demonstrated to vary
from 1 to 3 mm corresponding to ca 4mgcm
)2
.
Biorestoration treatment with whole bacterial cells
The ex situ biorestoration treatments were performed by
distributing a suspension of viable precultured bacterial cells
of P. stutzeri strain A29 on the altered fresco.
Of the three application modes, the best results were
achieved by putting a cotton wool layer soaked with microbial
suspension onto the fresco surface. In fact, in terms of
humidity preservation, this technique resulted in constant,
and favourable, microenvironmental conditions between the
viable cells and the fresco surface throughout the biological
treatment (time of contact and adhesion). The spraying and
brushing cell applications produced unfavourable rapid and
excessive drying, with a consequent reduction in bacterial cell
viability and activity. Furthermore, in both cases it was also
necessary to apply further cell suspension aliquots from time
to time after the first application, placing an extra burden on
our resources with regard to microbial biomass availability,
and on the restorers from the point of view of labour. Thus, on
the basis of these findings, the bacterial culture mode of
50 000
100 000
150 000
200 000
250 000
300 000
350 000
400 000
450 000
500 000
550 000
600 000
650 000
700 000
750 000
Time (min)
Abundance
12·00 14·00 16·00 18·00 20·00 22·00 24·00 26·00 28·00 30·00
ala
gly
val
IS
leu
ile
IS ser
pro
phe
asp
glu
lys
met hyp
tyr
Fig. 2 GC-MS chromatogram (acquired in
the SIM mode) of a sample taken from the
front of the fresco. Internal standards (IS),
alanine (ala), glycine (gly), valine (val), leucine
(leu), isoleucine (ile), methionine (met), serine
(ser), proline (pro), aspartic acid (asp), gluta-
mic acid (glu), lysine (lys), hydroxyproline
(hyp) and tyrosine (tyr)
12·00 14·00 16·00 18·00 20·00 22·00 24·00 26·00 28·00 30·00
20
000
40
000
60
000
80
000
120
000
100
000
140
000
160
000
180
000
200
000
220
000
240
000
260
000
280
000
Time (min)
Abundance
ala
gly
val
IS
leu
ile
IS
ser
pro
phe asp glu
lysmet tyr
Fig. 3 GC-MS chromatogram (acquired in
the SIM mode) of a sample taken from the
back of the fresco. Internal standards (IS),
alanine (ala), glycine (gly), valine (val), leucine
(leu), isoleucine (ile), methionine (met), serine
(ser), proline (pro), aspartic acid (asp), gluta-
mic acid (glu), lysine (lys), hydroxyproline
(hyp) and tyrosine (tyr)
78 G. RANALLI ET AL.
ª2004 The Society for Applied Microbiology, Journal of Applied Microbiology,98, 73–83, doi:10.1111/j.1365-2672.2004.02429.x

application adopted was the culture-soaked cotton wool layers
(40 cm wide, 6Æ0 m long); these were placed over the whole
fresco surface, guaranteeing homogeneous conditions and
saving time during application and recovery.
Different tests showed the marked influence of tempera-
ture on the biological treatment. An environmental tem-
perature below 10C resulted in no evident biological
effects, however at ca 20C the collagen began to dissolve
8–24 h after the daily cell suspension application; this did
not happen in the control areas.
At 28C the cellular activity was so intense, just 8–12 h
after applying the bacterial suspension (log 8 CFU g
)1
that
means ca 100 million of viable cells per g), that the treated
areas were clean and freefrom the animal glue without any
structural damage.
Nevertheless the degree of glue removal depended on the
thickness of the glue layer at the start of the treatment, varying
between 80 and 100%; some areas were cleaned completely,
while others (where the glue was as thick as 3 mm) were left
with residues, but never greater than ca 0Æ5 mm.
When residues were still present after the treatment it was
found that prolonging the bacterial treatment gave the
possibility of obtaining complete removal, but there was
some risk to the fresco. In fact, for periods of longer than
15–17 h, the prolonged contact with the moist cotton led to
the swelling and detachment of paint fragments.
Figure 4 shows the removal of cotton layer enriched with
P. stutzeri strain A29 after 10-h biorestoration from ex situ
fresco altered by animal glue (Conversione di S. Efisio e
battaglia, XIV century) at Pisa Camposanto Monumentale
(Italy).
Biorestoration treatment with enzymes
After removing the gauze through bacterial activity, small
organic matter residues (the glues themselves) still remained
on the surface. These were treated by enzyme solutions.
The purified enzymes, Collagenase Type IA and Protease
Type XIX, used separately and in mixture (data not shown
here), showed the highest removal efficiency, both on native
animal glue and on animal glue cross-linked (fragment of
altered fresco), as reported in Table 2.
However, because of the high cost of the collagenase-
based enzyme, we adopted Protease Type XIX (1Æ0g/
100 ml
)1
Tris buffer solution, pH 7Æ8–8Æ2, at 38C, for 10–
15 min).
During application a portable heating device was used to
maintain optimal temperatures for the fresh enzyme solu-
tions (38C). Throughout the fresco biorestoration process
and during the enzymatic treatment, the recorded indoor
environmental temperature showed an average of 28 ± 3C.
Of the three different enzyme application modes, swab,
paper-disc and brush, enzyme distribution was best by
brush; in fact its efficiency in removing residual organic
matter and animal glue was probably favoured by the gentle
pressure of the bristles of the brush and the repetitive and
gentle mechanical pressure of the manual application.
Monitoring of the biorestoration process
An important aspect of bioremediation is the process
monitoring. This was performed initially on the laboratory
scale using a microbiological approach. Instead the in situ
application called for more rapid, although not always
specific, techniques such as the determination of ATP
content by bioluminescence.
During the biorestoration process on a full scale fresco we
assessed how time and the environment affected the
Fig. 4 Removal of cotton layer enriched with Pseudomonas stutzeri
strain A29 after 10-h biorestoration from ex situ fresco altered by
animal glue (Conversione di S. Efisio e battaglia, XIV century) at Pisa
Camposanto Monumentale (Italy)
Table 2 Enzymatic activities on native animal glue and on animal glue
cross-linked (altered fresco)
Enzymes
Native animal glue
Animal glue
cross-linked
Specific
activity
Feller
test
Specific
activity
Feller
test
Collagenase Type IA 5* +++ 2* +
Collagenase Type V 4* +++ 1* +/)
Protease Type XIV 4++ 1+/)
Protease Type XIX 5+++ 2+
Lipase Type VII 1+/)0+/)
Assay of specific activity: *chymotrypsin; trypsin; lipase
+/), very low; +, low; ++, middle; +++, high.
Reaction of low intensity (value of 0–1: 0–5 nm of hydrolysed
substrate); moderate intensity (value of 2–4: 10–30 nm of hydrolysed
substrate); high intensity (value of 5: ‡40 nm of hydrolysed substrate).
BIOTECHNOLOGY APPLIED TO CULTURAL HERITAGE 79
ª2004 The Society for Applied Microbiology, Journal of Applied Microbiology,98, 73–83, doi:10.1111/j.1365-2672.2004.02429.x

dynamics of P. stutzeri A29 bacterial cell survival, measured
as log CFU ml
)1
and ATP content and compared with a
control. The results from a repeated series of tests showed
that there were no significant statistical differences in the
first 12 h and 14 h of the biological process. However a
significant variation appeared in the parameters of the 16
and 24-h treatments using the P. stutzeri strain A29. The
recovered cells (fresco surface) showed a final mean value in
cell viability of log 4–log 5 CFU cm
)2
, compared with
log 8 CFU cm
)2
in the control set (initial fresco surface
enriched by viable bacterial cell suspensions). The total ATP
content showed little variation as a function of treatment
duration. In fact, at hours 6, 10 and 12, the ATP content was
comparable with the initial values, whereas at 16 h, it was
slightly lower than the control.
The pH values were continually monitored by a probe
located between the fresco surface, in direct contact with the
altered patina, and the cotton wool layer enriched with the
selected bacterial cell suspension. After 12 h of biorestoration,
only a slight reduction in pH value was noted (from 7Æ2to6Æ7).
At the end of the microbial cell cleaning of the fresco
surface, and after removing the remaining bacteria by a final
washing with distilled sterile water (three soft manual
sponge applications), microbiological analyses were made on
the fresco surface to check for the possible presence of
residual microbial cells. The lack of growth on the
inoculated plates of the replica-plating technique confirmed
that no microbial cells were detectable with the methods
used. The data were confirmed by the free and total ATP
content values.
Figures 5 and 6 show the effects of biocleaningand
biorestorationprocess obtained with P. stutzeri strain A29
cells and Protease enzyme on Conversione di S. Efisio e
battaglia, fresco (XIV century) at Pisa Camposanto Monu-
mentale, (Italy) (top and bottom, before and after treatment
respectively).
Economic evaluation of the biorestoration
An analysis of the costs of the bioremediation process shows
that the use of bacterial cultures is more convenient
economically than the use of enzymes like Protease and
Collagenase.
In fact, the total cost can be quantified as being ca 50 l
)1
for bacterial culture, 150 l
)1
for the Protease and 500 l
)1
for
the Collagenase, using a litre of the bacterial or enzymatic
suspension for ca 2m
2
of fresco surface. Thus the cost ratio
was 1 : 3 : 10.
On comparing the costs of the bacterial biorestoration
process and the more frequently used chemico-physical
techniques, it is evident that the latter are less convenient;
this can be seen in terms of both the total time needed for
the restoration and the number and duration of the
application times (repetitive treatments every 6–8 h) as well
as in the greater costs involved for specialized personnel,
restorers and finally environmental safety.
DISCUSSION
The success of the viable bacterial cell application, compared
with that of enzymes is, presumably, attributable to the wide
versatility of the bacteria and their activity. Bacteria are
known to produce not only constitutive but also inducible
enzymes that attack and degrade different types of mole-
cules. The synthesis of inducible enzymes takes place only in
the presence of a substrate, creating a regulatory effect.
Thus the use of micro-organisms is more effective than just
the use of a single enzyme that attacks only specific linkages.
In fact, the alkaline properties of formalin, a common
component of the organic matter, are conducive to poly-
merization, protein cross-linking and reticulation during
drying, resulting in the formation of insoluble compounds.
Fig. 5 Effects of biocleaningand biorestorationprocess obtained
with Pseudomonas stutzeri strain A29 cells and Protease enzyme on
Conversione di S. Efisio e battaglia, fresco (XIV century) at Pisa
Camposanto Monumentale, (Italy) (top and bottom, before and after
treatment, respectively)
80 G. RANALLI ET AL.
ª2004 The Society for Applied Microbiology, Journal of Applied Microbiology,98, 73–83, doi:10.1111/j.1365-2672.2004.02429.x

Thus in the case of frescoes where the glue and casein
compounds are protein weathered, and altered by the
presence of formalin, the effectiveness of the micro-organ-
isms is really important as the enzymes, which are highly
specific, may not be useful. Moreover, the constitutive and
induced enzymes of viable cells of versatile bacteria like
Pseudomonas are able to break up many organic compound
chemical linkages.
These heterotrophic bacteria were applied to the Spinello
Aretino fresco as it was known that Spinello used only
inorganic pigments; thus his frescoes would not be damaged
in any way during the short application time of the bacterial
solution.
Many factors such as temperature, relative humidity, pH,
carbon and energy sources, etc. can influence metabolic
activity. Thus, to optimize the metabolic efficiency we
expected on the fresco, we had to check these factors
accurately in the laboratory, our assumption being that
if biorestoration could not be achieved under optimal
conditions on the laboratory scale, it would be unlikely to
succeed in an uncontrolled environment.
Thus we had to verify whether our successful laboratory
results would be confirmed under real, on site, environ-
mental conditions, especially for outdoor artwork.
In artwork bio-remediation, bacterial treatment requires
the application, by spray, brush or compress, of selected
micro-organisms to the artwork surface. The way the
application is carried out depends on the type of alteration,
the artwork material, the location of the areas undergoing
treatment, and the metabolic activity of the selected
microflora (aerobic and anaerobic).
Indeed the effectiveness of the biological process is
defined by the time course of the treatment. Long treatment
times require a high biomass concentration and favourable
contact between the micro-organisms and the surface.
Consideration must also be given to an adequate cell-carrier
support; in the case of ourfresco the choice was cotton
wool. A previous laboratory study carried out on stone
artwork had shown the best support matrix for the bacteria
to be sepiolite (Ranalli et al. 2000).
All artwork biorestoration requires a final, accurate
cleaning phase. Indeed, at the end of specific biological
processes all residual microflora should be removed carefully
to avoid the continuation of undesired metabolic processes.
Thus an adequate strategy for artwork protection must be
planned; if the micro-organisms remain alive and active they
can cause material loss or damage, either directly or as a
consequence of their catabolic by-products.
The proposed biological technique, that constitutes an
accurate cleaning phase, was a fundamental mustfor the
side of the fresco called intelaggio(the painted side),
although long contact with a water solution could have
posed an indirect risk to the safety of the fresco. In this
phase it was very important to reduce to the minimum both
the volume of water addition and its contact time on the
fresco surface.
Possible damage because of prolonged bacterial treatment
has, till now, only been evidenced by Ranalli et al. (2000),
who discussed the removal of black crusts from stone. In this
case, sulphate-reducing bacteria reduced the sulphates and,
consequently, removed the crusts; however prolonged
contact between the bacteria and the stone caused the
precipitation of sulphide salts. Thus it is very important to
control the duration of the application.
Nevertheless, when artwork undergoing microbial reme-
diation has peculiarities that permit other forms of inter-
vention, adequate surfactant solutions and weak biocides can
be used.
In order to identify optimal conditions to develop a
bioremediation approach for the use of selected microflora
on artwork, it is most important to monitor the microbial
activity during, and at the end of, the biological process.
Fig. 6 Effects of biocleaning,biorestoration, process obtained with
Pseudomonas stutzeri strain A29 cells and Protease enzyme on
Conversione di S. Efisio e battaglia, fresco (XIV century) at Pisa
Camposanto Monumentale, (Italy) (top and bottom, before and after
treatment, respectively)
BIOTECHNOLOGY APPLIED TO CULTURAL HERITAGE 81
ª2004 The Society for Applied Microbiology, Journal of Applied Microbiology,98, 73–83, doi:10.1111/j.1365-2672.2004.02429.x

Monitoring techniques are routinely used to carry out
microbial counts based on culturing methods and direct
counts by microscope observation. In fact SEM and optical
microscope observations give important information on the
presence of microflora and their degree of adhesion to the
surface of the treated materials. Rapid analyses are funda-
mental to the monitoring of biorestoration processes, and
fast feedback in the monitoring of microbial activity can be
achieved through bioindicators like ATP content by low
light imaging and dehydrogenase activity (DHA) (Ranalli
et al. 2000, 2003a).
The greatest advantage of bacterial/enzymatic bioresto-
ration, compared with traditional methods (chemical,
physical and mechanical), is that this new method is not
destructive and removes only extraneous substances or
altered compounds from the fresco. Moreover, microbial
cultures (bacteria) have non-specific activity, while the use
of enzymes is highly specific and limited. Another advant-
age is the use of safe micro-organisms (not pathogenic
bacteria or yeasts, not spore-forming bacteria) for both the
operators and the environment. Finally, the evaluation of
costs and convenience has revealed the transferability of
this technological innovation to other fields outside cultural
heritage.
Organic matter (i.e. collagen) on the surface of altered
frescoes can be removed, under appropriate conditions, by
treating the fresco surface with selected aerobic heterotro-
phic microbial cultures (bioaugmentation) and/or selected
pure enzymes. As a consequence of the bacterial activity,
carbon dioxide is released from the pollutants without the
use of toxic compounds, so the method must be considered
environment-friendly.
After optimizing the parameters for the biological process
on the laboratory scale, and subsequently checking the
results on real outdoor samples, we consider that the
application of the method in wider fields would require: (i)
an abundant activated biomass produced under environ-
mentally controlled conditions; (ii) a rapid and sensitive
method to monitor the biological activity and avoid undes-
irable effects; (iii) the correct and complete removal of
residual biological activity.
We are presently researching ways of applying, in situ,
biological methods to several works of art, and to reduce the
limitations and possible risks of such methods. Our results
have confirmed the potential of biorestoration processes as
soft innovative biotechnology. Such restoration, aimed at the
recuperation of degraded artwork, is nondestructive, and
uses micro-organisms and their metabolic activity in associ-
ation with purified enzymes.
The positive results we have achieved have led the
Technical Commission for Restoration (Pisa, Italy) to accept
and approve the use of a P. stutzeri cell culture suspension
and purified Protease enzyme solutions for biological
applications; thus the innovative biorestorationbioaug-
mentation process is already being used.
The biorestored fresco discussed here is an important case
study on the real, true to life scale where soft biotechno-
logies can play a positive role.
We foresee that appropriate micro-organisms will be used
for the remediation and recovery of artistic stonework
altered by atmospheric pollutants like nitrogen and sulphur
oxides, compounds that rapidly accelerate artwork deterior-
ation.
It has been shown that cultural heritage restoration now
has a new and innovative technology: biorestoration through
the use of viable cells of selected bacteria; such restoration
poses no risk at all to human health, or the environment, and
has shown great potential for removing undesirable com-
pounds from the surfaces of works of art.
ACKNOWLEDGEMENTS
The work was carried out with partial financial support from
the Opera Primaziale Pisana, Pisa, Ing. G. Bentivoglio. The
authors wish to thank Clara Baracchini, Soprintendenza
Beni Culturali ed Ambientali, Pisa, Italy; Restorer Sig.
Gianni Caponi of Conservazione and Restauro, Pisa;
Dr Lucia Maiuro, CSIM (for SEM observations). Finally,
the authors are grateful to Barbara Carey for editing the
manuscript.
REFERENCES
Bellucci, R. and Cremonesi, P. (1994) L’uso degli enzimi nella
conservazione e nel restauro dei dipinti. Kermes 21, 45–64.
Bonaduce, I. and Colombini, M.P. (2003) Gas Chromatography/Mass
spectrometry for the characterization of organic materials in frescoes
of the Monumental Cemetery of Pisa (Italy). Rapid Communications
in Mass Spectrometry 17, 2523–2527.
Bonomi, R. (1994) Utilizzo degli enzimi per il restauro di una scultura
in terracotta policroma. OPD Restauro 6, 101–107.
Colombini, M.P., Fuoco, R., Giacomelli, A., Muscatello, B., Baracch-
ini, C. and Caponi, G. (1995) Pulitura a tergo di affreschi:
caratterizzazione chimica del materiale proteico. In Scienza e Beni
Culturali: La Pulitura delle Superfici dell’Architettura. Vol. XI. pp.
227–236. Padova, Italy: Libreria Progetto Editore.
Colombini, M.P., Modugno, F. and Giacomelli, A. (1999) Two
procedures for suppressing interference from inorganic pigments in
analysis by gas chromatography-mass spectrometry of proteinaceous
binders in painting. Journal Chromatography Acta 846, 101–111.
Colombini, M.P., Bonaduce, I. and Gautier, G. (2003) Molecular
pattern recognition of fresh and aged shellac. Chromatographia 58,
1–8.
Cremonesi, P. (1999) L’uso degli enzimi nella pulitura di opere
policrome. In I Talenti ed. Cremonesi, P. pp. 5–11. Padova: Il Prato.
Feller, R.L., Stolow, N. and Jones, E.H. (1985) On Picture Varnishes
and Their Solvents. Washington, DC: National Gallery of Art.
82 G. RANALLI ET AL.
ª2004 The Society for Applied Microbiology, Journal of Applied Microbiology,98, 73–83, doi:10.1111/j.1365-2672.2004.02429.x

Makes, F. (1988) Enzymatic consolidation of the portrait of Rudolf II
as Vertumnusby Giuseppe Arcimboldo with a new multi-enzyme
preparation isolated from Antarctic Krill (Euphasia superba). Acta
Universitatis Gotheburg 23, 98–110.
Principi, P., Ranalli, G., da Borso, F., Pin, M., Zanardini, E. and
Sorlini, C. (2003) Microbiological aspects of humid husk compost-
ing. Journal of Environmental Science and Health,38B, 645–661.
Ranalli, G., Chiavarini, M., Guidetti, V., Marsala, F., Matteini, M.,
Zanardini, E. and Sorlini, C. (1996) The use of microorganisms for
the removal of nitrates and organic substances on artistic stoneworks.
In Proceedings 8th International Congress on Deterioration and
Conservation of Stone ed. Riederer, J. pp. 1415–1420. Berlin: Mo
¨ller.
Ranalli, G., Recchiuti, J.E. and Grazia, L. (1998) Bioluminescence and
impedance monitoring to detect the activity of starter cultures
during frozen storage. Annali di Microbiologia ed Enzimologia 48,
169–180.
Ranalli, G., Matteini, M., Tosini, I., Zanardini, E. and Sorlini, C.
(2000) Bioremediation of cultural heritage: removal of sulphates,
nitrates and organic substances. In Of Microbes and Art. The Role of
Microbial Communities in the Degradation and Protection of Cultural
Heritage ed. Ciferri, O., Tiano, P. and Mastromei, G. pp. 231–245.
New York: Kluwer Academic/Plenum Publishers.
Ranalli, G., Belli, C., Baracchini, C., Caponi, G., Pacini, P., Zanardini,
E. and Sorlini, C. (2003a) Deterioration and bioremediation of
frescoes: A case-study. In Molecular Biology and Cultural Heritage
ed. Saiz-Jimenez, C. pp. 243–246. Lisse, the Netherlands: A.A.
Balkema Publishers.
Ranalli, G., Zanardini, E., Pasini, P. and Roda, A. (2003b)
Rapid biodeteriogen and biocide diagnosis on artwork: a biolumi-
nescent low-light imaging technique. Annals of Microbiology 53,
1–13.
Saiz-Jimenez, C. (1995) Deposition of anthropogenic compounds on
monuments and their effect on airborne microorganisms. Aerobio-
logia 11, 161–175.
Tiquia, S.M. (2002) Evolution of extracellular enzyme activities during
manure composting. Journal of Applied Microbiology 92, 764–775.
Wolbers, R. (2000) Cleaning Painted Surface: Aqueous Methods.
London: Archetype.
BIOTECHNOLOGY APPLIED TO CULTURAL HERITAGE 83
ª2004 The Society for Applied Microbiology, Journal of Applied Microbiology,98, 73–83, doi:10.1111/j.1365-2672.2004.02429.x