Sampling techniques and genomic analysis of biological material from artworks

1 INTRODUCTION

Forensic analysis of artworks is used extensively to gain insights into methodology, technique and provenance [1], and to facilitate restoration efforts [2]. However, most scientific analyses of artwork have focused on direct physical examination using spectroscopic or analytical chemical techniques [2-6]. Advances in molecular biology have facilitated the development of approaches to study ancient microbial and human DNA sampled from archeological sites and burial sites [7-9] as well as environmental DNA (eDNA) [10]. These studies have provided significant insights regarding the genetics of human speciation, global migration, and health history. In principle, similar tools could also be used for the analysis of microbial (or human) DNA sampled from artworks to facilitate restoration and preservation, as well as provenance and attribution. However, systematic studies of DNA sampling and analysis from artworks have not been extensively reported. Some examples of sampling technique publications include a microbial biofilm [11] and a comparison study with swabs, scraping, eraser rubbings, and micro-aspiration [12].

Our overall aim was to advance methodologies to facilitate the sampling of DNA from artworks or other cultural/historical artifacts. We compared three sampling techniques to recover biological material suitable for microbial DNA sequencing, the human RNase P gene, and Y-STRs (Y-chromosome short tandem repeats) from metal points and chalk drawings on laid paper. We first compared the yields of DNA from samples obtained using three sampling methods with increasing degrees of invasiveness: swabbing with dry or wet cotton applicators, wet vacuuming with a commercial device, and core punch biopsies. The wet vacuuming technique was postulated to provide human DNA information from within the artwork in addition to the surface contaminants that would only be recovered from swabbing. The 16S rRNA gene, specific for targeting and amplifying bacterial DNA from the isolated genomic DNA (gDNA) of low biomass samples, was sequenced. Our results provide a methodological framework for collecting samples from artworks on paper to obtain biological material for microbial genomic DNA analysis and total copies of the human RNase P gene.

2 MATERIALS AND METHODS

2.1 Artwork

Since our aim is to develop and benchmark minimally invasive and non-destructive sampling techniques and compare them with more invasive and potentially destructive methods, we focused on centuries-old artworks on paper that were judged to hold limited historical or cultural significance as they might be damaged or destroyed in the course of the work. The five artworks used in this study were acquired from reputable e-commerce platforms or from private collections and to our knowledge had not been recently displayed. All works were obtained unframed and were stored in a portfolio folder with plastic sleeve separators at room temperature in the dark. The artworks studied included two metal points and three chalk drawings. Metadata (if known) is provided for each piece including the artist's name and time period, the support matrix and medium, and the size and paper thickness (Table 1). Prior to sampling, high-resolution digital photographs were recorded of the front and back of each drawing using a Sony ILCA-77 M2 camera in raw format (6000 × 4000 pixels) with a Sigma 35-mm f/1.4 DG HSM Art Lens (SIGMA, Kanagawa, Japan) under a variety of lighting conditions using a Bower SFD-RL71 Large Digital LED Ring Light (S. Bower, Inc., Queens, NY, USA). Images were scaled in Adobe Photoshop 2022 (Adobe, San Jose, CA, USA) based on a ruler photographed together with the artwork. A lightbox (LitEnergy A4S LED Copy Board Light Tracing Box, LitEnergy, Guangdong, China) was particularly useful for backlighting and highlighting surface irregularities, textures, watermarks, and translucent elements within the artwork.

TABLE 1. Artworks used in study.

Unique ID	E1	E4	K1	K3	K9
Type of object	Metal point drawing	Chalk drawing	Silverpoint drawing	Chalk drawing	Chalk drawing
Material/technique	Silverpoint on prepared paper	Red chalk drawing mounted onto board	Silverpoint on prepared paper with ink and blue wash	Red chalk drawing on laid paper	Red chalk with white highlights on laid paper
Support/ground	Laid paper	Laid paper	Laid paper	Laid paper	Laid paper
Measurement [height × width]	180 × 140 mm	205 × 112 mm	133 × 102 mm	296 × 212 mm	260 × 172 mm
Thickness (μm)	150 μm	300–325 μm	Not measured	Not measured	Not measured
Inscription/markings	Possible watermark	None obvious	Honeycomb or pinecone	Encircled cross with “V/B” on either side; possible honeycomb between the letters; possible tiny crucifixion above	None obvious
Distinguishing features	Wavy-laid line patterns may indicate water damage or poorly laid paper; white ground is thick and heavily cracked	Subject's right hand appears anatomically anomalous	Drawing is silhouetted, cut out, and integrated into a second sheet of laid paper	“And Sacchi” inscribed on verso	Thick, wrinkled, wavy paper; ~3 cm chain lines; ~ 1 mm laid lines
Title	“Old master drawing”	“Allegory of Music”	“St. John in a Classical Robe by a Pillar”	“Man Caught in the Wind”	“Seated Gentleman at Writing Table”
Subject	Face of a young boy; similar to “Head of a Young Boy” by Donato Creti		Classical contrapposto position	Possible study for celebrant in “Birth of Pope Urban VII” by Sacchi
Date or period	c.1500 C.E. ± 100	unknown	c.1501 CE	c. 1625 CE	c. 1740 CE
Maker	Unknown (— – —)	Unknown (1690–1710)	Filippo Lippi (1457/58–1504)	Andrea Sacchi (1599–1661)	Charles J. Flipart (1721–1796)
Photo versos
Photo rectos
Photo versos Backlight
Photo rectos Backlight RB or Photo rectos Raking light RR	RR			RB
Photo versos Raking light

2.4 Sampling

A systematic approach to sampling similar areas of the artworks using different sampling techniques was defined. Areas sampled were defined using virtual triangular zones on the artwork for the purpose of collecting samples from predefined regions. Each of the triangular zones was assigned a letter label. When orienting the artwork for display, the two upper corners (left and right) and an equal area of the artwork in the center were sampled. Using a template of precut plastic and the natural right angle of the artwork, an isosceles right triangle was drawn onto the artwork to guide the prototype sampling. For actual sampling procedures, a template was drawn onto an acetate film sheet placed between a lightbox and the transilluminated drawing artwork. Two equal isosceles right triangles (together making a square) were used to define the sample area of the center of the artwork. The isosceles right triangle at the upper left corner was divided in half to create two smaller right triangles. The leftmost of the two triangular regions from the upper left was labeled “A” and the rightmost triangular portion was labeled “B.” The upper right's corner isosceles right triangle was labeled “C.” The square in the center of the artwork was divided into two right isosceles triangles, divided from the upper left corner to the lower right corner. The lower isosceles right triangle was divided in half to create two smaller right triangles “D” and “E.” The upper isosceles right triangle was labeled “F.” (Figure 1). Triangles A, B, D, and E were measured to be 4.25 × 4.25 × 6 cm (perimeter = 14.5 cm; area = 9 cm²); triangles C and F were measured to be 6 × 6 × 8.5 cm (perimeter = 20.5 cm; area = 18 cm²). This geometric approach to sampling facilitated systematic collection of samples from various regions of the artwork, including the upper corners, the central area, and smaller regions within the central square for the different sampling techniques. The use of labels for each triangle ensured that the collected samples could be properly identified and associated with specific areas on the artwork for documentation and further analysis. Due to the different sizes of the artworks, if this configuration was not possible, then a reasonable version of it was applied.

The collection techniques employed included surface swabbing, core punches taken with a disposable biopsy punch that cuts through the solid support and media, and wet vacuuming the surface, which theoretically can extract samples from within the solid support and the media. Swabbing involved the use of sterile swabs to gently rub and collect samples from the surfaces of artworks. This relatively noninvasive technique allowed for the retrieval of surface materials, potentially including biological samples. The use of disposable biopsy punches allowed for the core punches of the artwork, both support and media. This technique was suitable for extracting samples from deeper layers or within the solid support material itself. Wet vacuuming is a specialized technique employed to extract samples from the solid support and media of the artworks. It theoretically has the capacity to collect material not only from the surface but also from within the solid support and media. While wet vacuuming retrieves materials from deeper layers of the support, the biological stratigraphy of those materials will be lost. The nomenclature used for each sampling is described in Table 2.

TABLE 2. Nomenclature of sampling from each triangular area.

Triangles A and D	Swabs
1	Front of artwork
2	Back of artwork
a	Wet swab
b	Dry swab
Pu-RB	Puritan swab reagent blank

Triangles B and E	Core punches
1	1.2 mm Harris punch
2	2 mm Integra® Miltex® punch
3	4 mm Integra® Miltex® punch
4	6 mm Integra® Miltex® punch
a	Punch 1
b	Punch 2
c	Punch 3

Triangles C and F	M-Vac filters
a, b, or c	Filter quadrants
a1	Filter quadrant, flash freeze

2.4.1 Swabbing techniques

A common technique used to collect samples from artworks is the “double-swabbing technique” [14, 15]. For the double-swabbing technique, a drop of molecular biology grade water, or ~ 50 μL, was added to one swab. The surface of the area to be sampled was rubbed with the swab while swirling. Then, the same surface area was sampled with a second dry swab using the same technique. After air drying, under conditions to avoid environmental contamination, both sampling swabs were placed in a sterile tube for storage. As a negative control, a drop of water was added to a swab and a second dry swab was also included. Depending on the artworks tested and the regions sampled, the double-swabbing technique, which involves the application of water to the artwork, was generally considered to be AS4–AS6.

2.4.2 Core punches

The collection of core punches from artworks is a destructive technique that removes samples from specific areas for analysis, including both the surface of the artwork (i.e., the art medium) and from the support (i.e., paper, canvas, wood support). Core punches in this study were hypothesized to contain all available biological material that could be potentially extracted from a given full-thickness area and were used as controls. Supplies needed to obtain core punch samples include the 1.2-mm Harris Uni-Core punch (MilliporeSigma, Burlington, MA) or the disposable biopsy punches (2, 4, or 6-mm diameters) (Integra LifeSciences Production Corporation, Mansfield, MA). The targeted area was identified, the artwork was placed on the barrier, and even downward pressure was applied at a right angle. While stabilizing the artwork with the free hand, the punch was advanced with a semi-circular rotation until the tip passed all the way through the artwork's support. After coring was complete, the punch/cutter tip was placed over a sterile storage tube and the plunger was depressed to drop the core sample into the tube. For some samples, sterile PEEK tubing (readily available in the laboratory) (Amersham Biosciences AB, Uppsala, Sweden) was used to push the core sample into the sterile storage tube. For quality control, collection of samples on the edges or folds of the artwork were collected to serve as background controls. Core punches were classified as AS7–AS9, depending on the punched area of the artwork (whether central or peripheral) and the size of the punch. For example, a punch on the periphery of the artwork was assigned AS7, but the same size punch in the center of the art was assigned AS9, the maximum AS possible.

2.4.3 Wet vacuum sampling

The M-Vac® System (M-Vac Systems, Inc., Sandy, UT) [16] is a wet vacuum system that dispenses sterile Surface Rinse Solution (SRS) (M-Vac Systems, Inc.) onto the artwork and uses a vacuum mechanism from a sampling head to collect not only the SRS, but also any suspended cellular material. This technique is advantageous for sampling larger areas and from various substrates. With the fluid barrier's impermeable polyethylene coating side up, its surface was cleaned with 70% ethanol and then allowed to air dry. The artwork was placed on the fluid barrier and the M-Vac instrument was powered on. The vacuum operator collected biological samples using a unidirectional sampling method. After the collection of the SRS containing potential biological material, a Nalgene 250-mL filter unit fitted with a 0.45-μm PES filter disc was connected to a vacuum pump. The SRS was slowly poured into the middle of the filter unit with the vacuum pump powered on. The filtered SRS was then poured back into the sample collection bottle, swirled to suspend any cellular material on the sides of the bottle, and filtered again. The use of double filtration enhances the concentration and purity of the collected cellular material. The Nalgene 0.45-μm PES filter was allowed to dry in a laminar flow hood with the lid to the Nalgene unit slightly ajar, ensuring a controlled and sterile environment. Once the filter was fully dry, the filter disc was carefully removed using a sterile scalpel and clean tweezers. The filter was then cut into quadrants, three quadrants were placed in a single sterile tube and stored at room temperature and the fourth quadrant was placed in a separate sterile tube and subjected to flash freeze [17] and stored at −80°C. Depending on the artworks tested and the region sampled, an AS7–AS9 was assigned for the M-Vac wet vacuuming technique. There was undoubtedly alteration to the artwork merely by adding liquid directly onto the solid support and media.

3 RESULTS AND DISCUSSION

This study reports an approach for sampling artwork for biological and genomic studies. We focused on centuries-old metal point and chalk drawings on paper that were acquired and utilized specifically for the study. We first recorded high-resolution images before any sample collection and used the images to define and describe the areas for sample collection. Since the upper corners of drawings tend to be handled most frequently by the artist, those areas were emphasized for sampling (areas A, B, and C). Further, since the center portion of the artwork is normally not handled post-production, this area was also chosen for sampling (areas D, E, and F). For the collection of samples, we compared double swabbing, core punching (1.2, 2, 4, and 6 mm), and M-Vac wet vacuuming. After samples were collected, gDNA was isolated using the QIAGEN DNeasy PowerSoil Pro Kit [25]. Finally, full-length 16S (V1–V9) sequencing was carried out using the PacBio Sequel II/IIe system with SMRT Cell 8 M on a HiFi sequencing run. In addition to collecting samples from the artwork, we also isolated and sequenced DNA from negative controls, the scientists handling the artworks, and the art curators to define contaminants from the environment and recent human handlers (Figures S1 and S2).

To determine if there were major differences in the DNA yields from the three sampling approaches, we first analyzed the total isolated DNA content from swabs, core punches, and M-Vac filters. Total DNA (ng/μL) levels obtained from swabs, pooled core punches (1.2, 2, 4, and 6 mm), or the M-Vac filters were compared. There were no significant differences between the yields obtained from the swabs and the pooled core punches; however, a significant increase in DNA yield was observed from the M-Vac filters (Figure 2A). Of note, when assessing the total read count estimates based on the full-length (V1–V9) 16S rRNA gene copy number after quality filtering of the sequencing data, similar trends in DNA concentration were observed (Figure 2B).

We then aimed to determine the differences between the DNA yields obtained from the four different core punch sizes (1.2-, 2-, 4- and 6-mm). A t-test was performed using the 1.2-mm core punch as a reference to compare against the mean read count of the other core punches at 2-, 4-, and 6-mm sizes. The results showed that only the higher surface area of the 6-mm core punch led to significantly more bacterial DNA (Figure 3). As with the M-Vac wet vacuuming technique, the higher AS7 to AS10 values for the 4- and 6-mm core punches may produce irreparable damage to the artwork, as the recovered medium is destroyed during DNA isolation. Therefore, the core punch technique should only be used if preservation of the artwork is not necessary or if restoration is possible by replacing the medium where the core punch was sampled. Repairing damaged artworks is a common practice. The reparation of core punches would be relatively easy since the exact size of the punch is known and the tools used for the sampling punch would be readily available to prepare a replacement core piece of the medium. Alternatively, in some cases, sampling can be targeted to areas beneath the framed regions of the artworks.

To determine if the location of the collected samples on the drawing affected the overall microbial readout obtained, we compared the total read counts for the full-length 16S rRNA obtained from the corners and the center of the artwork (Figure 1A). Comparisons between the corner and center samples for each collection technique (swab, core punch, and M-Vac filter) showed no significant difference in the mean read count (Figure 1B). These results suggest that while different collection approaches may improve the recovery of DNA, the microbial DNA content readout is similar irrespective of the localization within the artwork. This result was surprising since we expected more bacterial contamination in the corners and on the edges of the drawings. Furthermore, as noted later, there were no significant differences in the types of bacterial taxa found in the different locations, although the detection of diverse taxa was dependent on the sampling technique. These results suggest that the recovery of microbial material may be driven by a shared exposure to the environmental conditions across the medium (the different locations) or possibly the manufacturing process of the laid paper. Future studies might include studying similar laid paper from the same time period to rule out the input of microbial material from the medium even after its extensive bleaching, washing, and processing during the manufacturing process.

To define the bacterial profiles that are associated with each of the collection techniques, we determined the relative abundance of all bacteria identified in the full-length 16S sequencing data. A total of 150 operational taxonomic units (OTUs) were detected in the sequenced samples and are presented in an abundance plot (Figure 4). Interestingly, the number of OTUs varied based on the sample collection technique. Only 12% of the total OTUs observed were detected by all three sampling techniques. The wet vacuum technique detected 53% of the total OTUs observed, which was the highest among the three sampling techniques that were tested. Hence, these findings strongly indicate that sampling a larger surface area and retrieving sample from within the layers of the artworks will undoubtedly lead to the identification of more bacterial species. As shown by Alpha diversity index (Figure 6A), wet vacuum significantly yielded higher species richness compared with other sampling techniques.

Relative abundance of all bacteria (150 OTUs) identified in 16S full-length data showed differences in bacteria profiles based on the sample type collected (Figure 5). The top 15 most abundant taxa identified in the samples collected from the different techniques ranged from environmental-associated to host-associated bacteria (Figures S3–S5). Of note, there were some observable differences in taxa associated with specific collection techniques, suggesting that each technique may provide access to different regions of the artwork and different biological readouts. Different species of bacteria were visibly abundant in the different sample collection techniques, for instance, Turicibacter sp. from swabs, Bacillus sp. from M-Vac filters, and Anaerococcus sp. from core punches. Of note, human commensal skin-associated bacteria, Cutibacterium acnes and Staphylococcus hominis [26], potentially originating from artwork handlers were abundant in M-Vac filters and core punches. In addition, these results may also suggest that combinatorial sample collection techniques may provide a more complete microbial profile of the tested artwork.

We then assessed the overall microbial diversity changes among the collection methods. The Alpha diversity index (Chao1 and Shannon) showed a significant (t-test-based p < 0.05) increase in species richness of samples collected using M-Vac filters compared with swabs and core punches, with no significant changes in diversity observed between the swabs and core punches (Figure 6A). Beta diversity, a measure of the variability in the community composition, showed that the overall microbiome of each of the collection techniques was similar based on sample clustering (Figure 6B). Using a PERMANOVA test, we aimed to define the association of microbial composition with variables such as sample collection techniques and the location of the sample collected. The test showed a significant difference (R² = 6% and p-value 0.01) in the artwork microbiome based on the sampling technique (M-Vac filters, core punches, and swabs), as observed in Figure 5 and Figure 6. However, the location of the sample collected within the artwork did not significantly contribute to differences in the artwork microbiome (R² = 3% and p-value 0.06) (Figure 6C).

Bacteria from the genera Lactobacillus, Enterococcus, Pseudomonas, Bacillus, and Streptococcus identified in this study are known to produce hydrogen peroxide mainly for antimicrobial defense [27]. The presence of large quantities of hydrogen peroxide in paper can lead to bleaching and degradation of cellulose fibers, decreasing the integrity of the paper [28]. Similar to hydrogen peroxide, other primary and secondary metabolites from bacteria may alter the structural integrity of the visual features of art [15]. Thus, understanding the microbial signatures present in art may lead to new preservation strategies.

Lastly, we assessed the human DNA by gene RNase P and Y-STR analysis among the collection methods. Log copies per milliliter were tested via quantitative RT-PCR for the RNase P gene. No statistical difference was detected between the groups by one-way ANOVA (Figure 7). To determine if STRs could be detected, we isolated DNA from a second sampling via double swabs from artworks K1, K3, and K9; these samples were amplified for Yfiler Plus. Some Y-STR results were obtained from each artwork, but none produced a full single-source profile. The Y-STRs for K1 resulted in a major peak at 14 Y-STRs with some low-level peaks observed at four of these Y-STRs suggesting a mixture of three or more males; the larger Y-STRs (300 bp) did not amplify. The Y-STRs for K3 resulted in a major peak at three Y-STRs with some low-level peaks observed at five of the Y-STRs suggesting a mixture of two or more males. The Y-STRs for K9 resulted in low-level peaks at 11 Y-STRs with a mixture of two or more males (Figure S6).

4 CONCLUSIONS

This study provides a framework for sampling biological materials for DNA testing from artworks that may be useful for provenance, preservation, and authentication. We show that both microbial DNA and human DNA can be recovered from artworks on paper. As is also observed with forensic evidence, recovery of DNA from artworks can be challenging. Utilizing techniques of forensic science, we explored different sampling and extraction techniques to measure the levels and nature of DNA from artworks. The invasiveness of sampling techniques must be balanced with the potential for damage that may be caused to the artworks in light of the value of the information that can be recovered. The relative efficacy of various invasive and less invasive collection techniques to generate biological data from artwork was explored. In general, less invasive techniques resulted in less DNA recovery.

The wet vacuum approach was observed to alter the structural support of the artwork, the double-swab technique was sometimes observed to smear and remove some of the chalk from the red chalk drawings, and the core punches were most invasive as expected. All three techniques were efficient in providing organic material for DNA isolation and sequencing. Of note, our data did not find any significant differences in the mean read count (microbial biomass) between the corners and the center of the artwork, suggesting that outside regions away from the main focus of the artwork may be productive areas to sample to reduce damage to the artwork while still obtaining genetic material. Overall, our results show that double swabbing or small core punches provide an efficient route for sample collection for future studies of microbial material from artwork. However, it is still to be defined if these approaches can be used for the isolation of human DNA in artwork. Since in most cases, it is essential to utilize the most noninvasive, non-destructive sampling technique possible to preserve the integrity of the artwork being sampled, our results suggest that the double-swabbing technique should be the primary choice for sampling artwork for genomic analysis.

FUNDING INFORMATION

Financial support was provided by the Richard Lounsbery Foundation and the Achelis and Bodman Foundation.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.