An explorative study of polymers for 3D printing of bioanalytical test systems

Abstract

BACKGROUND:

The 3D printing is relevant as a manufacturing technology of functional models for forensic, pharmaceutical and bioanalytical applications such as drug delivery systems, sample preparation and point-of-care tests.

OBJECTIVE:

Melting behavior and autofluorescence of materials are decisive for optimal printing and applicability of the product which are influenced by varying unknown additives.

METHODS:

We have produced devices for bioanalytical applications from commercially available thermoplastic polymers using a melt-layer process. We characterized them by differential scanning calorimetry, fluorescence spectroscopy and functional assays (DNA capture assay, model for cell adhesion, bacterial adhesion and biofilm formation test).

RESULTS:

From 14 tested colored, transparent and black materials we found only deep black acrylonitrile-butadiene-styrene (ABS) and some black polylactic acid (PLA) useable for fluorescence-based assays, with low autofluorescence only in the short-wave range of 300–400 nm. PLA was suitable for standard bioanalytical purposes due to a glass transition temperature of approximately 60°C, resistance to common laboratory chemicals and easy print processing. For temperature-critical methods, such as hybridization reactions up to 90°C, ABS was better suited.

CONCLUSIONS:

Autofluorescence was not a disadvantage per se but can also be used as a reference signal in assays. The rapid development of individual protocols for sample processing and analysis required the availability of a material with consistent quality over time. For fluorescence-based assays, the use of commercial standard materials did not seem to meet this requirement.

Keywords

3D printing cell adhesion autofluorescence polylactic acid PLA polyethylene terephthalate glycol PETG acrylonitrile-butadiene-styrene ABS thermoplastic polyurethane elastomers TPU medicine pharmaceutical

1 Background

In the research and development there is a high demand for innovative concepts with which complex bioanalytical systems can be developed or adapted to experimental conditions. 3D printing technology is used for this task. The first contributions to 3D printing technology were made in the early 1980 s, by Charles Hull [1]. Since then 3D printing (3DP), also called additive manufacturing (AM), rapid prototyping (RP), or solid-freeform technology (SFF) is becoming increasingly relevant in industry and science. That is especially the case if many optimization steps are necessary for the development [2]. With this technology, three-dimensional, functional models can be produced independently by applying a material layer by layer. There are a variety of materials for that can be used for diagnostics applications and 3DP, with polylactic acid (PLA) and acrylonitrile-butadiene-styrene (ABS) being the most widely used [3, 4]. Other materials include PET (polyethylene terephthalate) and thermoplastic polyurethane (TPU).

Since 3DP enables the production of objects with different geometries, it also offers advantages over conventional manufacturing processes. In addition to drug production and customization (incl. personalized-dose medicines, sophisticated and complex dosage forms, on-demand manufacturing [5 –7]), the technology is also used in pharmaceutical drug manufacturing and the development of drug delivery systems with non-traditional geometries for pharmaceutical applications (e.g., drug-eluding implants) [8 –10]. For biomedical applications, devices were printed for the synthesis of nanoparticles, calorimetry, cell growth monitoring, tissue regeneration, blood evaluation, detection of pathogenic bacteria, detection of dopamine, forensics, or the detection of DNA and protein cancer biomarkers [7 , 11–15]. Often equipment for sample preparation and novel reaction vessels, lateral flow and microfluidic test systems are important for (smartphone-based) point-of-care testing, for the processing of (blood) samples, for the analysis of platelet aggregation and thrombosis kinetics as well as DNA and protein arrays are produced by 3DP [11 , 15–20].

The development of PLA dates back to 1932. As a pure substance, PLA has a glass transition temperature (T_g) of circa 55°C and a melting temperature (T_m) of approximately 180°C [21]. PLA is made from lactide and is therefore hydrophobic. It is one of the filaments that can be used for many medical applications like tissue engineering and orthopedic implants because of its ability to degrade to lactic acid [15, 22].

As ABS can be dissolved in acetone, it was used, for instance, as a template for the production of 3D microchannel networks surrounded by poly(dimethylsiloxane) (PDMS) [23]. The material is characterized by a high resistance to cold and heat and is very resistant to various chemicals [24].

PET is the most frequently used thermoplastic polymer in many areas, which is also ideally suited for biomedical applications due to its high mechanical resistance, food safety, high chemical resistance and biocompatibility [25 –27]. In the glycol copolymerized version (PETG), this material has good properties for fused deposition modeling (FDM) 3DP, with a lower melting temperature and viscosity, and combines the advantages of easy handling of PLA with the high durability and temperature resistance of ABS.

In addition to relatively solid plastics, which should retain their shape after printing and withstand mechanical forces, there are flexible materials such as urethane-based TPU. Flexible thermoplastic elastomers offer the advantage of yielding to compressive or tensile forces even at room temperature (RT). Therefore, they are suited as damping and sealing functions or as a flexible support structure for creating artificial skin or vessels [28, 29].

Due to technological progress and competition, the FDM procedure has become relatively inexpensive. Modern 3DP systems also achieve high print quality and accuracy down to 25 μm range under certain conditions. As defined by Beauchamp et al. (2017) 3D printed structures for millifluidic (>1 mm), sub-millifluidic (0.5–1.0 mm), large microfluidic (100–500 μm), or truly microfluidic (smaller than 100 μm) applications are feasible [30]. This allows projects to be carried out reproducible even in very small dimensions. Whilst the resolution of the 3D printer can only compete to a very limited extent with high-end lithography regarding surface roughness and size, researchers use these tools to design prototypes of microfluidic devices [2, 31].

Important factors during the production of polymer molds with plastics are their melting behavior, crystallinity and heat resistance. Melting behavior and crystallinity are important parameters for producing a product with stable mechanical properties [32]. The heat resistance is important to be able to use the material under certain conditions (e.g., DNA hybridization at high temperatures) or to autoclave the material. Also, the effects of additives have to be considered [33]. It is known for PLA that the addition of dyes has an influence on the crystallization rate and tensile strength [34].

After printing, material properties such as autofluorescence, melting behavior and stability (mechanical, chemical) are decisive for the optimal printing process and the applicability of the product. The autofluorescence of a material exists when a fluorescence is emitted after illumination with light of a certain wavelength. This fluorescence can limit the applicability of a material for qualitative and quantitative bioanalytical applications if the autofluorescence is in the range or higher than the biomarker signal to be measured [15]. The behavior of biomaterials needs to be tested in a broad spectrum of in vitro experiments [35].

Albeit there are numerous publications in the scientific literature dealing with the application of 3DP, we wondered about the limitations on the use of commercial filament materials and whether there are materials that should be preferred. Here we report on our experiences with the applicability of commercially available filaments. We purchased fourteen colored, transparent and black materials (Fig. 1, top row). Colored materials were purchased because the compounds causing the color to have an impact on the physical properties of the material [34]. For bioanalytical assays, an exact knowledge of the material properties is vital.

Fig.1

Experimental design of the study. In our studies we investigated various 3D printing filaments (top) for their suitability for the production of bioanalytical test systems. We used differential scanning calorimetry (DSC) and fluorescence spectroscopy to analyze thermal and spectral properties and functional assays to investigate possible applications such as cell and bacterial adhesion, fluorescence in situ hybridization (FISH) and DNA detection.

Filament materials of the same color are not easily distinguished by their optical appearance, although they give the impression that they should have the same properties. Since fluorescence-based processes are sensitive and easy to use, we investigated which filaments have an autofluorescence to conclude if they can be used in a certain spectral range. Then selected filaments were used to produce bioanalytical devices for the detection of single-stranded nucleic acids as DNA capture assays and functional studies using adhesion assays of human cells, fluorescence in situ hybridization (FISH) as well as biofilm formation.

2 Material and methods

2.1 Characteristics of the benchtop FDM 3D printers

To produce 3D printed devices, we used commercially available benchtop FDM 3D printers with different designs from different price segments. The main machine was a Raise3D Pro2 dual-nozzle printer equipped with a 0.4 mm nozzle, a direct feeder system for 1.75 mm filaments and a fully enclosed, large print chamber with heated bed. This printer was chosen because of its ability to print many materials up to temperatures of 300°C and its particularly high accuracy with a 0.78125 μm positioning resolution on X/Y axis and a minimal layer thickness of 10 μm.

For comparing prints in lower resolution, the low-cost 3D printer Anycubic I3 Mega printer (0.4 mm nozzle) was used. It has an open and accessible architecture, a heatable, movable build plate and a simple Bowden extruder.

2.2 3D printed samples and devices

Fourteen filament polymers (PLA: n = 9, ABS: n = 1, PETG: n = 2, TPU: n = 2; 1.75 mm diameter) from five manufacturers were used to produce sample material and devices. Detailed information on the source and properties can be found in Table 2. The production of 3D objects from a digital model involves the processing steps planning, drawing a digital design, translation into machine commands and execution of the printing process [1].

Our digital 3D models were created using the 3D CAD/CAM software Fusion 360 (Autodesk, USA) and converted as a triangular surface mesh into a *.STL (STereo-Lithography) file. The slicing to executable machine movements was done for the Raise3D Pro2 with the software ideaMaker (Raise3D, China) and for the Anycubic I3 Mega with the open-source software Cura (Ultimaker, Netherlands). The final print information was stored then in *.GCODE files. Both the *.STL and *.GCODE files used can be found in the supplement (Supp. 3D).

For the material studies we created and printed different 3D objects, which are represented in Fig. 2. For our functional studies we selected just four filaments (see Table 2: materials C, D, H, I) that were best suited based on further results described below.

Fig.2

CAD-created and printed 3D constructs involved in this study. This figure shows the three 3D objects created within this study (as digital model, scheme and printed model). They were created with the CAD software Fusion 360 (Autodesk) and produced with the examined polymers by 3D printing. The dimensions of the constructs are drawn accordingly. A: representative test slide for autofluorescence measurement, 10 replicates for each material (n = 14); B: 10-well slide for cell adhesion, FISH and biofilm tests; C: 2-chamber chip for DNA capture assay; B/C were printed from 4 selected PLA (black/ matt black, Das Filament; black, Elegoo) and ABS (black, Tronxy) filaments (see Table 2: materials C, D, H, I).

Ten replicates of slides (76×26×0.5 mm) labeled by A to N were printed to analyze the variation of autofluorescence of each material (Fig. 2A). Since the melt process may affect the spectral properties of the polymers [36] all materials were analyzed after printing.

Functional assays: Experiments on cell adhesion, bacterial biofilm formation as well as FISH experiments were performed using printed 10-well slides with a growth area of about 37.8 mm² in each well (Fig. 2B). For detection of short fluorescent labeled single-stranded DNA lateral flow experiments were performed in a custom-made two-chamber chip (Fig. 2C). Two punched out circular filter paper and a rectangular filter strip promoted the liquid flow in the chamber.

2.3 Analysis of autofluorescence and spectral characteristics

Autofluorescence was investigated by printing (print settings and 3D models can be found in the supplement Supp. 3D) 3D objects with the size of a standard slide (Fig. 2A). Ten technical replicas of each polymer were produced to assess the variance. The fluorescence spectra were measured with the FluroMax-4 (Horiba, Japan) spectrofluorometer. Test slides were placed vertically and centrally in a solid sample holder and fixed without cover glasses. The focal point was set on the sample and the front face was placed 30° to the excitation beam to prevent the excitation light from entering the emission slits and to avoid increasing interference from stray light.

The measurement was carried out automatically as a batch experiment in six steps. After successive irradiation with excitation light of different wavelengths (350, 405, 488, 561, 640, 750 nm) fluorescence light was captured by the emission monochromator. All excitation wavelengths were chosen to correspond to laser lines of confocal laser scanning microscopes to investigate the most common spectral ranges that play a role in fluorescence-based experiments (e.g., DAPI, FITC, Cy3, Cy5). All measurements were done using entrance and exit slit widths of 5 nm and an increment of 1 nm for emission detection.

Spectra data were analyzed as described before [37] using the R statistical programming language [38] using the R Commander GUI [39] and the RKWard GUI/IDE [40]. Principal component analysis (PCA) of normalized fluorescence spectra was performed using the prcomp() function to identify fluorescence wavelengths at which maximum variation between the filaments was observed and to identify similarities between the filaments and visualized as biplot [41]. The analysis is available in the supplement (Supp. PCA).

2.4 Differential scanning calorimetry

All samples were characterized by differential scanning calorimetry (DSC). The DSC experiments were performed with a DSC 1/700 and a DSC 1/500 from Mettler Toledo in closed Al pans (40 μL) under nitrogen atmosphere. The calorimeter was calibrated with n-octane and gallium as reference materials at heating rates of 1, 2, 5, 10 and 20 K·min^-1. After calibration, temperature and heat flux were verified with n-decane, gallium and indium. For these materials, the fusion properties were in perfect agreement with the data listed in NIST [42]. The characteristic temperatures derived from the DSC curves are extrapolated; melting was determined to be endothermic and freezing to be exothermic. Cold crystallization effects of glassy substances showed exothermic signals during the heating process. Several temperature regimes described in Section 3.2 were used to investigate the thermal behavior of the filaments. The mass of the samples was about 6–16 mg.

2.5 Cell adhesion assay

Cell adhesion was tested on 10-well slides (replicas) printed from PLA (black/ matt black, Das Filament; black, Elegoo) and ABS (black, Tronxy). Before usage, the printed slides were incubated 2 h in 70 % ethanol at RT. Afterwards, the slides were rinsed in 50 mL water and dried at RT.

The three commonly used cell lines SiHa & CaSki [43], HeLa [44] were incubated (5 vol.-% CO₂, 37°C) in Dulbecco's modified Eagle’s medium (DMEM; 10 % fetal calf serum (FCS), 2 mM L-glutamine, 100 U·mL^-1 penicillin/ streptomycin). Cells were seeded at 0 (medium control) 1.25·10³, 2.5·10³, 5.0·10³ and 1.0·10⁴ cells per well (Fig. 2B) and incubated for 24 h. Next, cells were fixated by 15 min incubation with 4 % paraformaldehyde (PFA) at 4°C. Nuclei were stained with 4^′,6-diamidino-2-phenylindole (DAPI) at a concentration of 50 μg· μL^-1 and afterwards rinsed three times with 50 μL phosphate buffered saline (PBS; 140 mM NaCl, 2.7 mM KCl, 1 mM Na₂HPO₄ · 2H₂O, 2 mM KH₂PO₄, pH 7.4) and stored at 4°C in PBS until analysis. An area of 2363.9 μm² around the center of each well was captured via fluorescence microscopy.

Cell adhesion was tested on printed TPU (approximately 1 cm² pieces) flat slides (Fig. 2A). The slides were placed in a 6-well plate to prevent evaporation. Either no (medium control), 2.5·10⁵ or 5.0·10⁵ cells in 2 mL medium were seeded per well and incubated 48 h. Finally, samples were investigated by fluorescence microscopy using a confocal laser scanning microscope (CLSM) LSM800 (Zeiss, Germany).

2.6 Fluorescence in situ hybridization

The adherence of bacteria to different filaments was assessed by FISH [45]. Prior to use, the printed slides were incubated 2 h in 70 % ethanol at RT to avoid contaminations. Afterwards, the slides were rinsed in 50 mL sterile water and dried at RT. Half of the slides were coated with cell immobilization solution (Medipan GmbH, Germany), the other half remained uncoated. On these, 5·10⁶ colony forming units (CFU)·mL^-1 bacteria (Staphylococcus aureus (S. aureus) strain 6095 and strain 6096, isolate from cystic fibrosis patient) were seeded per well and adhered at RT for one hour. Unbound bacteria were removed by washing with PBS. Bacteria were fixed for 1 h at 4°C with 4 % PFA and then washed three times with PBS. To ensure the accessibility of the FISH probe, S. aureus was incubated with 1 mg·mL^-1 lysozyme in 10 mM 2-Amino-2-(hydroxymethyl)-1,3-propanediol hydrochloride (TRIS/HCl) for 30 min at 37°C. All bacteria were then incubated with precooled methanol for 10 min at –21°C to stop the lysozyme reaction and permeabilize the bacteria. After a short drying time a fluorescent labeled (ATTO 647N) 16 S rRNA probe (EUB338 [18], 5‘-GCWGCCWCCCGTAGGWGT-3'; 5 ng· μL^-1 in hybridization buffer (5 M NaCl, 1 M TRIS/HCl, 10 % sodium dodecyl sulfate (SDS), 15 % formamide)) was hybridized for 1 h at 46°C. Afterwards, samples were washed with preheated wash buffer (5 M NaCl, 1 M TRIS/HCl, 10 % SDS) at 48°C for 10 min. Finally, bacteria were counterstained with 50 μg· μL^-1 DAPI solution and washed three times with PBS. Images were captured using a confocal laser scanning microscopy (CLSM) and analyzed by bioimage informatics (Section 2.9).

2.7 Biofilm formation

Printed slides made of PLA (black/ matt black, Das Filament; black, Elegoo) and ABS (black, Tronxy) were tested for their biofilm formation capacity. Before usage, the slides were incubated 2 h in 70 % ethanol at RT. Afterwards, the slides were rinsed in 50 mL water and dried at RT. For biofilm formation, S. aureus strain 6095 and Escherichia coli MG1655 F’tet strain 6101 [46] were seeded on the slides (3·10⁶ CFU·mL^-1) and incubated for 24 h at 37°C. The biofilm was then washed with PBS and stained with DAPI (50 μg· μL^-1). The slides on which the biofilm was formed were analyzed with a CLSM and evaluated as described in Section 2.9.

2.8 DNA capture assay

Four PLA (black/ matt black, Das Filament; black, Elegoo) and ABS (black, Tronxy) filaments were selected to build a two-chamber chip (Fig. 2C) for the DNA capture assay since they exhibited the lowest autofluorescence. Figure 3 illustrates the preparation and use of the chip. A nitrocellulose membrane (GE Healthcare, USA) with a pore size of 0.2 μm was used as the matrix in the chip. We used this classical approach to capture fluorescent labeled single-stranded DNA molecules in a suspension in a dose-dependent manner [47, 48]. Two punched out circular filter paper and a rectangular filter strip were inserted into the chip and closed with a lid. Four concentrations (0.05, 0.1, 1, 10 μM) of a fluorescent labeled (ATTO 647 N) 16 S rRNA: EUB338 probe suspension as well as a negative control (buffer only) were pipetted into the reservoir and incubated for 15 min at RT. The excess of the probe was removed by three times washing with Tris buffered saline with Tween (TBS-T; 50 mM TRIS, 150 mM NaCl, 0.05 % Tween^® 20). The signals were detected with the fluorescence imager Biostep Celvin S (Biostep, German, exposure: 50 % fluorescence intensity, 5 sec).

Fig.3

Preparation of a 3D printed 2-chambered chip for DNA detection experiments. A: An empty chip (2 parts) is first loaded with a rectangular filter strip and next with circular nitrocellulose membranes (diameter: 6 mm, prepared with a punch plier). All membranes are held in position after closing with the lid. B: For illustration, an eosin solution (orange) was pipetted into the sample inlet. During the incubation time of five minutes, the solution diffuses via the filter strip to the sample chambers. In DNA capture experiments fluorescent probes were used and evaluated with a fluorescence imager (Biostep Celvin S).

2.9 Bioimage informatics on fluorescence images

The image data taken was analyzed using a bioimage informatics data analysis pipeline in Python (v. 3.7.3) [49, 50]. In short, the data was imported as a TIFF image and analyzed using functions from the scikit-image [51] package (v. 0.14.2). Statistical analysis was performed using the statsmodels package (v. 0.9) [52].

As an indicator for the adhesion of cells to the slide surface, the average pixel intensity was calculated. Statistical significance between the different cell seeding numbers was determined by fitting a linear regression model and performance of ANOVA by using the Tukey’s honest significance test (Tukey’s HSD) as implemented in the Python module statsmodels. Additionally, the cell number on each image was estimated using the blog_log() function of scikit-image. To suppress the detection of false positive nuclei, the images were converted to grayscale and blurred using a median filter (r = 3.5) before analysis. Cell lumps were detected by calculating the interquartile range (IQR) for the sizes of detected nuclei. All nuclei larger than radius (r)≥75 % quartile + 1.5 * IQR μm were excluded from counting. To estimate the covered area on biofilm images, otsu’s thresholding method [53] was used. To analyze Chip assays, first the well position on the images were estimated by thresholding the images to suppress 98.7 % of all pixels, followed by calculation of a Euclidean distance map and determination of local maxima. Those maxima were used to create binary masks to calculate the average pixel intensity for each assay plate. To account for autofluorescence of the used material, a reference point was subtracted from the determined result (position was determined as the center between the two wells of each plate translated –30 pixels on the y-axis). The analysis is available in the supplement as Jupyter Notebook (Supp. IA).

3 Results and discussion

Variations in the added substances and effects due to the continuous optimization of the products can lead to major changes in the overall properties [54]. Plastics occasionally have an autofluorescence when excited by light (e.g., near UV light). They may also have a strong light-absorbing property. This is due, among other things, to additives that introduce fluorescent species and highly absorbent species. But also, the thermal treatment appears to change the properties. As a result, this often leads to suboptimal detection limits in optical measurements [36, 55].

It is important to bear in mind that filament manufacturers do not specify the exact chemical composition of their filaments. However, no spectra were available for any of the materials. Thus, it was not possible to assess whether there were any of these materials that had zero or only low autofluorescence in certain areas. Due to the lack of information, we have examined 14 commercial filaments of the material classes PLA, ABS, PETG and TPU with different colors. The filaments were used without any further pre-processing steps. In our study we only examined samples from one batch with several technical replicas after printing. We were particularly interested in the melting behavior and the spectral properties of 3DP filaments for use in fluorescence-based bioanalytical tests.

3.1 Based on autofluorescence analyses black filaments are well suited for fluorescence applications

Regarding whether the materials can be used for optical bioanalytical applications first we investigated to what extent the materials exhibit autofluorescence after printing. As described for the analysis of autofluorescence and spectral characteristics, we took emission measurements after excitation with six different wavelengths (see Fig. 4A for a comparison of mean emission data of all materials (λ _ex = 350, 488, 640 nm)). As already mentioned, the measurements were made for all materials with the same parameters. All settings such as excitation, gain and sensitivity of the detector were selected accordingly to capture all materials within the limits and compare them relative to each other. The measured fluorescence intensity in cps is therefore instrument or measurement specific and a relative value. Regarding later applications, in which the measurement system, filters and fluorophores have an influence on the signal-to-noise ratio, an absolute fluorescence value is not necessarily decisive, but above all the spectral profile and ranges in which most peaks appear. For comparison of the materials we defined the different levels of autofluorescence no or very low (below 1·10⁶ cps), medium (up to 5·10⁶ cps) and high (above 5·10⁶ cps).

Fig.4

Fluorescence spectra of the 3D printing polymers. Test slides (10 technical replicas) of each 3D printing filament (each from 1 batch) were printed and their spectral properties examined by fluorescence spectroscopy. Emission spectra were recorded after excitation with 6 different wavelengths. A shows the mean fluorescence spectra of the materials for three different excitation wavelengths (350, 488, 640 nm). B Representative spectra of black PLA filament (Das Filament), from four wavelength ranges typically used in biological assays. This material is usable in fluorescence-based assays with red and near-infrared dyes. The colors correspond to the emitted light at the respective wavelength. Shown is the mean and the confidence interval from 10 measurements.

The fluorescence spectrometric analysis showed that all materials contain substances which fluoresce in a wide range. Albeit the fluorescence intensity varied for other than black materials. Fluorescence was particularly pronounced at lower wavelengths (350–561 nm), making some materials unusable for fluorescence-based assays, where no background is desired. The materials PLA (Surreal), PETG (Sunlu), TPU (Enotepad) had a particularly high autofluorescence over a broad range.

Some materials had partly distinct peaks (e.g., white PLA, Janbex or green TPU, Enotepad), as well as no (black/ matt black PLA, Das Filament) or very low (black PLA, Elegoo or black ABS, Tronxy) autofluorescence even with excitations with high-energy radiation between 300–400 nm.

Transparent filaments exhibited strong autofluorescence, too. As found by others this is presumably due to additives. Substances intended to improve optical properties like color clarity or transparency of the materials can cause strong autofluorescence. Pigments that appear colored to the unaided eye have an influence on this property. For example, autofluorescence can be enhanced overall, occur only in certain wavelength ranges, or be quenched accordingly [56, 57].

Among the filaments examined, the four black PLA and ABS filaments (Das Filament, Elegoo, Tronxy) had low autofluorescence over all spectra, and we defined these as relevant for our study. Their properties and applications will be discussed in more detail below. All other filaments were excluded from following functional fluorescence-based assays.

For most biological samples, the autofluorescence decreases with increasing excitation wavelength, thus increasing the detectability by higher signal-to-noise ratios [58]. This was also observed for the filaments (>550 nm), which means that most filaments are usable with red and near-infrared (NIR) dyes.

Figure 4B shows the emission data of a material (black PLA, Das Filament) in the most relevant fluorescence ranges (see 405 nm - DAPI, 488 - FITC, 561 nm - Cy3, 640 nm - Cy5). As can be seen from the graphs, this filament material exhibits little or no autofluorescence over the entire wavelength range considered and makes it interesting for the desired applications. Based on our analytical data, this PLA appears to be the most suitable of the materials tested for fluorescence experiments.

These analyses and representations were prepared for all materials for all six wavelengths and can be found in the supplement (Supp. FS).

The first two principal components explained > 90 % of the total variance for the normalized fluorescence spectra at all excitation wavelengths. The discrimination between the filaments generally decreased with increasing excitation wavelength (Fig. 5).

Fig.5

Biplot of principal component analysis (PCA) for the excitation wavelength 350 nm. Red arrows show PCA loadings at annotated wavelengths (wl) (upper abscissa and right ordinate axes), letters indicate filament PC1 vs. PC2 scores (lower abscissa and left ordinate axes). An arrow directed towards a series of filament scores indicates positive correlation. That filament autofluorescence was more pronounced at the given wavelength. The opposite direction marks negative correlation. The angle between the arrows indicates correlation between autofluorescence intensities. Identical direction identifies positive correlation, opposite direction negative correlation and a right angle (90°) no correlation. A: PLA - Premium White (Raise3D); B: PLA - Matt White, C: PLA - Matt Black, D: PLA - Black, E: PLA - Natural, F: TPU - Natural, G: PETG - Natural (Das Filament); H: ABS - Black (Tronxy); I: PLA - Black (Elegoo); J: PLA - White (Janbex); K: PLA - Transparent (Sienoc); L: PLA Fluo Orange (Surreal); M: PETG - Orange (Sunlu); N: TPU - Green (Enotepad).

At an excitation wavelength of 350 nm, filaments PLA Fluo Orange (Surreal) and TPU Green (Enotepad) exhibited elevated autofluorescence at 550 nm, and filaments PETG Natural (Das Filament) and PLA White (Janbex) showed elevated autofluorescence at the wavelengths 400, 470 and 550 nm (Fig. 5). Elevated autofluorescence alleviates the image contrast with fluorescent objects, thus lowering the limits of detection. The materials PETG Natural (Das Filament), PLA White (Janbex), PLA Fluo Orange (Surreal), PETG Orange (Sunlu) and TPU Green (Enotepad) formed separate clusters. The analysis of the emission data for the other excitation wavelengths also showed elevated autofluorescence which can be used for differentiation Table 1. However, this was not as clear as at λ _ex = 350 nm (see Supp. PCA Fig. 5B-F). Overall, autofluorescence was more pronounced at excitation wavelengths <500 nm, where the filament types PLA White (Janbex), PLA Fluo Orange (Surreal), PETG Orange (Sunlu) and TPU Green (Enotepad) were most affected.

Table 1

Elevated autofluorescence of filaments. Wavelengths at which elevated autofluorescence was observed are given for excitation wavelengths studied

Filament		Excitation [nm]
Code	Type	350	405	488	561	640	750
G	PETG Natural (Das Filament)	400	·	·	·	·	795
J	PLA White (Janbex)	470	465	550	·	660	830
L	PLA Fluo Orange (Surreal)	550	560	625	700	·	·
M	PETG Orange (Sunlu)	·	620	625	740	·	·
N	TPU Green (Enotepad)	550	560	550	·	·	·

3.2 Thermal analyses revealed ABS as favorable filament to high temperature applications

The manufacturers of commercial 3DP filaments only provide rough information on the respective processing temperatures of the materials, that are not enough to estimate the appropriateness for the fabrication of bioanalytical devices. There is no documentation of exact thermal properties or even of specific temperature ranges in which those materials can be used. In addition, little is known about exact material compositions and production conditions, which leads to difficulties in material selection for bioanalytical systems. A total of 14 polymer samples were analyzed. After the spectral analysis, the black, matt black PLA (Das Filament), natural PETG, TPU (Das Filament), black ABS (Tronxy), black PLA (Elegoo) remained for DSC analyses (Table 2), which are described in more detail below. The other matrices had strong autofluorescence over the entire spectrum (see Fig. 4 and Supp. FS) and were therefore not considered further.

Table 2
Comparison of filament properties for 3D printing of bioanalytical test systems

Code Type Color Manufacturer Suitable fluorescence combinations (for common dyes &filter sets) Thermal properties Functional assays

Cells adhesion Biofilm formation FISH DNA capture

DAPI FITC, Cy2 TRITC, Cy3 Cy5 T_g T_r [°C] T_m [°C]

A PLA White Raise3D × × ✓ ✓ · · · · · · ·

B PLA Matt White Das Filament ✓ (✓) ✓ ✓ · · · · · · ·

C PLA Matt Black Das Filament ✓ ✓ ✓ ✓ 62.5 118.5 149.5 ✓ ✓ ✓ ✓

D PLA Black Das Filament ✓ ✓ ✓ ✓ 63.8 119.8 150.5 ✓ ✓ ✓ ✓

E PLA Natural Das Filament × (✓) ✓ ✓ · · · · · · ·

F TPU Natural Das Filament × (✓) ✓ ✓ –16.1¹ / / · · · ·

G PETG Natural Das Filament × (✓) ✓ ✓ 78.1 / / · · · ·

H ABS Black Tronxy ✓ ✓ ✓ ✓ 104.2² / / ✓ ✓ ✓ ✓

I PLA Black Elegoo ✓ ✓ ✓ ✓ 62.4 100.6 168.6 (✓) (✓) (✓) ✓

J PLA White Janbex × × ✓ ✓ · · · · · · ·

K PLA Transparent Sienoc × (✓) ✓ ✓ · · · · · · ·

L PLA Fluo Orange Surreal × × × ✓ · · · · · · ·

M PETG Orange Sunlu × × × (✓) · · · · · · ·

N TPU Green Enotepad × × ✓ (✓) · · · · · · ·

Code	Type	Color	Manufacturer	Suitable fluorescence combinations (for common dyes &filter sets)	Thermal properties	Functional assays
A	PLA	White	Raise3D	×	×	✓	✓	·	·	·	·	·	·	·
B	PLA	Matt White	Das Filament	✓	(✓)	✓	✓	·	·	·	·	·	·	·
C	PLA	Matt Black	Das Filament	✓	✓	✓	✓	62.5	118.5	149.5	✓	✓	✓	✓
D	PLA	Black	Das Filament	✓	✓	✓	✓	63.8	119.8	150.5	✓	✓	✓	✓
E	PLA	Natural	Das Filament	×	(✓)	✓	✓	·	·	·	·	·	·	·
F	TPU	Natural	Das Filament	×	(✓)	✓	✓	–16.1¹	/	/	·	·	·	·
G	PETG	Natural	Das Filament	×	(✓)	✓	✓	78.1	/	/	·	·	·	·
H	ABS	Black	Tronxy	✓	✓	✓	✓	104.2²	/	/	✓	✓	✓	✓
I	PLA	Black	Elegoo	✓	✓	✓	✓	62.4	100.6	168.6	(✓)	(✓)	(✓)	✓
J	PLA	White	Janbex	×	×	✓	✓	·	·	·	·	·	·	·
K	PLA	Transparent	Sienoc	×	(✓)	✓	✓	·	·	·	·	·	·	·
L	PLA	Fluo Orange	Surreal	×	×	×	✓	·	·	·	·	·	·	·
M	PETG	Orange	Sunlu	×	×	×	(✓)	·	·	·	·	·	·	·
N	TPU	Green	Enotepad	×	×	✓	(✓)	·	·	·	·	·	·	·

Explanations: ✓ - well suited, (✓) - limited application,×- not suited, · - not tested, / - no data; T_g - glass transition temperature, T_r - recrystallization temperature, T_m - melting temperature; ¹values derived from midpoint by curve fitting, ²values taken from cooling curves.

Two of the black PLA samples (Das Filament) showed a high degree of homogeneity in the DSC curve, both within the replicas and between the filaments. Glass transition, recrystallization and melting ranges can be characterized very well and correspond to the previous assumptions based on the material class. Although on average slightly lower glass transition (62.5±0.6°C) and recrystallization temperatures (118.5±0.9°C) were determined for the matt black variant, both materials are well suited for use up to at least 50°C and can be processed without any problems at comparatively low melting temperature of 150°C. We assume that the difference was due to the matting additives.

Black PLA (Elegoo) had a similar glass transition temperature (62.4±0.9°C) as the other PLA samples. Figure 6 shows the DSC curves for black PLA and matt black PLA filaments (Das Filament). The DSC curves for black PLA (Elegoo), TPU, PETG and ABS are available in the supplement (Supp. DSC). Surprisingly the recrystallization temperature and melting temperature were about 20°C lower or higher than the other PLAs, respectively (T_r,Elegoo = 100.6±6.8°C, T_m,Elegoo = 168.6±2.3°C). Irrespective of the heating rate, additional effects occurred which are presumably caused by impurities or additives. Such effects can also be distributed irregularly in the material and can impair a homogeneous melting behavior. However, this is decisive for the printing process and thus for a high-quality result.

Fig.6

Thermal Analyses of two 3D printing filaments. The plots show the DSC curves for matt black PLA (A, red line, Das Filament) and black PLA (B, black line, Das Filament). The curves represent the average of the data measured during three heating cycles. Peak analysis was done to find glass transition, recrystallization and melting temperatures.

Attributes such as glass transition and fluorescence are also important for later application, which is why materials should be used whose properties can be easily characterized and show no variations within a batch.

The PETG filament examined showed a uniform behavior and a glass transition temperature of almost 80°C (78.1±2.1°C) within the repetitions. However, the DSC method used was not able to determine precise measured values regarding recrystallization or melting temperature. Only above 400°C a melt effect could be observed shortly before degradation.

An average glass transition temperature of 104.2±0.1°C could be determined from the cooling curves for ABS; this is the highest of the six DSC analyzed materials. However, recrystallization and melting temperatures could not be clearly determined for ABS in the DSC. We hypothesize that this was due to the additives.

TPU is a special case since this material is very flexible at room temperature. This property makes it interesting for applications, where elastic deformation or sealing is needed. Due to its special form, this type of material showed a difficult behavior when performing the DSC analyses, resulting in more complex curves. More accurate measurements with changed conditions are needed to explain the underlying additional effects. Therefore, apart from an average glass transition temperature of –16.1±3.0°C, no critical temperature ranges or general processing conditions could be precisely specified here.

3.3 Cell adhesion analysis

Cell adhesion on the materials black/ matt black PLA (Das Filament) and black ABS (Tronxy) could be observed for all cell lines (see Figs. 7 and 8). The used algorithm generally performed on images with separated nuclei with enough contrast to the background (high signal-to-noise ratio) but failed on images were the cells were overly cluttered. The analysis must also consider the surface topography caused by the printing process, which leads to a partially directed distribution as well as to a slightly different position of cells in the Z-axis. Despite this fact, different adhesion behavior was observed when examining entire wells. Whilst SiHa and HeLa cells showed an overall uniform spread over the surface that was comparable for all materials, CaSki cells showed no adhesion pattern and the formation of cell lumps. A detachment of the cell layer from the material could also be observed in CaSki cells. This indicates that CaSki cells, in contrast to SiHa and HeLa cells, do not adhere to the materials. The analysis also showed that all cell lines adhered to TPU, too.

Fig.7

Determination of cell density of cells adhered on printed slides. Nuclei of SiHa, HeLa and CaSki cells were automatically counted. Top row: The cells were incubated 24 h at 37°C after seeding (4230.3 cells·mm^-2 seeded), PFA-fixed and stained with DAPI. Magenta circles indicate the detected nuclei as well as their approximate diameter. Bottom row: Adhesion pattern of the cell lines on TPU (520, 833 cells·mm^-² seeded in 6 well plate containing an approximately 1 cm² TPU patch). The cells were incubated 48 h at 37°C after seeding, fixated using 4 % PFA and stained with DAPI. Scale bar: 200 μm.

Fig.8

Adhesion pattern of cell lines SiHa, HeLa and CaSki on printed slides. In printed well-slides of PLA (black/ matt black, Das Filament and ABS (black, Tronxy)) cells were seeded with 2115 cells·mm^-², incubated for 24 h, fixated using 4 % PFA and stained with DAPI. The lines SiHa and HeLa showed a pattern of densely adherent cells with an even cell distribution on the materials. CaSki cells showed no adhesion pattern and the formation of cell lumps. Scale bars: 200 μm.

No qualitative difference in the adhesion pattern of the cell lines was observed, in contrast to the other materials. The lesser adhesion of CaSki cells on both PLA and ABS was also confirmed by the analysis of mean image intensity (i_{CaSki,blackPLA} = 5.02±4.82 (p = 0.2104), i_CaSki,ABS = 10.09±3.14 (p = 0.2638) compared to i_{SiHa,blackPLA} = 22.13±3.62 (p = 0.001), i_SiHa,ABS = 23.26±2.07 (p = 0.001) and i_{HeLa,blackPLA} = 31.47±9.88 (p = 0.001), i_HeLa,ABS = 25.96±6.93 (p = 0.001)) as well as the estimated cell counts (n_{CaSki,blackPLA} = 84.29±78.98 (p = 0.111), n_CaSki,ABS = 125.96±99.33 (p = 0.076) compared to n_{SiHa,blackPLA} = 574.16±99.58 (p = 0.001), n_SiHa,ABS = 495.89±45.67 (p = 0.001) and n_{HeLa,blackPLA} = 771.39±234.84 (p = 0.001), n_HeLa,ABS = 586.85±122.83 (p = 0.001)) (see Figs. 9 and 10), which were both much lower for CaSki cells for all tested materials and dilutions. No gradient between the different seeded cell numbers was observed. For all tested cell numbers, no significant difference to the negative control for both, average image intensity as well as estimated cell number, was observable and further indicating weak cell adhesion of the cells to the tested materials. Due to the dense clustering of CaSki cells on higher seeded cell numbers (2115.15 cells·mm^–2 and more), the used counting method performed poorly for distinguishing cells. Thus, the average image intensity was deemed a more reliable indicator for this kind of growth pattern.

Fig.9

Average DAPI fluorescence intensity of the cell lines SiHa, HeLa and CaSki on printed slides. Shown are results for PLA (black/ matt black, Das Filament) and ABS (black, Tronxy). The cells were incubated 24 h at 37°C after seeding, fixated using 4 % PFA and stained with DAPI. For matt black PLA and black PLA, the highest average intensities were observed for HeLa cells. SiHa cells showed lower average intensities than HeLa for both PLA filaments. Average intensities for both HeLa and SiHa cells were comparable for ABS. CaSki showed the least average intensities matt black PLA, black PLA and ABS with comparable intensities for all cell seeding numbers, indicating less adhesion of the cells to the materials. NC = negative control, n = 4, error bars indicate the standard deviation, * = p < 0.05 compared to negative control.

Fig.10

Approximation of the cell number of the cell lines SiHa, HeLa and CaSki on printed slides. Shown are results for PLA (black/ matt black, Das Filament) and ABS (black, Tronxy) for all tested cell numbers. The cells were incubated 24 h at 37°C after seeding, fixated using 4 % PFA and stained with DAPI. For both tested kinds of PLA, the highest cell counts were obtained for HeLa cells. SiHa cells showed generally lower cell counts than HeLa for both kinds of PLA. The cell counts obtained for both HeLa and SiHa cells were comparable for ABS. CaSki cells were omitted from counting because of the formation of densely packed cell lumps, which negatively influence the used algorithm. NC = negative control, n = 4, error bars indicate the standard deviation, * = p < 0.05 compared to negative control.

Approximate cell counts for SiHa and HeLa cells showed similar results as the observed average intensity. The highest number of cells was counted on matt black PLA for HeLa and SiHa cells (n_{SiHa,mattblackPLA} = 605.46±170.08 (p = 0.001), n_{HeLa,mattblackPLA} = 1344.18±185.76 (p = 0.001)), with a generally higher number of cells for HeLa cells. Lowest cell counts were determined for black ABS (n_SiHa,ABS = 495.89±45.67 (p = 0.001), n_HeLa,ABS = 586.85±122.83 (p = 0.001)). Both cell lines showed a gradient for all dilutions tested. A significant difference in cell count to the negative control was observable for all seeded cell numbers except on black PLA (Das Filament) for SiHa cells (significant difference only for 2115.15 and 4230.3 cells·mm^–2) and matt black PLA (Das Filament) for HeLa cells (significant difference only for 4230.3 cells·mm^–2). The determined average image intensity showed a similar trend.

3.4 Fluorescence in situ hybridization on 3D printed slides

Since human cells adhere to PLA and ABS (materials C, D, H and I, see Table 2) we tested if bacteria also adhere to them and how they perform in FISH experiments. Therefore, we analyzed the average fluorescence intensities of nuclei staining (DAPI) as measure for adherence and FISH probe binding to 16 s rRNA (ATTO 647 N) as a crude measure for viability [59].

The two S. aureus strains showed similar adhesion pattern (Fig. 11). There was no difference in bacterial adhesion between slides and slides coated with cell immobilization solution. ABS had significantly higher intensities for the genomic DNA results (p = 0.001) compared to the other materials. Differences in measurements of 16 S rRNA were non-significant. Due to our selected magnification (10×objective) we were not able to count single bacteria but estimated their number as events·mm^-2 for the comparison of the filaments. A higher magnification could not be used, because the lenses hit the corrugated edges on the printed slides and therefore no focusing was possible. We found a significant difference between ABS (Tronxy, n_events = 4405.2±660) and PLA (black, Elegoo, n_events = 1236.2±114) for nuclei staining (p = 0.008). However, the 16 S rRNA results do not show a uniform picture and cannot be compared optimally. FISH usually achieves higher fluorescence signals in the detection of genomic DNA than 16 S rRNA. This is based on the viability of the bacteria. The more metabolically active the bacteria, the more ribosomal RNA is present.

Fig.11

Average fluorescence intensity of nuclei staining (DAPI), 16 S rRNA signals (ATTO 647 N) of S. aureus strain 6095 and strain 6096. Uncoated slides made from PLA (black / matt black, Das Filament; black, Elegoo) and ABS (black, Tronxy) were compared. A: Strains 6095 and 6096 can adhere to all materials except black PLA (Elegoo). Black PLA and matt black PLA (Das Filament) showed pattern of adherent bacteria. ABS, on the other hand, shows pattern of tightly adhering bacteria and B: the most relevant results due to the signal distribution from genomic DNA to 16 S rRNA. n = 4, standard deviation between the two strains tested is shown above the bars. Scale bars = 200 μm.

Comparing the measured fluorescence signals of both detection molecules, ABS showed the most relevant results, since the signal for genomic DNA is higher than that for 16 S rRNA. To perform further detection using FISH, one should confine oneself to the material ABS and consider optimizations in printing. ABS showed advantages in handling compared to other materials. The sample volume was distributed more homogeneously within a cavity and remained exclusively in the wells and did not run over the outer edges.

In summary ABS showed pattern of densely adherent cells and performed best in this assay type. In turn, PLA (Elegoo) was the material without adhesion pattern of bacteria and did not appear to be suitable in comparison to the other materials.

The statistical test results for the other materials can be found in the supplement (Supp. IA).

3.5 Biofilm formation on 3D printed slides

With the four materials, bacterial species (E. coli, S. aureus) adhered to the ABS polymer (Tronxy) and colonize larger and denser surface areas (Fig. 12). Biofilm formation was also observed on the three PLAs (black/ matt black, Das Filament; black, Elegoo), although to a lesser extent.

Fig.12

Covered area by biofilms on printed slides. Shown are results for biofilms (E. coli, S. aureus) on slides made of PLA (black/ matt black, Das Filament; black, Elegoo) and ABS (black, Tronxy). A: fluorescence images, B: image analysis based on genomic DNA signals (DAPI). The bacteria E. coli and S. aureus adhere and form a biofilm on all materials. ABS, on the other hand, is the material with the densest bacterial population. n = 4, standard deviation is shown on the bars, * = p < 0.05 compared to negative control.

Besides material properties such as the chemical properties of the polymer (unknown additives like pigments), the printing process must also be considered in detail. When evaluating the colonization area, not only molecular interactions (depending on the material composition), but also the composition of the micro- and macro-surfaces are of importance [60]. We observed irregularities on the surface due to contamination or damage to the printer nozzle. This can provide microorganisms with a greater attack surface for adhesion and colonization. The typical groove structure caused by the linear movement of the print head is problematic. Bacteria in the valleys seemed to show a stronger colonization than bacteria on the higher round mountain areas of the surface. At a sufficiently high distance, a quantitative difference in bacteria can be assumed as well as a mountain-valley-gradient of available nutrients. This may lead to irregular results when testing for antibiotic efficacy on biofilms. For particularly smooth and homogeneous surfaces, care must be taken to ensure tight overlapping and fusion of the printed material. These can be further improved by appropriate after-treatments (e.g., steaming with acetone for ABS).

A subsequent coating with a gel matrix and biomolecules could be used to reduce even minor unevenness and additionally increase the adhesion of microorganisms. In addition to chemical and physical conditions, optimum compromises must also be made in the measuring method. The recorded layer thickness plays a special role and, especially for biofilms on uneven surfaces, stack images of different focal planes should be used. Since we wanted to clarify the question of a basic colonization capability and biofilm formation on 3D printed devices, we limited ourselves to larger, single-layer overview images. Our data are an approximation based on the analysis of the colonized area and a basis for extended studies in which more precise investigations can be carried out to enable individual cell quantification. The fluorescence properties of the materials are suited for adhesion and colonization experiments.

Even after staining with DAPI, no noticeable background fluorescence was observed with the selected exposure settings. The materials were selected for fluorescence detection applications. Although transmitted light images cannot be evaluated, this format offers the possibility to investigate the adhesion and colonization of polymers. This is an application used to investigate infections of implant material [22]. Long time experiments in bioanalytical devices are limited because degradation of biodegradable polymers occurs at different speed depending on the experimental conditions [61]. In our biofilm experiments, a strong, culture-untypical odor was noticed, which could probably be attributed to a beginning biological degradation.

3.6 DNA capture analysis in chip format

The application of fluorescent-labeled DNA probes in 3D printed chips is possible with the four PLA (black / matt black, Das Filament; black, Elegoo) and ABS (black, Tronxy) filaments (Fig. 13). None of the PLA materials showed an advantage over the other. ABS showed the best reproducibility in DNA detection with the lowest standard deviations. The values for the two lowest concentrations (50 nM and 100 nM) were at the detection limit and therefore varied. Considering the handling of the detection procedure, some steps such as washing and uniform filling must be optimized by changes in printing or using other materials or buffer systems.

Fig.13

Normalized fluorescence intensity of ATTO 647N-labeled DNA probes on nitrocellulose membranes in 3D printed chips. The 2-chamber chips were made of the materials PLA (black / matt black, Das Filament; black, Elegoo) and ABS (black, Tronxy). It was possible to detect a fluorescence-labeled DNA probe in chips with all materials. The signals of the two lowest concentrations 50 nM and 100 nM were at the detection limit and therefore varied. ABS represented the values with the lowest standard deviation and is therefore best suited for detection with fluorescence-labeled DNA probes. n = 2, * = p < 0.05 compared to negative control.

3.7 Summary of filaments properties

Fourteen ready-to-use polymer filaments from five manufacturers were tested for their thermal and spectral properties as well as the production of bioanalytical test systems. In summary, we provide more detailed information on their properties and give recommendations for their applicability based on our investigations in Table 2.

Apart from the more difficult processing conditions in 3DP, ABS proved to be the overall favorite in the various investigations and functional assays. In addition to its high temperature resistance, we found that it can be used almost indefinitely in fluorescence applications as well as in experiments with microorganisms or cells.

4 Conclusion

3DP technology is used to cost-effectively prototype devices for bioanalytical and pharmaceutical applications (incl., implants, microneedles, drug delivery systems). Based on the number of publications and patents in recent years, it can be concluded that there is a large active scientific and consumer community [21]. 3DP is a technology that could certainly be used more widely in science. The advancing use of 3DP is certain, and more innovations and applications will come in the next 30 years. But besides affordable 3D printers, other aspects are important for scientists.

We currently see chemically and physically precisely specified filaments as a bottleneck. Filament manufacturers provide useful information about printing conditions and some material properties. But users of functional bioanalytical test systems are also interested, for example, in whether the materials contain toxic substances or how the materials modulate gene expression. Cell adhesion assays are a prominent example in which the material/organism interface plays an important role [31, 62]. Parylene coating of 3D structures is often used to prevent cytotoxic effects [31]. However, members of the parylene family may have rapidly decreasing autofluorescence profile under UV excitation and thus has a low fluorescence background in microscopy, resulting in a low signal-to-noise ratio for appropriate optical detection [56].

We are particularly interested in autofluorescence, which could have a negative effect on the development of new systems. Therefore, we have addressed some of these issues and collected data for further use by third parties. Our aim is also to answer questions about how the apparent disadvantages can be exploited. For example, we see autofluorescence as a way of reproducibly referencing signals. So far, only little information can be found in the scientific literature. In this work, short single-stranded DNA probes with a fluorescence label should be detected on nitrocellulose membranes. Dealing with the autofluorescence was challenging because the autofluorescence of the material diminished the signal of the probes. There are detection methods that could be used to generate higher signals of the samples. Tyramide signal amplification (TSA), for example, is a universally applicable immunohistochemical detection method, which is used in various applications to quantify biomolecules due to its high sensitivity [63].

Future work should deal with quantitative variations of the fluorescence intensity of the materials under continuous long-term UV illumination. More detailed information about this would also be valuable for other disciplines, such as forensics. Since the 3DP is so easily accessible, it can in principle also be used for criminal purposes, such as printing a weapon [64]. Determining the material type and characteristics by inherent equipment distortion of a 3DP object is helpful in identifying its origin [65, 66]. Although we were able to achieve a classification of materials based on spectra and PCA, this was not sufficient for the individualization that would be necessary for court-proof forensic analysis.

We investigated if the materials could be used as provided by the vendor in functional cell-based tests. The use of polymer materials in bioanalytical test systems is particularly critical if the materials or their additives have a cytotoxic effect. It should also be borne in mind that the surface properties also have an influence on the cells. We investigated this in three human tumor cell lines (SiHa, CaSki, HeLa; cervical carcinoma) and observed that each cell line behaved differently on the materials. Based on the number of cells seeded, we concluded that the effect was systematic. The CaSki cells showed lower a cell density on black and matt black PLA as well as on ABS than HeLa and SiHa cells (Figs. 7–10). HeLa cells seemed to be least influenced by the material.

We would like to emphasize that we cannot assess the influence of unknown additives. However, precise knowledge is required for applications in pharmacy, medicine and bioanalytics to exclude inhibitory toxic influences and to be able to assess (cell) modulatory effects. These cell-type-specific behaviors coincide with studies of other authors. It has been shown that ABS can exhibit cytotoxic effects on human neuroblastoma cells and alter the functionality of cortical neurons. However, in related studies no effect was observed in HUVEC (human umbilical vein endothelial cells) or PCLS (precision cut liver slices). For PLA, which is known for its biocompatibility, it has been shown to affect cells of the central nervous system and the skin [24, 67]. Based on primary human adult skin epidermal keratinocytes and bone marrow mesenchymal stromal cells it has been described that even without direct material contact metabolic activity, gene expression and other biological processes can be significantly influenced [68].

Thus, studies with any polymer material must be considered critically. Study results can only be generalized if it is clearly stated in the methods and materials, which conditions were present (e.g., additives), so that this can be reproducibly repeated by third parties.

5 Funding

This work has been funded in part by the Pilot-Strukturwandel-Verbundprojekt “FISHng - FISH für die Tumor- und Erregerdiagnostik mittels VideoScan (03PSZZF1A)” (Federal Ministry of Education and Research (BMBF)) and the COST action CA16101 “MULTI-modal Imaging of FOREnsic SciEnce Evidence (MULTI-FORESEE)”.

Conflict of interest

The authors have no conflicts of interest to declare.

Footnotes

Acknowledgments

None.

The supplementary material is available in the electronic version of this article: .

References

Gross

, Erkal

, Lockwood

, Chen

, Spence

. Evaluation of 3D Printing and Its Potential Impact on Biotechnology and the Chemical Sciences. Anal Chem. 2014;86(7):3240–53. 10.1021/ac403397r.

Tang

, Vaze

, Rusling

. Automated 3D-printed unibody immunoarray for chemiluminescence detection of cancer biomarker proteins. Lab Chip. 2017;17(3):484–9. 10.1039/C6LC01238H.

González-Henríquez

, Sarabia-Vallejos

, Rodriguez-Hernandez

. Polymers for additive manufacturing and 4D-printing: Materials, methodologies, and biomedical applications. Progress in Polymer Science. 2019;94:57–116. 10.1016/j.progpolymsci.2019.03.001.

Rödiger

, Liebsch

, Schmidt

, Lehmann

, Resch-Genger

, Schedler

, et al. Nucleic acid detection based on the use of microbeads: A review. Microchim Acta. 2014;181(11-12):1151–68. 10.1007/s00604-014-1243-4.

Norman

, Madurawe

, Moore

CMV

, Khan

, Khairuzzaman

. A new chapter in pharmaceutical manufacturing: 3D-printed drug products. Advanced Drug Delivery Reviews. 2017;108:39–50. 10.1016/j.addr.2016.03.001.

Trenfield

, Awad

, Goyanes

, Gaisford

, Basit

. 3D Printing Pharmaceuticals: Drug Development to Frontline Care. Trends in Pharmacological Sciences. 2018;39(5):440–51. 10.1016/j.tips.2018.02.006.

Liaw

C-Y

, Guvendiren

. Current and emerging applications of 3D printing in medicine. Biofabrication. 2017;9(2):024102. 10.1088/1758-5090/aa7279.

Park

, Choi

, Moon

, Kim

, Bajracharya

, Min

, et al. Pharmaceutical applications of 3D printing technology: Current understanding and future perspectives. Journal of Pharmaceutical Investigation. 2018; 10.1007/s40005-018-00414-y.

Capel

, Rimington

, Lewis

, Christie

SDR

. 3D printing for chemical, pharmaceutical and biological applications. Nat Rev Chem. 2018;2(12):422–36. 10.1038/s41570-018-0058-y.

10.

Palo

, Holländer

, Suominen

, Yliruusi

, Sandler

. The next generation of crime tools and challenges: 3D Printing. Expert Review of Medical Devices. 2017;14(9):685–96. 10.1080/17434440.2017.1363647.

11.

Kadimisetty

, Mosa

, Malla

, Satterwhite-Warden

, Kuhns

, Faria

, et al. 3D-printed supercapacitor-powered electrochemiluminescent protein immunoarray. Biosensors and Bioelectronics. 2016;77:188–93. 10.1016/j.bios.2015.09.017.

12.

Hossain

MdA

, Canning

, Ast

, Rutledge

, Jamalipour

. Early warning smartphone diagnostics for water security and analysis using real-time pH mapping. Photonic Sens. 2015;5(4):289–97. 10.1007/s13320-015-0256-x.

13.

Ebert

LChr

, Thali

, Ross

. Getting in touch— 3D printing in Forensic Imaging. Forensic Science International. 2011;211(1):e1–6. 10.1016/j.forsciint.2011.04.022.

14.

Pfeil

, Dangelat

, Frohme

, Schulze

. Smartphone based mobile microscopy for diagnostics. Journal of Cellular Biotechnology. 2019;4(1-2):57–65. 10.3233/JCB-180010.

15.

Song

, Feijen

, Grijpma

, Poot

. Tissue engineering of small-diameter vascular grafts: A literature review. Clinical Hemorheology and Microcirculation. 2011;(1-4):357–374. 10.3233/CH-2011-1486.

16.

Chan

, Coen

, Hardick

, Gaydos

, Wong

K-Y

, Smith

, et al. Low-Cost 3D Printers Enable High-Quality and Automated Sample Preparation and Molecular Detection. Romesberg F, editor. PLOS ONE. 2016;11(6):e0158502. 10.1371/journal.pone.0158502.

17.

Klicks

, von Molitor

, Ertongur-Fauth

, Rudolf

, Hafner

. In vitro skin three-dimensional models and their applications. Journal of Cellular Biotechnology. 2017;3(1):21–39. 10.3233/JCB-179004.

18.

Dinter

, Burdukiewicz

, Schierack

, Lehmann

, Nestler

, Dame

, et al. Simultaneous Detection and Quantification of DNA- and Protein Biomarkers in Spectrum of Cardiovascular Diseases in a Microfluidic Microbead Chip. Anal Bioanal Chem. 2019; 10.1007/s00216-019-02199-x.

19.

Chen

, Fu

, Li

, Xie

, Ke

, Song

, et al. A smartphone colorimetric reader integrated with an ambient light sensor and a 3D printed attachment for on-site detection of zearalenone. Analytical and Bioanalytical Chemistry. 2017;409(28):6567–74. 10.1007/s00216-017-0605-2.

20.

Chen

, Townsend

, Hayter

, Birk

, Sell

, Martin

. Insert-based microfluidics for 3D cell culture with analysis. Analytical and Bioanalytical Chemistry. 2018;410(12):3025–35. 10.1007/s00216-018-0985-y.

21.

Hamad

, Kaseem

, Yang

, Deri

, Ko

. Properties and medical applications of polylactic acid: A review. Express Polymer Letters. 2015;9(5):435–55. 10.3144/expresspolymlett.2015.42.

22.

Barton

, Sagers

, Pitt

. Bacterial adhesion to orthopedic implant polymers. Journal of Biomedical Materials Research. 1996;30(3):403–10. 10.1002/(SICI)1097-4636(199603)30:3<403::AID-JBM15>3.0.CO;2-K.

23.

Bhattacharjee

, Urrios

, Kang

, Folch

. The upcoming 3D-printing revolution in microfluidics. Lab on a Chip. 2016;16(10):1720–42. 10.1039/C6LC00163G.

24.

Duong

, Chen

P-C

. Simple and low-cost production of hybrid 3D-printed microfluidic devices. Biomicrofluidics. 2019;13(2):024108. 10.1063/1.5092529.

25.

Smith

, Alexander

, Enterline

. Delayed stenosis following placement of a polyethylene terephthalate endograft in the cervical carotid artery. Journal of Neurosurgery. 2003;98(2):421–5. 10.3171/jns.2003.98.2.0421.

26.

Wang

, Huang

, Yang

, Leng

, Sun

, Liu

, et al. The effects of amorphous carbon films deposited on polyethylene terephthalate on bacterial adhesion. Biomaterials. 2004;25(16):3163–70. 10.1016/j.biomaterials.2003.10.010.

27.

Reinthaler

, Johansson

, Braune

, Al-Hindwan

HSA

, Lendlein

, Jung

. Shear-induced platelet adherence and activation in an in-vitro dynamic multiwell-plate system. Clinical Hemorheology and Microcirculation. 2019;71(2):183–91. 10.3233/CH-189410.

28.

Lind

, Busbee

, Valentine

, Pasqualini

, Yuan

, Yadid

, et al. Instrumented cardiac microphysiological devices via multimaterial three-dimensional printing. Nature Materials. 2017;16(3):303–8. 10.1038/nmat4782.

29.

Feng

, Ye

. Morphologies and mechanical properties of polylactide/thermoplastic polyurethane elastomer blends. Journal of Applied Polymer Science. 2011;119(5):2778–83. 10.1002/a32863.

30.

Beauchamp

, Nordin

, Woolley

. Moving from millifluidic to truly microfluidic sub-100-μm cross-section 3D printed devices. Analytical and Bioanalytical Chemistry. 2017;409(18):4311–9. 10.1007/s00216-017-0398-3.

31.

Van den Driesche

, Lucklum

, Bunge

, Vellekoop

. 3D Printing Solutions for Microfluidic Chip-To-World Connections. Micromachines. 2018;9(2):71. 10.3390/mi9020071.

32.

Boyd

. Relaxation processes in crystalline polymers: Experimental behaviour — a review. Polymer. 1985;26(3):323–47. 10.1016/0032-3861(85)90192-2.

33.

Brzozowska-Stanuch

, Rabiej

, Fabia

, Nowak

. Changes in thermal properties of isotactic polypropylene with different additives during aging process. Polimery. 2014;59(04):302–7. 10.14314/polimery.2014.302.

34.

Wittbrodt

, Pearce

. The effects of PLA color on material properties of 3-D printed components. Additive Manufacturing. 2015;8:110–6. 10.1016/j.addma.2015.09.006.

35.

Roch

, Ma

, Kratz

, Lendlein

. Cell-based detection of microbial biomaterial contaminations. Clinical Hemorheology and Microcirculation. 2015;60(1):51–63. 10.3233/CH-151939.

36.

Piruska

, Nikcevic

, Lee

, Ahn

, Heineman

, Limbach

, et al. The autofluorescence of plastic materials and chips measured under laser irradiation. Lab Chip. 2005;5(12):1348–54. 10.1039/B508288A.

37.

Fischer

. PCA-based supervised identification of biological soil crusts in multispectral images. MethodsX. 2019;6:764–72. 10.1016/j.mex.2019.03.013.

38.

R Core Team. R: A Language and Environment for Statistical Computing [homepage on the Internet]. Vienna, Austria: R Foundation for Statistical Computing; 2018.

39.

Fox

. The R Commander: A Basic-Statistics Graphical User Interface to R. Journal of Statistical Software. 2005;14(1):1–42. 10.18637/jss.v014.i09.

40.

Rödiger

, Friedrichsmeier

, Kapat

, Michalke

. RKWard: A comprehensive graphical user interface and integrated development environment for statistical analysis with R. Journal of Statistical Software. 2012;49(9):1–34. 10.18637/jss.v049.i09.

41.

Gabriel

. The biplot graphic display of matrices with application to principal component analysis. Biometrika. 1971;58(3):453–67. 10.1093/biomet/58.3.453.

42.

NIST Office of Data and Informatics. NIST Chemistry WebBook [homepage on the Internet]. NIST Standard Reference Database Number 69. 2018 [cited 2019 Aug 30]. 10.18434/T4D303. Available from: https://webbook.nist.gov/chemistry/

43.

Baker

, Phelps

, Lindgren

, Braun

, Gonda

, Howley

. Structural and transcriptional analysis of human papillomavirus type 16 sequences in cervical carcinoma cell lines. J Virol. 1987;61(4):962–71.

44.

Masters

. HeLa cells 50 years on: The good, the bad and the ugly. Nat Rev Cancer. 2002;2(4):315–9. 10.1038/nrc775.

45.

Kolenda

, Burdukiewicz

, Schiebel

, Rödiger

, Sauer

, Szabo

, et al. Adhesion of Salmonella to Pancreatic Secretory Granule Membrane Major Glycoprotein GP2 of Human and Porcine Origin Depends on FimH Sequence Variation. Front Microbiol. 2018;9. 10.3389/FMICB.2018.01905.

46.

Schiebel

, Böhm

, Nitschke

, Burdukiewicz

, Weinreich

, Ali

, et al. Genotypic and phenotypic characteristics in association with biofilm formation in different pathotypes of human clinical Escherichia coli isolates. Applied and Environmental Microbiology. 2017;AEM.01660-17. 10.1128/AEM.01660-17.

47.

Tonkinson

, Stillman

. Nitrocellulose: A tried and true polymer finds utility as a post-genomic substrate. Front Biosci. 2002;7:c1–12. 10.2741/A755.

48.

van Hal

NLW

, Vorst

, van Houwelingen

AMML

, Kok

, Peijnenburg

, Aharoni

, et al. The application of DNA microarrays in gene expression analysis. Journal of Biotechnology. 2000;78(3):271–80. 10.1016/S0168-1656(00)00204-2.

49.

Schneider

, Weiss

, Ruhe

, Jung

, Roggenbuck

, Stohwasser

, et al. Open source bioimage informatics tools for the analysis of DNA damage and associated biomarkers. Journal of Laboratory and Precision Medicine. 2019;4(0):1–27. 10.21037/jlpm.2019.04.05.

50.

Millman

, Aivazis

. Python for Scientists and Engineers. Computing in Science & Engineering. 2011;13(2):9–12. 10.1109/MCSE.2011.36.

51.

van der Walt

, Schönberger

, Nunez-Iglesias

, Boulogne

, Warner

, Yager

, et al. scikit-image: Image processing in Python. PeerJ. 2014;2:e453. 10.7717/peerj.453.

52.

Seabold

, Perktold

. Statsmodels: Econometric and statistical modeling with python. In: 9th Python in Science Conference. 2010.

53.

Sezgin

, Sankur

. Survey over image thresholding techniques and quantitative performance evaluation. JEI. 2004;13(1):146–66. 10.1117/1.1631315.

54.

Salentijn

GIJ

, Oomen

, Grajewski

, Verpoorte

. Fused Deposition Modeling 3D Printing for (Bio)analytical Device Fabrication: Procedures, Materials, and Applications. Anal Chem. 2017;89(13):7053–61. 10.1021/acs.analchem.7b00828.

55.

Young

EWK

, Berthier

, Beebe

. Assessment of Enhanced Autofluorescence and Impact on Cell Microscopy for Microfabricated Thermoplastic Devices. Anal Chem. 2013;85(1):44–9. 10.1021/ac3034773.

56.

Jervis , Daniel

. Optical brighteners: Improving the colour of plastics. Plastics, Additives and Compounding. 2003;5(6):42–6. 10.1016/S1464-391X(03)00049-7.

57.

Choochottiros

, Chin

I-J

. Potential transparent PLA impact modifiers based on PMMA copolymers. European Polymer Journal. 2013;49(4):957–66. 10.1016/j.eurpolymj.2012.12.008.

58.

Lakowicz

, editor. Fluorophores. In: Principles of Fluorescence Spectroscopy. Boston, MA: Springer US; 2006 [cited 2019 Aug 20]. pp. 63–95. 10.1007/978-0-387-46312-4_3 .

59.

Amann

. Methodological Aspects of Fluorescence In Situ Hybridization. Bioscience and Microflora. 2000;19(2):85–91. 10.12938/bifidus1996.19.85.

60.

Yoda

, Koseki

, Tomita

, Shida

, Horiuchi

, Sakoda

, et al. Effect of surface roughness of biomaterials on Staphylococcus epidermidis adhesion. BMC Microbiology. 2014;14(1). 10.1186/s12866-014-0234-2.

61.

Kucharczyk

, Pavelková

, Stloukal

, Sedlarík

. Degradation behaviour of PLA-based polyesterurethanes under abiotic and biotic environments. Polymer Degradation and Stability. 2016;129:222–30. 10.1016/j.polymdegradstab.2016.04.019.

62.

Vitte

, Benoliel

A-M

, Pierres

, Bongrand

. Regulation of cell adhesion. Clinical Hemorheology and Microcirculation. 2005;33(3):167–88.

63.

Jurischka

, Dinter

, Sowa

, Noack

, Schiebel

, Roggenbuck

, et al. Tyramide signal amplification as universal detection method on protein coated microbeads. Journal of Cellular Biotechnology. 2018;4(1-2):15–22. 10.3233/JCB-189003.

64.

Chase

, Laporte

. The Next Generation of Crime Tools and Challenges: 3D Printing. NIJ Journal. 2018;NCJ250697(279):10.

65.

Eden

. Print Technology and Forensics. Stevenson University Forensics Journal. 2014;5:8–13.

66.

Peng

, Yang

, Lin

Z-X

, Long

. Source identification of 3D printed objects based on inherent equipment distortion. Computers & Security. 2019;82:173–83. 10.1016/j.cose.2018.12.015.

67.

Rosenzweig

, Carelli

, Steffen

, Jarzem

, Haglund

. 3D-Printed ABS and PLA Scaffolds for Cartilage and Nucleus Pulposus Tissue Regeneration. International Journal of Molecular Sciences. 2015;16(7):15118–35. 10.3390/ijms160715118.

68.

Schmelzer

, Over

, Gridelli

, Gerlach

. Response of Primary Human Bone Marrow Mesenchymal Stromal Cells and Dermal Keratinocytes to Thermal Printer Materials In Vitro. J Med Biol Eng. 2016;36(2):153–67. 10.1007/s40846-016-0118-z.