Solid‐phase silica‐based extraction leads to underestimation of residual DNA in decellularized tissues

Abstract Decellularization of animal tissues is a novel route to obtain biomaterials for use in tissue engineering and organ transplantation. Successful decellularization is required as animal DNA causes inflammatory reactions and contains endogenous retroviruses, which could be transmitted to the patient. One of the criteria for successful decellularization is digestion (fragmentation) and elimination (residual quantity) of DNA from the tissue. Quantification of DNA can be done in many ways, but it has recently been shown that silica‐based solid‐phase extraction methods often do not completely purify in particular small DNA fragments. In the context of decellularization, this means that the measured DNA amount is underestimated, which could compromise safety of the processed tissue for in‐patient use. In this article, we review DNA quantification methods used by researchers and assess their influence on the reported DNA contents after decellularization. We find that underestimation of residual DNA amount after silica‐based solid‐phase extraction may be as large as a factor of ten. We therefore recommend a direct assessment of DNA amount in tissue lysate using dsDNA‐specific binding dyes, such as Picogreen, due to their higher accuracy for small fragment detection as well as ease of use and widespread availability.

remaining in decellularized tissues. However, processing of decellularized samples for DNA quantification may affect the results of such quantification, depending on the method. In the following paragraphs, common methods for DNA extraction are explained in more detail.
Before the amount of residual DNA in decellularized tissues can be assessed, the tissue is digested, the cells are lysed (if any remain), and the DNA is dissolved in the buffer solution. Thereafter, the amount of DNA can be directly assessed by adding a fluorescent probe to the digested sample. Alternatively, the DNA can be further purified from the sample using an extraction procedure (Figure 1), exploiting its physicochemical properties. Phenol/chloroform-extraction of DNA exploits differences in solubility of DNA vs. proteins and lipids in water-/oil-based solvents, respectively. A mixture of 25:24:1 phenol/chloroform/isoamyl alcohol is added to the sample, which is then vortexed for emulsification before centrifugation to ensure phase separation. While lipids are dissolved in the organic phase, proteins remain at the interphase and DNA in the aqueous supernatant, which can be transferred into a new vessel for quantification and further analysis. 8,9 Alternatively, high concentrations of salt can be used to precipitate proteins and cellular debris, due to their hydrophobicity, while the DNA remains in the supernatant. This method is often preferred over the phenol/chloroform-based extraction, as is does not rely on hazardous chemicals. 8,10 Solid-phase extraction exploits interactions of DNA with a solid substrate, such as silica resin/beads in the presence of chaotropic salts, allowing for rapid purification of DNA from digested samples.
Immobilization of DNA to the silica-surface is based on electrostatic interactions, only allowing for release in the presence of hypotonic buffers. Especially for sfDNA, however, this does not recover the total amount. Investigations into the recovery of sfDNA from solid-phase extraction kits have shown that for DNA fragments < 50 bp and < 100 bp, only about 16.5% and 27.7% (median across various extraction kits) are recovered, respectively. [11][12][13] In the context of decellularized tissues, the method chosen for sfDNA extraction therefore biases the interpretation of results. This has subsequent consequences for suitability for in-patient applications due to possible immunological side-effects. Here, we review DNA extraction methods used in decellularization studies, discuss their effect on clinically safe use and identify suitable methods for DNA quantification in decellularized tissues.

| RE SULTS AND D ISCUSS I ON
For this study, PubMed was searched for papers on decellularization methods for tissues by searching for "decellulariz*" OR "decellularis*" in title/abstract AND "DNA" as text word (see Appendix S1). 387 publications were reviewed for their DNA quantification approach for decellularized tissues. Over the past 20 years, a clear trend can be seen with an ever-increasing number of publications describing protocols for decellularization of various tissues and uses thereof in regenerative medicine ( Figure 2). Reducing the amount of residual DNA in these decellularized tissues is crucial for their further clinical application.
A large portion of research groups quantify residual DNA levels after extracting the DNA from the tissue lysate (ca. 70%), with a steady popularity of spin-column silica-based solid-phase extraction Modifying the surface structure of the solid phase used, may improve sfDNA recovery, however, the investigated fragment sizes are often still > 100 bp and the studies' results might not necessarily apply to even smaller fragments of DNA as found in decellularized tissues. 17,18 One study investigated sfDNA adsorption onto silica-coated magnetite particles, achieving ≈54% adsorption of available sfDNA (80-160 bp). 19 To the best of our knowledge, none of these modified substrates are readily commercially available.
Careful adjustment of buffer-conditions may increase yield of fragments ≥ 20 bp, but are not discussed in commercially available kits and therefore not routinely used. 20 Complicating the situation is the lack of available information on the smallest extractable sfDNA using commercially available extraction kits (Table 1). Most groups use silica-membrane-based extraction kits, optimized for genomic DNA extraction from tissues, that is, optimized for large DNA fragment recovery. Moreover, these kits often do not state a lower limit of extraction concerning DNA fragment sizes. This highlights the need for either determining and correcting for the relative loss of sfDNA prior to using commercially available kits, or employing extraction methods that are more suitable for small fragment recovery.
Our search did not result in studies examining potential bias toward certain DNA fragment sizes based on their solubility in presence of high ion-concentrations. The suitability of salting-out proteins for purification of DNA is therefore difficult to judge. From a usability standpoint, while organic extraction protocols do seem to enable sfDNA recovery, 13 they employ hazardous chemicals unfit for F I G U R E 2 Increasing number of decellularization protocols published over the last 20 years. Note that PubMed was searched for decellularization studies in March 2020, that is, the 2020 bar doesn't represent the full year. 387 studies were identified in total F I G U R E 3 (Non-)extracting DNA methods prior to quantification. Of the 387 identified studies, 186 employ solidphase-based DNA extraction (93% of these solid phases are silica-based), 106 quantify DNA directly in tissue lysate, 59 perform organic extraction of DNA, while 23 studies utilize salting-out protocols protocols (note that some studies used several methods). 21 studies did not specify the quantification method F I G U R E 4 Silica-based solid-phase extraction of DNA from digested decellularized anterior cruciate ligament samples severly depletes DNA before quantification. Porcine anterior cruciate ligament (ACL) was decellularized based on a previously published protocol employing freeze-thaw cycles, washes in detergent or ultrapure water, and enzymatic digestion of DNA (see Appendix S1). 53 Samples were then handled either according to the DNeasy Blood & Tissue kit (Qiagen, Venlo, Netherlands), or digested overnight at 60°C using 140 mg/mL papain (Sigma-Aldrich, Zwijndrecht, Netherlands) prior to DNA quantification using the Qubit platform (Invitrogen, Fisher Scientific, Landsmeer, Netherlands). More information available in Appendix S1. Native sample n = 5, decellularized samples n = 8. Values for remaining DNA in anterior cruciate ligaments across different quantification groups stem from the same samples. Statistical differences were investigated with a pairwise Wilcoxon test, assuming P < .05 as a significant difference between groups. * P < .05, ** P < .01. Red-line marks the 50 ng/mg dry weight recommended limit. 35 Blood & Tissue kit Typically, DNA concentration after extraction and purification is measured spectrophotometrically, assessing the absorption value at 260 nm. Alternatively, colorimetric quantitation is used, and more sensitive in the sub-µg range. 21 A different approach utilizes quantitative real-time PCR for DNA quantification, of which the reproducibility is however dependent on the initial DNA extraction method chosen, as well as potential interference from non-DNA components in the sample itself. 12,22 Probably easiest is the direct quantification of DNA in digested tissue samples. This has obvious consequences for the detection method, as other tissue components and the homogeneity of the lysate will affect spectrophotometric approaches. Addition of a fluorophore, however, has been demonstrated to be a highly sensitive and reproducible approach for DNA detection in whole blood, serum, urine, and in the presence of proteins 23 and glycosaminoglycans. 24 Especially sensitive for detection of DNA in low amounts are PicoGreen and SYBR Green, contrary to ethidium bromide and Hoechst-based dyes. 25 The binding site sizes of all probes are smaller than the DNA fragments produced by commonly used DNases used in decellularization protocols (eg, Benzonase cleaves DNA to fragments of ca. 5 bp in size, 26 while DNase I leaves fragments of ≥ 10 bp size 27 ), enabling them to detect even small fragments to varying degrees. Although these fluorescent dyes exhibit sequence-dependent specificity, with Hoechst and SYBR Green preferentially binding to AT-rich sequences whereas PicoGreen binds more often to GC-rich regions, 28-33 this effect is most likely negligible in the context of whole (cleaved) genome detection. More important is the use of an appropriate control sample of sfDNA that exhibits similar fragment size compared with samples obtained from decellularized tissues to account for differences in dye saturation of small versus large DNA fragments. 34 There are several commercial kits available utilizing dsDNA-binding fluorophores like Picogreen with high specificity.
Some of these are designed in a 96-well format, that is, enabling high-throughput testing.
An often-cited limit for acceptable DNA levels is 50 ng/mg dry weight of decellularized tissue with <200 bp in fragment length. 35 The fragment length limit is derived from the smallest generally observed fragment length in apoptosis and extracellular DNA length in healthy individuals. 36 Investigating the fragment size distribution of dilute DNA can be performed after concentrating the residual DNA from the tissue lysate. Usage of centrifugal filters for this purpose is quick, easy, and reliable. These filters are also used in cleanup of PCR-products and specific retention of DNA fragments based on their size. We propose that a molecular weight cut-off of <30 kDa is suitable for concentration of DNA from decellularized, digested tissue samples for subsequent gel electrophoresis. 37 The origins of the acceptable absolute amount to define "decellularization" are somewhat nebulous. So far, studies on extracellular cell-free DNA mostly focus on its abundance in serum or plasma, where it functions as a reporter of various diseases. 38 Also here, the use of (non-)extracting approaches to DNA quantification results in vastly different reported values. 39  A key player in extracellular dsDNA-recognition and downstream signaling is interleukin 26 (IL-26; Figure 5). 42

| CON CLUS ION
Immunological sensing of DNA is one possible adverse reaction to xenotransplants in vivo. Accurate determination of DNA amount and fragment size distribution is therefore paramount in assessing the clinical suitability of decellularized tissues. From the currently available facts, DNA extraction from decellularized tissues via silica-based approaches is not advisable due to depletion of sfDNA, leading to an underestimation of total DNA content. More suitable are solvent-based extraction methods utilizing, for example, phenol/chloroform, or methods selectively precipitating proteins and cell debris for DNA isolation. Alternatively, direct assessment of DNA in tissue lysate can be performed. As no extraction procedure is performed, no bias in DNA detection is given, and the obtained value is expected to more accurately reflect residual DNA in the sample.

ACK N OWLED G M ENTS
TCS gratefully acknowledges funding from the European Commission's Horizon 2020 funding program for the iPSpine project 676338.

CO N FLI C T S O F I NTE R E S T
The authors declare no conflicts of interest.

O RCI D
Tara C. Schmitz https://orcid.org/0000-0003-4247-1620 F I G U R E 5 Proposed inflammatory signaling pathways in response to high amounts of uncleaved extracellular DNA from xenografts