Decellularized Porcine Cartilage Scaffold; Validation of Decellularization and Evaluation of Biomarkers of Chondrogenesis

Osteoarthritis is a major concern in the United States and worldwide. Current non-surgical and surgical approaches alleviate pain but show little evidence of cartilage restoration. Cell-based treatments may hold promise for the regeneration of hyaline cartilage-like tissue at the site of injury or wear. Cell–cell and cell–matrix interactions have been shown to drive cell differentiation pathways. Biomaterials for clinically relevant applications can be generated from decellularized porcine auricular cartilage. This material may represent a suitable scaffold on which to seed and grow chondrocytes to create new cartilage. In this study, we used decellularization techniques to create an extracellular matrix scaffold that supports chondrocyte cell attachment and growth in tissue culture conditions. Results presented here evaluate the decellularization process histologically and molecularly. We identified new and novel biomarker profiles that may aid future cartilage decellularization efforts. Additionally, the resulting scaffold was characterized using scanning electron microscopy, fluorescence microscopy, and proteomics. Cellular response to the decellularized scaffold was evaluated by quantitative real-time PCR for gene expression analysis.


Introduction
Osteoarthritis (OA) is defined as a chronic, debilitating, and painful disease. It is estimated to be one of the leading causes of disability worldwide [1][2][3][4]. Sports, recreational activities, and even daily movements can contribute to the formation of cartilage lesions. Lesions or chondral defects, when left untreated, can lead to degenerative joint disease that may include an inflammatory response [4,5].
Knee OA is the most common type of OA and accounts for 70% of arthritis-related hospital admissions and 23% of clinical visits [2]. Given the anatomical position of the knee, it acts as a shock absorber by withstanding both tension and compression [6]. Hyaline cartilage within articular joints is located at the ends of long bones. Cartilage lacks nerve fibers and is avascular. It is tough, but flexible, and contains large amounts of glycosaminoglycans (GAGs) such as chondroitin sulfate and hyaluronic acid (HA), which interact with type II/IX/XI heterotypic collagen fibrils. Proteoglycans, such as aggrecan, are predominant molecular constituents of articular cartilage [2,4,7]. After a subsequent 24 h hyaluronidase treatment [23], followed by a 24 h 37 • C DNase and RNase treatment [19,23], we evaluated the tissues by histology and scanning electron microscopy. Freeze-thaw cycles were repeated followed by DI water incubation, and SDS treatment at 37 • C, with subsequent DNase digestion for 72 h at 37 • C with agitation. The overall decellularization process is depicted in Figure 2. Results from histological analysis are shown in Figure 3a,b. Histological analysis showed a decrease in cellular structures with the preservation of the extracellular matrix. Hoechst staining showed DNA content to be reduced (Figure 3c,d).  To quantify the depletion of DNA in the samples, samples were subjected to a DNA extraction process before and after decellularization. We found that the decellularization process removed 98.8% of the DNA associated with cells ( Figure 4). . DNA content before, during sequential stages, and after decellularization process. DNA was extracted and quantified spectrophotometrically. Quantitative measurements of DNA within scaffolds before, during, and after decellularization process indicated that residual DNA was at approximately 24.76%, 2.63%, 0.66%, and finally 0.22% of the original content. Error bars: Mean ± standard error of the mean. N = 6. Scanning electron microscopy was performed to monitor tissue throughout the decellularization process. Surface features are visualized in Figure 5. Prior to the decellularization process, cells, although present within the tissue, are not visible at the surface of the cartilage samples at either low (Figure 5a) or high magnification (Figure 5b). The surface features revealed by the decellularization process and scanning electron microscopy demonstrate an increase in surface complexity and exposure of the collagenous fibrillar matrix existing within the tissue (Figure 5c,d). (c) final decellularized cartilage disc; (d) higher magnification of decellularized cartilage disc. Appearance after decellularization indicates a rough surface with increased surface area and exposed collagen fibrillar networks. Scale bars = 20 µm.
To complement the analysis of DNA depletion after treatment, we carried out mass spectrometry to analyze the protein content of the scaffold before and after treatment for decellularization. Nuclear proteins were considered for their potential as suitable biomarkers for decellularization. We investigated a profile of 20 nuclear proteins that included APTX, UIMC1, DMRT1, H3F3A, MX1, ISG20, IREB2, MYOCD, NFATC1, NROB1, PTBP1, POU5F, SORBS2, SRPK3, SREBF1, HISTONE H4, STAT5A, HNF1B, DHX16, and RAG1. Of these, 15 were depleted to a level that rendered them no longer detectable by mass spectrometry. Five of these (HISTONE H4, STAT5A, HNF1B, DHX16, and RAG1) were depleted by 83-96%, indicating that there may be biochemical interactions mediating molecular interactions independent of the cellular compartment. We propose a nuclear protein profile of 15 proteins that may be used to assess and evaluate the efficiency of the decellularization process. The nuclear proteins comprising this profile are listed in Table 1. In addition to nuclear biomarkers, we also analyzed organellar cellular proteins including those associated with mitochondria, Golgi, and endoplasmic reticulum. We investigated five Golgi-specific proteins, 29 mitochondrial proteins, and 20 endoplasmic reticulum proteins that were depleted as a result of the decellularization process. We found that the Golgi proteins B3GALNT1, MAN1A1, FUT2, and MGAT4C represented a protein profile that may be suitable to monitor cellular depletion during decellularization processes, however the Golgi protein B3GNT5 was not fully depleted in our experiments potentially due to secondary interactions. Golgi proteins and the extent to which depletion was observed are listed in Table 2. We investigated mitochondrial proteins to determine a protein profile of mitochondrial biomarkers that may serve as a reference set to provide more reliable indicators of decellularization. We measured the protein content of 29 mitochondrial proteins and found that 19 of these (ACO2, AKAP10, GOT2, CPT1B, CYP11A1, ATC4D, CYBB, COX17, CYP11B1, GPAM, GATM, MUT, RHOT2, UCP2, UCP3, MT-ND5, SDHA, CUCLG1, and VARS2) were efficiently depleted through the decellularization process. In contrast, some mitochondrial proteins were detected after the decellularization process, indicating that they may not be reliable indicators of decellularization. Mitochondrial proteins and their extent of depletion are listed in Table 3. We investigated endoplasmic reticulum proteins by mass spectrometry before and after decellularization and found that a protein profile comprising 16 prevalent endoplasmic reticulum proteins may be considered as reliable biomarkers for decellularization processes. These include RPS13, RPS3, CYP8B1, RPL14, RPL6, CRYBB1, RPN2, HSPA5, FOLH1, HSPA1A, HSPA1B, HSPA1L, HMOX1, GANAB, ATP2A2, and VCP. Other endoplasmic reticulum proteins detected by mass spectrometry were not as efficiently removed from the tissue by the decellularization process and may indicate non-specific interactions that would render these proteins unreliable as indicators of decellularization. Endoplasmic reticulum proteins and the extent to which they were depleted during the decellularization process are listed in Appendix A, Table A1.
Analysis of cytosolic proteins showed that decellularization depleted the samples of these proteins as well. As shown in Appendix A, Table A3, we identified 40 cytosolic proteins that were depleted to a level exceeding 78%. These profiles may be useful to others that desire to track the efficiency of the decellularization process.
We utilized mass spectrometry to characterize the composition of the scaffold resulting from our process. We quantified extracellular matrix proteins before and after decellu-larization. As expected, we found that many of the noncollagenous extracellular matrix proteins were significantly depleted during the process, as shown in Table 4. However, collagens were retained within the scaffold, as shown in Figure 6.  Collagen alpha chains were analyzed before and after decellularization to assess collagenous composition of the resulting decellularized scaffold. The collagen alpha chains were detected in native cartilage by mass spectrometry prior to decellularization. After the decellularization process, nine of these were no longer detectable as shown in Figure 6. Blue horizontal bars indicate the composition of cartilage before the decellularization process, and red bars show the composition of the scaffold after the decellularization process. The resulting decellularized scaffold (red) contained COL2A1, COL1A1, COL6A3, COL1A2, COL11A2, COL11A1, COL4A5, COL3A1, COL5A2, COL16A1, COL5A1, COL4A2, COL4A3, COL5A3, COL27A1, COL13A1, COL4A1, COL12A1, and COL17A1. Minor contributions by COL7A1, COL22A1, COL28A1, COL4A4, COL9A2, COL6A2, COL14A1, COL8A2, COL18A1, COL23A1, COL6A6, and COL20A1 were detected in the final decellularized cartilage scaffold ( Figure 6). The final decellularized cartilage scaffold did not contain detectable levels of COL5A1, COL10A1, COL8A1, COL21A1, COL25A1, COL9A1, COL6A5, COL15A1, and COL26A1.
To test the biocompatibility of our scaffold, we seeded C28/I2 human chondrocyte cells onto the scaffold and maintained these constructs in culture. After one week, we observed cells associated with the surface of our scaffold, as shown in Figure 7. Scanning electron micrographs showed that the cells were maintained on the surface of the scaffold. Cellular clusters were observed after 8 months as shown in Figure 7b. Cellular response was evaluated using quantitative real-time PCR to monitor the expression of genes related to chondrogenic differentiation. Five candidate housekeeping genes were compared for all experimental conditions used in this study to identify those that remain constant and may therefore serve as appropriate housekeeping genes. GAPDH and HPRT1 were selected as the housekeeping gene for normalization in these experiments based on comparison to three other candidate housekeeping genes and were found to be stably expressed independent of experimental conditions based on minimal variance as shown in Figure 8.
The expression levels of genes related to chondrogenic differentiation were determined relative to housekeeping genes and these relative abundance values were reported as mean plus/minus standard deviation. Correlation analysis of gene expression was carried out to compare gene expression over time on conventional tissue culture plastic (Figure 9a) comparing day 7 to day 0 in culture. Correlation analysis was also carried out to compare gene expression of cells grown on the scaffold to the gene expression levels on tissue culture plastic after 7 days in culture (Figure 9b). The diagonal line indicates the trend expected if there is no change between the cartilage scaffold and plastic. Data points above the line reflect genes expressed at higher levels on the cartilage scaffold compared to plastic (Figure 9b). Data points below the diagonal line indicate genes that were expressed at higher levels on tissue culture plastic compared to the cartilage scaffold. Data points that fall on the line were not changed based on culture conditions. Figure 8. Analysis of variance of difference to determine most suitable housekeeping genes. Candidate housekeeping genes were considered in a pairwise fashion to determine which ones were the most consistently expressed independent of experimental condition within this study. GAPDH and HPRT1 displayed the minimum variance of difference over all experimental conditions and timepoints. Figure 9. Correlation analysis of gene expression during growth on tissue culture plastic and cartilage scaffold. C28/I2 cells were grown under conventional conditions on tissue culture plastic for one week (a) and compared to cells grown on cartilage scaffold for 7 days (b). The diagonal line in (a,b) indicates the trend expected if there was no change between conditions; day 0 versus day 7 on tissue culture plastic in (a) and day 7 on tissue culture plastic versus day 7 on cartilage scaffold in (b). Data points above the line reflect genes expressed at higher levels on the cartilage scaffold compared to plastic. Data points below the line indicate genes that were expressed at higher levels on plastic compared to the cartilage scaffold. Data points that fall on the line were not changed.
A total of 77 genes were analyzed during C28/I2 chondrocyte cell culture and differentiation for 7 days. We identified 52 genes that were upregulated from day 0 to day 7 on standard tissue culture plastic and therefore serve as indicators of chondrocyte phenotype. Of those 52 genes, 33 genes also showed an increase in expression when cells were cultured on decellularized 3D porcine scaffold. A total of 19 genes were upregulated in cells maintained on tissue culture plastics but were not upregulated significantly during the same time period by cells cultured on decellularized porcine cartilage scaffold. A total of 25 genes were upregulated in cells grown on decellularized scaffold that were not observed to be upregulated on tissue culture plastic under standard 2D culture conditions and these are summarized in Table 5 and Figure 10.

Discussion and Conclusions
In this study, we used decellularization to create an extracellular matrix scaffold that supports chondrocyte cell attachment and growth. We evaluated the decellularization process histologically and molecularly. Our conclusions identify new and novel biomarker profiles that may aid future cartilage decellularization efforts. The resulting scaffold was characterized using scanning electron microscopy, fluorescence microscopy, and proteomics. Cellular response to the decellularized scaffold was evaluated by quantitative real-time PCR and mass spectrometry for gene expression and proteomic analysis to analyze collagen content. Our approach demonstrated effective decellularization of a porcine cartilage scaffold by monitoring DNA content before and after decellularization.
Osteoarthritis is one of the leading causes of disability worldwide [1][2][3][4]. Many studies have investigated the use of various scaffolds as provisional chondro-inductive matrices [1,2,5]. In addition to biocompatibility, criteria for tissue engineering composites include (1) resorbability, (2) the ability to resist mechanical stresses, and (3) clinical relevance. Scaffolds must support cell differentiation and maintenance of a mature cartilage phenotype. However, there is no standard decellularization method to date. Previously published approaches include chemical, physical, or combinative methods [6][7][8]19].
The ultimate goal of decellularization is to remove all native genetic information and cellular components from the ECM. Surfactants, acid and bases, and enzymes make up the chemical and enzymatic portion of the process. Mechanical agents are also under study to determine the effectiveness for decellularization of a tissue or organ. These agents typically work by lysing cells through disrupting the phospholipid bilayer of the cell membrane. Ionic surfactants are widely used to remove cells and genetic material [19,20]. Treatments should be applied with continuous shaking [75,76]. Sodium dodecyl sulfate (SDS) currently meets the standard requirements of complete cell removal and elimination of at least 90% DNA [23]. SDS has been shown to damage structural properties if used at high concentration for long durations [6,7,20,21,23]. A comparison of five different decellularization treatments showed that several methods resulted in a significant reduction of DNA. Treatment with 2% SDS for eight hours resulted in the greatest decrease of DNA; with only minor decreased collagen content [6,7,20].
Some conflicting information exists in the literature regarding the duration of 1% SDS washes. The time to reach desired decellularization results range from 24 h to seven days [23,75,77]. The reported results indicate that the number of cells could be significantly reduced from engineered constructs. Higher or lower levels of DNA are most likely related to the thickness of tissue and the concentration and duration of specific detergents [20,23,77].
Only a few studies have explored decellularization of whole cartilage scaffolds for joint regeneration [23,75]. In this study, we used ethanol to defat samples and guanidine hydrochloride and sodium acetate to denature and remove noncollagenous components from cartilage dissected from the ear. Guanidine hydrochloride and sodium acetate have been shown to be effective for denature and remove noncollagenous components [75].
In this study, sodium hydroxide (NaOH) was used to inactivate cellular proteins and pathogens and denature DNA and RNA. Previous studies have shown that NaOH is an effective means of inactivating cellular proteins and pathogens and denaturating DNA and RNA. NaOH treatment removes cells and helps increase the porosity of the tissue [23,75].
Our study included the use of freeze-thaw cycles to help increase the porosity by forming more pores Freeze-thaw cycles have been shown to result in the formation of pores after ice crystal formation in addition to contributing to the disruption of resident chondrocytes. These cycles are often conducted in phosphate-buffered saline (PBS) solution to maintain physiological pH and osmolality, which additionally helps remove the residual reagents [7,19,20,23,[76][77][78][79].
In this study, DNase and RNase treatments were used to remove DNA and RNA. DNase and RNase may require as many as three cycles to accomplish complete depletion of native genetic material [7,19,20,23,[76][77][78][79]. Removal of 99% of genomic material was observed after a six-day wash cycle in our studies [23]. To quantify the DNA present in cartilage samples before and after decellularization, we used a DNA extraction process [21,79]. The DNA content of the sample was also assessed using Hoechst stain [21,23]. We concluded that the decellularization wash cycles successfully resulted in decellularizing the porcine cartilage scaffold. Traditionally, laboratories have monitored DNA content to determine if cells have been removed since all cells contain DNA. However, DNA may be present even when cells no longer exist due to potential interaction between the matrix and the DNA. Therefore, mass spectrometry was used to measure the removal of cellular proteins including those from the nucleus, mitochondria and Golgi, cytosol, rough endoplasmic reticulum, and plasma membrane. These additional measures more thoroughly demonstrated that the cells were removed at least to a level below the threshold of detection. In this study, lyophilization was used to help with cell disruption and removal of cellular components [78,79]. All samples were sterilized before using for cell culture.
To our knowledge, this study represents the first demonstration of biomarker analysis for decellularization of porcine ear for cartilage regeneration. Additionally, PCR was used to quantify mRNA expression of key chondrogenic differentiation markers expressed by cells after seeding on the scaffold. Chondrogenic marker genes were analyzed for significant changes between control cells grown on tissue culture plastic and those grown under conditions provided by the decellularized scaffold [76]. The cells expressed genes, as shown in PCR and proteomic data. There have been limited proteomic studies of cartilage, which may be because of the difficulty in determining the amount of protein contribution by the cells relative to the total protein contributed by the ECM [80]. The collagen fibers of the ECM were maintained during the process. Characterization of the scaffold showed that thirty-nine collagen alpha chains were detected in the cartilage prior to decellularization by mass spectrometry. After the decellularization process, nine of these were decreased below the limit of detection. While most of the collagens were maintained throughout the decellularization process, nine were depleted to the extent that they were no longer detectable by mass spectrometry.
SEM was used to visualize the surface topology of the decellularized scaffold [19,23,75,81]. The scaffold was shown to be less smooth and displayed exposed collagenous fibrillar networks after decellularization in SEM and more porous by histology when compared to the original material. SEM images showed cells could attach and proliferate on the surface of the scaffold.
Previous work has not fully analyzed what is left behind after decellularization. Future work should be carried out to better understand the material remaining after decellularization as it is important when evaluating laboratory-generated cartilage for patient-specific biocompatibility. This study is novel because it generates new knowledge by analyzing the protein profile of the decellularized tissue and contributes biomarkers other than well-established COL2A1 and ACAN. It provides an improved approach to monitoring decellularization and insight into acceptable markers for decellularization by looking at the protein profile of markers instead of just one or two proteins. The use of patient-specific MSCs or pre-chondrocytes will advance this line of research. A limitation of this study was that it focused on short-term changes. The scaffolds grown up to eight months showed an increase in cellular adhesion and proliferation. Future experiments in cell culture should extend duration as well as monitor protein synthesis and accumulation over time by proteomics, since protein expression does not always follow RNA expression. In this case, we will identify newly synthesized human proteins based on the number of unique peptides for human proteins for quantification. Additionally, it will be important to test biocompatibility. Another limitation may be the dense nature of native cartilage which restricts cell migration into the matrix. To address this limitation, future studies will include introduction of pores to allow penetration of cells into the center of the scaffold. An alternative approach that will be considered for future studies is devitalized cartilage to deliver higher concentrations of endogenous growth factors that may aid in cellular differentiation. In our procedure, soluble growth factors were removed due to the harsh conditions used.
Overall, this novel research shows promise that laboratory-generated cartilage may be a future alternative treatment option for individuals suffering from OA. Cartilage discs can be made to fit specific cartilage lesions. This approach aims to restore the patient's natural anatomy and prevent the need of a joint replacement. Using decellularization to create biomaterials can generate biocompatible scaffolds. Patient-specific chondrocytes may promote formation of a tissue that has superior compatibility for replacement or healing of damaged tissue.

Materials
Four pig ears were acquired from Wakefield Meats in Melba, Idaho. The pigs were estimated to be one year of age and of adult size.

Methods
The pig ears were shaved to remove hair, and a scalpel was used to remove remaining skin without damaging the underlying cartilage layer. Samples were placed in a 0.5 M NaOH bath overnight.
Tissue was transferred to a 1.0 M NaOH solution for three hours followed by transfer to a 70% ethanol solution and incubated at 40 • C with heating for three hours.
Pig ear cartilage was converted into 8 mm circular discs using a tissue punch. Twenty 8 mm punches were frozen to be used for characterization of the cartilage-derived scaffold before decellularization.
Residual cells were identified based on H&E staining. SEM was performed on the nondecellularized and decellularized tissues at this intermediate step. A second series of washes was carried out to remove the residual cells. Samples underwent another freezethaw cycle in DI water and 2% SDS with agitation. Samples were treated with DNase for 72 h at 37 • C with agitation, and samples were analyzed by histology, Hoechst staining, and SEM.
A third series of decellularization washing cycles was carried out followed by lyophilization. DNA removal was confirmed by extraction and purification of total DNA using DNeasy Kit for purification of total DNA from animal tissue (Qiagen).

Histology
The cartilage samples were fixed in 4% paraformaldehyde (PFA) for 1 h and then stored in 35% ethanol at 4 • C and then dehydrated, cleared in Histoclear, and embedded in paraffin. The tissue block was sectioned to achieve both transverse and cross-section configuration on the slide and stained with H&E and Hoechst stain. Hoechst stain (1 µg/mL) was placed on the sample for 5 min, then rinsed 3 times with PBS. The samples were then imaged on a Zeiss Confocal LSM 510 Meta microscope.

SEM Preparation and Imaging
Cartilage discs were fixed in 2.5% glutaraldehyde and 1% osmium tetroxide in Nanopure water. After fixation, samples underwent dehydration with 50%, 70%, 90%, and 100% ethanol. Critical point drying was performed for 10 cycles at 5 • C then heated to 35 • C. Samples were positioned onto an aluminum stub and sealed under vacuum. Gold sputtering was performed at 0.15 megabar (mbar) and 10 milliamp (mA) for 15 cycles of 60 s sputtering and 60 s of rest. Prepped samples were examined at an accelerating voltage of 15 kilovolts (kV) using the secondary electron detector.

PCR
RNA from cells seeded on cartilage scaffold or from cells grown on conventional tissue culture plastic was extracted using TRIzol. Complementary DNA (cDNA) synthesis was carried out using the RT2 First Strand Kit (Qiagen) followed by qRT-PCR in a 96-well plate using a Roche LightCycler96 ® . The samples underwent one cycle for 10 min at 95 • C and then 45 cycles of 15 s at 95 • C and 1 min of 60 • C. Analyzed genes included extracellular matrix proteins, matrix remodeling enzymes, and cell adhesion molecules. PCR primers were purchased from Qiagen (Cat. no. 330231 PAHS-013ZA). Relative gene expression levels, mean of three replicates plus/minus standard deviation, were expressed with respect to housekeeping genes determined empirically for this study.

Protein Concentration Determination
The total protein concentration of the homogenate for all samples was determined using Pierce TM BCA (Bicinchoninic Acid) Protein Assay Kit, Thermo Scientific ® .

Mass Spectrometry and Proteomics
Proteins from nondecellularized, decellularized, and recellularized scaffolds were homogenized and extracted using RIPA buffer (Millipore, Billerica, MA, USA) Twenty micrograms of total protein from each sample was digested with Trypsin/Lys C mix (Promega, Madison, WI, USA) following the manufacturer's instruction. Resulting peptide mixtures were chromatographically separated on a reverse-phase C18 column (10 cm × 75 µm, 3 µm, 120 Å) and analyzed on a Velos Pro Dual-Pressure Linear Ion Trap mass spectrometer (Thermo Fisher Scientific).
Peptide spectral matching and porcine and human protein identification were achieved by database search using Sequest HT algorithms in a Proteome Discoverer 2.2 (Thermo Fisher Scientific). Raw spectrum data were searched against the UniProtKB/Swiss-Prot and TrEMBL porcine protein databases and Swiss-Prot human protein database (25 May 2019). The main search parameters included trypsin, maximum missed cleavage site of two, precursor mass tolerance of 1.5 Da, fragment mass tolerance of 0.8 Da, and variable modification of oxidation/hydroxylation of methionine, proline, and lysine (+15.995 Da). A decoy database search was performed to calculate a false discovery rate (FDR). Proteins containing one or more peptides with FDR ≤ 0.05 were considered positively identified and reported. For all proteins, the total number of peptide spectral matches (PSMs) reported by the Protein Discoverer 2.2 was used for quantification. The mass spectrometry analysis used three samples at each condition and time point.
Twenty-four-well plates were prepared with 300 µL of agarose gel in the bottom of each well and sterilized cartilage scaffold discs were seeded with 500,000 C28/I2 cells each. Characterization took place at seven days for PCR analysis and one week and 8 months for SEM analysis.

Statistical Analysis
Selection of housekeeping genes for qRT-PCR was based on pairwise analysis of variance for differences between cycle threshold values for five candidate housekeeping genes from 15 samples within this study. Relative expression of genes of interest was analyzed relative to average values for GAPDH and HPRT1 and expressed as mean plus/minus standard deviation. Log-transformed gene expression data were subject to a paired T-test to determine if the differences in mean values for relative gene expression were statistically significant with p < 0.05.

Data Availability Statement:
The datasets used and/or analyzed during the current study are available from the corresponding author upon request.

Conflicts of Interest:
The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.