https://orcid.org/0000-0001-9852-773X
Figures
Abstract
Idiopathic pulmonary fibrosis (IPF) is a severe lung disease with a poor prognosis and few treatment options. In the most widely used experimental model for this disease, bleomycin is administered into the lungs of mice, causing a reaction of inflammation and consequent fibrosis that resembles the progression of human IPF. The inflammation and fibrosis together induce changes in gene expression that can be analyzed with reverse transcription quantitative real-time PCR (RT-qPCR), in which accurate normalization with a set of stably expressed reference genes is critical for obtaining reliable results. This work compares ten commonly used candidate reference genes in the late, fibrotic phase of bleomycin-induced pulmonary fibrosis and ranks them from the most to the least stable using NormFinder and geNorm. Sdha, Polr2a and Hprt were identified as the best performing and least variable reference genes when alternating between normal and fibrotic conditions. In order to validate the findings, we investigated the expression of Tnf and Col1a1, representing the hallmarks of inflammation and fibrotic changes, respectively. With the best three genes as references, both were found to be upregulated relative to untreated controls, unlike the situation when analyzed solely with Gapdh, a commonly used reference gene. We therefore recommend Sdha, Polr2a and Hprt as reference genes for RT-qPCR in the 4-week bleomycin challenge that represents the late fibrotic phase.
Citation: Norman O, Koivunen J, Mäki JM, Pihlajaniemi T, Heikkinen A (2022) Identification of suitable reference genes for normalization of reverse transcription quantitative real-time PCR (RT-qPCR) in the fibrotic phase of the bleomycin mouse model of pulmonary fibrosis. PLoS ONE 17(10): e0276215. https://doi.org/10.1371/journal.pone.0276215
Editor: Kyung-Wan Baek, Gyeongsang National University, REPUBLIC OF KOREA
Received: August 12, 2022; Accepted: October 2, 2022; Published: October 17, 2022
Copyright: © 2022 Norman et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the manuscript and its Supporting Information files.
Funding: This study was supported by grants from the Sigrid Jusélius Foundation (TP) (https://www.sigridjuselius.fi/en/), the Jane and Aatos Erkko Foundation (TP) (https://jaes.fi/en/) and the Emil Aaltonen Foundation (ON) (https://emilaaltonen.fi/). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Idiopathic pulmonary fibrosis (IPF) is the most common of the interstitial pneumonias and is a progressive incurable disease with a poor prognosis. Its yearly incidence in Europe and North America is estimated to be 2.8–18 cases per 100 000 persons, which is similar to diseases such as stomach or brain cancer [1]. The cause of the disease has been viewed as being a combination of various genetic and environmental cues that lead to the proliferation of fibroblasts and the accumulation of pathological extracellular matrix, a process in which changes in gene expression are instrumental, ultimately leading to the alveolar space being replaced with connective tissue. Multiple attempts have been, and continue to be, made to develop an anti-fibrotic drug, but despite pre-clinical studies showing positive effects, most compounds have not shown comparable performance in clinical trials [2] and to date only two anti-fibrotic pharmaceuticals that retard the progression of IPF have been approved for use [3]. Thus we still need a better understanding of the disease and more effective therapy, as median survival remains at the level of 2–4 years after diagnosis [1], thereby empowering further investigations.
Reverse transcription quantitative real-time PCR (RT-qPCR) is a widely used method for assessing gene expression, the key to its accuracy being valid normalization of the expression of the gene of interest to a set of stably expressed reference genes in order to minimize the effects of differences in the quantity and quality of the RNA between individual samples. The reference genes are often housekeeping genes that have a role in the cell’s basic vital activities such as energy metabolism or protein synthesis. A systematic review by Chapman and Waldenström [4] shows that the most commonly used reference genes are beta-actin (Actb), glyceraldehyde-3-phosphate dehydrogenase (Gapdh) and the ribosomal RNA 18S subunit (Rn18s), although even their expression does not always remain stable but often varies between experimental groups. Thus the optimal reference genes have to be verified for the experiment in question in order to achieve accurate results. Gapdh, in particular, has numerous pseudogenes that may confound the analysis [5, 6]. In addition, it is upregulated under hypoxic conditions and most inflammatory tissues exhibit a degree of hypoxia due to edema [7]. In the case of Actb attention should be paid to primer design, as it shares 91% of its sequence with leucine-rich repeat-containing 58 (Lrrc58) [6].
Different methods have been developed to determine which genes are most suitable for normalization, some of the most commonly used algorithms being NormFinder [8] and geNorm [9]. NormFinder employs a model-based approach that estimates intra- and intergroup variance for each candidate gene. This approach relies on the assumption that the average expression of the candidate genes will be the same in both experimental groups. geNorm uses a pairwise comparison approach in which the candidate genes are ranked according to the similarity of their expression profiles. This method requires that the candidate genes be regulated independently of each other. Unlike NormFinder, geNorm does not compare the expression levels of the candidate genes between experimental groups but instead comparisons are made between all the samples. RNA sequencing has also been used for finding stably expressed genes, but Sampathkumar et al. [10] recently demonstrated that the results are similar to the conventional approach.
Bleomycin is an anti-neoplastic drug that has a potential to cause lung fibrosis when used in large doses [11]. The histological changes it causes in mouse lungs are relatively similar to those seen in human IPF and thus the bleomycin model is the most widely used experimental animal model for studying pulmonary fibrosis [12]. The reaction to the administration of bleomycin to the lungs consists of an initial phase of acute lung injury, inflammation and the development of fibrosis that lasts for around two weeks, followed by a phase of established fibrosis from two to four weeks after bleomycin administration and ultimately slow resolution of the fibrosis [2]. A previous study has explored suitable reference genes in the initial inflammatory phase of the bleomycin mouse model [13], but to the best of our knowledge there are no reports on reference genes suitable for the normalization of RT-qPCR in the fibrotic phase of the bleomycin model by comparison with the untreated state. We present here data on the expression of ten commonly used reference genes during the fibrotic phase of the bleomycin mouse model and rank them for their utility in normalizing gene expression.
Materials and methods
Ethics approval
Permission for the administration of bleomycin and the induction of fibrosis was obtained from the Finnish Animal Care and Use Committee (ESAVI/8179/04.10.07/2017). All the mouse experiments complied with the European Community Council Directive on the protection of animals used for scientific purposes (September 22, 2010; 2010/63/EEC), national legislation and the regulations for the care and use of laboratory animals.
The bleomycin mouse model of pulmonary fibrosis
The mice used here were part of a larger study on the effect of collagen XIII deficiency on pulmonary fibrosis and were wild-type littermates of collagen XIII (Col13a1) knockout mice of both sexes. The generation of the mice is described in detail by Latvanlehto et al. [14]. The genetic background of the mice originates from sv129 embryonic stem cells followed by cross breeding with the C57BL/6JOlaHsd strain for 17 generations and the C57BL/6NClr strain for 3–4 generations. The mice were maintained in a specific pathogen free (SPF) facility at +21°C with a 12-hour light-dark cycle and were fed Teklad global 18% protein rodent diet with ad libitum access to food and water. From 6 weeks of age onwards the mice were transferred to animal research facilities and the diet was supplemented with banana and hazelnut cocoa spread to minimize malnutrition and mortality during the development of pulmonary fibrosis. 8-week-old mice were anesthetized with isoflurane and given 1,25 U/kg bleomycin (Baxter) diluted with sterile isotonic saline to a volume of 1.67 microliters/g of body weight by oropharyngeal aspiration as described by Barbayianni et al. [15]. Untreated mice served as controls. After 4 weeks of follow-up the mice were sacrificed by injection of fentanyl, midazolam and medetomidine and exsanguination. The left lung was fixed with intratracheal fixation with 4% paraformaldehyde (PFA) at a pressure of 25 cm H2O for 2 minutes to open the airways with the right bronchi ligated to prevent fixation. The right lung lobes were dissected, snap frozen in liquid nitrogen and stored at -80°C. The right postcaval lobe was used for extraction of RNA. After dissection of the right lobes, the left lobe was immersed in 4% PFA for 24 h, embedded in paraffin and used for histological analysis.
Total RNA extraction and cDNA synthesis
Total RNA was isolated from the frozen right postcaval lung lobes using an RNeasy kit (Qiagen) according to the manufacturer’s instructions. All the samples were processed in one batch. The quantity and purity of the RNA were assessed by measuring absorbances at 260 and 280 nm with a Nanodrop spectrophotometer (Thermo Scientific). The quality of the RNA was assessed by capillary electrophoresis using a Bioanalyzer 2100 (Agilent). Based on their fibrotic histological appearance and adequate RNA quality, 9 (4 male, 5 female) bleomycin-treated samples and 10 (5 male, 5 female) untreated samples were selected for use in the analyses. cDNA was synthetized with an iScript cDNA Synthesis kit (Bio-Rad) according to the instructions, using 1000 ng of total RNA and including all the samples in one batch. Genomic DNA was not digested in order to assess whether it affected the results.
Quantitative real-time PCR
The RT-qPCR reactions were performed using 5 μl iTaq Universal SYBR Green Supermix (Bio-Rad), 1.2 μl of forward and reverse primers at 2.5 μM, 2 μl of cDNA template diluted 1:10 and 0.6 μl of water to a final volume of 10 μl. All the reactions were performed in triplicate and those for a given target gene were all performed on the same plate to minimize batch effects. Efficiency analyses were mostly performed separately from the sample analyses. The instrument used was a CFX 96 Real-Time PCR system (Bio-Rad). The thermocycling protocol consisted of 3 minutes at 95°C followed by 40 cycles of denaturation for 15 seconds at 95°C and annealing for 45 seconds. The annealing temperature was 55°C for the primers for the succinate dehydrogenase complex subunit A (Sdha), 64°C for the RNA polymerase II subunit A (Polr2a) and 60°C for all the other targets. Melt curves were created by heating from 55°C to 95°C with 5-second increments of 0.5°C and used to assess the specificity of the PCR.
Primers
The primer sequences were mostly collected from previous publications (Table 1). To minimize unspecific amplification, the primers should be specific to the gene of interest, not form dimers and the amplicon should stretch across an exon-exon junction to distinguish possible genomic DNA amplification. The primer specificity was verified with Primer-BLAST [16]. The primers for Gapdh and peptidylprolyl isomerase A (Ppia) both amplified several pseudogenes according to BLAST, whereas the other primers were specific to the gene of interest. Amplification of a series of four consecutive 1:5 dilutions of cDNA pooled from the samples with three replicates for each dilution was used to generate a standard curve with quantification cycle (Cq) values on the Y-axis and Log10 of the dilution on the X-axis, a line fitted to the points and primer efficiency (E) calculated with the equation: [17]. The standard curves for all the primers, were within the linear dynamic range as indicated by R2 of the standard curves, and the Cq values of all the samples fell inside the linear dynamic range.
Histological analysis
After fixation the left lungs were embedded in paraffin in the frontal plane and cut into 5 μm sections at 200 μm intervals. The sections were stained with Masson’s trichrome. One section immediately before the main bronchus and one after were used for histological analysis. The slides were scanned at 40 x magnification with a NanoZoomer S60 slide scanner (Hamamatsu). Regions of interest with only lung tissue in them were drawn with QuPath [25] and the images downsampled by a factor of 5 with a script [26] to enable processing with ImageJ. Ten 0.9 x 0.9 mm fields were randomly sampled from each section using an ImageJ script modified from [27] and graded using the modified Ashcroft scale [28] independently by 2 observers blinded to the treatment. The scores from all the observers for a given field were averaged, after which the scores for all the fields of the sample were averaged to obtain the score for the entire sample.
Data analysis
Bio-Rad CFX Maestro 2.2 software was used to analyze the raw data. Automated functions within the software were used for baseline correction, thresholding and Cq determination. For Fig 3, the raw Cq values were corrected for primer-specific PCR efficiency using the formula . The melt curves were inspected to confirm the specificity of the amplification. NormFinder for R (https://moma.dk/normfinder-software) [8] and GeNorm [9] for Bio-Rad CFX Maestro were used according to their respective instructions. For the NormFinder analysis the raw Cq values were transformed to relative quantities on a linear scale using the formula
, where
. Expression of the collagen I alpha 1 chain (Col1a1) and tumor necrosis factor (Tnf) normalized to the combination of Sdha, Polr2a and hypoxanthine guanine phosphoribosyl transferase (Hprt1) was calculated using CFX Maestro with the formula:
, where GOI = gene of interest, as proposed by Pfaffl [29] and Vandesompele et al. [9], and expression of Col1a1 and Tnf normalized to Gapdh was calculated using the formula
.
Results
Level of pulmonary fibrosis
To confirm the suitability of the samples for investigating pulmonary fibrosis, the degree of fibrosis in the lungs was graded with a modified Ashcroft score [28]. The bleomycin-treated samples showed clear fibrotic changes with a mean score of 3.2 (Standard deviation (SD) 0.98), while the untreated mice had a score of 0.4 (SD 0.086) and exhibited no signs of pulmonary fibrosis (Fig 1).
a) Masson’s trichrome staining of an untreated left lung and b) of a bleomycin-treated left lung. Fibrotic changes are indicated by arrows. Scale bars 1 mm. c) Example of a 0.9 mm wide close-up image of a bleomycin-treated lung produced by the random sampling procedure, with a modified Ashcroft score of 5. A nonlinear tone curve adjustment was applied to all images to correct white balance. d) Modified Ashcroft scores for the experimental groups, untreated (round symbols) and bleomycin-treated (square symbols). Lines at means. n = 9–10. Mann-Whitney U test.
RNA integrity
Given that RNA integrity (RIN) values above 5 are considered sufficient for RT-qPCR analysis [30], the average RIN values recorded here were 6.6 (SD 0.45) for untreated and 6 (SD 0.51) for bleomycin-treated samples (Fig 2). The values were on average 10%, or 0.6 units, lower in the bleomycin-treated group (Fig 2), possibly reflecting different conditions in the fibrotic tissue, since the sample collection and RNA extraction procedures were the same for all the samples. Nevertheless, given the similarity of the values, the samples should be readily comparable.
RIN values for the untreated (round points) and bleomycin-treated samples (square points). Lines at means. n = 9–10. Mann-Whitney U test. Dashed line at RIN = 5.
Expression characteristics of the candidate genes
The primer efficiency was between 85% and 110% for all the primer pairs, which is considered sufficient for RT-qPCR [30]. This efficiency was taken into account when correcting the Cq values and when calculating the relative quantities. All the primer pairs amplified products that had efficiency-corrected Cq values between 13 and 30 (Fig 3), and each primer pair showed uniform single-peak melting curves for all the samples. No template control (NTC) Cq values were over 38 for all the primers except for Rn18s, for which they were above 30, and for Sdha, for which they were above 33, all indicating sufficiently low background signal levels. No reverse transcription (NRT) controls for Gapdh had a Cq value of 22.13 while the highest Cq for a sample was 21.21 and Ppia had a Cq of 23.90 with the highest Cq for a sample being 22.0. NRT controls for all other genes had Cq values at least 7 cycles higher than the sample with the highest Cq, indicating that the primers did not significantly amplify genomic DNA. The NRT values for Gapdh and Ppia indicate that up to 35% and 22%, respectively, of the expression detected with RT-qPCR could originate from pseudogenes or other unspecific amplification and cause variation in the results. Neither the primers for Gapdh nor those for Col1a1 were exon junction-spanning, but as the NRT control for Col1a1 did not exhibit unspecific amplification from genomic DNA, the unspecific amplification in the NRT control for Gapdh is most likely to have originated from pseudogenes or similar sequences.
Untreated: untreated samples. Bleomycin: bleomycin-treated samples. Each box has lines for minimum, maximum and mean values. X-axis truncated; 40 cycles were run. n = 9–10.
Reference gene stability
Normfinder and geNorm for Bio-Rad CFX Maestro algorithms were used to rank the reference genes in terms of their stability, and averages were calculated from these ranks to obtain the overall ranking (Fig 4). Sdha performed best in both analyses, and Ppia, Polr2a and Hprt were the three next best genes, although the potential pseudogene problem of Ppia makes it a suboptimal choice. The most commonly used reference gene, Gapdh, showed the poorest performance according to both programs, and Actb and Rn18s performed less well than most other candidates. Thus we suggest that Sdha, Polr2a and Hprt would form the best combination of reference genes for assessing bleomycin-induced pulmonary fibrosis at the 4-week time point.
Values from a) NormFinder and b) geNorm, and c) average of ranks based on the two programs.
Validation of candidate reference genes
In order to validate the findings, the expression levels of Col1a1 and Tnf, which are expected to be increased under fibrotic conditions, were compared between the experimental groups using Sdha, Polr2a and Hprt as the reference genes (Fig 5). The expression of Col1a1 appears to be slightly upregulated in the bleomycin-treated samples and Tnf clearly so, as expected. When only Gapdh was used as a reference gene, however, neither Col1a1 nor Tnf showed any significant differences between the groups.
Changes in the expression of Col1a1 (a, c) and Tnf (b, d) in bleomycin-treated mice (square points) relative to untreated mice (round points) when Sdha, Polr2a and Hprt are used as reference genes (a, b) and Gapdh as reference (c, d). Lines at medians. n = 9–10. Mann-Whitney U test.
Discussion
We present here a strategy for determining the optimal reference genes for RT-qPCR normalization in the widely used bleomycin model of pulmonary fibrosis at 4 weeks post-treatment, i.e. the late fibrotic stage. Accurate normalization is the key to obtaining reliable results, and selecting stably expressed reference genes for normalization can help avoid excessive variation in the results. As established in many earlier studies, a single reference gene is rarely enough for reliable results, so that at least three genes should be used after determining the invariability of their expression [9, 31, 32].
The candidate genes we selected include some of the most popular reference genes used in RT-qPCR analyses, genes having various functions in the cell’s basic activities that are integral to its survival, and therefore genes that may be expected to be expressed steadily under different conditions. We picked primer sequences that were found to work well in previously published studies and were careful to select primers that were specific to the gene of interest, especially with Actb. This is because some of the primers used previously, and also some commercially available primers, were predicted to be unspecific and also to amplify Lrrc58, with which Actb shares 91% of its sequence. The primers for Gapdh and Ppia, however, were found to be nonspecific, as seen from their high NRT values, and hence were deemed unreliable and nonoptimal for acting as reference genes. The fact that the NTC for both genes failed to show significant amplification argues against primer dimerization, and thus the amplification may have been derived from pseudogenes. The pseudogene problem attached to Gapdh has been demonstrated earlier [5], but its continued ubiquitous use as a reference gene warranted its inclusion in this study. The results further demonstrate that the use of Gapdh as the sole reference gene is not appropriate without determining the invariability of its expression by comparing it with a few alternatives, although we would rather recommend reference genes that are free of pseudogene issues. As shown in Fig 5, inappropriate normalization could lead to false results. Actb and Rn18s, the other most commonly used reference genes [4], showed poor performance compared with the remaining candidates studied here. The main issue with Rn18s is that the product is rRNA instead of mRNA and thus its expression is governed by different factors from those lying behind the genes of interest [32].
In this study we used two different algorithms to determine the stability of 10 candidate genes in the late fibrotic phase of bleomycin-induced pulmonary fibrosis in mice. Although the two algorithms represent different approaches to calculating reference gene stability, the best-performing genes are the same in both programs, a fact that highlights their stability. A previous study focusing on the initial inflammatory phase in the bleomycin-induced pulmonary fibrosis model found Hprt, Ppia and Sdha to be the most stable reference genes [13], and our results confirmed the very same genes, with the addition of Polr2a, as acting reliably as reference genes in RT-qPCR at a later stage in pulmonary fibrosis. Taken together, we propose that Hprt, Polr2a and Sdha form a reliable set of reference genes for RT-qPCR in the bleomycin model of pulmonary fibrosis.
Supporting information
S1 Table. Data underlying the findings described.
Cq values and plots of standard curves used for determination of primer efficiency, raw Cq values for all samples, efficiency-corrected Cq values and relative quantities, datafile used for NormFinder analysis, and individual values behind the graphs and results.
https://doi.org/10.1371/journal.pone.0276215.s001
(XLSX)
Acknowledgments
The authors thank Jaana Peters and Päivi Tuomaala for technical assistance and Oulu Laboratory Animal Center (OULAC) for care of the animals. The work was supported by Biocenter Oulu, Sequencing Center, and the Transgenic and Tissue Phenotyping Core Facility endowed by the University of Oulu, Finland, and Biocenter Finland.
References
- 1. Richeldi L, Collard HR, Jones MG. Idiopathic pulmonary fibrosis. The Lancet. 2017;389: 1941–1952.
- 2. Jenkins RG, Moore BB, Chambers RC, Eickelberg O, Königshoff M, Kolb M, et al. An Official American Thoracic Society Workshop Report: Use of Animal Models for the Preclinical Assessment of Potential Therapies for Pulmonary Fibrosis. Am J Respir Cell Mol Biol. 2017;56: 667–679. pmid:28459387
- 3. Glass DS, Grossfeld D, Renna HA, Agarwala P, Spiegler P, DeLeon J, et al. Idiopathic pulmonary fibrosis: Current and future treatment. Clin Respir J. 2022;16: 84–96. pmid:35001525
- 4. Chapman JR, Waldenström J. With Reference to Reference Genes: A Systematic Review of Endogenous Controls in Gene Expression Studies. PLoS ONE. 2015;10: e0141853. pmid:26555275
- 5. Sun Y, Li Y, Luo D, Liao DJ. Pseudogenes as Weaknesses of ACTB (Actb) and GAPDH (Gapdh) Used as Reference Genes in Reverse Transcription and Polymerase Chain Reactions. PLOS ONE. 2012;7: e41659. pmid:22927912
- 6. Mehta A, Dobersch S, Dammann RH, Bellusci S, Ilinskaya ON, Braun T, et al. Validation of Tuba1a as Appropriate Internal Control for Normalization of Gene Expression Analysis during Mouse Lung Development. Int J Mol Sci. 2015;16: 4492–4511. pmid:25723738
- 7. Graven KK, Yu Q, Pan D, Roncarati JS, Farber HW. Identification of an oxygen responsive enhancer element in the glyceraldehyde-3-phosphate dehydrogenase gene. Biochim Biophys Acta. 1999;1447: 208–218. pmid:10542317
- 8. Andersen CL, Jensen JL, Ørntoft TF. Normalization of Real-Time Quantitative Reverse Transcription-PCR Data: A Model-Based Variance Estimation Approach to Identify Genes Suited for Normalization, Applied to Bladder and Colon Cancer Data Sets. Cancer Res. 2004;64: 5245–5250. pmid:15289330
- 9. Vandesompele J, De Preter K, Pattyn F, Poppe B, Van Roy N, De Paepe A, et al. Accurate normalization of real-time quantitative RT-PCR data by geometric averaging of multiple internal control genes. Genome Biol. 2002;3: research0034.1. pmid:12184808
- 10. Sampathkumar NK, Sundaram VK, Danthi PS, Barakat R, Solomon S, Mondal M, et al. RNA-Seq is not required to determine stable reference genes for qPCR normalization. PLOS Comput Biol. 2022;18: e1009868. pmid:35226660
- 11. Della Latta V, Cecchettini A, Del Ry S, Morales MA. Bleomycin in the setting of lung fibrosis induction: From biological mechanisms to counteractions. Pharmacol Res. 2015;97: 122–130. pmid:25959210
- 12. Carrington R, Jordan S, Pitchford SC, Page CP. Use of animal models in IPF research. Pulm Pharmacol Ther. 2018;51: 73–78. pmid:29981850
- 13. Della Latta V, Cabiati M, Burchielli S, Frenzilli G, Bernardeschi M, Cecchettini A, et al. Lung inflammation after bleomycin treatment in mice: Selection of an accurate normalization strategy for gene expression analysis in an ex-vivo and in-vitro model. Int J Biochem Cell Biol. 2017;88: 145–154. pmid:28526615
- 14. Latvanlehto A, Fox MA, Sormunen R, Tu H, Oikarainen T, Koski A, et al. Muscle-Derived Collagen XIII Regulates Maturation of the Skeletal Neuromuscular Junction. J Neurosci. 2010;30: 12230–12241. pmid:20844119
- 15. Barbayianni I, Ninou I, Tzouvelekis A, Aidinis V. Bleomycin Revisited: A Direct Comparison of the Intratracheal Micro-Spraying and the Oropharyngeal Aspiration Routes of Bleomycin Administration in Mice. Front Med. 2018;5: 269. pmid:30320115
- 16. Ye J, Coulouris G, Zaretskaya I, Cutcutache I, Rozen S, Madden TL. Primer-BLAST: a tool to design target-specific primers for polymerase chain reaction. BMC Bioinformatics. 2012;13: 134. pmid:22708584
- 17.
Rasmussen R. Quantification on the LightCycler. In: Meuer S, Wittwer C, Nakagawara K-I, editors. Rapid Cycle Real-Time PCR. Berlin, Heidelberg: Springer Berlin Heidelberg; 2001. pp. 21–34. https://doi.org/10.1007/978-3-642-59524-0_3
- 18. Han LQ, Yang GY, Zhu HS, Wang YY, Wang LF, Guo YJ, et al. Selection and use of reference genes in mouse mammary glands. Genet Mol Res. 2010;9: 449–456. pmid:20391330
- 19. Gong K, Qu B, Wang C, Zhou J, Liao D, Zheng W, et al. Peroxisome Proliferator-Activated Receptor α Facilitates Osteogenic Differentiation in MC3T3-E1 Cells via the Sirtuin 1-Dependent Signaling Pathway. Mol Cells. 2017;40: 393–400. pmid:28614912
- 20. Dong Y-F, Chen Z-Z, Zhao Z, Yang D-D, Yan H, Ji J, et al. Potential role of microRNA-7 in the anti-neuroinflammation effects of nicorandil in astrocytes induced by oxygen-glucose deprivation. J Neuroinflammation. 2016;13: 60. pmid:26961366
- 21. YOKOYAMA T, OMOTEHARA T, HIRANO T, KUBOTA N, YANAI S, HASEGAWA C, et al. Identification of reference genes for quantitative PCR analyses in developing mouse gonads. J Vet Med Sci. 2018;80: 1534–1539. pmid:30175754
- 22. Petäistö T, Vicente D, Mäkelä KA, Finnilä MA, Miinalainen I, Koivunen J, et al. Lack of collagen XVIII leads to lipodystrophy and perturbs hepatic glucose and lipid homeostasis. J Physiol. 2020;598: 3373–3393. pmid:32449518
- 23. Aikio M, Elamaa H, Vicente D, Izzi V, Kaur I, Seppinen L, et al. Specific collagen XVIII isoforms promote adipose tissue accrual via mechanisms determining adipocyte number and affect fat deposition. Proc Natl Acad Sci. 2014;111: E3043–E3052. pmid:25024173
- 24. Hossen MJ, Yang WS, Kim D, Aravinthan A, Kim J-H, Cho JY. Thymoquinone: An IRAK1 inhibitor with in vivo and in vitro anti-inflammatory activities. Sci Rep. 2017;7: 1–12.
- 25. Bankhead P, Loughrey MB, Fernández JA, Dombrowski Y, McArt DG, Dunne PD, et al. QuPath: Open source software for digital pathology image analysis. Sci Rep. 2017;7: 16878. pmid:29203879
- 26. Bankhead P. Script of the Day: Extracting image regions to ImageJ. In: Pete’s blog [Internet]. 7 Mar 2018 [cited 29 Mar 2022]. Available: https://petebankhead.github.io/qupath/scripting/2018/03/07/script-extracting-image-regions-to-imagej.html
- 27. Mason D. Random Access MemoROI. In: Post-Acquisition [Internet]. 21 Jun 2018 [cited 29 Mar 2022]. Available: https://postacquisition.wordpress.com/2018/06/21/random-access-memoroi/
- 28. Hübner R-H, Gitter W, El Mokhtari NE, Mathiak M, Both M, Bolte H, et al. Standardized quantification of pulmonary fibrosis in histological samples. BioTechniques. 2008;44: 507–511, 514–517. pmid:18476815
- 29. Pfaffl MW. A new mathematical model for relative quantification in real-time RT–PCR. Nucleic Acids Res. 2001;29: e45. pmid:11328886
- 30. Fleige S, Pfaffl MW. RNA integrity and the effect on the real-time qRT-PCR performance. Mol Aspects Med. 2006;27: 126–139. pmid:16469371
- 31. Dheda K, Huggett JF, Chang JS, Kim LU, Bustin SA, Johnson MA, et al. The implications of using an inappropriate reference gene for real-time reverse transcription PCR data normalization. Anal Biochem. 2005;344: 141–143. pmid:16054107
- 32. Kozera B, Rapacz M. Reference genes in real-time PCR. J Appl Genet. 2013;54: 391–406. pmid:24078518