Introduction

Bats are increasingly recognized as reservoir hosts of highly-virulent pathogens, such as Filoviruses, Lyssaviruses, Paramyxoviruses and Coronaviruses, including severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), which causes the current global pandemic1. Egyptian fruit bats, Rousettus aegyptiacus, have been identified as putative reservoir hosts of Marburg virus2, Kasokero virus3 and Sosuga virus4, and were shown to be susceptible to experimental challenge with SARS-CoV-25 and Rift valley fever phlebovirus6. These viruses may cause severe diseases with high mortality rates in humans. However, bats show minimal and often even no clinical manifestation upon natural infection. Under experimental conditions, Rousettus bats infected with high doses of Ebola7, SARS-CoV8 or SARS‐CoV‐25 only support transient viral replication and display limited pathology. Several hypotheses could explain the reservoir potential of bats. As the only mammals capable of powered flight, their body temperature can reach 42 °C during flight9. Hence, ‘flight-as-fever’ has been postulated as a unique mechanism conferring effective immune defence6. Further, the black flying fox, Pteropus alecto, constitutively expresses IFN-α10, and gene loci of type I interferons (IFN-I) in R. aegyptiacus, are markedly expanded11, indicating a potential contribution of boosted IFN-I signalling to antiviral immunity in bats. In addition to the enhanced antiviral immunity, bats can suppress excessive inflammation which may explain their lack of symptoms during viral infection1,12. For example, TNFα expression in Eptesicus fuscus cells is abolished upon poly(I:C) stimulation13, NLRP3-mediated inflammasome activation is impaired in P. alecto14 and Myotis davidii12, and R. aegyptiacus does not upregulate pro-inflammatory genes (CCL8, FAS and IL6) upon Marburg virus infection15. Enhanced antiviral immunity along with reduced inflammation likely explains the ability of bats to harbour high-impact pathogens in absence of clinical disease. Confirmation of these findings in distinct bat species as well as elucidation of novel immune mechanisms contributing to the reservoir potential of bats require accurate monitoring of their immune responses.

The knowledge about the exceptional immune system of bats has significantly advanced during the past decade16. However, experimental tools to systematically investigate bat immune responses, such as species-specific or cross-reactive antibodies, are largely missing17,18,19. Accordingly, investigations on host immunity heavily rely on gene transcription profiling by qRT-PCR. This method is sensitive, specific, highly reproducible and accurate20,21. A critical step in qRT-PCR setup is selection of several stable reference genes. The inclusion of such reference genes is crucial for gene expression normalization and subsequent data interpretation. The ideal reference genes should maintain stable expression levels across diverse tissues and cell types as well as under different experimental conditions22. Considering that bats display unique physiological features, notably oscillating metabolic rates and core body temperature depending on flying and roosting phases23,24,25,26, the expression stability of reference genes must be evaluated in context of these physiologically relevant conditions. Multiple reference genes, including glyceraldehyde-3-phosphate dehydrogenase (GAPDH), actin-beta (ACTB), small nuclear ribonucleoprotein Sm D3 (SNRPD3) and 18S ribosomal RNA (18S rRNA) have been employed in gene expression studies of P. alecto, E. fuscus, M. davidii and other bats13,14,27,28,29. However, a comprehensive analysis of reference genes, particularly their expression stability under oscillating temperatures, has not been performed in bats, including the model bat R. aegyptiacus.

Here, we provide a first in-depth validation of four reference gene candidates, including ACTB, GAPDH, eukaryotic translation elongation factor 1 alpha 1 (EEF1A1) and phosphoglycerate kinase 1 (PGK1) for R. aegyptiacus. Our findings support EEF1A1 as an appropriate reference gene for normalization of qRT-PCR data in the Egyptian fruit bat and call for caution when using other candidate genes, i.e. GAPDH and ACTB, due to their instability under specific conditions.

Results

Performance of PCR primers targeting reference gene candidates

To evaluate the performance of the qRT-PCR assay, we first examined the specificity of the primer pairs with melting curve analysis, agarose gel electrophoresis and sequencing. Melting curve analysis revealed single peaks for all primer pairs (Fig. S1). Agarose gel electrophoresis further demonstrated single bands for all the PCR products with the predicted sizes, indicating high specificity of all primer pairs (Fig. S1). PCR products were sequenced and the specificity of the primers was confirmed (Fig. S2). To evaluate the amplification efficiency, standard curves with ten-fold dilution steps were generated (Fig. S3) and subsequently the linear dynamic range (LDR) and precision of each primer pair were assessed following the MIQE guideline30. Amplification efficiencies of all tested primers met the validation criteria, notably 100.51% for ACTB, 99.53% for EEF1A1, 106.52% for GAPDH and 106.44% for PGK1 (Table 1). The correlation coefficient (R2) of all candidates was above 0.99, suggesting excellent linearity of the standard curves. The LDR values of all primers were in the range of 3 to 300,000 copies and precision values varied from 0.31 (EEF1A1) to 1.27 (ACTB) (Table 1). Thus, all the primers for reference gene candidates demonstrated satisfactory specificity and efficiency in qRT-PCR.

Table 1 Summary of the performance of primers employed in this study.
Full size table

Expression profile of reference gene candidates in various tissues from R. aegyptiacus

To investigate the expression of reference gene candidates in tissues from R. aegyptiacus, qRT-PCR was performed with pooled cDNA from nose (nasal epithelium), trachea, lung, blood, spleen and duodenum. Threshold cycle (Ct) values were employed to determine the expression levels of the candidate reference genes. Their expression levels varied and EEF1A1 displayed the highest expression levels across different tissues as indicated by the lowest Ct values. The overall Ct values of EEF1A1, ACTB, GAPDH and PGK1 were 23 ± 1.5, 33 ± 3, 27 ± 2 and 27 ± 1.5, respectively (Fig. 1). EEF1A1 and PGK1 showed the lowest variability in Ct values, suggesting that these two genes display the most stable expression across diverse tissues from R. aegyptiacus.

Figure 1
figure 1

Ct values of selected reference genes in various tissues from Rousettus aegyptiacus. RNA was extracted from nose, trachea, lung, blood, spleen and duodenum from 12 animals. Expression of reference gene candidates were determined via qRT-PCR. Data show mean ± SD from 3 independent experiments.

Full size image

Expression of candidate reference genes in primary bat fibroblasts upon IFN-I stimulation and incubation at various temperatures

To investigate the expression stability of candidate reference genes, bat primary fibroblasts were incubated at 35 °C, 37 °C or 40 °C in the presence or absence of universal type I interferon (uIFN) for 4 h. We employed 35 °C, 37 °C and 40 °C to mimic the physiological daily oscillation of body temperature in R. aegyptiacus 9. The expression levels of all candidate reference genes under these conditions were assessed by qRT-PCR. ACTB showed a broad variation in Ct values, ranging from 29.7 to 39.01 at 40 °C. The Ct values of ACTB were lower at 35 °C, 31.5 ± 0.7 compared to 33.5 ± 1 at 37 °C, suggesting an unstable expression of ACTB upon temperature changes (Fig. 2). The expression levels of EEF1A1, GAPDH or PGK1 were comparable at 35 °C, 37 °C and 40 °C, demonstrating stable expression of these potential reference genes under temperature oscillations. Since uIFN activates IFN-I pathway in bats31, we employed this cytokine to further investigate the expression of the candidate reference genes upon IFN-I stimulation. The mean Ct values of GAPDH decreased upon IFN-I stimulation at 35 °C, 37 °C and 40 °C, indicating increased expression of GAPDH by IFN-I. The Ct values of EEF1A1 and PGK1 remained unchanged following stimulation with IFN-I at 35 °C, 37 °C or 40 °C, suggesting their stable expression under these experimental conditions (Fig. 2). These findings indicate that expression stabilities of ACTB and GAPDH are impaired by temperature oscillations and IFN-I, respectively.

Figure 2
figure 2

Expression stability of candidate reference genes under different conditions. Bat primary fibroblasts were incubated at 35 °C, 37 °C or 40 °C in the presence or absence of 1,000 U/ml universal type I interferon (uIFN). Gene expression levels of ACTB, EEF1A1, GAPDH and PGK1 were determined by qRT-PCR. Data show mean ± SD from 3 independent experiments.

Full size image

Expression stability analysis of reference gene candidates under different conditions

To quantitatively evaluate the expression stability of candidate reference genes, qPCR results were analysed with the statistical algorithms BestKeeper and NormFinder. BestKeeper enables pairwise correlation, regression analysis32,33 and calculations of standard deviation (SD) of all Ct values [SD (± Ct)] as well as of the standard deviation of absolute regulation coefficients [SD (± x-fold)]. Both parameters are indicators of expression variability. A suitable reference gene should display values of < 1 for SD (± Ct), < 2 for SD (± x-fold), and have a coefficient of correlation (r) close to 1 34. According to this algorithm, the [SD (± Ct)] values for ACTB, EEF1A1, GAPDH and PGK1 were 1.40, 0.34, 0.76 and 0.36, along with their corresponding r-values of 0.52, 0.96, 0.78 and 0.88. Thus, expression stability for candidate reference genes ranks as following: EEF1A1, PGK1, GAPDH, and ACTB (Table 2).

Table 2 Stability analysis of reference gene candidates based on pairwise correlations by BestKeeper.
Full size table

NormFinder was employed as the second algorithm to determine gene stability, since it allows evaluation of the overall stability as well as the individual stability for each condition35. The most stable reference genes display stability values close to 0 according to this algorithm. Based on this method, we calculated the overall and individual stability values. The total stability values for ACTB, EEF1A1, GAPDH and PGK1 were 0.097, 0.013, 0.05 and 0.015, respectively, suggesting EEF1A1 as the most stable reference gene under all tested conditions. EEF1A1 also displayed the highest expression stability under high temperature or upon IFN-I treatment (Table 3). Altogether, both algorithms suggest that EEF1A1 is the reference gene with the highest expression stability under the selected conditions, followed by PGK1, GAPDH, whereas ACTB has the lowest stability.

Table 3 Stability analysis of reference gene candidates based on estimated expression variations by NormFinder.
Full size table

GAPDH expression is induced by IFN-I in R. aegyptiacus cells

To evaluate the relative expression levels of ACTB, GAPDH and PGK1 under different conditions, we normalized each candidate to the most stable reference gene, EEF1A1 (Fig. 3A–C). The relative expression levels of PGK1 remained stable at all temperatures evaluated, irrespective of the IFN-I treatment. ACTB expression was also stable at 35 °C and 37 °C in the presence of IFN-I. However, the expression level was significantly reduced following a 2 h IFN-I stimulation at 40 °C (Fig. 3C). Intriguingly, the expression of GAPDH was significantly increased in cells treated with IFN-I at all selected temperatures, suggesting that the induction of GAPDH by IFN-I is temperature independent in R. aegyptiacus (Fig. 3A–C). To investigate whether GAPDH induction by IFN-I is specific to R. aegyptiacus, we stimulated human fibroblasts with IFN-I at 37 °C and 40 °C (Fig. 3D, E). Both human GAPDH and ACTB have been previously used as reference genes in several qRT-PCR studies36,37. Indeed, both genes displayed good expression stability at 37 °C, yet the variation of Ct values at 40 °C argued against stability of human GAPDH and ACTB at 40 °C. The relative expression of human GAPDH normalized against ACTB remained unchanged upon IFN-I treatment either at 37 °C or 40 °C, revealing that human GAPDH expression in fibroblasts is not modulated by IFN-I (Fig. 3D, E). Overall, we conclude that IFN-I triggers GAPDH expression specifically in R. aegyptiacus.

Figure 3
figure 3

GAPDH expression is induced by IFN-I in R. aegyptiacus. Relative gene expression of ACTB, GAPDH and PGK1 normalized to EEF1A1 upon uIFN treatment at 35 °C (A), 37 °C (B) or 40 °C (C). (D, E) Ct values of human ACTB and GAPDH (D) and relative expression of human GAPDH normalized to ACTB (E) in human fibroblasts upon uIFN treatment and incubation at 37 °C or 40 °C. Bat (A, B, C) or human (D, E) primary fibroblasts were stimulated with PBS (ctrl) or 1000U/ml universal type I interferon (uIFN) for indicated time at 35 °C, 37 °C or 40 °C. Gene expression levels of ACTB, GAPDH and PGK1 were determined with qRT-PCR. Data show mean ± SD from 3 independent experiments. The statistical significance was calculated using one-way ANOVA with Holm-Šidák’s post-hoc test. (*) P ≤ 0.05, (**) P ≤ 0.01, (***) P ≤ 0.001, (****) P ≤ 0.0001.

Full size image

Discussion

Accurate gene transcription measurements require selection of reference genes that maintain high stability under various experimental conditions. In this study, we selected ACTB, GAPDH, EEF1A1 and PGK1 as reference gene candidates due to their wide applications in other species14,32,36,39,42,43,44,45,46,47,48 and evaluated the suitability of these candidates in R. aegyptiacus under the physiological relevant conditions, notably temperature oscillation and IFN-I stimulation. By employing BestKeeper and NormFinder, we calculated the stability and validated EEF1A1 as the most stable reference gene in R. aegyptiacus under conditions relevant for the biology of this species36,38,39,40,41. By normalizing the qRT-PCR data to EEF1A1 we observed that expression of GAPDH was significantly induced by IFN-I, suggesting unsuitableness of using this commonly used reference gene in R. aegyptiacus. Consistent with our results, many reports have demonstrated that the expression of GAPDH and ACTB is unstable in various tissues or cells from mice and humans and upon certain stimulations, such as IL-2 and hypoxia 37,49,50,51,52. Whether GAPDH expression is induced by IFN-I, and whether GAPDH and ACTB are suitable reference genes in other bat species, require further investigations.

Our finding that the expression of GAPDH, one of the key enzymes in glycolysis, is induced by IFN-I in R. aegyptiacus may have implications for the immunometabolism of bats. It is well established that metabolic reprogramming in cells controls their immune responses53,54. Glycolysis is commonly utilized by various immune cells to enable prompt responses to infections. In bone marrow derived macrophages, aerobic glycolysis promotes IL-1β production upon LPS stimulation or Bordetella pertussis infection 55. In CD4+ and CD8+ T-cells56,57,58,59,60, as well as in NK cells61,62, glycolysis is required for their effector functions, such as IFN-γ production and cytotoxicity. In human plasmacytoid dendritic cells (pDCs) and monocyte-derived DCs (moDC), glycolysis promotes IFN-I production upon TLR9 or RIG-I activation, respectively63. Further, IFN-I induces a metabolic shift towards glycolysis, contributing to the antiviral activity in fibroblasts and antigen presentation in DCs64,65. On the other hand, lactate, the end metabolite of glycolysis, directly binds to mitochondrial antiviral-signaling protein (MAVS), and consequently inhibits its activation and IFN-I production66. Thus, outcomes of such metabolic shift vary in diverse cells or under different stimulations. The nectarivore bat Glossophaga soricina employs high rates of glycolysis to generate ATP during flight67. As a fruit bat, R. aegyptiacus could also directly utilize dietary sugars to fuel both roosting and flight metabolism68. Whether induction of GAPDH expression in this species impacts on metabolic reprogramming towards glycolysis, needs to be clarified. Moreover, whether and how such metabolic shifts affect IFN-I signalling or other immune pathways in R. aegyptiacus remains to be uncovered.

In addition to its roles in glycolysis, GAPDH also modulates cell death, RNA export and cytoskeleton dynamics69. Upon serum deprivation and DNA damage, GAPDH translocates to the mitochondria and interacts with voltage-dependent anion channel (VDAC), leading to apoptosis70. It can also bind to the 3' untranslated region of TNFα mRNA and represses TNFα expression in human monocytes and macrophages71. Hence, GAPDH upregulation by IFN-I may contribute to apoptosis induction and TNFα repression in R. aegyptiacus, which could represent novel mechanisms for the prevention of excessive inflammation during viral infections.

Overall, our study provides an extensive analysis of reference genes and identifies EEF1A1 as the most stable reference gene in R. aegyptiacus under temperature changes and IFN-I stimulation, which allows us to perform accurate gene transcription studies in this species. Our findings also open new investigation avenues by showing that GAPDH is regulated by IFN-I which has a broad relevance in context of immunometabolism.

Material and methods

Selection of reference gene candidates and design of primer pairs

Reference gene candidates (ACTB, EEF1A1, GAPDH and PGK1) for R. aegyptiacus gene expression studies were selected based on their utilization in other bat species13,14,17. Primers against these reference genes were designed using the PrimerQuest tool (Integrated DNA Technologies, Inc.). The criteria for primer design were as follows: primer lengths around 17–30 bp, GC content of 40–55%, optimal melting temperature at 62 °C, and amplicon lengths within a range of 100–250 bp. Derived primer pairs were evaluated using the OligoAnalyzer tool to exclude primers with hairpin structures and homo- and/or heterodimer formation (Integrated DNA Technologies, Inc.). Primer sequences were also blasted using the NCBI BLAST tool to ensure their specificity for R. aegyptiacus. The primer pairs meeting all criteria were selected for further experiments. The characteristics of these primers are shown in Table 4.

Table 4 Summary of selected reference gene candidates.
Full size table

Cells, tissues and stimulation experiments

Bat fibroblasts were derived from the lung of a female R. aegyptiacus bat from the R. aegyptiacus breeding colony at the Friedrich-Loeffler-Institut. Sampling was performed in accordance with current European and National Animal Welfare regulations, after ethical review and approval by the authority of the Federal State of Mecklenburg-Western Pomerania, Germany (file number 7221.3–2-042/17) and the experiments were carried out according to ARRIVE guidelines (https://arriveguidelines.org). The lung tissue was dissected into small pieces and digested with trypsin overnight at 4 °C. Dissociated cells were seeded in cell culture dishes in DMEM medium (DMEM high glucose medium, 10% fetal bovine serum (FBS), 2 mM glutamine and 100U/mL of Penicillin–Streptomycin) for 2 h, and only adherent cells were propagated. Confirmation of bat fibroblast identity was carried out by examining fibroblast activation protein (FAP) expression via PCR (data not shown).

Tissue samples from 12 individual bats were obtained from an animal experiment published before76. For IFN-I stimulation, bat fibroblasts and human dermal fibroblasts (#C0045C, Thermo Fisher Scientific) were incubated with 1000U/ml universal IFN-I (uIFN) (#11200-1, PBL Assay Science) for 1 h, 2 h and 4 h at either 35 °C, 37 °C or 40 °C, respectively.

RNA extraction and cDNA synthesis

Cells were lysed in homemade Trizol solution and RNA was extracted as published before77. Purified RNA was quantified using NanoDrop 2000c spectrophotometer (#ND-2000c, Thermo Fisher Scientific) and 800 ng RNA were subsequently utilized for cDNA synthesis with the LunaScript RT SuperMix Kit (#E3010L, New England BioLabs).

qRT-PCR

qRT-PCR reactions were carried out with EvaGreen Fluorescent DNA stain (# PCR-379, Jena Bioscience), ROX as an internal reference dye (#PCR-351, Jena Bioscience) and GoTaq Polymerase (#M3001, Promega) according to manufacturers’ instructions. The reaction setup was as follows: 95 °C for 2 min; (95 °C for 30 s, 62 °C for 30 s, 72 °C for 1 min) for 40 cycles; 72 °C for 10 min and infinite hold at 4 °C. Unless stated otherwise, each qPCR reaction was performed with 100 ng cDNA. To minimize pipetting errors, the template was diluted and 5 µl were used for each qRT-PCR reaction. Each primer pair was added separately into different wells. Measurements were performed with the QuantStudio 6 Flex real-time PCR system (#4485691, Applied Biosystems). Melting curves were performed within the temperature range from 60.16 °C to 94.885 °C in steps of 0.193 °C, respectively.

Establishment of standard curves and examination of amplification efficiency via qRT-PCR

Standard curves of all reference genes in the qRT-PCR reaction were generated with copy numbers from 300,000 to 3 copies in ten-fold dilution steps. To achieve accurate copy numbers, amplicon sizes of each reference gene were used to calculate the specific weight of each amplicon78, resulting in 2.12 × 10–19 g for ACTB, 1.78 × 10–19 g for EEF1A1, 1.32 × 10–19 g for GAPDH and 1.62 × 10–19 g for PGK1. The amplification efficiency of all primer pairs was subsequently determined with the slope of the standard curve according to the equation 10–1/slope-1. The amplification efficiency of favourable primers ranges between 90–110%. Linear dynamic range (LDR) is described as the highest to the lowest quantifiable copy numbers from standard curves. LDR should cover at least 3 orders of magnitude, ideally 5–6 orders. Precision refers to intra-assay variation and is defined as standard deviation (SD) of technical replicates30.

Amplicon purification and sequencing

Amplicons of all candidate reference genes were visualized in 1.5% agarose gels and bands were cut and purified using the QIAquick gel extraction kit (#28506, Qiagen). Purified amplicons were subsequently sequenced using the Eurofins tubeseq platform.

Stability analyses of reference gene candidates

To investigate the expression stability of the four reference gene candidates, two statistical algorithms were used: BestKeeper33 and NormFinder35. The highest stability was defined as the lowest variation of expression levels under all the selected experimental conditions33,35. In brief, for BestKeeper, raw Ct values without any normalization are subjected to the calculation and the parameters of interest are SD (std dev [± CP]) and the Pearson coefficient of correlation (r). NormFinder utilizes normalized Ct values (2−ΔCt) to provide a direct readout for the estimated expression variation and can separately calculate the stability under each condition (IFN-I treatment, temperature) or total stability.

Statistical analysis

Statistical analysis was performed with GraphPad Prism 8 (GraphPad Software Inc., USA). To determine statistical significance among investigated groups, one-way analysis of variance (ANOVA) with Holm-Šidák’s post-hoc test was performed. A P value of < 0.05 was considered to be significant.