Daily supervisor
Ing. Q.D. Gunst
Senior supervisor
Dr. M.J.B. van den Hof
Bachelor Project for BSc Biomedical Sciences
Marijn Bult 11617349 March 2020
Transcriptional adaptation via mutant mRNA degradation
does not affect phenotype of Fstl1 knockout mice
Abstract
Introduction – The Fstl1 knockout mouse model exhibits a relatively mild phenotype compared to Fstl1 knockdown models of zebrafish and chicken. A novel mechanism might explain this phenotype. This so called transcriptional adaptation is based on accelerated degradation of the mutant mRNA that results from knockout. The mRNA fragments on its turn, are able to upregulate genes that show sequence similarity to the knocked out gene, thereby compensating the loss of gene function. Possibly this mechanism lessens the severity of the Fstl1 knockout phenotype.
Material and methods – Hearts were isolated from E11.5 and E18.5 mouse embryos, from both wild type and knockout. From these hearts RNA was isolated. First, to confirm accelerated degradation of the mutant Fstl1 transcript, the abundancy of nuclear pre-mRNA and mRNA of Fstl1 was determined with qPCR. Next the expression level of 15 genes was measured with qPCR. These genes were identified with BLASTN, having a bit-score >40 and a stretch of sequence similarity of at least 20 nucleotides with the Fstl1 mRNA.
Results – Fstl1 pre-mRNA is present at the same level in wild type and knockout hearts, whereas the Fstl1 mRNA is 16 times more abundant in wild type hearts than in knockout hearts. This suggests accelerated degradation of the mutant Fstl1 mRNA. However, none of the evaluated genes was diferentially expressed in wild type hearts compared to knockout hearts.
Discussion – The genes that show sequence homology to Fstl1 are not upregulated in knockout hearts. Based on these result, loss of Fstl1 is not compensated by the evaluated genes. Taken together, these observations suggest that the phenotype of Fstl1 knockout mice is primarily the consequence of functional disruption of Fstl1.
Follistatin-like 1 (Fstl1) is a secreted glycoprotein. It is known to be involved in cardiovascular disease, cancer and inflammation (for review see: Matiotti et al., 2018). Furthermore, through knockout experiments, its importance during embryonic development of heart, lung, bone and ureter has become clear (Sylva et al., 2011; Geng et al., 2011; Xu et al., 2012).
Despite the apparently crucial role of Fstl1, there are no congenital disorders known that are correlated to or caused by mutations in Fstl1 (Prakash et al., 2019). Moreover, after analysis of 120.000 human exomes, no homozygous loss of function variations were found and only 35 heterozygous individuals were identified (Prakash et al., 2019). Together, this suggests that Fstl1 is a highly conserved gene and that alterations are not tolerated nor compatible with life.
The importance of Fstl1 during embryonic development is underscored in knockdown models of zebrafish and chicken. In zebrafish, knockdown of Fstl1 during late gastrulation induces an expansion of the chordamesoderm (Esterberg et al., 2008) and knockdown of Flik in chicken, the Fstl1 homologue, severely afects neural induction (Towers et al., 1999). However, when Fstl1 is functionally disrupted in mice, homozygous knockout mice are born alive at the expected Mendelian ratio (Geng et al., 2011; Sylva et al., 2011). Nevertheless, all homozygous neonatal mice die within hours due to respiratory insufficiency (Geng et al., 2011; Sylva et al., 2011; Liu et al., 2017). Although this is a severe phenotype, it is surprising that all knockout pups are born alive, considering the phenotype observed in zebrafish and chicken knockdown models and the fact that homozygous mutations of Fstl1 are not tolerated in the human population.
Such large diferences in phenotype between knockdown and knockout models have been observed in a number of model systems, in which the knockdown models generally present a more severe phenotype than the knockout models (El-Brolosy & Stainier, 2017). In zebrafish it was shown that loss of a gene by knockout, sometimes unexpectedly results in upregulation of proteins with a similar function (Rossi et al., 2015). Until recently such a compensatory mechanism was not described in knockout mouse models. However, lately in mouse and zebrafish a mechanism was described, referred to as transcriptional adaptation. This mechanism is observed in knockout models, in which gene function is altered such that no or a non-functional protein is produced. Transcriptional adaptation is a compensatory mechanism that is mediated by the rapid degradation of the aberrant mRNA through nonsense-mediated decay (NMD) (El-Brolosy et al. ,2019; Ma et al., 2019). NMD was found to induce upregulation of genes that exhibit short stretches of approximately 20-180 base pair sequence similarity to the knocked out gene in the respective knockout model. Analysis of the mechanism revealed that the COMPASS complex plays a crucial in transcriptional adaptation. The COMPASS complex binds to the RNA fragments and mediates H3K4me3 histone modifications at the homologous position in the genome (Wang et al., 2011). H3K4me3 histone modifications are observed in the promotor and enhancer regions of upregulated genes in the knockout models, suggesting their role in the increase in transcription of the respective genes (El-Brolosy et al., 2019; Ma et al., 2019). Altogether, this suggests a mechanism where short RNA sequences, derived from degraded mutant mRNA, function as a guide for the COMPASS complex, afecting the methylation status of the targeted gene and as a result its transcription.
This mechanism might also explain why Fstl1 knockout mice show a milder phenotype than one would expect based on the phenotype of Fstl1 knockdown models of other species. From a wider perspective, this novel mechanism needs to be evaluated in existing mouse models and has to be taken into account whenever a new knockout model is created. Currently, knockout models are often
used to identify the function of a gene. However, if the loss of a gene is compensated by other genes, the conclusions drawn from these experiments might not be accurate.
In order to investigate whether such a compensatory mechanism is also operational in Fstl1 knockout mouse hearts, we will first examine whether the degradation of the mutated Fstl1 mRNA is enhanced compared to wild type Fstl1 mRNA. Next, we will identify genes that have sequence similarity to Fstl1 and determine whether their expression level is afected using quantitative PCR (qPCR).
Materials and methods
Identifying genes exhibiting sequence similarity
In order to identify genes that show sequence similarity to Fstl1, a BLASTN (NCBI) search was performed using mus musculus Fstl1 mRNA (NM_008047.5) as the query sequence, searching in the Reference RNA Sequence database of mus musculus. Because optimal alignment parameters are still under debate and difer per gene, all sequences that returned with a bit-score ranging from 40 till 200 and an aligned sequence stretch from 20 and 180 nucleotides in length were included (El Brolosy et al., 2019).
Primer design
For the genes that were identified using BLASTN, a primer pair suitable for the qPCR was designed using NCBI primer-BLAST. Primers were designed using the following criteria: (1) the primer pair should be located within the last three 3’ exons, (2) the melting temperature (Tm) should be 60 °C ± 1 °C and the diference between the forward and reverse primer should be ≤ 1 °C, (3) the amplicon should be 70-200 base pairs in length and (4) each primer should show minimal similarity to other parts of the genomic DNA. If possible, at least one primer of the pair should span an exon-exon junction. If this criterion could not be met, the primers should be positioned in two diferent exons that are separated by an intron with a length of at least 500 base pairs. To asses homo- and heterodimer formation of the primer pair, primers were analyzed using the OligoAnalyzer Tool 3.1 of Integrated DNA Technologies (IDT). As parameters for the analysis an oligo concentration of 0.25 µM, a Na+ concentration of 50 mM, a Mg++ concentration of 2 mM and a dNTPs concentration of 0.25 mM was used. If primers were found to form a hairpin with a Tm lower than 50 °C and the stability (∆G) of potential homo- and heterodimers was ≤ -5 kcal/mole, the primer pair was considered to meet all criteria. Subsequently, the primer pair was tested in a qPCR experiment using total embryo cDNA as a positive control, genomic DNA as a control for specificity on cDNA, and as negative controls on a -Reverse Transcriptase (RT) cDNA synthesis and water (NTC). The primer sequences that passed all test, are summarized in Supplementary table 1.
To measure the expression level of Fstl1 pre-mRNA, a gene specific RT primer was designed that targeted intron 4. To prevent this RT primer from interfering with the amplification process, a slightly lower Tm of 50 °C was chosen. The accompanying qPCR primers were designed according to the same criteria as the primers for mRNA, except that the forward and reverse primer target intron 3 and exon 4, respectively. A schematic representation of these primers is shown in Supplementary figure 1. These qPCR primers were validated on genomic DNA.
After qPCR, the reactions were analyzed using melting curve analysis and agarose gel electrophoresis. When the primer pair amplified the product of the expected size and the melting curve analysis showed a single sharp peak at the same temperature as the positive control, the primer pair was considered to target the proper sequence in the qPCR reaction.
Mouse Fstl1-KO strain
Fstl1 heterozygous KO mice were crossed, to isolate embryos being homozygous Fstl1 KO or WT. Fstl1 is functionally disrupted by removal of the 2nd exon, resulting in a mRNA from which no Fstl1 protein can be produced (Sylva et al. 2011). The hearts of embryos were isolated at 11.5 and 18.5 days post fertilization. For each time point and genotype three embryos were collected and processed separately.
cDNA synthesis
RNA was isolated using the ReliaPrep RNA Tisssue Miniprep system (Promega) according to the protocol of the manufacturer. The quality and amount of the RNA was assessed using the Nanodrop. 1 g of isolated RNA was mixed with 1 µl of oligodT12VN primers (125 pmol/µl) and 1 µl of dNTPs (10mM). The RNA was denatured at a temperature of 65 °C for 20 minutes using a Biometra TAdvanced Thermocycler (Westburg), after which the tubes were placed on ice allowing the oligodT12VN primers to anneal. 1 µl of SuperScript II Reverse Transcriptase enzyme was added together with 2 µl DTT (0.1M) and 4 µl 5x First Strand bufer (all Thermo Fisher). The cDNA was synthesized at 42 °C for 60 minutes, followed by heat inactivation of the enzyme for 15 minutes at 72 °C. The final mixture was diluted to 5 ng/µl cDNA solution.
To prepare cDNA of nuclear pre-mRNA the Fstl1 specific RT primer was used, targeting the 4th intron. After melting the RNA at 65 °C for 20 minutes, the primer was annealed at 50 °C for 10 minutes, using 1 µl of the gene specific primer (1 µM). cDNA was synthesized as described above.
Real time quantitative PCR
For RT-qPCR analysis the Roche LightCycler 480 II was used. Each qPCR reaction consisted of 5 µl Sybrgreen Master mix (Roche), 1 µl of cDNA (equivalent to 5ng RNA), 1 µl of the primer pair mix, containing forward and reverse primers both in a concentration of 2.5 ng/µl and 3 µl MilliQ water. A qPCR run consisted of a pre-incubation step at 95 °C for 5 minutes, followed by 45 cycles, each comprising of a denaturation step at 90 °C for 10 seconds, an annealing step at 60 °C for 20 seconds, and an elongation step at 72 °C for 20 seconds. At the end of the elongation step, the fluorescence was measured. Each run ended with a melting curve analysis protocol.
qPCR reactions which reached the fluorescence threshold (Cq) after 35 cycles were repeated. In these reactions the maximal amount of cDNA was used (5 µl of cDNA equivalent to 25 ng RNA). Adding the maximal amount of cDNA to the reaction would result in reaching the Cq 2 cycles earlier. For E11.5 samples, this was the case targeting Cacna1b, Fstl5 and Grik4, and for E18.5 samples besides Cacna1b, Fstl5 and Grik4 also for Cplx1 (3 µl of cDNA equivalent to 15 ng).
qPCR data processing
LinRegPCR (Version 2018.0) was used to determine the expression level of each targeted gene. The program uses PCR efficiency per target to calculate the starting concentration (Ruijter, 2016). Inter plate variation was accounted for using Factor qPCR (Version 2016.0). The mean expression level of each gene was measured in both WT and KO hearts. The data was normalized by dividing expression levels by the geometric mean of the expression levels the reference genes, Rpl32 and Ppia (Ruiz-Villalba et al., 2017). The fold change for each gene was calculated as follows:
expressionlevel KO
expressionlevel WT
.A student t-test was performed, a p-value smaller than 0.05 was considered significant. Calculations were performed in Excel, Summary Statistics (Version 2018.0) and R.
The error bars of the fold change comparing WT and KO (Figure 3C and 4C) were calculated according to the formula below, giving the relative standard error (RSE) for each gene:
SEM KO
mean expression KO
¿
¿
(
SEM WT
meanexpression WT
)
2+
¿
RSE=fold change∗
√¿
Results
To verify whether the KO Fstl1 mRNA is more rapidly degraded than the WT Fstl1 mRNA, the expression level of both Fstl1 nuclear pre-mRNA and mature mRNA was measured in E18.5 KO and WT hearts. As shown in Figure 1A, the expression level of pre-mRNA is the same in KO and WT hearts. However, as depicted in Figure 1B the mRNA transcript of Fstl1 is approximately 16 times more abundant in WT hearts than in KO hearts. This strongly suggests that the non-functional Fstl1 mRNA, created by removal of exon 2, is rapidly degraded in KO mice.
Transcripts of mRNA species that are rapidly degraded, are suggested to activate transcriptional adaptation. To evaluate the existence of mRNA transcripts in the mouse genome that exhibit similarity to Fstl1, a BLASTN search was performed using the Fstl1 mRNA sequence as query. This search returned 103 mRNA transcripts that showed similarity to the Fstl1 transcript. Of these, 15 genes met the criteria of a bit-score ≥ 40 and stretches of similarity of 20-180 base pairs in length (El-Brolosy et al., 2019). Moreover, alignment of these 15 genes with the Fstl1 mRNA, revealed that the regions of similarity clustered to 5 regions in Fstl1. A schematic overview of these findings is shown in Figure 2. Detailed information on these genes and their alignment are given in Supplementary table 2.
Figure 1 (A) Expression of Fstl1 pre-mRNA in E18.5 WT and KO mouse hearts. (B) Expression of Fstl1 mRNA in E18.5 WT
Figure 3 (A) Bar plots of the expression of each separate gene in E11.5 mouse hearts (WT n = 3, KO n = 3). Error bars
indicate SEM. Note that for Cacna1b, Fstl5 and Grik4 25 ng of cDNA was used instead of 5 ng (see Material and Methods). (B) Table with the excluded genes and reason of exclusion. (C) Fold change of expression in KO compared to WT, calculated as follows: expression KO/expression WT. Fold change values are displayed on top of the bars. The y-axis is on a log2 scale. Error bares indicate SRE.
To evaluate the expression level of each of these genes, qPCR primers were designed. The expression levels of these genes were measured in both WT and KO hearts of E11.5 and E18.5 embryos. The normalized expression level of each separate gene is shown in Figure 3A and 4A, respectively. The fold change between KO and WT hearts for each gene are displayed in Figure 3C and 4C.
Figure 4 (A) Bar plots of the expression of each separate gene in E18.5 mouse hearts (WT n = 2, KO n = 3). Error bars
indicate SEM. Note that for Cacna1b, Fstl5 and Grik4 25 ng of cDNA and for Cplx1 15 ng was used (see Material and Methods). (B) Table with the excluded genes and reason of exclusion. (C) Fold change of expression in KO compared to WT, calculated as follows: expression KO/expression WT. Fold change values are displayed on top of the bars. The y-axis is on a log2 scale. Error bares indicate SRE.
At E11.5, none of the genes exhibiting similarity to Fstl1 mRNA showed significant diferential expression in the KO hearts compared to the WT hearts. Cadps2 shows the strongest trend of upregulation; being 1.8 times more abundant in KO than in WT hearts (p=0.073). Though not significant, Efna5 (p=0.422), Fstl5 (p=0.372), and Mdga2 (p=0.332) show a trend of downregulation. Above all, Cplx1 shows the most convincing trend of downregulation with a p-value of 0.070. Cacna1b (p=0.940), Desi2 (p=0.454), Frg1 (p=0.218), Fstl4 (p=0.808), Grik4 (p=0.859), Hmgcs1 (p=0.792), Lamp2 (p=0.515), Vangl2 (p=0.629) and Zfp426 (p=0.954) are expressed at similar levels in both KO and WT hearts
Similar results were found in E18.5 hearts, meaning that none of the genes evaluated showed significant diferential expression between KO and WT hearts. Cadps2 (p=0.480), Desi2 (p=0.375), Efna5 (p=0.365), Frg1 (p=0.179), Lamp2 (p=0.599), Mdga2 (p=0.109), Vangl2 (p=0.388) and Zfp426 (p=0.608) show a trend of being slightly upregulated in KO hearts (≤ 1.3 times), whereas Fstl4 (p=0.926) and Hmgcs1 (p=0.295) show a trend for being slightly downregulated in KO hearts (≥ 0.75 times).
Originally, three biological samples of both WT and KO hearts of E18.5 embryos were analyzed. However, one of the WT samples consistently had an expression level about 5-10 times higher for most target genes and to a far lesser extend for the reference genes. As a consequence normalization did not correct for these high expression levels (Supplementary figure 1). The cause of this aberrant behavior of the sample is not known and is considered an outlier. The sample was therefore excluded from further analysis of the results. Due to time restrictions and the fact that no additional fetal material was available, a novel E18.5 sample could not be included in this analysis.
Not all qPCR measurements met the criterion of reaching the Cq within 35 cycles. Furthermore, for some genes one of the technical replicates gave no signal with the qPCR. When less than 2 samples of either WT or KO were measured properly, it was not possible to calculate a fold change for these genes. Spink12 transcripts were undetectable in any of the evaluated samples. An overview of the excluded genes and the reason of exclusion can be found in Figure 3B and 4B.
Discussion
This experiment was performed to evaluate whether the observed phenotype of the Fstl1 knockout mouse (Sylva et al., 2014) is primarily due to inactivation of Fstl1 or due to secondary efect caused by the recently described mechanism of transcriptional adaptation. Transcriptional adaptation refers to a mechanism, in which loss of the inactivated gene is compensated by upregulation of homologous genes, mediated through nonsense mediated decay of the mutant mRNA (El-Brolosy et al., 2019; Ma et al., 2019). To assess whether the mutated Fstl1 mRNA shows enhanced degradation compared to the wild type mRNA, the expression level of the pre-mRNA and mature mRNA was determined. The Fstl1 knockout mice express similar levels of Fstl1 pre-mRNA as in wild type mice, meaning that the production of Fstl1 mature mRNA is equal. However, the mature Fstl1 mRNA is present at very low levels in knockout hearts compared to wild type hearts. This finding strongly suggests that the mutated mRNA is subjected to rapid degradation, meaning that mRNA fragments might initiate transcriptional adaption, thereby saving the phenotype of Fstl1 knockout mice (El-Brolosy et al., 2019; Ma et al., 2019).
However, none of the genes that exhibit homology to Fstl1, was observed to be significantly diferentially expressed in Fstl1 knockout hearts compared to wild type hearts. This was the case in both E11.5 and E18.5 hearts. Therefore, it can be concluded that the function of Fstl1 in the heart is
consequence of transcriptional adaptation as described by El-Brolosy et al. (2019) and Ma et al. (2019).
This finding is not in line with previous research, that suggest that fragments of the degraded mRNA can upregulate genes based on their homology to the knocked out gene (El-Brolosy et al., 2019; Ma et al., 2019). El-brolosy et al. (2019) do note that genes that exhibit similarity to the 3’ untranslated region (UTR) are not afected by transcriptional adaptation. 11 of the 15 genes that were selected through BLASTN, mapped to the 3’ UTR of the Fstl1 transcript (see Figure 2). Consistent with this finding, these genes are not afected by transcriptional adaptation. Nevertheless the 4 genes that do align to the coding domain sequence (CDS), being Fstl4, Fstl5, Grik4 and Spink12, are also not significantly afected in the heart.
Some of the investigated genes did not reach the Cq within 35 cycles, meaning that the qPCR reaction started with less than 10 transcripts. At low concentrations, the Poisson distribution induces high variance within biological and technical replicates (Rutledge & Stewart, 2010). This makes it more difficult to find significant results. Due to the Poisson distribution, there is a possibility of no transcript being added to reaction at all, which might explain why for some genes only one of the duplicate measurements gave a signal. As a consequence, these measurements had to be excluded. We tried to decrease the Poisson efect by adding more cDNA. The results of this are displayed in Supplementary table 3. For most genes, it lowered the Cq and increased the number of samples that gave a signal. However, the variance for Fstl5 and Grik4 remained high in E11.5 samples. For E18.5, adding more cDNA had less positive efects, as the Cq of those genes stayed above 35. This indicates that the concentration of these transcripts is extremely low or that these genes are not at all expressed in the evaluated tissue. However, in the face of this research, where the aim is to find genes that save a severe phenotype, an upregulation resulting in an expression level just above the detection limit, is most likely not having severe efects on the organism.
Another observation is that both El-Brolosy et al. (2019) and Ma et al. (2019) only report genes that are already expressed in wild type samples and are upregulated after knockout. But actually, a gene that is undetectable in wild type samples but detectable in knockout samples would be the most evident prove for a mechanism that induces upregulation. In this experiment, the gene Spink12, that exhibits homology to the CDS, was the only gene from which absolutely no signal could be detected in the wild type samples, but it remained undetectable in the knockout samples as well. This observation suggests that transcriptional adaption is not able to overrule strong downregulation, but this has to be further elucidated.
Furthermore, the negative efects of Fstl1 knockdown in chicken and zebrafish occurred early in embryonic development, during gastrulation (Esterberg et al., 2008; Towers et al., 1999). The time points currently used are much later in embryonic development. It would be interesting to measure the expression levels in the knockout mice during early embryonic development as to see whether transcriptional adaption is active during these stages. The possibility that transcriptional adaptation is developmental stage dependent, is supported by the observation that the patterns of fold changes between knockout and wild type hearts are not similar in E11.5 hearts compared to E18.5 hearts (Figure 3C and 4C). Also, Fstl1 has been described to play a role in development of other organs than the heart (Matiotti et al., 2018), so evaluation of transcriptional adaptation in other organs or even the entire embryo would be an interesting approach.
Moreover, the current idea that the Fstl1 knockout mouse model displays a relatively mild phenotype, is based on the comparison of the efect of Fstl1 knockdown in zebrafish and chicken
(Esterberg et al., 2008; Towers et al., 1999). However, the method of knocking down genes with morpholinos is controversial as it has been reported to have of-target efects and unwanted toxicities (Jackson & Linsley, 2010). It is reasonable to think that these side-efects add to the severity of the phenotype of Fstl1 knockdown zebrafish and chicken, meaning that the phenotype of the knockout mice is ‘normal’ rather than relatively mild. This however, is contradicted by the fact that no congenital disorders are associated with mutations of Fstl1 and that homozygous alterations are not tolerated in humans (Prakash et al., 2019). These observations strongly suggest that alterations in Fstl1 must have severe efects on embryos, as they do not seem to be compatible with life.
In conclusion, it remains unclear why transcriptional adaptation via homology of degraded mRNA fragments is not playing a role in Fstl1 knockout mice. Nevertheless, this compensatory mechanism is real and should be taken into account when a knockout model is created. This research points out that the compensatory mechanism is not active in every instance, therefore indicating that the criteria for transcriptional adaptation are not yet fully understood. For future research with knockout models it might be advisable to use RNA-less knockouts. El-Brolosy et al. (2019) found that removal of the entire promotor region using CRISPR-Cas9, in which no mRNA is formed, does not initiate transcriptional adaptation. Even with the uncertainty of transcriptional adaption being relevant in a certain model, this method enables it to be ruled out anyway.
In our case, it can be concluded that the phenotype of Fstl1 knockout mice is primarily the consequence of the depletion of Fstl1. These results do not ask for rejection of the current Fstl1 knockout mouse model, as there is no indication that the legitimacy of this model is compromised by transcriptional adaption. It does remain unexplained why the Fstl1 knockout mice show a mild phenotype. Therefore, it would be interesting to create an RNA-less Fstl1 knockout model, and compare this phenotype with the current knockout model to rule transcriptional adaptation out completely.
El-Brolosy, M. & Stainier, D. (2017). Genetic compensation: A phenomenon in search of mechanisms. Public Library of Sciences Genetics, 13, e1006780, https://doi.org/10.1371/journal.pgen.1006780. El-Brolosy, M., Kontarakis, Z., Rossi, A., Kuenne, C., Günther, S., Fakuda, N. et al. (2019). Genetic compensation triggerd by mutant mRNA degradation. Nature, 568, 193-197.
Esterberg, R., Delalande, J. & Fritz, A. (2008). Tailbud-derived Bmp4 drives proliferation and inhibits maturation of zebrafish chordamesoderm. Development, 135, 3891-3901.
Geng, Y., Dong, Y., Yu, M., Zhang, L., Yan, X., Sun, J. et al. (2011). Follistatin-like 1 (Fstl1) is a bone morphogenetic protein (BMP) 4 signalling antagonist in controlling mouse lung development. Proceedings of the National Academy of Sciences of the United States of America, 108, 7058-7063. Jackson, A. & Linsley, P. (2010). Recognizing and Avoiding siRNA Of-Target Efects for Target Identification and Therapeutic Application. Nature Reviews Drug Recovery, 9, 57-67.
Liu, X., Liu, Y., Li, X., Zhao, J., Geng, Y. & Ning, W. (2017). Follistatin like-1 (Fstl1) is required for the normal formation of lung airway and vascular smooth muscle at birth. Public Library of Science One, 12, e0177899, https://doi.org/10.1371/journal.pone.0177899.
Ma, Z., Zhu, P., Shi, H., Guo, L., Zhang, Q. Chen, Y. et al. (2019). PTC-bearing mRNA elicits a genetic compensation response via Upf3a and COMPASS components. Nature, 568, 259-263.
Matiotti, A., Prakash, S., Barnett, P. & Van den Hof, M (2018). Follistatin-like 1 in development and human diseases. Cellular and Molecular Life Sciences, 75, 2339-2354.
Prakash, S., Mattiotti, A., Mulder, B., Postma, A. & Van den Hof, M., (2019). Identifying pathogenic variants in the Follistatin-like 1 gene (FSTL1) in patients with skeletal and atrioventricular valve disorders. Molecular Genetics & Genomic Medicine, 7, e00567, https://doi.org/10.1002/mgg3.567. Rossi, A., Kontarakis, Z., Gerri, C., Nolte, H., Hölper, S., Krüger, M. et al. (2015). Genetic compensation induced by deleterious mutations but not gene knockdowns. Nature, 524, 230-233.
Ruijter, J. (2016). Why is the PCR amplification efficiency still ignored? The gene expression blog, https://blog.qbaseplus.com/why-is-the-pcr-amplification-efficiency-still-ignored.
Ruiz-Villalba, A., Mattiotia, A., Gunst, Q., Cano-Ballosteros, S., Van den Hof, M. & Ruijter, J. (2017). Reference Genes for Gene Expression Studies in the Mouse Heart. Scientific Reports, 7, https://doi.org/10.1038/s41598-017-00043-9.
Rutledge, R. & Stewart, D. (2010). Assessing the Performance Capabilities of LRE-Based Assays for Absolute Quantitative Real-Time PCR. Public Library of Science One, 5, e9731, https://doi.org/10.1371/journal.pone.0009731.
Sylva, M., Li, V., Buffing, A., Van Es, J., Van den Born, M., Van der Velden, S. et al. (2011). The BMP Antagonist Follistatin-Like 1 Is Required for Skeletal and Lung Organogenesis. Public Library of Science One, 6, e22616, https://doi.org/10.1371/journal.pone.0022616.
Towers, P., Patel, K., Withington, S., Isaac, A. & Cooke, J. (1999). Flik, a Chick Follistatin-Related Gene, Functions in Gastrular Dorsalisation/Neural Induction and in Subsequent Maintenance of Midline Sonic Hedgehog Signalling. Developmental Biology, 214, 298-317.
Wang, K., Yang, Y., Liu, B., Sanyal A., Corces-Zimmerman, R., Chen, Y. et al. (2011). A long noncoding RNA maintains active chromatin to coordinate homeotic gene expression. Nature, 472, 120-124. Xu, J., Qi, X., Gong, J., Yu, M., Zhang, F., Sha, H. et al. (2012). Fstl1 Antagonizes BMP Signaling and Regulates Ureter Development. Public Library of Science One, 7, e32554, https://doi.org/10.1371/journal.pone.0032554.
Appendix
Supplementary table 1 Sequences of qPCR primers and the specific RT primer for Fstl1 pre-mRNA, together with the
expected amplicon length and the melting temperature (Tm) of the amplicons. The expected amplicon lengths were
obtained from NCBI primer blast and verified by gel electrophoresis. The Tm was determined with a melting curve assay
Supplementary table 2 Genes identified with BLASTN with a BIT score of 40 or higher. The start and end position and
the length of the similarity stretches are indicated on Fstl1 mRNA and mRNA of the respective genes. Based on these results, every gene could be assigned to on of the five regions on the Fstl1 mRNA.
amount of cycles needed to reach the detection threshold (mean Cq). For Cacna1b, Fstl5 and Grik4 5 of cDNA µl was used (equivalent to 25 ng cDNA), whereas for Cplx1 3 µl of cDNA was used (equivalent to 15 ng cDNA). n indicates the amount of samples that gave a signal. 12 is the maximum amount as there were 3 biological replicates for both WT and KO and 2 technical replicates per sample (6 x 2 = 12).
E11.5
Before After
Gene Mean Cq n Gene Mean Cq n
Cacna1b 35,14989 12 Cacna1b 32,6207 12 Fstl5 36,4338 5 Fslt5 29,56235 12 Grik4 36,76344 12 Grik4 34,39577 12
E18.5
Before After
Gene Mean Cq n Gene Mean Cq n
Cacna1b 37,79344 2 Cacna1b 37,31009 6 Cplx1 37,76942 4 Cplx1 34,67003 10 Fstl5 37,55582 5 Fstl5 35,57454 12 Grik4 37,745 12 Grik4 39,87544 12
Supplementary figure 1 Schematic depiction of the Fstl1 gene on chromosome 16, with the location of the RT primer
Supplementary figure 2 To illustrate that one of the wild type samples showed aberrant expression, the following plots
are shown. (A) Expression of Fstl1 is much higher in wild type 7 than in the other two wild types. (B-C) The higher expression is also visible in normalization genes Ppia and Rpl32, albeit much less pronounced. (D-F) Examples of genes in which WT7 disturb a reliable calculation of the mean expression in wild type mice.
