Development of a Uniform Biomarker Signature in Calves Heart and Lung to Detect the Abuse of Different Anabolic Substances

Anabolic agents like steroid hormones or β-agonists are used in food-producing animals in order to increase muscle mass and reduce fat content [1,2]. The application of specific anabolic agents in animal husbandry is allowed and common in different countries, e.g. USA, Canada, South Africa, etc. Due to proven side effects of substance residues for the consumer, the use of all growth promoting agents is forbidden within the European Union (EU) (Directives 96/22/EC, 96/23 EC, 2008/97/EC). To monitor this ban, routine controls are performed in animal Abstract


Special Issue: Food Safety & Hygiene
samples from stables and slaughterhouses.Within this control system, substance residues are detected in different organs and matrices, like blood, urine, hair, eyeballs and other tissues using immunoassays or chromatography methods combined with mass spectrometry [3].But new xenobiotic substances or treatment regimens still present a problem for the control laboratories.For example, the distinction between endogenous hormones and natural hormones that are applied exogenously is methodically demanding.Another challenge is the detection of hormone cocktails, where a mixture of different growth promoters is applied, and each hormone in a concentration lower than the detection limit of the certified control methods [4].For this reason, the search for alternative innovative detection methods is an important research field.Thereby, the approach of identifying molecular biomarkers is very promising [5,6].This approach is based on the analysis of physiological changes caused by the treatment with growth promoting agents.The intended physiological effect is an increased muscle growth and a decreased fat content, regardless of the growth-promoting substance.Also other organs like liver, kidney, heart, lung etc. are physiologically influenced as well by growth-promoting agents and therefore these are further potential target tissues for molecular biomarker screening.There are different levels on which such biomarkers are detectable, namely the transcriptome, the proteome or the metabolome.Up to now, identified biomarkers are still substance specific [3,[7][8][9][10].In order to be applicable a screening method for treated animals, the ideal biomarker pattern should be independent of the applied anabolic substances, or other conditions, like breed, age, gender or even species.
We could already identify a first biomarker signature based on changes on the level of the transcriptome in the liver of calves treated with a steroid implant or clenbuterol respectively, which enabled to distinguish untreated from treated animals, independent of the applied drug [11].
It is known, that the applied substances of this trial also had an influence on the physiology of heart and lung.Clenbuterol is bound to β-adrenergic receptors, which are also present in heart tissue [12].Different studies have already shown that, β-agonists alter cardiovascular function, e.g.influencing heart rate, contractility and blood pressure [13].An effect on cardiac muscle growth could also be shown [14,15].Steroid hormones also have proven effects on heart tissue.Abuse of anabolic steroids seems to induce cardiac arrhythmia, hypertrophy, thrombogenesis or congestive heart failure [16].
Clenbuterol was originally designed for the treatment of asthma as a bronchospasmolytic agent.It mainly binds to β2adrenergic receptors, which are present in the bronchial tubes and in lung tissue.It has a direct effect on smooth muscles in the lung [17].In contrast, there is not much known about the effect of steroid hormones on lung tissue, yet.
As minimally one of the applied substances has a direct physiological influence on the heart or lung, the gene expression in those tissues was quantified, in order to identify additional biomarkers on the transcriptional level.

Animal Experiment
21 male, 6 month old Holstein Friesian calves were separated into three equal groups.One group was untreated (control), the second group was treated with a combination of progesterone plus estradiol benzoate (steroid) and the last group was treated with clenbuterol.A detailed description of the animal trial is given in Riedmaier et al. 2014 [11].

Gene Expression Analysis
Gene expression analysis was performed according to the MIQE guidelines [18].
RNA was extracted from heart and lung samples using the miRNeasy Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer´s protocol.RNA purity was calculated using the OD 260/280 ratio.RNA quality was determined using the Eukaryotic total RNA Nano Assay on the 2100 Bioanalyzer (Agilent Technology, Palo Alto, USA).
For cDNA synthesis, constant amounts of 500 ng integer total RNA were reverse transcribed as already described before [19].
To analyze the expression of candidate genes, qPCR analysis was done using the iQ5 detection system (Bio-Rad, Munich, Germany) as already described before [11].
Candidate genes were chosen by screening the current literature for the effects of steroid hormones and clenbuterol on the respective organs and by analyzing different biochemical pathways, where the respective substance or factors that showed significant regulation are involved.Selected genes can be summarized in the functional groups of hormone receptors, transcription factors, proliferation factors, regulators of angiogenesis, apoptosis, blood pressure, protein, glucose & lipid metabolism, immune factors, oncogenes, structural proteins and different other factors.In total 80 genes were quantified in heart tissue and 90 genes were quantified in lung tissue, whereas 48 of those genes were quantified in both tissues.A detailed list of quantified genes and traits of all primer pairs is given in supplemental Tables 1 and 2.
The analysis of significantly regulated genes and multivariate data analysis was done as described before [11,20].Dynamic PCA was performed with the normalized gene expression data from each tissue separately and in combination using GenEx version 6 (MultiD Analyses AB, Gothenburg, Sweden).Within this method, an ideal biomarker signature can be chosen using step by step exclusion of genes due to a declining p-value or a rising distance of regulation [11].
Additionally Orthogonal Projections to Latent Structures Discriminant Analysis (OPLS-DA) was performed using SIMCA software (Umetrics, Umea, Sweden) in order to build statistical models.OPLS-DA is a well-known statistical method used in 'omics research' like genomics, proteomics or metabolomics [21,22].OPLS removes undesired variability in complex data sets and can though be used in multivariate calibration and development of various filters.It separates non-correlated and correlated variation and decreases the total number of components used.OPLS-DA is performed to sharpen the separation between groups and to filter for variables carrying the class separation information [23].Within the resulting scatter plots, horizontal direction gives information about inter-class variation and vertical direction shows variations within the classes.SIMCA software provides two parameters that give information about the quality of the OPLS-DA models.R2 (cum) describes how well the model fits the X data and thereby how well the model is able to separate between the groups.The closer the value to 1, the better the model fits the X data.Q2 (cum) describes how well the generated model will predict new data.A value >0.5 indicates a good predictability of the designed model.In contrast to PCA analysis, groups have to be defined before OPLS-DA performance in order to find the best parameters to build a statistical model for group separation.

RNA quantity and quality
The quantity and purity of the extracted total RNA were determined using the Nano Drop Photometer (Peqlab, Erlangen, Germany).Mean RNA concentration was 768 ± 63 ng/µL for heart tissue and 1380 ± 430 ng/µL for lung tissue, respectively.The OD 260/280 ratio is an indicator for RNA purity whereas a ratio >1.8 is considered as adequate for RT-qPCR experiments.Mean OD 260/280 was 2.14 ± 0.009 for heart tissue and 2.05 ± 0.01 for lung tissue, respectively.
RNA quality and integrity was determined using the Eukaryotic total RNA Nano Assay on the 2100 Bioanalyzer (Agilent Technologies).RNA Integrity Number (RIN) >7 can be considered as good quality RNA that is usable for RT-qPCR experiments.The mean RIN for the heart samples was 7.25 ± 0.24 and for the lung samples 8.2 ± 0.4, indicating intact RNA [24].

Influence of treatment on gene expression in heart
Both substance groups that were applied to the calves in this study are known to have a physiological influence on heart tissue.

Influence of treatment on gene expression in lung
Clenbuterol was originally designed for the treatment of asthma as bronchospasmolytic agent, highlighting the direct effect on lung tissue.However, there is not much known about the effect of steroid hormones on lung tissue.Therefore, the expression of the gene set that was chosen for clenbuterol was investigated in samples obtained from steroid-treated animals.

Multivariate data analysis
To evaluate the potential of those gene expression results to develop a biomarker pattern, dynamic PCA and OPLS-DA were In lung tissue, a signature of 4 genes (FHIT, LPL, p53, PTGDS) could be identified for steroid treated animals and as signature In both tissues, a first biomarker pattern could be identified, but separation by PCA was not perfect meaning that control animals still group within the treated animals or vice versa.Therefore, qPCR data from lung and heart were combined in one dynamic PCA to evaluate, if a united biomarker signature can be identified.For the separation of control animals from steroidtreated animals, a signature of 17 genes, namely, ANGPT1, BMP2R, CAST, FLT-1, GRα, IRβ, RB1 and SMAD2 measured in heart tissue and ADRB2, FHIT, GDPD1, LPL, p53, PKC, SerpinE1, SMAD2 and TGFβ measured in lung tissue could be identified.For clenbuterol treatment, a signature of 9 genes could be identified, namely CEBPD, eNOS and SRF measured in heart tissue and AK3L1, CSF2, GRα, HRH, IGF-1R and IRβ measured in lung  tissue.The dynamic PCA for both treatments is shown in Figure 3. Additionally, a biomarker pattern of 23 genes (BMPR2, CAST, ENO1, NOS3, ERα, FAS, FLT-1, GRα, IGF-1, IRβ, PKB, RB-1, SMAD2 and TIMP-2 measured in heart tissue and ADRB2, Casp8, FHIT, GDPD1, GRα, IRα, PKC, PTGDS and SerpinE2 measured in lung tissue) that enabled separation of treated and untreated animals, independent of the applied substance, could be identified (Figure 3C).

Development of a Uniform Biomarker Signature in Calves Heart and Lung to Detect the Abuse of Different Anabolic Substances
To evaluate the predictive power of the biomarker signatures obtained by combining the results from both tissues, OPLS-DA analysis was performed.Regarding the resulting scatter plots, horizontal direction gives information about variations between the defined groups and vertical direction shows variations within the groups.Blue dots represent animals in the control group, red dots represent animals in the steroid group, light green dots represent clenbuterol-treated animals and dark green squares represent treated animals, independent of treatment.Figure 4 shows that, horizontal separation was achieved in both groups regarded separately (Figure 4A), whereas separation between control and the clenbuterol group was better (R2 (cum) = 0.929; Q2 (cum) = 0.824) than the separation of control group Vs the steroid group (R2 (cum) = 0.793; Q2 (cum) = 0.392) (Figure 4B).In the steroid group, one control animal, grouped with the treated animals.Regarding the variation within the groups, no difference could be observed.The predictive power of the third biomarker pattern (Figure 4C), enabling separation of treated from untreated animals independent of the applied substance (R2 (cum) = 0.830; Q2 (cum) = 0.520) was superior than for the steroid group regarded separately, but not as good as the pattern identified for clenbuterol alone.

Discussion
The misuse of growth-promoting substances in animal husbandry is an everlasting problem.Controls that are performed in routine monitoring are based on the detection of substance residues in different matrices, like urine, blood, hair, eyes or other tissues.But the identification of unknown xenobiotic drugs or hormone cocktails is still a problem [4].An additional challenge is to distinguish between endogenous natural hormones and natural hormones that are applied exogenously.The idea of the identification of molecular biomarkers that are based on the investigation of physiological changes caused by the treatment with growth promoting agents has come into focus during the last years.Efforts in that field have already been made on the level of the metabolome [25], the proteome [7,8] and the transcriptome [3, [9][10][11]19,26,27].
Within this study, new transcriptomic biomarker candidates for the detection of treatment with steroid hormones and clenbuterol in bovine heart and lung were quantified.Therefore, the targeted approach was applied, meaning that a list of potential biomarkers was selected by screening the literature for already published candidates or by choosing key factors from biochemical pathways that are known to be influenced by the treatments.The mRNA expression of those biomarker candidates was measured by RT-qPCR.
Regarding the physiological effects of treatment monitored by gene expression changes, it can be concluded that both treatments cause an increase in cell proliferation, an increased protein metabolism and an increase in the mobilization of lipid tissue.These effects go in line with the anabolic effect of the applied drugs.
To evaluate, if the quantified genes can be combined to a biomarker signature that enables the separation of treated from untreated animals, dynamic PCA was applied.In both tissues, a gene pattern could be specified for both treatments.All four biomarker patterns were not a simple combination of    could be identified in heart and lung tissue to detect the illegal use of clenbuterol or steroids.The combination of the results of both tissues showed even better results than regarding each organ individually.Additionally, a biomarker pattern for the separation of untreated vs. treated animals independent of the applied substances could be identified.In liver tissue, similar results -a biomarker pattern independent of treatment -could be achieved, whereas in liver tissue, additional biomarkers could be identified using the non-targeted RNA-Sequencing approach [6].The results of the current study showed that combining results from more tissues could also be a target-aimed.
To verify those biomarker signatures, more validation studies with other anabolic substances or application strategies like low dose cocktails will be necessary.Including gene expression data from other target organs, e.g.muscle tissue into the dynamic PCA analysis would also be a promising way to get a valid biomarker signature.significantly regulated genes.Both gene patterns in lung tissue consisted of genes that were significantly regulated or regulated by trend (p<0.1).In heart tissue, also genes that showed a p-value higher than 0.1 were included in the pattern.Using the dynamic PCA algorithm, genes can either be excluded by a decreasing p-value or by an increasing distance of regulation [11].Though, in both biomarker patterns identified in heart tissue, several genes were included, that has been chosen due to their distance of regulation.Significance was not reached for those genes due to inter-individual variability within the data set.The biomarker signatures identified in both tissues separately were not suitable to perfectly separate treated and untreated animals.To examine, if a better separation could be achieved by combining the results from both tissues, dynamic PCA was performed with qPCR data obtained in heart plus lung.Thereby a gene pattern could be identified to separate control animals from the steroid and clenbuterol-treated group.Additionally, a biomarker pattern that enabled separation of treated and untreated animals, independent of the applied substance was found.Identifying a substance independent biomarker pattern would be of advantage for establishing a first screening method in order to evaluate if an animal was treated with growth-promoting agents or not.Afterwards, an identification of the applied drug could be performed.
Predictive multivariate data analysis tools are helpful to determine the predictive power of a biomarker set.For that purpose, the use of OPLS-DA has already been shown to be adequate [21,22].Regarding the resulting OPLS-DA analyses, best results were achieved for the combination of the results from heart and lung tissue for clenbuterol treatment.Good predictive power could also be confirmed for the biomarker signature that was independent of treatment.Only the results obtained for steroid treatment showed lower predictability.This may result from the control animal that clusters within the steroid treatment group.
These results show the suitability of primary target organs of the applied drugs for biomarker identification.Clenbuterol was designed to act as bronchodilator for the treatment of asthma.It directly acts via β2-adrenergic receptors in lung tissue resulting in relaxation of the bronchial muscles [17].Another physiological effect of clenbuterol is an increased heart rate [28].Both tissues are direct target organs of clenbuterol and were therefore regarded as ideal targets for biomarker identification.That assumption could be confirmed by the obtained results.Heart tissue as a potential source for gene expression biomarkers was chosen due to already published effects of steroid hormones on heart muscle [29].During slaughter, increased heart sizes could also be observed (data not shown).As a consequence, major effects of steroid treatment were expected in heart tissue.Regarding lung tissue, effects of steroid hormones are rather unknown.However, a first biomarker pattern for the treatment with the steroid implant could be identified.

Figure 1 :
Figure 1: Dynamic principal component analysis of results obtained in heart tissue.Figure 1A shows separation of the control group and the steroid treated group and Figure 1B the separation of the control animals from the clenbuterol treated ones.Animals of the control group are represented by blue dots, animals of the steroid-treated groups are represented by red triangles and clenbuterol treated animals are shown by light green diamonds.

Figure 2 :
Figure 2: Dynamic principal component analysis of results obtained in lung tissue.Figure 2A shows separation of the control group and the steroid treated group and Figure 2B the separation of the control animals from the clenbuterol treated ones.Animals of the control group are represented by blue dots, animals of the steroid-treated groups are represented by red triangles and clenbuterol treated animals are shown by light green diamonds.

Calves
Heart and Lung to Detect the Abuse of Different Anabolic Substances.J Nutrition Health Food Sci 3(4): 1-8.DOI: http://dx.doi.org/10.15226/jnhfs.2015.00151Development of a Uniform Biomarker Signature in Calves Heart and Lung to Detect the Abuse of Different Anabolic Substances Copyright: © 2015 Riedmaier et al.

Figure 3 :
Figure 3: Dynamic principal component analysis of results obtained by the combination of qPCR data from heart plus lung tissue.Figure 3A shows separation of control group and the steroid treated group, Figure 3B the separation of the control animals from the clenbuterol treated ones and Figure 3C the separation of control animals from treated individuals, independent of the applied substance.Animals of the control group are represented by blue dots, animals of the steroid treated group are represented by red triangles and clenbuterol treated animals are shown by light green diamonds.Ellipses were drawn by hand to clarify a clear separation of treated and untreated animals.
Figure 3: Dynamic principal component analysis of results obtained by the combination of qPCR data from heart plus lung tissue.Figure 3A shows separation of control group and the steroid treated group, Figure 3B the separation of the control animals from the clenbuterol treated ones and Figure 3C the separation of control animals from treated individuals, independent of the applied substance.Animals of the control group are represented by blue dots, animals of the steroid treated group are represented by red triangles and clenbuterol treated animals are shown by light green diamonds.Ellipses were drawn by hand to clarify a clear separation of treated and untreated animals.

Figure 4 :
Figure 4: Dynamic principal component analysis of results obtained by the combination of qPCR data from heart plus lung tissue.

Table 1 :
List of significantly regulated genes in heart and lung tissue.