A Combined Metabonomic and Transcriptomic ... - ACS Publications

May 5, 2010 - Investigate Metabolism during Development in the Chick ... Biomolecular Medicine, Department of Surgery and Cancer, Faculty of Medicine,...
0 downloads 0 Views 4MB Size
A Combined Metabonomic and Transcriptomic Approach to Investigate Metabolism during Development in the Chick Chorioallantoic Membrane Rachel Cavill,† Jasmin K. Sidhu,†,⊥ Witold Kilarski,‡,§ Sophie Javerzat,‡,§ Martin Hagedorn,‡,§ Timothy, M. D. Ebbels,† Andreas Bikfalvi,*,‡,§ and Hector C. Keun*,† Biomolecular Medicine, Department of Surgery and Cancer, Faculty of Medicine, Sir Alexander Fleming Building, Imperial College London, London, SW7 2AZ; INSERM, U920, 33405 Talence, France; and Universite´ Bordeaux I, 33405 Talence, France Received January 15, 2010

The chick chorioallantoic membrane (CAM) is a powerful alternative to rodent models for the study of physiological or pathological angiogenesis. We investigated metabolic changes during the maturation of the CAM by 1H NMR-based metabolic profiling (metabonomics/metabolomics), allowing simultaneous measurements of many metabolites in an untargeted fashion. Specifically, we examined the time course of the measured metabolites to elucidate common patterns of regulation. Three clusters of metabolites were observed that correspond to essential biological processes active in the CAM with similar dynamics. The time courses common to the metabolite clusters distinguished specific stages of vessel growth, identifying waste product metabolites being stored in the CAM and energy-related substrates decreasing during embryonic growth. Using this top-down approach, combined with existing microarray data, we could link gene expression to metabolic consequences during the growth of a vascularized organ. For example, transcriptomic analysis demonstrated that many transcripts involved in the TCA cycle were down-regulated during CAM development, which correlated with the decrease in levels of TCA precursors and intermediates seen in the metabolite data. Taken together, this paper provides the first metabonomic study in an embryonic tissue where vessel development is the most active morphogenic process. Keywords: angiogenesis • chicken model • metabolomics

Introduction The chicken chorioallantoic membrane (CAM) is used as a model system for chemically induced and tumor angiogenesis assays. It has been used to test biocompatibility and toxicology of organic and inorganic materials1 and to evaluate the ability of bacterial strains or viruses to invade epithelial barriers.2,3 It has also been used for the evaluation of angiogenic and antiangiogenic compounds within its vasculature.4-8 Compared with other assays for this task the CAM assay is low-cost, allows more rapid throughput, and is easily applicable.4 The CAM has also been used to study growth, angiogenesis, and metastasis of human tumors.10,11 By implanting tumor cells onto the CAM 10 days after fertilization of the egg, a model that simulates key features of human tumor growth can be generated within a few days,5 thus allowing rapid research on human tumor progression and preclinical drug screening. * To whom correspondence should be addressed. E-mail addresses: [email protected]; [email protected]. † Imperial College London. ⊥ Present address: West Midlands Cancer Intelligence Unit, Public Health Building, The University of Birmingham, Birmingham B15 2TT, U.K. ‡ INSERM. § Universite´ Bordeaux I.

3126 Journal of Proteome Research 2010, 9, 3126–3134 Published on Web 05/05/2010

Between developmental days 5 and 6 (E5 and E6), the CAM undergoes rapid growth and increases its size 20-fold,6 and this growth continues until around E10.7 The allantois connects the embryo to the chorion, the bloodstream runs through the allantois, and the allantoic veins and artery are connected to the chick’s heart. These preformed blood vessels then migrate round the embryo and form capillaries. The speed of this capillary formation increases until day E11,8 when it starts to decrease very rapidly, and as this occurs, the mitotic index of the cells also declines.6,14 By E14, the allantois will completely surround the embryo, and by E18, the vascular system will have reached its final arrangement.8,9 In the CAM, blood vessels are formed by three mechanisms called vasculogenesis, sprouting, and intussusception. Vasculogenesis is the formation of new blood vessels from angioblasts, which occurs very early in development. This mechanism is different from sprouting (sprouting angiogenesis) where vessels are formed from preexisting vessels (E3-E7). Both mechanisms exist in the developing CAM, and the molecular mechanisms have been extensively studied.10 Intussuception is a nonsprouting mechanism that consists of vascular expansion through incorporation of tissue pillars (E7-E13).11 Intussuception includes three modes of vascular growth: in addition 10.1021/pr100033t

 2010 American Chemical Society

Metabolism during Development in the Chick CAM to intussusceptive microvascular growth (IMG ) expansion of capillary networks), intussusceptive arborization (IAR ) formation of feeding vessels from capillaries) and intussusceptive branching remodeling (IBR intussusception at bifurcations).12 This mechanism originally identified in the CAM is also operating in other normal and pathological tissues. The CAM is involved in several important biological processes.13 First, it allows exchange of oxygen and carbon dioxide. Second, it is also involved in the storage of waste products such as uric acid, urea, and ammonia that are produced by the embryo. Finally, it is involved in calcium uptake from the shell through the chorionic epithelium. This leads to a strong increase in calcium concentrations in plasma and embryonic tissues from E10. Metabonomics20-22 aims to study the profile of metabolites present within a living system and their changes over time or in response to experimental conditions. It uses analytical techniques such as NMR or mass spectrometry, which allow the levels of hundreds of metabolites to be measured simultaneously. Since NMR and mass spectrometry produce large amounts of complex data, the use of pattern recognition techniques to aid the interpretation of the data is common, so that further analysis and identification of metabolites can focus on those features of most interest. While several studies have used NMR or the related technology of MRI to image the chick,23 the CAM has never been specifically examined by NMR despite its importance for vascular biology. We therefore undertook a systematic study to determine key metabonomic signatures of the developing CAM by NMR spectroscopy and combined the interpretation of these with existing transcriptomic profile data. Because vascular development is the most active process occurring in the developing CAM, these observations may be primarily related to morphogenesis of the vasculature in this tissue.

Materials and Methods Chick Embryo Culture, Tissue Extraction, and NMR Analysis. Fertilized chicken eggs (Morriseau, France) were grown in a humidified atmosphere at 37 °C in egg incubators (Ehret, Emmendingen, Germany) as described previously.5 For metabonomic analysis, CAMs were isolated every day from embryos between developmental day E6 and E19 (n ) 5 embryos/day) and directly snap-frozen in liquid nitrogen. Aqueous soluble metabolites were separated from cell membrane lipids for each CAM sample using a methanol-chloroformwater extraction. All CAM tissues (∼50-800 mg wet weight) were twice extracted, beginning with manual homogenization using 300 µL of chloroform/methanol (2:1). The homogenate was placed on ice, and 300 µL of water was then added. This mixture was vortexed and then centrifuged at ∼16 000g for 10 min. The upper aqueous layer was carefully separated from the lower organic layer, and organic solvent was allowed to evaporate overnight. Residual water in the aqueous phase was removed by lyophilization the following day. Aqueous extracts were reconstituted in phosphate buffer (pH 7.4, 100% D2O, 1 mM TSP, and 3 mM NaN3) for 1H NMR spectroscopy. A 550 µL aliquot of each aqueous extract was pipetted into 5 mm NMR tubes. Aqueous extracts were frozen at -40 °C until NMR analysis. 1 H NMR spectra for all aqueous extracts were collected on a Bruker AVANCE 600 spectrometer (Bruker Biospin, Rheinstetten, Germany) at a frequency of 600.44 MHz and temperature of 300 K. Samples were automatically inserted into a 5 mm BBI probe, and gradient shimming was performed prior

research articles to the acquisition of each spectrum. The spectra were acquired using a 1D Carr-Purcell-Meiboom-Gill (CPMG) pulse sequence [RD-90°-(τ-180°-τ)n-acquire],14 which is edited by T2 relaxation times to attenuate residual protein resonances. Specifically τ ) 400 µs and n ) 80 was used. Presaturation of the water resonance was applied during the 2 s RD (relaxation delay). Spectra were the sum of 256 free induction decays (FIDs) collected into 32 000 complex data points, using a spectral width of 12 019 Hz (20 ppm). Data were zero filled by a factor of 2 to 64 000 points with an exponential multiplication equivalent to 1.0 Hz (for aqueous extracts) line width applied prior to Fourier transformation (XWINNMR software, Bruker Biospin, Rheinstetten, Germany). The Fourier-transformed spectra were then phased, baselinecorrected, and referenced to TSP (0 ppm). Unless otherwise stated all processing was conducted using in-house routines written in Matlab by Rachel Cavill, Hector C Keun, Timothy M. D. Ebbels, and Jake T. M. Pearce (version 2007b, The Mathworks, Natwick). Spectral assignments were performed using 2D 1H-1H COSY spectra and the Chenomx NMR Suite Professional software package, version 6.0 (Chenomx Inc., Edmonton AB). Metabonomic Data Analysis. All analysis was performed on the spectra normalized using probabilistic quotient normalization15 with the reference spectrum being defined as the median of all spectra. The spectra were reduced to a set of 127 variables by integrating the NMR intensity in regions corresponding to a peak, a resonance, or a set of overlapped peaks, selected by visual inspection. Peaks were required to be double the noise level in all samples for at least one day in order to be included in the final list of integrals. Using this approach, we were able to simplify spectral interpretation and increase the signal-tonoise. The average size of these regions was 0.04 ppm (range 0.01-0.11) and in total 51.8% of the spectral width was integrated (excluding the water and TSP regions). To examine correlation in the intensity variation between metabolite resonances, we calculated the Pearson correlation of all spectral variables to each to other and used the subsequent correlation matrix as input into a hierarchical clustering routine, clustering according to single linkage and a Euclidean distance metric. The method used (based on median fold change) minimizes normalization errors because it removes any gross linear variation in intensity across samples (e.g., as a result of variable dilution). The choice of the median spectrum as a reference is fairly arbitrary, and the method has been proven to be robust to the exact nature of the reference spectrum chosen.15 This normalization procedure compensates for spurious linear variation introduced in metabolic profiles by, for instance, sample dilution or variation in total biomass. To evaluate the significance of the variation in metabolite resonances, we performed one-way ANOVA testing whether the means were significantly different for each region across each day of the experiment. The results (p-values) from this analysis are available online in Supplementary Table 2, Supporting Information. Transcriptomic Data Analysis. The CAM transcriptomic data used has been published, and all technical procedures are detailed.16 In brief, embryonic days E5, E7, E10, and E14 were compared using Affymetrix chicken GeneChips followed by significance of microarrys (SAM) analysis.17 SAM analysis utilizes a Wilcoxon-test statistic and repeated sample-label permutation (n ) 200) to evaluate statistical significance between sample groups. Selecting a false discovery rate (FDR) Journal of Proteome Research • Vol. 9, No. 6, 2010 3127

research articles

Cavill et al.

Figure 1. 1H NMR metabolic profiles of the CAM during development. A section of overlaid 1H NMR CPMG spectra of the aqueous extracts of CAM tissue, colored according to day of collection. Labels: 1, R-hydroxyisovalerate; 2, isoleucine; 3, valine; 4 β-hydroxybutyrate; 5, lactate; 6, alanine; 7, acetate; 8, glutamate; 9, citrate; 10, aspartate; 11, creatine; 12, choline; 13, phosphocholine; 14, betaine; 15, glucose; 16, leucine; 17, arginine; 18, glutamine; 19, dimethylglycine; 20, proline; 21, glycine; 22, myo-inositol.

Results

during CAM development. Having established the identities of several metabolites in each of the three clusters (Table 1), we wished to examine qualitatively how their levels changed over time. Typical time courses of normalized levels of selected metabolites are shown in Figure 3. Metabolites in “cluster 1” generally increased over time, while those in “cluster 3” exhibited decreases from day 6. Metabolites in “cluster 2” appeared to possess a biphasic trajectory, increasing until day 8-13 and then decreasing. A complete description of the integral regions, the cluster membership, and the significance of interday variation (via ANOVA) are available online in Supplementary Table 2, Supporting Information. The metabolites from the three clusters can be seen in Table 1, along with a full list of each region and the cluster in which it is found, in Supplementary Table 2, Supporting Information.

In this study, we sought to determine the metabolite profile in the CAM over the course of development, taking samples from the CAM daily from day E6 to day E19 for metabonomic analysis by 1H NMR spectroscopy. Visual inspection of the NMR spectra of aqueous extracts from these CAM samples revealed the presence of several metabolites in high abundance (Figure 1). Spectral assignments were performed using publicly available data, commercial databases (Chenomx NMR Suite Professional), and 2D-homonuclear NMR experiments. A complete table of annotated resonances can be found online in Supplementary Table 1, Supporting Information. To examine whether there were changes in the relative metabolite levels over the course of CAM development, spectra were colored according to the developmental day. This presentation demonstrated that there were not only major changes in the metabolite levels over the time course studied but also different patterns of change between different metabolites through time. To define the groups with a similar time course more precisely, we performed hierarchical clustering on the metabolite correlation matrix, which summarizes the similarity between the variation in each integrated region and every other region (Figure 2). The heatmap and dendrogram in Figure 2 clearly show three well-defined clusters, each potentially representing groups of metabolites with different dynamics

The observed metabolic profile dynamics are likely to reflect changes in both metabolic reactions within the CAM and transport of metabolites between the CAM and the embryo. In order to examine in more detail the source of these changes, we examined the dynamics in the transcription of enzymes in pathways related to the identified metabolites, using data from a microarray profiling study that had been reported previously.16 This study compared gene expression between each pairwise combination of time points (E5, E7, E10, and E14). Using significance of microarrays (SAM) analysis,17 sets of genes were identified with significant differences in expression between time points (controlling for 5% false discovery). The sets from each comparison contained 162 upregulated genes and 98 downregulated genes between E5 and E7, 537 upregulated and 26 downregulated between E7 and E10, 178 upregulated and 250 downregulated between E10 and E14, 2109 upregulated and 1346 downregulated between E5 and E10, and 1258 upregulated and 1797 downregulated between E5 and E14. Several points of correspondence of gene transcripts and metabolite levels were observed. Restricting our analysis to those genes defined in KEGG18 as being members of the key metabolic pathways processing the selected metabolites of interest, we observed that many experienced significant alterations at later time points when compared with E5 (Table 2). For example, three genes unambiguously identified as associ-

of 5%, we identified sets of transcripts significantly different between pairs of time points. Data integration was performed by first defining pathways from KEGG18 in September 2008 that contained the selected metabolites of interest. For each pathway, genes with known chicken orthologues (Table 2) were compared with sets of significant genes from SAM analysis to define corresponding pathway effects. All microarray data files were submitted to the US National Center for Biotechnology Information, Gene Expression Omnibus (GEO) (accession number GSE11636; sample numbers GSM294982, GSM294983, GSM294984, GSM294985, GSM294986,GSM-294987,GSM294988,GSM294989,GSM294990, GSM294991, GSM294992, and GSM294993.)

3128

Journal of Proteome Research • Vol. 9, No. 6, 2010

research articles

Metabolism during Development in the Chick CAM

Figure 2. Clustered heatmap and dendrogram of metabolite signals. The hierarchical clustering dendrogram was generated using single linkage and a Euclidean distance metric. The accompanying heatmap shows Pearson correlations between the levels of metabolites. The clustering shows three distinct groups of metabolites, which are labeled as clusters 1, 2, and 3. Table 1. Metabolites Identified in Each Cluster Resulting from Analysis of Intensity Correlationsa cluster 1

myo-inositol (δ 3.64) Creatine (δ 3.05, 3.93) glycine (δ 3.57) dimethylglycine (δ 2.94) acetate (δ 1.98) β-hydroxybutyrate (δ 1.20, 2.31, 2.41) betaine (δ 3.28) formate (δ 8.47) proline (δ 3.36) adenosine (δ 6.11, δ 8.25, 8.35) choline (δ 3.21) R-hydroxy-isovalerate (δ 0.84)

cluster 2

cluster 3

valine (δ 1.05) isoleucine (δ 0.95, δ 1.02) leucine (δ 0.96, δ 0.97) glutamine (δ 2.46) glucose (δ 3.48, 3.50, 3.54, 3.84, 3.86, 3.91, 4.66, 5.25) taurine (δ 3.25, δ 3.43) tyrosine (δ 6.92, 7.20) phenylalanine (δ 7.34, 7.38) lactate (δ 1.34, 4.12) glycerophosphocholine (δ 3.25)

citrate (δ 2.53, δ 2.68) aspartate (δ 2.82) ADP/AMP/ATP (δ 6.25, δ 8.27, δ 8.54, δ 8.61) hypoxanthine (δ 8.11, δ 8.12) glutamate (δ 2.13) phosphocholine (δ 3.23, δ 3.60)

a Three clusters of metabolites were identified as shown in Figure 3. For each cluster, this table gives identified metabolites and the chemical shifts of observed resonances in the 1H-NMR spectra, referenced to TSP at 0 ppm. The metabolites are ordered in decreasing significance of interday variation (ANOVA p value; see Supplementary Table 2, Supporting Information).

ated with the TCA cycle showed significant downregulation over time. Taken alongside the reducing levels of citrate (a key pathway intermediate) and aspartate (a precursor of citrate via another intermediate, oxaloacetate), this seems to indicate a significant change in the activity of the TCA cycle pathway occurring over the time period studied (Figure 4). Another pathway where multiple genes are differentially expressed between time points is taurine metabolism (Figure 5). In this pathway, GAD1, which catalyzes a reaction that produces taurine, decreases slowly throughout the time course, initially at a greater rate (2-fold decrease between days E5 and E7). GGT, which can further metabolize taurine downstream of GAD1, increases sharply toward the final time point of comparison (>4-fold increase between E10 and E14). These observations are consistent with a model where downregulation of taurine production and upregulation of taurine conjugation could lead to the observed reduction in taurine levels between E10 and E14. Although beyond our analysis of traditional intermediary metabolism, we also observed that three genes coding for tRNA

ligases were significantly decreased in expression over the time course. Subsequently other tRNA ligase genes were also examined and 11/37 (CARS, DARS, DARS2, EPRS, FARSB, IARS, LARS, RCJMB04_4k14, TARSL2, WARS, YARS2) showed a decrease in expression between day 5 and either day 10 or day 14, which could account for some of the perturbation to amino acid levels via a global change in the levels of protein synthesis. In addition to the activity of enzymes in metabolic pathways, the levels of metabolites will also be affected by the expression of transporter genes. Therefore, we examined the gene expression of several solute carrier (SLC) families. SLC family 16 are monocarboxylate transporters and therefore may affect the levels of metabolites such as lactate, although the metabolite specificity of many transporters is as yet unclear. Four of these transporters were seen to increase in expression over the time course, specifically, SLC16A1, SLC16A2, SLC16A5, and SLC16A7, which code for proteins MOT1, MCT8, MCT6, and MCT2, respectively. Metabolites involved in the TCA cycle are transported by SLC25 family members, the mitochondrial transporters, such as SLC25A1, a tricarboxylate transporter, Journal of Proteome Research • Vol. 9, No. 6, 2010 3129

research articles

Cavill et al.

Figure 3. Example time courses from the different clusters of metabolites. The metabolites shown, with the integral region used, are as follows: column 1, acetate (δ1.96-1.99), formate (δ8.45-8.48), β-hydroxybutyrate (δ1.155-1.25); column 2, valine (δ1.03-1.06), β-glucose (δ3.48-3.515), tyrosine (δ6.89-6.94); column 3, citrate (δ2.5-2.56), phosphocholine (δ3.215-3.24), asprtate (δ2.79-2.85). All data is from the median fold change normalized spectra, and the intensity is shown in arbitrary units (a.u.). Error bars show a single standard deviation across all replicates at each point.

which increased in expression throughout the time course. Other mitochondrial transporters that changed in expression are SLC25A30 (increase over time), SLC25A15 (decrease over time), and SLC25A36 (decrease over time). Hence it is possible that altered transporter activities contribute to the metabolite dynamics observed in this study.

Discussion Our study is the first metabonomic 1H NMR spectroscopic analysis of a developing highly vascularized embryonic organ, the chicken chorioallantoic membrane. We have assigned many of the resonances in the resulting spectra to specific metabolites using both 1D and 2D spectra and have identified clusters of coexpressed metabolites, using hierarchical clustering. Having observed tight clusters of metabolite level patterns over the time course, we correlated these patterns to gene regulation data to confirm changes in specific pathways within the CAM; the metabolites, which were clustered together, may be tentatively associated with biological functions. For instance, the metabolites seen in cluster 1 include salts such as acetate and formate, which would be commonly found in the waste products stored in the CAM. Other metabolites grouped into 3130

Journal of Proteome Research • Vol. 9, No. 6, 2010

this cluster may show similar time courses for different reasons; for instance, β-hydroxybutyrate is an energy substrate formed in fatty acyl decomposition; therefore the increasing levels of this substrate reflect the intense lipid metabolism active in the egg toward the end of the development period. The more complex time course seen in the second cluster follows closely the known time course for the speed of angiogenic growth in the CAM.8 These metabolites often show a bimodal maximum value with peaking levels at E8 and E11. It is known that blood vessels are still forming in the allantois at these time points and that the speed of capillary formation in the CAM increases until E11.8 The metabolic responses seen in cluster 2 could reflect substrate use possibly during vessel development or that these substrates are being transported away from the CAM by the newly formed vascular system to other parts of the embryo, which need them for further development. The TCA cycle intermediates and precursors were found in cluster 3, and the dynamics observed along with changes in TCA cycle pathway gene expression provide strong evidence for decreased activity in this pathway as the availability of substrates is decreased.

research articles

Metabolism during Development in the Chick CAM a

Table 2. Genes Related to Metabolites Exhibiting Alterations during Development in the CAM pathway

glucogenesis

enzyme

name

6.2.1.1

acyl-CoA synthetase

1.2.1.3

aldehyde dehyrdogenase

1.2.1.5

aldehyde dehydrogenase

3.6.1.7

acylphophastase

3.1.2.6

hydroxyacylglutathione hydrolase

2.6.1.44

alanine-glyoxylate transaminase

2.6.1.2

alanine transaminase

6.1.1.7

alanine tRNA ligase

4.1.1.15

glutamate decarboxylase 1

selenoamino acid metabolism glyoxylate and dicarboxylate metabolism

4.4.1.16 6.3.4.3 3.5.1.9

selenocysteine lyase formate-tetrahydrofolate ligase arylformidase

glycine, serine and threonine metabolism

1.1.99.1

choline dehydrogenase

2.1.1.5

betaine homocysteine s-methyltransferase

2.1.1.2

guanidinoacetate N-methyltransferase

valine, leucine and isoleucine degradation

2.7.3.2 2.6.1.42

creatine kinase branched chain amino-acid transanimase

glycolysis pentose phosphate

5.1.3.3 1.1.1.47

aldose 1-epimerase glucose 1-dehydrogenase

galactose metabolism

3.2.1.23 3.1.20

β-galactosidase R-glucosidase

phenylalanine metabolism

1.11.1.7 4.1.1.28

peroxidase aromatic L-amino-acid decarboxylase

6.1.1.20

phenylalanine tRNA ligase

4.1.1.28

aromatic L-amino-acid decarboxylase

1.11.1.8

thyroid peroxidase

6.1.1.1

tyrosine tRNA ligase

2.3.2.2

γ-glutamyltransferase

6.1.1.4

leucine tRNA ligase

pyruvate metabolism

alanine and aspartate metabolism

streptomycin biosynthesis

genes

ACSS1 LOC416714 ACSS2 LOC419158 LOC423347 ALDH2 ALDH3A ALDH7A1 LOC426812 ALDH3B1 v LOC428813 ACYP2 ACYP1 HAGHL HAGH AGXT2 LOC431666 GPT2 LOC415746 AARS RCJMB04_28n18 AARS2 GAD1 V GAD SCLY MTHFD1 AFMID LOC769005 CHDH V LOC415993 BHMT LOC416371 GAMT LOC770737 CKMT1A BCAT1 LOC418193 GALM H6PD LOC428188 GLB1 GAA v LOC416462 LOC417691 PRDX6 DDC LOC420947 FARSB V FARS2 DDC LOC420947 TPO LOC776120 YARS2 V LOC418131 YARS GGT5 LOC416943 GGT1 v GGT7 LOC419121 LARS V LARS2 LOC420702

Journal of Proteome Research • Vol. 9, No. 6, 2010 3131

research articles

Cavill et al.

Table 2. Continued pathway

citrate cycle

enzyme

name

4.2.1.3 2.3.3.8

citrate (pro-S)-lyase aconitate hydratase

4.2.1.2 1.3.5.1

fumarate hydratase succinate dehydrogenase

6.2.1.4

succinate-CoA ligase

2.3.1.61

dihydrolipamide S-succinyltransferase

1.2.4.2

oxoglutarate dehyrogenase-like

1.1.1.42

isocitrate dehydrogenase 1

1.1.1.41

isocitrate dehydrogenase 3

1.8.1.4 4.1.3.6

dihydrolipoamide dehydrogenase citrate-lyase

genes

ACLY ACO1 ACO2 FH SDHA SDHB SDHD SUCLG2 v SUCLG1 RCJMB04_39i8 DLST OGDHL LOC423618 RCJMB04_19j12 v OGDH IDH1 LOC424112 RCJMB04_7e11 v LOC431056 IDH3A RCJMB04_9n20 IDH3B DLD CLYBL LOC418775

a Details of the pathways, enzymes, and gene names that were examined to find metabolite-related changes. The pathway names and enzyme details are taken from KEGG,18 in September 2008. Where an enzyme is involved in multiple pathways, it is shown in the table only once. Enzymes shown in bold were significantly changed in SAM analysis between E5 and at least one of the later time points; the arrows following these enzymes show the direction of change over time.

Figure 4. Gene transcription and metabolic variation in the TCA cycle during CAM development. Green is used to mark downregulated elements whose expression decreases with time, and red is used to mark upregulated elements whose expression increases with time. Metabolites and genes that do not show any significant changes or for which there are no measurements are shown in black. Variation of mean relative citrate and acetate levels over time are also shown. Error bars indicate one standard deviation from the mean.

Around E10, endothelial cell proliferation starts to decrease, potentially causing some of the observed changes. In particular, the noted downregulation of tRNA ligase genes from E10 is likely to lead to a decrease in cell proliferation. We have suggested different functional features of the CAM to which these clusters of metabolites may be related, namely, the storage of waste products, vessel formation, and energy substrates particularly those relating to the TCA cycle. An important outcome of our study is that we have linked a subset 3132

Journal of Proteome Research • Vol. 9, No. 6, 2010

of these changes to transcript data by identifying transcript and pathway alterations that account for modifications of specific metabolites such as creatine, aspartate, and taurine, demonstrating that the alterations seen in these metabolite levels over time originate from changes in the metabolic pathways within the CAM. Our study has limitations; for instance, it does not provide specific information on the different cell types, because both the microarray data and the metabolic data are obtained from the whole CAM. However, the CAM is a very complex

research articles

Metabolism during Development in the Chick CAM

Figure 5. Gene transcription and metabolic variation in the taurine metabolism during CAM development. Green is used to mark downregulated elements whose expression decreases with time; red indicates increased expression. Metabolites and genes that do not show any significant changes or for which there are no measurements are shown in black. Variations of mean relative taurine levels over time are also shown. Error bars indicate one standard deviation from the mean.

tissue built of three layers including the chorionic epithelium of ectodermal origin, the intermediate mesenchyme consisting of both ecto- and endodermal components, and the endodermal allantoic epithelium in contact with the allantoic cavity containing urine. Different cell types are present within the CAM; these include villus cavity (VC) cells and capillary covering (CC) cells, which are the two major components of the chorionic epithelium. Both cell types pass through the chorionic epithelium to face the eggshell and are involved in Ca2+ mobilization and resorption.19 Other cell types present in the chorionic epithelium layer are basal cells20 and pericytelike cells.21,22 Furthermore, capillaries are in close contact with the chorionic epithelium. The allantoic epithelium also consists of three different cell types including granule-rich cells, mitochondria-rich cells, and basal cells.21 Nevertheless, vessel development is the most prominent process occurring in the developing CAM and thus the metabonomic changes we observed may primarily reflect this process. This may be linked to specific modes of vascular development in the CAM such as sprouting angiogenesis, intussuception, and expansion. The earliest event is vasculogenesis, which is followed from E5 to E7 by sprouting angiogenesis.8 From E7 to E13 intussuception and expansion of the CAM is observed. Before E10, endothelial cell proliferation is high with approximately 20% of cells in proliferation;23 this decreases after E10. Thus, earlier metabonomic changes before E7 observed within clusters 2 and 3 are possibly linked to sprouting angiogenesis. Metabonomic changes observed before day 10 (cluster 2) may reflect the high mitotic rate of endothelial cells at this time point. Finally, metabonomic changes observed between E7 and E13 (clusters 2 and 3) may also be linked to intussuceptive growth. However, these contentions should be taken with caution since our study does not specifically discriminate between these different modes of vessel development in the CAM, because of the complexity of the CAM tissue and partially overlapping time frames. Taken together, this paper provides the first metabonomic study in an embryonic tissue where vessel development is the most active morphogenic process. In particular, we have studied the kinetics of metabonomic changes linked to gene

profiling and identified key signatures during CAM development. It will be important to determine metabonomic changes of growing tumors implanted into the CAM in comparison to tumor mouse models in order to evaluate its use as an alternative to murine models.

Acknowledgment. R.C. is supported by the EU carcinoGENOMICS project (Contract No. PL037712). This work has also been supported by a fellowship from the Lefolon-Foundation to W.K. and by INSERM (recurrent funding) and the “Ligue Nationale Contre le Cancer”. Supporting Information Available: Assignment of resonances in 600 MHz 1H NMR spectra of aqueous extracts of the chorioallantoic membrane and complete description of the integral regions used in this study. This material is available free of charge via the Internet at http://pubs.acs.org. References (1) Azzarello, J.; Ihnat, M.; Kropp, B.; Warnke, L.; Lin, H. Assessment of angiogenic properties of biomaterials using the chicken embryo chorioallantoic membrane assay. Biomed. Mater. 2007, 2 (2), 55– 61. (2) Adam, R. d.; Mussa, S.; Lindemann, D.; Oelschlaeger, T. A.; Deadman, M.; Ferguson, D. J. P.; Moxon, R.; Schroten, H. The avian chorioallantoic membrane in ovo - a useful model for bacterial invasion assays. Int. J. Med. Microbiol. 2002292, (3-4), 267–275. (3) Woolcock, P. R. Avian influenza virus isolation and propagation in chicken eggs. In Avian Influenza Virus; Spackman, E., Ed.; Methods in Molecular Biology, Vol. 36; Humana Press: Totowa, NJ, 2008; pp 35-46. (4) Taraboletti, G.; Giavazzi, R. Modelling approaches for angiogenesis. Eur. J. Cancer 200440, (6), 881–889. (5) Hagedorn, M.; Javerzat, S.; Gilges, D.; Meyre, A.; de Lafarge, B.; Eichmann, A.; Bikfalvi, A. Accessing key steps of human tumor progression in vivo by using an avian embryo model. Proc. Natl. Acad. Sci. U.S.A. 2005, 102 (5), 1643–1648. (6) Melkonian, G.; Munoz, N.; Chung, J.; Tong, C.; Marr, R.; Talbot, P. Capillary plexus development in the day five to day six chick chorioallantoic membrane is inhibited by cytochalasin D and suramin. J. Exp. Zool. 2002, 292 (3), 241–254. (7) Ribatti, D.; Vacca, B. N. A.; Roncali, L.; Burri, P. H.; Djonov, V. Chorioallantoic membrane capillary bed: A useful target for studying angiogenesis and anti-angiogenesis in vivo. Anat. Rec. 2001, 264 (4), 317–324. (8) Ausprunk, D. H.; Knighton, D. R.; Folkman, J. Differentiation of vascular endothelium in chick chorioallantois - structural and autoradiographic study. Dev. Biol. 1974, 38 (2), 237–248.

Journal of Proteome Research • Vol. 9, No. 6, 2010 3133

research articles (9) Wilting, J.; Brand-Saberi, B.; Kurz, H.; Christ, B. Development of the embryonic vascular system. Cell. Mol. Biol. Res. 1995, 41 (4), 219–32. (10) Drake, C. J.; Cheresh, D. A.; Little, C. D. An antagonist of integrin alpha v beta 3 prevents maturation of blood vessels during embryonic neovascularization. J. Cell Sci. 1995, 108 (7), 2655–2661. (11) Djonov, V. G.; Galli, A. B.; Burri, P. H. Intussusceptive arborization contributes to vascular tree formation in the chick chorio-allantoic membrane. Anat. Embryol. 2000, 202 (5), 347–357. (12) Djonov, V. G.; Kurz, H.; Burri, P. H. Optimality in the developing vascular system: Branching remodeling by means of intussusception as an efficient adaptation mechanism. Dev. Dyn. 2002, 224 (4), 391–402. (13) Bellairs, R.; Osmond, M. Atlas of Chick Development, 2nd ed.; Elsevier: London, 2005. (14) Meiboom, G. Modified spin-echo method for measurement of relaxation times. Rev. Sci. Instrum. 1958, 29, 688. (15) Dieterle, F.; Ross, A.; Schlotterbeck, G.; Senn, H. Probabilistic quotient normalization as robust method to account for dilution of complex biological mixtures. Application in 1H NMR metabonomics. Anal. Chem. 2006, 78 (13), 4281–4290. (16) Javerzat, S.; Franco, M. l.; Herbert, J.; Platonova, N.; Peille, A.-L.; Pantesco, V. r.; De Vos, J.; Assou, S.; Bicknell, R.; Bikfalvi, A.; Hagedorn, M. Correlating global gene regulation to angiogenesis in the developing chick extra-embryonic vascular system. PLoS ONE 20094, (11), e7856.

3134

Journal of Proteome Research • Vol. 9, No. 6, 2010

Cavill et al. (17) Tusher, V. G.; Tibshirani, R.; Chu, G. Significance analysis of microarrays applied to the ionizing radiation response. Proc. Natl. Acad. Sci. U.S.A. 2001, 98 (9), 5116–5121. (18) Ogata, H.; Goto, S.; Sato, K.; Fujibuchi, W.; Bono, H.; Kanehisa, M. KEGG: Kyoto encyclopedia of genes and genomes. Nucleic Acids Res. 1999, 27 (1), 29–34. (19) Akins, R. E.; Tuan, R. S. Transepithelial calcium transport in the chick chorioallantoic membrane. I. Isolation and characterization of chorionic ectoderm cells. J. Cell Sci. 1993, 105 (2), 369–379. (20) Hoshi, H.; Mori, T. The fine structure of the chorionic epithelium of chick embryo. Arch. Histol. Jpn. 1971, 33 (1), 45–58. (21) Coleman, J. R.; Terepka, A. R. Electron probe anlyisis of the calcium distribution in cells of the embryonic chick chorioallantoic membrance I. A critiical evaluation of techniques. J. Histochem. Cytochem. 1972, 20 (6), 401–413. (22) Patan, S.; Haenni, B.; Burri, P. H. Evidence for intussusceptive capillary growth in the chicken chorio-allantoic membrane (CAM). Anat. Embryol. 1993, 187 (2), 121–130. (23) Kurz, H.; Ambrosy, S.; Wilting, J.; Marme, D.; Christ, B. Proliferation pattern of capillary endothelial cells in chorioallantoic membrane development indicates local growth control, which is counteracted by vascular endothelial growth factor application. Dev. Dyn. 1995, 203 (2), 174–186.

PR100033T