Available online at www.sciencedirect.com

Theriogenology 77 (2012) 1767–1778

0d

Cellular and molecular characterization of the impact of laboratorysetup on bovine in vitro embryo production

Dany Plourdea, Christian Vigneaultb, Isabelle Laflammea, Patrick Blondinb,Claude Roberta,*

a Laboratory of Functional Genomics of Early Embryonic Development, Centre de recherche en biologie de la reproduction, Institut desnutraceutiques et des aliments fonctionnels, Faculté des sciences de l’agriculture et de l’alimentation, Pavillon des services, Université Laval

(Québec), Canada G1V 0A6b L’Alliance Boviteq Inc., 19320 Grand rang St-François, Saint-Hyacinthe, (Québec), Canada J2T 5H1

Received 20 May 2011; received in revised form 22 November 2011; accepted 17 December 2011

Abstract

One of the main objectives related to performing comparative analysis of embryonic transcriptomes is to share information with otherreproductive biologists or commercial service providers. Biological extracts influence performance of in vitro production systems andaffect the reproducibility of results between production sites; these sources of variation could impede the potential for knowledgetransfer. The objective of the present study was to assess the impact of the production site when sharing a common in vitro embryoproduction protocol. Biological extracts and semen were shared between production sites and thus removed as potential sources ofvariation. To remove the impact of blastocyst staging, all comparisons used expanded blastocysts. Although blastocyst yields and thenumber of Tunel positive cells per embryo differed between production sites, blastocysts were morphologically very similar in regardsto cell number, their allocation to either the trophoblast or inner cell mass, or their gender ratio. These observations were also confirmedat the gene expression level, as indicated by highly similar transcript abundances. Only 36 genes out of the 16,121 expressed duringbovine prehatching development were statistically differentially expressed, of which a large proportion were associated with theapoptotic process. These results highlighted the impact of laboratory set up, including personnel experience, when replicating an in vitroproduction system. Although inherent differences may arise, given the similarity of results between production sites, we concluded thatembryo production protocols have the potential to be transferred and shared.© 2012 Elsevier Inc. All rights reserved.

Keywords: Bovine blastocyst; In vitro production; Morphology; Gene expression; Microarray

www.theriojournal.com

1. Introduction

In cattle, the natural reproductive cycle leads tothe birth of one offspring per year. To enhance thedissemination rate of high genetic merit animals,assisted reproductive technologies are used, as they

* Corresponding author. Tel.: 1-418-656-2131; fax: 1-418-656-3766.

E-mail address: claude.robert@fsaa.ulaval.ca (C. Robert).

093-691X/$ – see front matter © 2012 Elsevier Inc. All rights reserved.oi:10.1016/j.theriogenology.2011.12.021

can increase the number of offspring per year. Stan-dard procedures involve increasing the amount offertilizable oocytes through ovarian stimulation, their fer-tilization in vivo through artificial insemination, recoveryof embryos through uterine flushing, and transfer intosurrogates. The in vitro production of embryos representsa method to further increase the number of transferableembryos by performing oocyte maturation, fertilizationand embryonic growth ex vivo, which allows rapid and

more frequent gamete collections. The repeated collection

ga1

1768 D. Plourde et al. / Theriogenology 77 (2012) 1767–1778

of immature oocytes for in vitro embryo production isestimated to yield almost three and a half times moretransferable embryos than multiovulatory embryo transferprocedures [1–3].

Commercial implementation of this strategy is pos-sible because of well-established in vitro productionsystems for cattle. However, it is well known thatcurrent in vitro conditions are suboptimal, as exempli-fied by lower blastocyst yields [4–8]. In addition, thereis a well-documented increase in the frequency of un-desirable phenotypes in in vitro produced embryos e.g.,lower cryotolerance [9], a skew in gender ratio in favorof males [10–12], and altered cellular counts betweenthe inner cell mass and the throphectoderm [13], whichmay culminate in development of the large offspringsyndrome [14].

To better understand the nature of these induceddeviations, comparative gene expression analyses havebeen widely performed [15–21]. So far, little is knownabout the physiological significance of these RNAabundance differences. In fact, heavy discordanceamong reports suggests a lack of agreement betweenresults produced by different laboratories. In this re-spect, we recently benchmarked every aspect of a typ-ical sample handling procedure for microarray analysisand reported a very high impact on the ensuing genelists [22]. In the current study, our concern was theimpact of the in vitro production systems themselves ongene expression. It is well accepted that although thenumber of in vitro systems in usage is limited, manyvariations have been introduced into each system bylaboratories. The interpretation of data produced bydifferent research teams relies on our knowledge ofthese sources of variations between in vitro embryoproduction systems. Furthermore, the integration of anyimprovement based on these studies is tightly linked tothe potential for knowledge transfer. It is alreadyknown that the source of biologically extracted com-ponents used in an in vitro production system, such asserum, albumin and hormones, can deeply influenceembryo production [23–25]. The current study aimed todetermine the extent of variance that could solely beattributed to laboratory set up and personnel. We hy-pothesized that the impact of the production site ismarginal when a common protocol is strictly followedby both sites and biological components i.e., BSA,serum and semen are shared. Additionally, ovaries werecollected from the same abbatoir to remove any impactthe source of oocytes on embryo production. Embryoswere compared on the basis of several phenotypic char-

acteristics, as well as at the gene expression level.

2. Materials and methods

All chemicals were from Sigma-Aldrich (St. Louis,MO, USA) unless specified otherwise.

2.1. In vitro embryo production

2.1.1. Oocyte recovery and selectionOvaries from dairy cattle were obtained from a com-

mercial abbatoir (the same for both production sites)and transported to the laboratory in saline (0.9% NaCl)containing 1% antimycotic agent. The interval fromovary collection to aspiration of cumulus-oocyte com-plexes (COCs) was similar for both production sites,with COC collection conducted within 2 h after ovariesarrived. The COCs were aspirated from 2 to 6 mmfollicles using an 18 G needle, 3.8 cm long, attached toa 10 ml syringe. Healthy COCs with at least five layersof cumulus cells were selected for maturation. Cumu-lus-oocyte complexes with fragmented cytoplasm, py-knotic cumulus, pale nuclei and abnormal morphologywere discarded.

2.1.2. In vitro maturationCumulus-oocyte complexes were placed in Hepes

buffered Tyrode’s media (TLH) solution (supplementedwith 10% bovine serum, 200 �M and 50 �g ml�1 ofentamicin) and washed thoroughly twice, to ensure thebsence of contamination from follicular liquid. Groups of0 healthy COCs were placed in 50-�l droplets of media

under 9 ml of filtered mineral oil. Maturation media wascomposed of TCM199 (Gibco 11,150-059; Invitrogen,Burlington, ON, Canada), 10% fetal bovine serum (Sterilefetal bovine serum for Cell Culture, Medicorp, Montréal,QC, Canada), 200 �M, 50 �g ml�1 of gentamicin and 0.1�g ml�1 follicle-stimulating hormone (FSH; Gonal-F, Se-rono, Canada Inc., Mississauga, ON, Canada). Cumulus-oocyte complex containing droplets were incubated for24 h at 38.5°C with 5% CO2, 20% O2 and high humidity.

2.1.3. In vitro fertilizationMatured COCs were washed twice in Hepes buff-

ered Tyrode’s media (TLH) solution. Groups of fivematured COCs were added to 50-�l droplets of mediaunder filtered mineral oil. Each droplet consisted ofmodified Tyrode lactate media, supplemented with0.6% BSA (Sigma fraction V), 40 mM pyruvate and 50�g ml�1 gentamicin. Then, 2 �L of PHE (1 mM hypo-taurine, 2 mM penicillamine, 250 mM epinephrine) wereadded to COCs containing droplets. Spermatozoa usedwere a cryopreserved pool of ejaculates from five Hol-stein bulls (Centre d’insémination artificielle du Qué-bec). Semen was thawed in 37 °C water, put on a

discontinuous Percoll gradient (2 mL of 45% Percoll

2

dHsslesepS

5

o

edadQcpipfp

2

mlSpttrsfds3ptclclt

2

PPtvMe

1769D. Plourde et al. / Theriogenology 77 (2012) 1767–1778

over a 2 mL of 90% Percoll) and centrifuged at 700 gfor 30 min at room temperature. The supernatant wasdiscarded and spermatozoa pellet were resuspended inin vitro fertilization media after sperm concentrationwas assessed using a hemocytometer to obtain a ratio of10 � 104 spermatozoa/COCs. Fertilization took placein an incubator for 15 to 18 h at 38.5 °C with 5% CO2,20% O2 and high humidity.

.1.4. In vitro cultureZygotes and unfertilized COCs were mechanically

enuded by repeated pipetting, and washed twice inepes buffered Tyrode’s media (TLH) solution to in-

ure complete removal of cumulus cells. Washed pre-umptive zygotes were then transferred to culture drop-ets in groups of 10. For standard culture conditions,mbryos were placed in 10-�l droplets of modifiedynthetic oviduct fluid (SOF; [26]) under filtered min-ral oil. Sequential SOF media were used as describedreviously [27]. The embryos were first placed inOF#1 (6 mM lactate, 0.2-mM glucose). The embryo

culture dishes were incubated 38.5 °C with 6.5% CO2,% O2 and high humidity. Embryos were transferred in

new droplets of synthetic oviduct fluid (SOF#2; 1 mM

lactate, 0.5-mM glucose) 72 h post-fertilization andnce again 120 h post-fertilization (SOF#3; 1 mM lac-

tate, 2.5-mM glucose) to prevent ammonium toxicitybecause of amino acid degradation.

2.2. Inner cell mass, trophoblast and apoptotic cellsstaining

Freshly collected expanded blastocysts were washedtwice in PBS containing 0.3% BSA. Trophectordermcells were permeabilized with 1% Triton X-100 andstained with propidium iodide for 25 s, and thenwashed twice in PBS. Blastocysts were then fixed with100% ethanol and nuclei stained with 0.25% Hoechst33,342 overnight. Fixed and stained embryos were thenwashed twice in PBS. Cells positive for DNA fragmen-tation were detected with TUNEL, using Terminal de-oxynucleotidyl transferase (TdT), which catalyzes po-lymerization of labeled nucleotides to free 3=-OH DNAends, (in situ Cell Death Detection Kit, fluorescein,Roche Diagnostics, Laval, QC, Canada). The TUNELlabeling was performed as recommended by the man-ufacturer, including negative and positive controls.Fluorescent labels incorporated in nucleotide poly-mers were detected and quantified by fluorescencemicroscopy using a Nikon Eclipse E600W with a

40� objective.

2.3. Determination of sex ratio

Embryos were lysed individually in a microtube byaddition of a 25 mM NaOH alkaline solution. Thequivalent of 10% of the embryo was used for sexiagnosis by real-time PCR using a LightCycler 1.5pparatus. The sexing assay is a proprietary methodeveloped at L’Alliance Boviteq, Inc, (St-Hyacinthe,C, Canada). Briefly, a gene fragment from the X

hromosome was amplified as internal control for theresence of DNA, and a Y chromosome fragment fordentification of male embryos. Different fluorescentrobes specific to each PCR fragment coupled to dif-erent dyes (LC640 and LC705) were used to detect theresence of each amplified fragment.

.4. Microarray description and experimental design

For high throughput RNA abundance measure-ents, the EmbryoGENE microarray platform was uti-

ized. The microarray printed by Agilent using theurePrint technology contains 44 k 60-mers oligo-robes [28]. This microarray contains 42,242 featuresargeting all known reference genes (21,139), in addi-ion to 9,322 novel transcribed and still uncharacterizedegions. The microarray probes were specifically de-igned to account for representation biases arisingrom sample amplification and also included featuresesigned to detect transcript isoforms, such as 3,677plice variants of the insertion-deletion (indel) type and,353 potential alternate polyadenylation sites that im-act the length of the untranslated region (3=UTR) ofranscripts. The experimental design involved two-olor hybridizations with full dye-swap technical rep-ication. For all biological contrasts, a simple directomparison was performed with four independent bio-ogical replicates (each composed of a pool of 10 blas-ocysts) per treatment.

.5. Sample extraction and amplification

Sample extractions were performed using theicoPure RNA kit (Molecular Devices, Downingtown,A, USA) following the manufacturer’s recommenda-

ions. Potential genomic DNA contamination was pre-ented by an on column DNase 1 treatment (Qiagen,ississauga, ON, Canada). Purified RNA was recov-

red in 11 �L. An aliquot of every sample was used todetermine RNA integrity by microfluidic profiling us-ing a 2100 BioAnalyzer (Agilent, Santa Clara, CA,USA). Only samples with an RNA Integrity Numberrate of 7.5 and more were further processed through

global amplification using the RiboAmpHSPlus kit

le1

Rpr

imer

sequ

ence

san

dre

actio

nco

nditi

ons.

nena

me

Gen

esy

mbo

lPr

imer

sequ

ence

s5=

–3=

Am

plic

onsi

ze(b

p)A

nnea

ling/

acqu

isiti

onte

mp

(°C

)A

cces

sion

num

ber

A(c

ytos

ine-

5-)–

met

hyltr

ansf

eras

e3

alph

aD

NM

T3A

F.A

AA

AG

GA

GA

CT

CC

CT

CA

GC

AA

GR

.A

AA

GC

GA

CT

TA

GA

CA

GG

AG

AG

C19

557

/82

XM

_002

6914

63

ulin

-lik

egr

owth

fact

orbi

ndin

gPr

otei

n7

IGFB

P7F.

AA

CA

GA

AC

TC

TT

GC

CT

GG

TG

AC

R.

CT

GT

CC

TT

GG

GA

AT

TG

GA

TG

154

55/8

3N

M_0

0110

2300

tallo

thio

nein

1EM

T1A

F.G

GA

CC

CG

AA

CT

GC

TC

CT

GR

.C

CC

CT

TT

GC

AG

AC

AC

AG

C15

557

/88

NM

_001

0404

92

P-gl

ucos

ece

ram

ide

gluc

osyl

tran

sfer

ase

UG

CG

F.G

AT

CA

GG

TG

GA

CC

AA

AT

TG

CR

.G

TC

AT

GG

AT

TC

AC

GG

AT

GA

AC

261

54/7

9N

M_0

0107

6850

rosi

ne3-

mon

ooxy

gena

se/tr

ypto

phan

5-on

ooxy

gena

seac

tivat

ion

prot

ein,

zeta

olyp

eptid

e

YW

HA

ZF.

GC

AG

AA

GA

AC

GG

CA

TC

AA

GG

TG

AA

R.

CT

TC

AG

CT

TC

GT

CT

CC

TT

GG

286

55/7

9N

M_1

7481

4

nser

ved

helix

-loo

p-he

lixub

iqui

tous

kina

seC

HU

KF.

TG

AT

GG

AA

TC

TC

TG

GA

AC

AG

CG

R.

TG

CT

TA

CA

GC

CC

AA

CA

AC

TT

GC

180

57/8

1N

M_1

7402

1

1770 D. Plourde et al. / Theriogenology 77 (2012) 1767–1778

(Molecular Devices) following the manufacturer’srecommendations. Following two rounds of amplifi-cation, samples were purified with the provided col-umns and the concentration of purified antisense-RNA (aRNA) was determined using a NanoDropND-1000 Spectrophotometer (NanoDrop, Wilming-ton, DE, USA).

2.6. Antisense RNA labelling and microarrayshybridization

The aRNA samples were labeled using the UniversalLinkage System (ULS) kit designed for Agilent oli-goarrays (Kreatech Biotechnology, Amsterdam, theNetherlands). Samples were split and fractions were,respectively labeled with Cy3 or Cy5. An amount of825 ng of labeled aRNA was used to hybridize the4x44k microarray. Hybridization occurred in a hybrid-ization oven for 17 h at 65 °C. Microarray slides werewashed successively in two wash buffers (Agilent GeneExpression Wash buffer Kit, Agilent) to eliminate en-vironmental contaminants, before being washed in ace-tonitrile and in a stabilization/drying solution to pre-serve fluorescence and to help protect from anyresidues that remained on the slides. Microarrays werethen scanned with the PowerScanner (Tecan, Männe-dorf, Switzerland) and analyzed with ArrayPro soft-ware (MediaCybernetics, Bethesda, MD, USA). Themicroarray dataset is publically available in the GEOOmnibus Repository (GEO accession number:GSE34958).

2.7. Quantitative rt-pcr

Three additional pools of five blastocysts per samplewere collected. Total RNA was extracted using Pico-pure columns (Molecular Devices). Reverse transcrip-tion was conducted using the qScript cDNA SuperMix(Quanta Biosciences, Gaithersburg, MD, USA) with anoligodT to prime the reaction according to the manu-facturer’s recommendations.

Primers for each candidate were designed using thePrimer3 Web interface (http://frodo.wi.mit.edu/cgi-bin/primer3/primer3_www.cgi). Primer sequence, anneal-ing and fluorescence acquisition temperatures, am-plicon size and GenBank accession numbers areshown in Table 1. The reaction mixture was com-posed of the LightCycler FastStart DNA MasterSYBR Green I kit components (Roche Diagnostics)and real-time measurements were conducted in aLightCycler 2.0 apparatus (Roche Diagnostics). Ourreal-time PCR amplification procedure has been de-

scribed in detail previously [22]. The nature of the T

abPC G

e

DN

Ins

Me

UD

Ty m p

Co

m

BtsLind

dp

3

3p

aarD(cda

TE

1771D. Plourde et al. / Theriogenology 77 (2012) 1767–1778

amplified products was confirmed using the meltingcurve profile and DNA sequencing.

2.8. Data processing and statistical analyses

P values for embryo quality parameters were calcu-lated using an unpaired two-tailed Student’s t test with

eans associated with embryos production runs.Microarray data were preprocessed as follows: 1)

ackground correction was performed by simple sub-raction of locally measured background noise for everypot; 2) within-array normalization was performed withoess; 3) Quantile was applied for interarray normal-

zation. The normalized dataset was subjected to sig-ificance testing using the LIMMA package of Biocon-uctor [29,30] with the TREAT algorithm for a better

control of false discoveries [31]. Differences in RNAabundance were considered significant at the 95% con-fidence level (P � 0.05) and a fold change of 2 (log2

Fig. 1. Picture of an expanded bovine blastocyst typically collected ateach production site. The expanded blastocyst was collected on Day7 post fertilization. Scale bar � 100 �m.

able 2fficacy of embryo production systems and gender proportions betw

Site Efficacy

No. Day 4 cleavagerate (%)

Day 48-cells rat(%)

A 2309 73 � 7 43 � 9B 1894 NA NA

ab within a column, rates without a common superscript differed (P � 0.0

fold change of 1). The Between Group Analysis wasperformed using the multivariate analysis of microarray(MADE4) package of Bioconductor [32,33]. Pathwayanalysis was conducted using Ingenuity Pathway Anal-ysis Software Version 8.6 (Ingenuity Systems, Inc.,Redwood City, CA, USA).

For quantitative RT-PCR data, normalization wasperformed using the spiked-in exogenous polyade-nylated transcript coding for the green fluorescentprotein (GFP) as described previously [34]. Testingof statistical significance of observed differences wasachieved using Prism software version 5.0 (GraphPadSoftware, La Jolla, CA, USA). One-way ANOVA wasone using Kruskal-Wallis and Dunn’s multiple com-arison tests.

. Results

.1. Performance of in vitro system betweenroduction sites

The in vitro production protocol selected for thisstudy uses the tissue culture media (TCM) to supportoocyte maturation and the modified SOF media to sus-tain embryonic growth. Common sources of serum andBSA were used to minimize the impact of biologicalcontaminants. Straws of frozen semen were also sharedbetween sites to eliminate the bull effect. To avoid theimpact of subjectivity from embryo staging, only ex-panded stage blastocysts present on Day 7 post-fertil-ization were collected for downstream analyses. Thetypical blastocysts morphology included a clear inner-cell mass, diameter �120 �m, a blastocelic cavityccounting for at least 50% of total embryonic volume,nd thinned zona pellucidae (Fig. 1). The blastocystate was determined by the percentage of blastocysts onay 7 post fertilization per oocyte put in maturation

Table 2). Overall, the blastocyst rate was higher at theommercial service provider (Site B) than at the aca-emia embryo production facilities (Site A; 30% � 8nd 20% � 3, respectively; Table 2).

two production sites.

Gender proportion

Day 7 blastocystrate (%)

No. Male(%)

Female(%)

20 � 3a 46 24 [52] 22 [48]30 � 8b 477 245 [52] 222 [48]

een the

e

5).

1772 D. Plourde et al. / Theriogenology 77 (2012) 1767–1778

3.2. Sex ratio

Sex ratio was determined as a potential indication ofdivergence between locations. However, productionsite showed no significant impact on sex ratios, as eachwas very close to an expected 50/50 ratio (Table 2).

3.3. Total cell count and cellular allocation

Total number of cells and cellular allocation to ei-ther the inner cell mass or to the trophoblast weredetermined by differential staining (Fig. 2). Both totalcell counts and trophoblast cell counts had relativelyhigh standard deviation in both laboratories (Table 3),with P values of 0.5467 and 0.8104, respectively. Innercell mass counts, ICM/TE ratio and ICM % showedlittle variation (P values of 0.2740, 0.3496, and 0.3496).

3.4. TUNEL positive cells

TUNEL staining was performed to account for thenumber of apoptotic cells (Fig. 2). Embryo obtained atthe academia site (Site A) had more apoptosis thanthose produced at the service provider (Site B), with Pvalues for apoptotic cells count and percentage of apop-totic cells of 0.0151 and 0.0023, respectively (Table 3).

3.5. Gene expression profiles

Overall, the correlations between microarrays werehigh (Fig. 3A). In fact, mean correlations between bi-ological replicates (94.5% � 1.5) were similar to theones observed between treatments (site of production)(93.3% � 0.8). The Between Group Analysis (Fig. 3B)also indicates that replicates for both production sitesare clustered together. The variance between replicatesillustrated by the vectors is as important as the distancebetween clusters. The distance and variance werehigher for samples collected in academia settings (SiteA) than those collected in commercial settings (Site B;Fig. 3B).

Based on expression profiles, out of the total 16,121protein encoding genes known to be involved in earlybovine development [28], only 42 were differentiallyabundant between the two production sites (Fig. 4 andSupplemental Table 1; the latter is present in the on-lineversion only). In addition to known reference genes, thecontrast highlighted an impact on RNA splicing events.Indeed, the production environment influenced the lengthof the 3=UTR for eight genes (LOC789567, PLOD2,TPM2, SLC25A21, CYP51A1, SLC1A4, TAOK3, IFIT5,SLC10A1), indicative of the use of an alternate polyade-nylation signal sequence (Fig. 4). Out of the 3,677 features

aimed at detecting potential indel type splicing events,

none was influenced by production sites, indicating thatno such exonic rearrangements occurred. The yellow

Fig. 2. Differential cellular staining of Day 7 expanded bovine blas-tocysts. (A) Trophoblast cells in red (PI) and inner cell mass in blue(Hoechst 33,342). (B) TUNEL positive cells in green. (C) Positivecontrol for TUNEL assay.

spots shown in Fig. 4 were a tandem of microarray

ef

3

ditlftascswf

4

cmpsmco

nwtpTwparchcgp

fctvcBola[s[ai[[pmrmtdimte

� 0.0

1773D. Plourde et al. / Theriogenology 77 (2012) 1767–1778

features that represented constitutive probes for allknown variants, as well as one of the two associatedfeatures designed to detect indel isoforms. The fluores-cence intensity of the other microarray feature aimed atdetecting alternate transcripts for this target was in thevicinity of the background, indicating that only one thesuspected alternate transcripts was differentially ex-pressed. Based on these results, we inferred that pro-duction site did not affect differential splicing, whichwould require both indel variants to be differentiallyexpressed. In addition to splice variant detection, themicroarray array also contained a large contingent offeatures detecting novel transcribed regions (NTR)(n � 9,322). In total, nine of these NTR were differ-ntially abundant between blastocysts produced at dif-erent sites (Fig. 4).

.6. Gene list interpretation

Based on pathway analysis, most genes showingifferential abundance between production sites weremplicated in different apoptosis processes, supportinghe TUNEL assay results. Four candidates were se-ected for validation by qRT-PCR based on their dif-erential expression detected by microarray, in additiono their known association to apoptosis. These RNAbundance measurements, performed on additionalamples, validated three out of the four candidates andonfirmed that the apoptosis pathway was indeed moretimulated in embryos produced at Site A (academia),hereas non-apoptosis-related candidates were unaf-

ected (Fig. 5).

. Discussion

Reports stating the significant influence of in vitroonditions on embryonic gene expression, either deter-ined through the candidate gene or microarray ap-

roaches, are abundant in the literature. Most of thesetudies have been conducted with the underlying com-on objective of better understanding the impact of

ulture conditions or procedures applied in the context

Table 3Embryo quality parameters for Day 7 expanded bovine blastocysts p

Site No. Total cellcount

Inner cell mass(ICM)

Trophoblast cell(TE)

A 44 111 � 23 30 � 9 81 � 17B 37 115 � 25 33 � 10 82 � 19

ac within a column, means without a common superscript differedbd within a column, rates without a common superscript differed (P

f assisted procreation. So far, this wealth of data has c

ot provided clear insights into mechanisms throughhich blastocysts react to their environment, nor did

hey enable us to distinguish between a normal com-ensative response and an aberrant ill-fated reaction.his apparent lack of true progress may arise from theide discrepancies between reports. We previously hy-othesized that variation in sample handling proceduresre the cause of these differences, to the extent ofendering most datasets incompatible. We have re-ently shown, by benchmarking the common sampleandling procedures leading to the generation of mi-roarray probes, that some specific steps, such as thelobal amplification process, can have a profound im-act on end point results [22].

In all logic, variance between reports can also ariserom embryo production procedures themselves. Thisonsideration is of prime importance for knowledgeransfer between research teams and commercial ser-ice providers. It is well known that culture mediaomponents extracted from biological sources, such asSA, serum or hormones represent an important sourcef variance because of contaminants that remain fol-owing purification [23–25]. The source of gametes islso known to influence developmental competence8,35–38]. The variance introduced by the source ofemen is most often referred to as the “bull-effect”39,40]. With these considerations came concernsbout the impact of the overall laboratory setup,ncluding the brand of petri dishes [41], incubators42], the purity level of gas, type of overlaying oil43], etc., along with the potential impact of theersonnel performing the work. In other words, howuch of our observations are associated with labo-

atory setup rather than with the experimental treat-ent itself? This question is important to determine

he level of reproducibility of studies performed inifferent laboratories and how it could influence thenterpretation of dataset. It is also pertinent to deter-ine whether knowledge transfer is possible once

hese studies have identified conditions to improvembryonic production. The present study was thus

d at two production sites.

M/TE ratio ICM % TUNEL positivecell count

TUNEL positivecells (%)

:2.8 � 0.6 27 � 5 17 � 8a 16 � 9b

:2.6 � 0.8 29 � 5 11 � 7c 9 � 6d

05).1).

roduce

s IC

11

(P � 0.

onducted to produce bovine blastocysts at two dis-

1774 D. Plourde et al. / Theriogenology 77 (2012) 1767–1778

Fig. 3. Relationship between treatments (production sites). (A) Distribution of net signal intensities between production sites. Correlationvalue is indicated in the upper left corner. (B) BGA showing the relationship of transcriptome between expanded blastocysts produced indifferent laboratories. The estimation of biological distance is shown by the distance between each cluster of replicates. Replication

associated variance is related to the length of the vector.

1775D. Plourde et al. / Theriogenology 77 (2012) 1767–1778

tinct sites using a common in vitro production pro-tocol while sharing all other conditions, such as thesource of biological extracts, BSA, serum and semen.For each type of analysis, all samples were processedby a single person and all samples were processedusing the same technological platform.

Another consideration that needed to be addressedpertained to the stage at which the blastocysts would becollected. By definition, a blastocyst is the embryonicstage characterized by the presence of a cavity. It is wellknown that blastocysts represent a very heterogeneouscohort of embryos. Practitioners routinely face this issuewhen selecting embryos to be transferred into recipients.As a tool, embryonic quality classification based on mor-phologic characteristics has been developed [44–46]. Inthe current study, the selection of a specific class of blas-tocysts was justified by the need to limit variance to theone introduced by the laboratory setup. As such, earlyexpanding blastocysts were used to negate the effect ofsubjective classification that could have arisen betweensites. It is noteworthy to mention that some experimentaldesigns involve a treatment that will influence blastocystmorphology and thus the selection of a specific morphol-ogy may not apply to every study.

Within this very controlled context, blastocysts pro-duced at both sites were very similar, both morpholog-ically and at the gene expression level. Although blas-tocyst yield was higher at one site, the numbers of cellsper embryo were not different between sites nor werecell allocation and sex ratios. Conversely, the numberof TUNEL positive cells and blastocyst rates were

Fig. 4. Contrasting the two embryo production sites at the geneexpression level. Volcano plot showing the distribution of differen-tially expressed microarray features. The microarray is composed offeatures designed to target different classes of transcripts or differentsplicing events. Protein coding and known pseudogenes (cyan); Al-ternate polyadenylation sites (blue), Exon skipping type of splicevariants (yellow) and Novel transcribed regions (green).

significantly different between sites. Blastocysts pro-

duced at the commercial service provider contained lessTUNEL positive cells than those produced in the aca-demia setup. These concerns were confirmed at thegene expression level (apoptosis related genes weredifferentially expressed). The selected candidates areknown to be involved with the apoptosis processes, astheir deregulation can trigger the apoptotic cascade andlead to cellular demise. Indeed, metallothioneins (namelyMT1A) are known to control intracellular zinc ion levels,of which an imbalance can lead to apoptosis [47]. More-over, the antiapoptotic role of MT1A may be of primeimportance for early development as it protects stemcells from losing their pluripotency [48]. The involve-ment of IGFBP7 in apoptosis seems to be regulated bya complex equilibrium, as both the addition of therecombinant protein and gene suppression by promoterhypermethylation inhibit cell proliferation, induce cel-lular senescence and enhance apoptosis [49,50]. In fact,gene inactivation of DNA methyltransferases, includ-ing DMNT3a, perturbs the establishment of epigeneticmarks, which can drive cell death [51,52]. Overall,three of four potential apoptosis triggering genes wereconfirmed to be differentially expressed between pro-duction sites, thus corroborating the phenotype ob-served at the cellular level.

Large intergenic non-coding RNAs represent an-other class of epigenetic factors that are involved ingene regulation through chromatin remodeling [53–55].Although their physiological involvement in embryo-genesis has not yet been elucidated, their abundantpresence has been reported [56,57]. The production siteonly had a marginal impact on the expression of theseuncharacterized RNAs (NTRs); only 9 of 9,322 weredifferentially expressed.

5. Conclusions

We inferred that subtle differences within the embryoproduction workflow can significantly impact embryonicyield, causing shifts in the blastocyst populations in re-gards to their quality grades. The level of traffic in theincubators affecting temperature and gas level stabilityand personnel skill level, which affects the time thatoocytes/embryos spend outside the incubator are amongstthe suspected causes. Overall, although differences weredetected, most features (cellular and molecular) were notsignificantly influenced. As such, results can be repeatedacross IVF laboratories when care is taken to comparesamples of defined morphology. These results supported

the potential for technology transfer between groups.

c

1776 D. Plourde et al. / Theriogenology 77 (2012) 1767–1778

Appendix A. Supplementary data

Supplementary data associated with this articlean be found, in the online version, at 10.1016/j.

Fig. 5. Validation of microarray data by quantitative RT-PCR. The mRNbetween production sites by microarray analysis were measured by quanmethyltransferase three alpha; IGFBP7, Insulin-like growth factor bindityrosine 3 - monooxygenase/tryptophan 5-monooxygenase activation pra,bBars without a common letter differed (P � 0.05).

theriogenology.2011.12.021.

Acknowledgments

The authors thank Dr. Julie Nieminen (Université La-val, Canada) for critical review of the manuscript and

dance levels of four gene candidates found to be differentially expressedRT-PCR. MT1A, Metallothionein 1E; DNMT3A, DNA (cytosine-5-)—ein 7; UGCG, UDP - glucose ceramide glucosyltransferase; YWHAZ,eta polypeptide; CHUK, Conserved helix-loop-helix ubiquitous kinase.

A abuntitativeng Prototein, z

language correction and also Dominic Gagné for his sup-

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

1777D. Plourde et al. / Theriogenology 77 (2012) 1767–1778

port for microarray hybridization and quantitative RT-PCR measurements. This work received funding from theEmbryoGENE Network, which is supported by the Nat-ural Sciences and Engineering Research Council of Can-ada (grant Number NETGP340825-06). Additional sup-port was provided by the Ministère du développementéconomique, innovation et exportation of the province ofQuébec from a grant obtained within the Programme desoutien à la recherche (grant number PSR-SIIRI-238).

References

[1] Bousquet D, Burnside EB, Van Doormaal BJ. Biotechnologiesof reproduction applied to dairy cattle production: Embryotransfer and IVF. Can J Anim Sci 1998;83:403–7.

[2] Bousquet D, Twagiramungu H, Morin N, Brisson C, CarboneauG, Durocher J. In vitro embryo production in the cow: Aneffective alternative to the conventional embryo production ap-proach. Theriogenology 1999;51:59–70.

[3] Durocher J, Morin N, Blondin P. Effect of hormonal stimulationon bovine follicular response and oocyte developmental com-petence in a commercial operation. Theriogenology 2006;65:102–15.

[4] Blondin P, Coenen K, Guilbault LA, Sirard MA. In vitro pro-duction of bovine embryos: Developmental competence is ac-quired before maturation. Theriogenology 1997;47:1061–75.

[5] Blondin P, Bousquet D, Twagiramungu H, Barnes F, SirardMA. Manipulation of follicular development to produce devel-opmentally competent bovine oocytes. Biol Reprod 2002;66:38–43.

[6] Bermejo-Alvarez P, Lonergan P, Rizos D, Gutierrez-Adan A.Low oxygen tension during IVM improves bovine oocyte com-petence and enhances anaerobic glycolysis. Reprod BiomedOnline 2010;20:341–9.

[7] Lonergan P, Fair T. In vitro-produced bovine embryos: Dealingwith the warts. Theriogenology 2008;69:17–22.

[8] Sirard MA, Coenen K. In vitro maturation and embryo produc-tion in cattle. Methods Mol Biol 2006;348:35–42.

[9] Rizos D, Gutiérrez-Adán A, Pérez-Garnelo S, De La Fuente J,Boland MP, Lonergan P. Bovine embryo culture in the presenceor absence of serum: Implications for blastocyst development,cryotolerance, and messenger RNA expression. Biol Reprod2003;68:236–43.

10] Gutiérrez-Adán A, Lonergan P, Rizos D, Ward FA, Boland MP,Pintado B, et al. Effect of the in vitro culture system on thekinetics of blastocyst development and sex ratio of bovineembryos. Theriogenology 2001;55:1117–26.

11] Kimura K, Iwata H, Thompson JG. The effect of glucosamineconcentration on the development and sex ratio of bovine em-bryos. Anim Reprod Sci 2008;103:228–38.

12] Kimura K, Spate LD, Green MP, Roberts RM. Effects of D-glu-cose concentration, D-fructose, and inhibitors of enzymes of thepentose phosphate pathway on the development and sex ratio ofbovine blastocysts. Mol Reprod Dev 2005;72:201–7.

13] Augustin R, Pocar P, Wrenzycki C, Niemann H, Fischer B.Mitogenic and anti-apoptotic activity of insulin on bovine em-bryos produced in vitro. Reproduction 2003;126:91–9.

14] McEvoy TG, Sinclair KD, Young LE, Wilmut I, Robinson JJ.

Large offspring syndrome and other consequences of ruminant

embryo culture in vitro: Relevance to blastocyst culture inhuman ART. Hum Fertil 2000;3:238–46.

15] Wrenzycki C, Herrmann D, Carnwath JW, Niemann H. Expres-sion of RNA from developmentally important genes in preim-plantation bovine embryos produced in TCM supplementedwith BSA. J Reprod Fertil 1998;112:387–98.

16] Wrenzycki C, Herrmann D, Keskintepe L, Martins A Jr., Siri-sathien S, Brackett B, et al. Effects of culture system and proteinsupplementation on mRNA expression in pre-implantation bo-vine embryos. Hum Reprod 2001;16:893–901.

17] Niemann H, Wrenzycki C. Alterations of expression of devel-opmentally important genes in preimplantation bovine embryosby in vitro culture conditions: Implications for subsequent de-velopment. Theriogenology 2000;53:21–34.

18] Lonergan P, Rizos D, Gutiérrez-Adán A, Fair T, Boland MP.Effect of culture environment on embryo quality and geneexpression - Experience from animal studies. Reprod BiomedOnline 2003;7:657–63.

19] Knijn HM, Wrenzycki C, Hendriksen PJ, Vos PL, Herrmann D,van der Weijden GC, et al. Effects of oocyte maturation regimenon the relative abundance of gene transcripts in bovine blasto-cysts derived in vitro or in vivo. Reproduction 2002;124:365–75.

20] Knijn HM, Wrenzycki C, Hendriksen PJ, Vos PL, ZeinstraEC, van der Weijden GC, et al. In vitro and in vivo cultureeffects on mRNA expression of genes involved in metabo-lism and apoptosis in bovine embryos. Reprod Fertil Dev2005;17:775– 84.

21] McHughes CE, Springer GK, Spate LD, Li R, Woods R, GreenMP, et al. Identification and quantification of differentially rep-resented transcripts in in vitro and in vivo derived preimplan-tation bovine embryos. Mol Reprod Dev 2009;76:48–60.

22] Gilbert I, Scantland S, Sylvestre EL, Dufort I, Sirard MA,Robert C. Providing a stable methodological basis for compar-ing transcript abundance of developing embryos using microar-rays. Mol Hum Reprod;16:601–16.

23] Batt PA, Miller BG. Development of sheep embryos in vitro ina medium supplemented with different batches of serum albu-min. Aust J Biol Sci 1988;41:371–6.

24] Kane MT. Variability in different lots of commercial bovineserum albumin affects cell multiplication and hatching of rabbitblastocysts in culture. J Reprod Fertil 1983;69:555–8.

25] Rorie RW, Miller GF, Nasti KB, McNew RW. In vitro devel-opment of bovine embryos as affected by different lots ofbovine serum albumin and citrate. Theriogenology 1994;42:397–403.

26] Holm P, Booth PJ, Schmidt MH, Greve T, Callesen H. Highbovine blastocyst development in a static in vitro productionsystem using SOFaa medium supplemented with sodium citrateand myo-inositol with or without serum-proteins. Theriogenol-ogy 1999;52:683–700.

27] Vigneault C, Gravel C, Vallée M, McGraw S, Sirard MA.Unveiling the bovine embryo transcriptome during the mater-nal-to-embryonic transition. Reproduction 2009;137:245–57.

28] Robert C, Nieminen J, Dufort I, Gagne D, Grant JR, Cagnone G,et al. Combining resources to obtain a comprehensive survey ofthe bovine embryo transcriptome through deep sequencing andmicroarrays. Mol Reprod Dev 2011;78:651–64.

29] Smyth GK. Linear models and empirical bayes methods forassessing differential expression in microarray experiments.

Stat Appl Genet Mol Biol 2004;3:Article3.

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

1778 D. Plourde et al. / Theriogenology 77 (2012) 1767–1778

[30] Smyth GK, Michaud J, Scott HS. Use of within-array replicatespots for assessing differential expression in microarray exper-iments. Bioinformatics 2005;21:2067–75.

[31] McCarthy DJ, Smyth GK. Testing significance relative to afold-change threshold is a TREAT. Bioinformatics 2009;25:765–71.

[32] Culhane AC, Perrière G, Considine EC, Cotter TG, HigginsDG. Between-group analysis of microarray data. Bioinformatics2002;18:1600–8.

[33] Culhane AC, Perriere G, Higgins DG. Cross-platform compar-ison and visualisation of gene expression data using co-inertiaanalysis. BMC Bioinform 2003;4:59.

[34] Vigneault C, McGraw S, Massicotte L, Sirard MA. Transcrip-tion factor expression patterns in bovine in vitro-derived em-bryos prior to maternal-zygotic transition. Biol Reprod 2004;70:1701–9.

[35] Palma GA, Sinowatz F. Male and female effects on the in vitroproduction of bovine embryos. Anat Histol Embryol 2004;33:257–62.

[36] Tamassia M, Heyman Y, Lavergne Y, Richard C, Gelin V,Renard JP, et al. Evidence of oocyte donor cow effect overoocyte production and embryo development in vitro. Reproduc-tion 2003;126:629–37.

[37] Lonergan P, Rizos D, Gutierrez-Adan A, Fair T, Boland MP.Oocyte and embryo quality: effect of origin, culture conditionsand gene expression patterns. Reprod Domest Anim 2003;38:259–67.

[38] Barnes FL, Sirard MA. Oocyte maturation. Semin Reprod Med2000;18:123–31.

[39] Machado GM, Carvalho JO, Filho ES, Caixeta ES, Franco MM,Rumpf R, et al. Effect of Percoll volume, duration and force ofcentrifugation, on in vitro production and sex ratio of bovineembryos. Theriogenology 2009;71:1289–97.

[40] Ward F, Rizos D, Corridan D, Quinn K, Boland M, Lonergan P.Paternal influence on the time of first embryonic cleavage postinsemination and the implications for subsequent bovine em-bryo development in vitro and fertility in vivo. Mol Reprod Dev2001;60:47–55.

[41] Schiewe MC, Schmidt PM, Bush M, Wildt DE. Toxicity po-tential of absorbed-retained ethylene oxide residues in culturedishes on embryo development in vitro. J Anim Sci 1985;60:1610–8.

[42] Avery B, Greve T. Impact of incubator type on the yield of in vitroproduced bovine blastocysts. Acta Vet Scand 1992;33:341–8.

43] Van Soom A, Mahmoudzadeh AR, Christophe A, Ysebaert MT,de Kruif A. Silicone oil used in microdrop culture can affectbovine embryonic development and freezability. Reprod Do-mest Anim 2001;36:169–76.

44] van Soom A, Ysebaert MT, de Kruif A. Relationship between

timing of development, morula morphology, and cell allocation

to inner cell mass and trophectoderm in in vitro-producedbovine embryos. Mol Reprod Dev 1997;47:47–56.

45] Farin PW, Britt JH, Shaw DW, Slenning BD. Agreement amongevaluators of bovine embryos produced in vivo or in vitro.Theriogenology 1995;44:339–49.

46] Donaldson LE. Matching of embryo stages and grades withrecipient oestrous synchrony in bovine embryo transfer. VetRec 1985;117:489–91.

47] Dutsch-Wicherek M, Sikora J, Tomaszewska R. The possiblebiological role of metallothionein in apoptosis. Front Biosci2008;13:4029–38.

48] Duval D, Trouillas M, Thibault C, Dembelé D, Diemunsch F,Reinhardt B, et al. Apoptosis and differentiation commitment:Novel insights revealed by gene profiling studies in mouseembryonic stem cells. Cell Death Differ 2006;13:564–75.

49] Hinoue T, Weisenberger DJ, Pan F, Campan M, Kim M, YoungJ, et al. Analysis of the association between CIMP and BRAF incolorectal cancer by DNA methylation profiling. PLoS ONE2009;4:e8357.

50] Nousbeck J, Sarig O, Avidan N, Indelman M, Bergman R,Ramon M, et al. Insulin-like growth factor-binding protein 7regulates keratinocyte proliferation, differentiation and apopto-sis. J Invest Dermatol 2010;130:378–87.

51] Sakaue M, Ohta H, Kumaki Y, Oda M, Sakaide Y, Matsuoka C,et al. DNA methylation is dispensable for the growth andsurvival of the extraembryonic lineages. Curr Biol 2010;20:1452–7.

52] Stancheva I, Hensey C, Meehan RR. Loss of the maintenancemethyltransferase, xDnmt1, induces apoptosis in Xenopus em-bryos. EMBO J 2001;20:1963–73.

53] Marques AC, Ponting CP. Catalogues of mammalian long non-coding RNAs: Modest conservation and incompleteness. Ge-nome Biol 2009;10:R124.

54] Wang KC, Yang YW, Liu B, Sanyal A, Corces-Zimmerman R,Chen Y, et al. A long noncoding RNA maintains active chro-matin to coordinate homeotic gene expression. Nature 2011;472:120–4.

55] Gupta RA, Shah N, Wang KC, Kim J, Horlings HM, Wong DJ,et al. Long non-coding RNA HOTAIR reprograms chromatinstate to promote cancer metastasis. Nature 2010;464:1071–6.

56] Bui LC, Evsikov AV, Khan DR, Archilla C, Peynot N, HénautA, et al. Retrotransposon expression as a defining event ofgenome reprogramming in fertilized and cloned bovine em-bryos. Reproduction 2009;138:289–99.

57] Côté I, Vigneault C, Laflamme I, Laquerre J, Fournier E, GilbertI, et al. Comprehensive cross production system assessment ofthe impact of in vitro microenvironment on the expression ofmessengers and long non-coding RNAs in the bovine blasto-

cyst. Reproduction 2011;142:99–112.

tide repeats 5 IFIT5 515091 0.0003 3.2040

1778.e1D. Plourde et al. / Theriogenology 77 (2012) 1767–1778

Supplementary Table 1List of targets differentially expressed between production sites.

Type of feature Gene name

Reference genes Metallothionein 1EInsulin-like growth factor binding protein 7DNA (cytosine-5-)-methyltransferase 3 alphUDP-glucose ceramide glucosyltransferaseADP-ribosylation factor 5Asparagine synthetaseActivating transcription factor 4 (tax-responelement B67)AXL receptor tyrosine kinaseBiliverdin reductase B (flavin reductase (NACholine phosphotransferase 1Carnosine dipeptidase 1 (metallopeptidase MCancer/testis antigen family 147, member BDystoninGTP binding protein overexpressed in skeleGIPC PDZ domain containing family, memHeme binding protein 2Insulin induced gene 1Similar to nucleosome assembly protein 1-lSimilar to germ cell-lessLumicanMyosin, light chain 6, alkali, smooth musclNidogen 1Plasminogen activator, tissuePolymerase (RNA) I polypeptide DPhosphoserine aminotransferase 1Phosphoserine phosphatasePyrroline-5-carboxylate reductase 1ScinderinSolute carrier family 10 (sodium/bile acid cfamily), member 1Solute carrier family 40 (iron-regulated tranSecreted protein, acidic, cysteine-rich (osteoStimulated by retinoic acid gene 6 homologTransmembrane protein 20Transmembrane protein 52VimentinChemokine (C motif) ligand 2

Probes designed for3’UTR length

Similar to steroid dehydrogenase homologProcollagen-lysine, 2-oxoglutarate 5-dioxygTropomyosin 2 (beta)Solute carrier family 25 (mitochondrial oxocarrier), member 21Solute carrier family 10 (sodium/bile acid cfamily), member 1Solute carrier family 1 (glutamate/neutral amtransporter), member 4Cytochrome P450, family 51, subfamily A,TAO kinase 3Interferon-induced protein with tetratricopep

Genesymbol

Gene ID Correctedp-value

Foldchange

MT1A 404071 0.0371 2.1737IGFBP7 616368 0.0002 2.0890

a DNMT3A 359716 0.0007 0.4479UGCG 514357 0.0002 0.4417ARF5 511918 0.0097 0.4013ASNS 514209 0.0014 3.2512

sive enhancer ATF4 509107 0.0284 2.2499

AXL 516598 0.0016 2.9526DPH)) BLVRB 281650 0.0018 2.4549

CHPT1 511291 �0.0001 0.350720 family) CNDP1 614200 0.0141 0.3853

1 CT47B1 787180 �0.0001 4.044DST 535297 0.0022 2.5459

tal muscle GEM 538437 0.0003 3.2788ber 2 GIPC2 518246 0.0001 0.2047

HEBP2 509223 0.0253 0.4057INSIG1 511899 0.0160 0.4198

ike 1 LOC786285 786285 0.0270 2.6021LOC786530 788930 0.0067 4.1673LUM 280847 �0.0001 5.0930

e and non-muscle MYL6 281341 0.0280 2.2395NID1 534319 0.0101 0.3927PLAT 281407 0.0472 2.4329POLR1D 616972 0.0013 0.3910PSAT1 533044 0.0113 2.5600PSPH 533630 0.0285 2.7032PYCR1 539606 �0.0001 3.7088SCIN 281478 �0.0001 0.3717

otransporter SLC10A1 532890 �0.0001 0.2456

sporter), member 1 SLC40A1 527023 0.0257 2.2615nectin) SPARC 282077 0.0488 2.2615(mouse) STRA6 515911 0.0012 4.2298

TMEM20 617598 0.0007 0.3843TMEM52 617403 0.0170 0.3801VIM 280955 0.0009 3.2593XCL2 319096 0.0266 2.8606LOC789567 789567 0.0004 0.3914

enase 2 PLOD2 533642 0.0092 2.4269TPM2 497015 0.0047 2.5850

dicarboxylate SLC25A21 513423 0.0063 0.3674

otransporter SLC10A1 532890 0.0006 0.2721

ino acid SLC1A4 326577 �0.0001 4.9894

polypeptide 1 CYP51A1 505060 0.0353 0.4050TAOK3 534620 0.0357 0.4368