.

,'.CSIROPUBLISHING

www.publish.csiro.auljoumals/rfd Reproduction. Fertility and Development, 2006,18,781-788

Sperm DNA damage is related to field fertility of semenfrom young Norwegian Red bulls

K. E. WaterhouseA,B,G, T. HauganA,B, E. KommisrutP, A. TverdaJD, G. F/atbergE,Jv. FarstaJB, D. P.EvensonF and P.M De AngelisE

ATeamSemin, PO Box 8146 Departmental Division, N-0033 Oslo, Norway,B-rheNorwegian School of Veterinary Science, Department of Production Animal Clinical Science,

. PO Box 8146DepartmentalDivision,N-0033Oslo,Norway.COeno Breeding' and AI Association, N-2326 Hamar, Norway,

D-rheNorwegian School of Veterinary Science, Department of Basic Sciences and Aquatic Medicine,PO Box 8146DepartmentalDivision,N-0033Oslo,Norway.

EInstitute of Pathology, Rikshospitalet, N-0f)27 Oslo, Norway.FSouth Dakota State University, Brookings, SD 57007, USA.GCorresponding author. Email: karin.waterhouse@veths.no

,'1

Abstract. Flow cytometrywas utilised for the fU'Sttime to independentlymeasurefive sperm parametersofindividual spermatozoa of bull ejaculates to differentiate between outcome successes after artificial insemination(AI). These parameters included plasma membrane and acrosome integrity, mitochondrial functionality and DNAdamage measured by sperm chromatin structure assay (SCSA) and terminal deoxynucleotide transferase-mediateddUfP nick end labening (TUNEL) assays. For each parameter, results of 142 ejaculates (30 bulls) were ranked intothree groups according to their flow cytometric measures: (1) ejaculates with the 25% lOwestmeasures; (2) the 50%middle measures; and (3) the 25% highest measures. In total, 20272 first-service inseminations (18 x 106sperma-tozoa per AI dose) were performed, where fertility was defined as non-return within 60 days after first insemination.Whileplasmamembrane.andacrosomeintegrity,and mitochondrialfunctionalitywerenotsignificantlyrelatedtofertility, data from SCSA and TUNEL assays were significantly associated with fertility. Ejaculates in SCSA group 1had.higher odds of AI success (1.07,95% CI= 1.02-1.12), whereas those in group 3 had lower odds of AI success(0,94, 95% CI = 0.89-0.99), compared with the average odds of all three groups. Ejaculates in group 2 did not havesignificantly higher odds of AI success compared with the average odds, For TUNEL-positive spermatozoa, theodds of AI success was higher in group I compared with the average odds (1.10, 95% CI = 1.02-1.13), whereasodds of AI success in groups 2 and 3 were not significant compared with the average odds. In conclusion, despitethe high number of spermatozoa per AI dose from high-quality bulls, both SCSA and TUNEL assays were valuablemeasures in this study for evaluating sperm quality in relation to fertility after AI.

Extra keywords: 60 day non-return rate, sperm DNA fragmentation. NOTICE:THISMAlERlALMAYBEPROTI;{:T;:'eVCOPYRIGHnAWI'hU;= Ii.

Introduction

Since approximately 1990, numerous studies have attemptedto predict fertility of semen from breeding bulls based onsperm quality. Precise and accurate estimates of field fertilityare very important when attempting to explain differences inthe potential fertility of a particular ejaculate or bull. Success-ful fertilisation and normal embryo development is a functionof many factors, the most important including quality offemale and male gametes, physiological and genetic femalefactors and artificial insemination (AI) and herd management(Amann 1989;Amann and Hammerstedt2002). Furthermore,factors such as the number of spermatozoa per AI dose andtime of AI in relation to ovulation will affect the fertility

Q CSIRO 2006

outcome. The reliability-ofthe field fertility data will dependon factors such as the number of AIs per ejaculate and howthe outcome of a given AI is reported (Amann 1989;Amannand Hammerstedt 2002). A recent article by Amann (2005)found deficiencies in 51 out of 67 papers that included fertil-ity data as an outcome measure, due to limits in and scarcityofthe field fertility data.

In Norway, semen from all YOUDgbulls entering the AIprogeny testing system is distributed to herds over the entirecountry. Registrationof fertility data from the field arebased on a controlled. systemwhere aU inseminationdataare reported into an AI database by the insemination techni-cians and veterinarians (AI personnel). The AI personnel are

IO.1071/RD06029 1031-3613/06/070781

782 Reproduction, Fertility and Development

financially credited only when the insemination data are reg-istered into the AI database, which is controlled by the Nor-wegian Dairy ijerd Recording System (Ranberg et 01.2003).

The overall sperm quality of one ejaculate-is not deter-mined by only one sperm attribute, but depends on severalsperm characteristics. The complex role of spermatozoa infertilisation and embryo development consequently compli-cates the prediction of a successful outcome. Hence, besidesaccurate and reliable fertility data, objective and repro-ducible laboratory methods for analysing sperm attributesare of utmost importance (Amann 1989). The introduc.tion and utilisation of flow cytometric assays for spermquality analyses have greatly enhanced the objectivity andreproducibility of the andrology laboratory analyses (Gra-ham 2001), the first being reported by Evenson et al.(1980) on bull sperm chromatin. Attributes such as mem-brane integrity (Evenson et al. 1982; Ericsson et af. 1993;Anzar et al. 2002), mitochondrial function (Evenson et 01.1982; Ericsson et al. 1993) and DNA quality (Evensonet 01. 1980: Ballachey et al. 1987, 1988: Karabinus et al.1990; Januslcauskas et oJ. 2001,2003; Anzar et 01. 2002)bave been studied. Except for the sperm chromatin structureassay (SCSA), few single sperm attributes measured by flowcytometry have been useful in relating sperm quality and fieldfertility (Gillan et of. 2005). The SCSA has shown a consis-tent relationship with fertility in several species (for review,see Gillan et al. 2005; Evenson and WIXon2006a). To date,few studies have included flow cytometric analyses of sev-eral sperm attributes in the same ejaculates and furthermore,limits in and scarcity of data from field fertility records areoften minimised or ignored (Amann and Hanunerstedt 2002;Amann 2005).

The present controlled insemination trial aimed to useestablished flow cytometric assays to assess variability in cer-tain sperm attributes among ejaculates from several bulls andstudy how the sperm quality associates wi~ field fertility ina commercial AI setting. We tested whether the chance 0tAI success differed between ejaculates with variable spermquality.

Materials and method!!

Semen processing

Semen from 30 young Norwegian Red (NRF) bulls entering the AIprogeny testing system (Geno Breeding and AI Association, Hamar,Norway) during SeptembeJ<..Novcmber of 2000 and 2001 in Norwaywas included. The bulls were kept at Geno Store ReeAI Centre (Hamar,Norway) under uniform feeding and housing conditions throughout thesemen production period. At the start of semen collection, the bu1ls were14-17 months old, except for two bulls that were 19 months o1d.Approx-imately 2200 AI doses were routinely used in the progeny testing of eachbull and semen was collected once a week for a period of3-7 wce1cstoachieve this number for the included bulls. On the day of collection, twoejaculates were routinely collected within 15 miD, and then the ejacu-lates were pooled and given the same ftceze code identification (bull mand ftcezing date, hereafter referred to as .fi'ceze code'). The semen wasdiluted to a final concenttation of 82 x 106 spermatozoaper mL in a

--

1(. E. Waterhouse et at.

skim-milk-based extcnderina two-step procedurcand frozen in 0.25 mLFrench mini-straws (IMY, I:Aigle, France), with -18 x 106 SPermato-zoa per straw, in liquid nitrogen (LN2) {Kommisrud et aI. 1996). Afterthe first dilution step, morphology IIIIdprogressive motility were deter.mined microscopica1ly at the AI station as part of the routine semenassessment Only semen with ~70% progressively motile spcr1nato-

70a and ;::80"10normal morphology was further processed to fteezing.Routinely, a motility control ~ done at the AI station after freczing-thawing (assessed -IOmin after thawing) and only freeze codes with~SO"I0progressively motile spcnnato7Oa were included in the progenytesting. In total, 149 freeze codes were collected from the included bulls.

One frecze code was excluded before freezing owing to low morphologyscore. Six fi'eeze codes, from five different bulls, were excluded after

fteezing-tbawing owing to low motility score (30-45%). Mean motilityof the 142 included freeze codes was 60% (range S0-70%). The meannumber of included frcezecodes fiom each bull was five, with a range ofthree to seven (the nwnber depending on the total nwnber oCAI dosesproduced per fteeze code). Frozen straws from a1l142 freeze codesincluded in the progeny testing were systematically collected and storedat the AI station Cor later sperm quality assessment by flow cytOmetry.

Field .fertility recordings

The fertility expressed by non-return (NR) 60 days after first inSCltl-ination was used as a measure of fertility on freeze code level.

Approximately 2200 AI doses were produced and distributed for all testbulls. On average, -..4()0,{,oC the cows and heifers are bred with youngtest bulls each year and an average of 1500 first, second and later insem-inations are reponed to the- Norwegian Dairy Herd Recording Systemper test bull (Ranbcrg et at. 2003). The freeze codes included in thepresent study were routinely mixed within bull (-2200 AI doses) and

batches of AI doses were riIndornly distributed to the four semen depotstations that cover all parts ofNorway. The batch size was detennined bythe number of animals covered by each depot. From each depot, batchesoC 10-20 AI doses from each bull were randomly distn'buted to dif-f=nt Al personnel, who further used these AI doses mndomly amongherds and animals. To be sure about the paternity, animals inseminatedwithin 3 days after first insemination were excluded from the pn:sentstudy. Animals older than 5th lactation were also excluded, as well asthose inseminations where infonnation aboUt parity, month of AI andAIpersonnel was not avaiJable. After exclusion, 20272 fust-servicc insem-inations were included in the dataset, with a mean of 143 Als per ftcezecode. Oftbe anill181sinseminated, 44% were heifers, 27% 1st lactation

cows and 29% multiparous cows, and the overall NR rate of the AIs withtho 142 freeze codes was 72.9% (range 63.0-83.S%).

Flow cywmetric analyses

All 142 freeze codes included in the progeny testing were subjectedto flow cytomctric analyses as described below. One straw from eachfi'ceze code was used for each of the flow cytometric assays and tworeplicate sample runs were prepared from the same straw (except for theterminal deoxynucleotidetrausfcrase-mediateddUTPnickcnd labelling(TUNEL) assay, where one replicate sample was run). The semen wasthawed at 37°C for I min. All flow cytometry-generated data were anal-ysed using EXP032 Analysis Software (Bcclanan Coulter, Fullerton,CA,USA).

Spe,.", chromatin structUre assay

The SCSA defines abnonnal chromatin structure as increased sus-

ceptibility of sperm DNA to acid-induced denaturation in situ (forreview. see Evenson el al. 2002). The method utilises the uniquemetachromatic properties ofacridinc orange (AO; Polysciences Europe,Eppclheim, Germany), which fluoresces green when intercalated intodouble-stranded DNA and red when boDnd to single-stranded DNA.

f. I. i

II

I. I

Fertility of bull semen and sperm quality

Semen was difutcd to 200 j.l.L in TNE buffer (0. IS M NaCI, 0.01 MTRI8-

HCl, I mM disodium ethyl~c acid (EDTA), pH 7.4,4°C) to -2 ~ 106 spermatozoa per mL and mixed with 4OOj.l.L aciddenaturation solution (0.08M HCI, 0.15M NaCl, 0.1% Triton X-lOO,

pH 1.2, 4°C). After 30 s, 1.2 rnL of staining solution (0.037 M citricacid, 0.126M Na2HP04, 0.0011 M disodium EDTA, O.ISM NaCI, pH6.0, 4°C). containing 6 j.l.gmL -1 AO, was added. The sample was then

subjected to flow cytometric analysis on a FACSCalibur flow cytome-ter (BD Biosciences Immunocytometry Systems, San Jose, CA, USA)equipped with a IS mW argon laser with excitation at 488mn. The sam-ple was run in setup mode until 3 min after the staining procedure badstarted, and then 5000 sperm events were collected, using a sample flowrate of 200evcntss-1 (low sampleprcssure). If the replicate sample orthe next fteeze code was not ready. the previous sample was run in setupmode to 1a:cpthe sampling line saturated withAO. A bull reference sam-ple was run for every fifth fteeze code to ensure thai the instrument and1aser remained stable throughout the experiment, with X-mean channelvalue of 125:1: 5 and Y-mean channel value of 425 :!: S. Light scatterand fluorescence data were collected in linear mode. Green fluores-

cencewas detected using a 51S-S4S nmbandpass (BP) fIIter(FLI) and

red fluorescence was detected using a 650 mn long pass filter (FL3). Acytogram of green vmus red DNA fluorescence was used to identifyspermatozoa with denatured DNA (Fig. I). A computer-defined gatewas set around the sperm sigoals with increased red DNA fluorescence

(denatured DNA) compared with the main population to determine thepercentages of spermatozoa with single-stranded (nagmented) DNA(Fig. I, region B). This percentage was calculated as a percentage ofthe total sperm population and was reported as the DNA ftagmentationindex (% DFI).

.'

TermilUll deOlC)11lucleotidetransferase-mediated dUTPniclc end labelling assay

The TUNEL assay is thought to specifically detect 3'-OH ends ofDNA strand breaks (Gavrie1i eI aI. 1992; ADzar et al. 2002). DNAfragmentation was assessed by treating permeabilised fIXed sperma-tozoa with exogenous terminal deoxynucleotidyl transferase (TdT)

. .~ .

1023

Fragmented DNA

FIg. 1. A typical cytogram of red (single-stranded DNA) v. green(double-stranded DNA) acridine orange (AO) fluorescence of thawedbull spermatozoa following treatment with acid denaturation solu-tion. Region A, spermatozoa having predominantly green AO fluores-cence. Region B, spermatozoa having increased red AO fluorescence(fragmented DNA). The percentage of spermatozoa in B relative tothe total sperm population (regions A + B) corresponds to the DNAfragmentation index (%).

- -

Reproduction, Fertility and Development 783

enzyme (TdT-kit, Roche Diagnostics, Oslo, Norway) in the pres-enceofbiotin-16-deoxyuridine triphospbates (biotin-dUTP). The 3'-OHends of DNA strand breaks serve as primers for the incorporationof biotin-dUTP, which is detected by the use of fluorescein isothio-cyanate (FITC)-conjugated streptavidin. Thawed semen was washed inphosphate-buffered saline (PBS) and fixed in 1% paraformaldehydefor IS min on ice. The spermatozoa were washed and stored in 100%

methanol at -2O"C until further processing. Approximately 3 x 106fixed spermatozoa were washed in PBS and permeabilised with 0.1 %Triton X-I 00 and 0.1 % sodium citrate in PBS for IOmin on ice. The sper-matozoa were washed again and incubated with 20 j.l.gmL-I proteinaseK (to degrade chromatin-associated proteins; Roche Diagnostics. Oslo,Norway) for 20 min at room temperature. Thereafter, spermatozoa werewashed and incubated at 37°C for 4 h (Anzar et aI. 2002) in SO~LTdT solution containing 5 units TdT, 5 j.l.L5x reaction buffer (suppliedwith TdT kit). l.SmM CoClz, O.5nM biotin-I6-dUTP, 0.1 111Mdithio-tbreitol and double-distilled water. The suspensions were agitated every30 miD. Negative staining controls received simi1ar treatment, exceptthat theTdT was omitted. Spermatozoa treated with 10j.l.gmL-t DNase(DNase I, Roche Diagnostics, Oslo, Norway) for 10min were usedas positive controls. After incubation, spermatozoa were washed and

incubated with 50 j.l.L 1150streptaVidin-FlTC in PBS containing 0.1%TritonX-loo aod3% dry milk solution for 30 mioat room temperature.The spennatozoa were washed again and stored at 4°C overnight Thenext day, 2j.1.grnL-I Hocchst 33258 (Amersham Pharmacia Biotech,Oslo. Norway) was added in order to discriminate and exclude.doublets,aggregates and debris from the analyses. using Hoecbst fluorescencepulse-width processing (Stokke et aI. 1998). Samples were ana1ysedon a FACSVantage SE (BD Bioseiences Immunocytometry Systems)and 10000 spermatozoa per sample were collected at low sample pres-sure. Hoechst 33258 was excited with a SOmW UV laser (35 1-365 DID

excita~!>n) and fluorescence was collected using the FL4 detector (402-446 am BP filter), whereas me was excited with a 350 mW laser(488 mn excitation) and flunrescence was collected usingthe FLI detec-tor (51'S-S45 DIDBP filter). A cytogram ofHoecbst 33258 fluorescence

(DNA content) versus mc fltiorescence (biotin-dUTP-positivity) wasused to determine the percentages of spermatozoa with DNA strandbreaks (TUNEL-positive spermatozoa) expressed as a percentage ofthe total sperm population (Fig. 2, region B).

Assessment of plasma membrane and acrosome integrity,and mitochondrialjimCttonality

The membrane-impermeable DNA-binding dye Yo-Pro-I (Molec-ular Probes Europe, Leiden, the Netherlands) was used to distinguishbetween plasma membrane-intact and -degenerated spermatoZOa. Acro-some integrity was assessed with the non-lipophilic peanut agglutininconjugated with FITC (PNA-FITC; Sigma Aldrich, Oslo, Norway).which only binds to glycoprotcins in the acrosome membrane ofacrosome-mlCted or -damaged spermatozoa (Arya and Vanba-Pcrttula1985). MitoTrackcr RcdCMXRos(MT-RED, Molecular Probes Europe,Leiden, the Netherlands), which accumulates in functionally polarisedmitochondria and leaks out of mitochondria with a dcpolarised poten-tial, was used to distinguish between spermatozoa with functional andnon-functional mitochondria (GadcUa and Harrison 2002), respectively,within the live acrosome-intact population. 'IWo replicate samples fromeach straw were diluted to -2 x 106 spermatozoa per mL in PBS keptat 37°C. stained with 25 OMYo-Pro-I, 0.1 JIogmL-1 PNA-FITC and25 nM MT-RED for IS min in the dark and then subjected to flow

cytometric analyses. Flow cytometric analyses were performed usinga Coulter EPICS XL flow cytometer (Beckman Coulter) equipped witha ISmW argon 1aserwith a 488nm excitation wavelength. Light seat-ter data were collected in linear mode, whereas fluorescence data were

collected in logarithmic mode. Side and forward light scatter parame-ters were used to identify sperm events and 10000 spermatozoa per

A .I.' B

j1/1

gGI

I0

0

784 Reproductioll. Fertility and Development

A.. "I. . .

Q

10° 10' . 102 103

Blotin-dUTP/streptailidln-FITC

Fig. Z. A typical cytogram of fluoteSCCin isothiocyanate (FITC)(corresponda to biotin-dUTP-positive spennatozoa) v. Hocchst 33258fluorescence (DNA content). Events inregionsA and B are spermatozoa,i.e. positively labeUed with Hocchst 33258. R.egionA. spermatozoa withnegative FlTC-f1uoresccnce intensities. Region B, spermatozoa that areTUNEL-positive, i.e. those spennatozoa with ftagmented DNA thatbave been end-labelled byTdT with biotin-dUTP/streptavidin-FlTC.

sample were collected at low sample pressure. Yo-Pro-I and PNA-mc fluorescence were detected simultaneously using a 50S-545 omBP filter (FLI HHarrison et al. 1996), and MT-RED fluorescence was

detected using a 605--635 om BP filter (FL3). Unstained samples \\Weused as negative fluorescence controls. The double negative populationin FLI represents the live acrosome.intact (LAI) spermatozoa (Fig. 3a,region A) and the positive populations rcpresentdead, acrosome-reactedor -damaged spermatozoa or both (Fig. 3a, region B). A cytogram ofMT.RED fluorescence versus Yo-Pro-l and PNA-FlTC fluorescence

was used to determine the percentages of LAI spermatozoa withfunctional mitochondria (LAI-MT-RED-positive) (Fig. 3b. region A).Populations of dead, acrosome-reacted or -damaged spennatozoa, orboth, with high MT-RED fluorescence (region B), live acrosome-intactspermatozoa with low. MT-RED fluoteSCCl1CC(region C) and dead,acrosome-reacted or-damaged spermatozoa, or both. with low MT-REDfluorescence (region D) arc also shown in the cytogram.

Statistical analyses

The mean value of the two replicate sample runs per fteeze code (ciceptfor the TUNEL assay. where one replicate sample was nm), for eachof the flow cytometric assays, was used in the analysis of the data.All ~ parameters arc presented as mean values of the 142 ftcezecodes with standard deviations in brackets. Statistical analyses wereconducted using SAS version 8.01 for Windows (SAS Institute 1999).Spearman rank correlations were used to test the relationships betweenthe spenn parameters. Logistic regression models were fit using theproc logistic procedure in SAS to assess the relationship between eachof the sperm quality attn"butes and field fertility. Results from the flowcytometric analyses of each sperm quality attribute were ranked intothree groups in the logistic regression models: (1) freeze codes with thelowest 25% measures; (2) freeze codes with the middle 50% measures;and (3) freeze codes with the highest 25% measures. In the logisticmodels, the three ranked groups were coded by deviation from meanscoding (Hosmer and Lcmcshow 2000). The three ranked freeze codegroups were tested against the average of all three freeze code groupsfor each sperm parameter. Results are presented as odda ratio of AI

'. - -. ..-

Ie. E. Waterhouse et al.

(a)

10' 102 103

Yo-Pro-1 and PNA-FITC

(b)

103

fi!a:

S102

10'

100

10' 102 103

Yo-Pro-1 and PNA-FITC

FIg. 3. (a) A typical histogramof Yo-Pro-l and peanut agglu-tinin (PNA)-fluorescein isothiocyanate (FlTC) fluorescence intensi-ties in post-thawed bull spennatozoa. Region A, live acrosome-intact(LAl) spermatozoa, negative forYo-Pro-I and PNA-FlTC. Region B,dead, acrosome-reacted or -damaged spermatozoa or both, positivefor Yo-Pro-I and/or positively labelled with PNA-mc. (b) A typ-ical cytogram ofYo-Pro-I and PNA-mc v. MT-RED fluorescencein post-thawed bull spermatozoa. Region A. LAI spermatozoa withhigh MitoTraclter Red CMXRos (MT.REO) fluorescence. Region B,dead, acrosome-reacted or -damaged spermatozoa, or both, with highMT-RED fluorescence. Region CrLAl spermatozoa with low MT-RED

fluorescence. Region D, dead, acrosome-reacted or -damaged spcnna-tozoa, or both, with low MT-RED fluorescence.

success defined as the change in odds betw=n each ranked group andthe average odds of all the three groups.

The sperm variables were entered one at a time in the logistic regres-sion with 60 days NR (0 = return, I = non-return) as the binary outcome

variable. The model contained the following explanatory variables: AIpersonnel (divided into veterinarian and AI technician); lactation num.ber (divided into beifers, 1st lactation and multiparous); month of AI;and the sperm parameter (divided into 25% lowest, 50% middle and25% highest freeze code measures from the flow cytometric analyses).Odds ratio estimates for AI success (return Yonon-return) in the threeranked groups compared with the average odds of all the three groups

Fertility of bull semcn and spcnn quality Reprodvction. Fertility and Development 785

Table 1. Correlallou coefllc1euh between sperm quaHty parameters assessed byRow cytomeb'y

LA!. percentagc of live acrosome-intact spermatozoa; LAI-MitoTrackcr Red CMXRos(MT-RED)-positivc, percentage of live acrosome-intact spermatozoa with fimctional

MT-RED tluoresc:cnce intensity; DFI, DNA fragmentation index measured by thesperm chromatin s1rUctllrc assay; DNA strand breaks, percentage of

TUNEL-positive spermatozoa

are presented for significant sperm parameters. P-valucs less than 0.05wen: considered statistically significant

The same type of model as described above was used to test the

relationship between multiple sperm parameters and 60 days NR withstepwise backward selection procedure. Sperm parameters with highcorrelation coefficients were tested in separate models.

From the logistic regression models that were significantly related tofertility, the 60 days NR rate (independent of parity) was also estimatedfor the three ranked groups of significant sperm parameters.

Results

Post-thaw sperm quality

Some spermatozoa with abnormal chromatin structure wereobserved in all included freeze codes, with a mean percent-age DFI of 6.1 (s.d. =3.1). Spermatozoa with fragmented,single-smmded DNA assessed as having increased red AO-fluorescence compared with the main speno populatio~ areshown in Fig. 1.

A typical cytogram of spermatozoa demonstrating biotin-dUTP':'positivity (FITC fluorescence), representing sperma-tozoa with DNA strand breaks, is presen~ in Fig. 2. TheTUNEL assay detected some spermatozoa with DNA strandbreaks in all the included freeze codes, with a mean percent-age of7.6 (s.d. =3.5) of TUNEL-positive spermatozoa. Thepercentage of spermatozoa having DNA strand breaks and% DFIwerehighlycorrelated(0.82,P < 0.001). .

A typical histogram with populations of LA! spermato-zoa and dead. acrosome-reacted or -damaged spermatozoa orboth are shown in Fig. 3a and a typical cytogIaIDofMT-REDstaining in LA! spermatozoa is shown in Fig. 3b. After freez-ing, the mean percentage of LAI spermatozoa was 54.7(s.d. = 10.2),whereasthemean percentageofLAI sperma-tozoa with positive MT-RED staining was 51.3 (s.d. =10.2).Almost all spermatozoa in the LA! population had dis-tinctly higher MT-RED staining intensity compared with thesame parameter in the dead. acrosome-reacted or -damagedspermatozoa, thus we were only able to detect Ii few LA!spermatozoa with dcpolarised potential over the inner mito-chondrial membrane (mean percentage of 3.0 (s.d. = 1.1),region C, Fig. 3b). Hence, the percentage ofLAI spermatozoaand the percentage of LAI-MT-RED-positive spermatozoawere highly correlated (Table 1).

Significant negative conelations were f01md between both

percentagesof DFI and DNA smmd breaks and the per-centage of LAI and LAI-MT-RED-positivespermatozoa(Table1).

Relationship between sperm parametersandfield fertility

Percentages of DFI and TUNEL-positive spermatozoa atfreeze code level were significantly associated with field fer-tility(P = 0.019 and P =0.024 respectively). Compared withthe average of all freeze code groups, group 1with DFI of 1.6-3.8% had an estimated 7% increase in odds of AI success

(60 days NR. P=O.OIO), whereas groupJ withDFI of 7.5-21.6% had an estimated 6% reduction in odds of AI success(P =0.011)(Table2). The odds ofAI success for group 2 with3.8-7.5% DFI was not significantly different from the averageof ail freeze code groups (P =0.848) (Table 2). For group 1with 2.2-4.8% TUNEL-positive spermatozoa, the odds of AIsuccess was 10010(P =0.006) higher compared with the aver.age of all freeze code groups, whereas the odds of AI successfor group 3 with 9.4-21.1 %TUNEL-positive spermatozoa orgroup 2 with 4.8-9.4% TUNEL-positive spermatozoa werenot significantly different from the average .(P =0.089 andP =0.202 respectively; Table 2).

Neither percentages of LAI nor MT-RED LA! at freezecode level were significantly associated with field fertilitywhen entered one at a time in the logistic regression models(P = 0.734 and P =0.870 respectively).

The backward stepWise selection procedure did not giveany significant multiple models where other sperm parame-ters than DFI or DNA strand breaks were included.

Estimated 60 days NR rate values based on the threeranked groups of DFI and DNA strand breaks measures arepresented in Table 2.

Discussion

Despite the high number of spermatozoa per AI dose andthe overall high field fertility of the included freeze codes,significant relationships between sperm quality attributes and

Sperm parameters LAI LAI-MT-RED-positive DNA strand breaks

LAI-MT-RED-positive 0.99.DFI -0.29. -0.31. 0.82.DNA strand breaks -0.40. -0.41.

786 Reproduction.FerliJitymul Development Ie. E. Waterhouse et al.

Table 2. Odds rado estimates of U1IftelallnsemlnatioD (AI) success and esdmated 60 daysDon-return (NR) rate of the three ranked freeze code groups of DNA fragmeutadoa assessed

by the sperm du'omatln structure assay (SCSA) and TUNEL assayDF!, DNA fragmentationindex measured by the SCSA;DNA strand breaks, percentage of

TUNEL-positive spermatozoa

ARaagcofthe 2S%lowcst &eezccodemcasures (group 1). IlRangeofthe SOO4middle freczecodemeasures

(group 2). CRange of the 2S% highest freeze code measures (group 3). DSixty days NR rate (independent

of parity) for tile tbr= ranked groups were estimated from the logistic regression mocIels. .Significant1y

diffcn:ntfromthe averageof allthree&cezecodegroups,p < O.OS. .

field fertility were revealed in the present study. In particular,the SCSA-derived % DFI and % TUNEL-positive spermato-zoa were significantly related to field fertility. The range in% DFI and % TUNEL-positive spennatozoa were not atthe level to separate freeze codes with an unacceptably lowfertility potential from those with a high fertility poten-tial. However, our findings are of significant practical andeconomical consequence for the AI industry (Economy andFertility in Cattle Software; personal communication withA. O. Refsdal, Geno Breeding and AI Association).

The SCSA has proven valuable as a way to distinguishbetween ferti1e, subfertile or infertile men (Evenson et al.1999, 2002; Evenson and Wixon 2006b). Bulls with poorsemen quality and potential sub fertility are removed fromthe breeding population and are therefore not the main issuehere. The question is rather if the SCSA can distinguishbetween semen from fertile bulls in the high range and whosespermatozoa have been pre-selected before AI (~5()o1cJmotilespermatozOa in the present study). Previous studies in bullswith a wide range in fertility of 46-83% (Januslcauskaset al.2001,2003) and 53-80% in NR rate (Karabinus et al. 1990)have reported significant correlations between % DFI andfield ferti1ity.Even though the range in NR for the AIs of thefreeze codes in the present study was rather narrow,a signifi-cant association between % DFI and field fertility was foundThus, the SCSA does seem able to distinguish between fer-tility levels of freeze codes with proven fertility in the highrange. .

Sailer et al. (1995) found a high correlation between% DFI and % TUNEL-positive spermatozoa in bull semen(r =0.78, P < 0.001). To our knowledge, our study is one ofthe first to confmn these findings. The high correlation indi-cates that DNA strand breaks most likely are responsible forthe increase in susceptibility of sperm chromatin to in situ

_ ~R ...J

denatUration(Sailer et al. 1995; Evenson and WIXon2006a).Our data suggest that the two assays are measuring the samesites of DNA damage, and thus the high correlation betweenSCSA and TUNEL measures justifies using only one of themethods for evaluating sperm DNA quality. The less expen.sive and time consuming method would be preferable, whichin this case would be the SCSA.

As with the DFI, the percentage of spermatozoa withDNA strand breaks assessed by the TUNEL assay was sig-nificantly associated with fertility. Only a limited numberof studies have looked at the relationship between DNAstrand breaks measured by the TUNEL assay and bull .fer-tility, and to our knowledge, the present study is the firstto demonstrate a significant relationship between fertility ofbull semen and DNA strand breaks of frozen-thawed sper-matozoa. Anzac et al. (2002) found a significant correlationbetween TUNEL-positive spermatozoa in ftesh semen andfertility (r:;: -0.90, P < 0.05), but no significant correlationwas found for frozen-thawed spermatozoa. It has been spec-ulated that DNA ftagmentation in mature spermatozoa is aresult of one oftbe foUowing: defects in the reorganisation ofchromatin during sperm maturation; insufficient protectionagainst reactive oxygen species (ROS) in the epididymis; orthe presence of an 'abortive apoptosis' mechanism (Sakkas

.et aI. 1999; Shen et al. 2002). Recently, it has been shown thatDNA ftagmentation of human spermatozoa was not related toearly apoptotic markers, thus ROS might be a stronger can-didate (Henkel et al. 2004). Spermatozoa with fragmentedDNA may in fact have normal motility and morphology andthus may still be able to fertilise an oocyte (Ahmadi and Ng1999; Fatehi et al. 2006). A recent study of bull spermato-zoa showed that DNA ftagmentation does not impair in vitrofertilisation or completion of the first cleavage stages, butblocks further embryonic development when the blastocyst

Sperm parameters Odds ratio 95% CI No. AIs Estimated 60 daysNRratc(%)D

DFI (%)Lower quarter (I.6-3.8A) 1.07. 1.02-1.12 4846 73.9

Middle haIf(3.8-7.SB) 1.00 0.96-1.04 10180 72.8

Upper quarter (7.S-21.6c) 0.94. 0.89-0.99 5246 71.6DNA strand breaks (%)

Lower quarter (2.2-4.8A) 1.07. 1.02-1.13 5018 74.2

Middle haIf(4.8-9.4B) 0.97 0.93-1.01 9787 72.9

Upper quarter (9.4-2I.1c) 0.96 0.99-1.01 5467 72.1

Fertility of bull semen and sperm quality

fonnanon is reached, by inducing apoptosis (Fatehi et al.2006).

A negative correlation was found between DNA frag~mentation, assessed by SCSA or TUNEL assay, and thepercentage of LAl spermatozoa, which may indicate thatfactors affecting DNA during sperm maturation in the epi-didymis also act at the plasma- and acrosome-membranelevel and make the spermatozoa more wlnerable to thefteezing and thawing process. The high level of polyunsat-urated fatty acids in mammalian sperm membranes makesthem especially vulnerable to ROS, which may be the com-mon factor that damages both membranes and DNA (Sikh2004). A rather high variation between the fteeze codes inthe percentage of LAI spermatozoa was observed. However,the percentage of LAI spermatozoa showed no significantrelation with field fet:tility.The literature is somewhat contra-dictory on this point. Anzar et al. (2002) showed a significantcorrelation (r = 0.87,P < 0.05) between the percentage ofplasma membrane-intact spermatozoa and 56 days NR ratefor fresh bull semen, but not for ftozen-thawed semen.Although lanuskauskas et al. (2003) found a significantcorrelation between fertility and the percentage of plasmamembrane-intact spermatozoa assessed byfluorometry, a sig-nificant relationship was not found between the percentageof plasma membrane-intact spermatozoa assessed by flowcytometry and fertility.

To our knowledge, the present study is the first inwhich mitochondrial membrane functionality has beenassessed simultaneously with plasma membrane and acro-some integrity, and related to field fertility. In aU fteezecodes, most LAl spermatozoa had staining correspondingto a functional potential over the inner mitochondrial mem-brane, reflected in the high correlation coefficient betweenthe percentage of LAl and LAI-MT-RED-positive sper-matozoa. Variation between freeze codes in percentage ofLAl~MT-RED-positive spermatozoa was observed, but nosignificant relation to fertility was found. These findings arein agreement with Ericsson et al. (1993), who found no sig-nificant correlation between fertility and the percentage oflive spermatozoa with functional mitochondria. .

It is crucial to have accurate and reliable fertility data instudies as described bere. Otherwise it will be difficult to

draw valid conclusions concerning relations between spermquality and fertility (Amann and Hammerstedt 2002; Amann2005). The Norwegian Dairy Herd Recording System, pro-viding fertility data for the present study, made it possibleto record accurate and reliable fertility data. The fertilitydata used in the present study were based on Als with asperm number per Al dose of -18 x 106, which is a num-ber commonly used in an AI setting, as well as a numberthat is likely to fall on the asymptote of the spermatozoa perAI dose-response curve (Amann and Hammerstedt 2002).While DNA ftagmentation is considered to be an uncom-pensable trait, plasma membrane and acrosome integrity and

Reproduction,Fertility and Development 787

motility of spermatozoa are considered to be compensableattributes (pace et al. 1981; Saacke et al. 2000). We sug-gest that the observed freeze code variation in % LA! and% LAl~MT-RED-positivespermatozoa is camouflaged by therelatively high number of spermatozoa per Al dose. Below ac;ertainthreshold we would expect a negative effect on fer-tility with use of AI doses with lowered sperm number, i.e.with reduced number of LAI spermatozoa in the Al dose. Itwould be interesting to use much lower sperm numbers. forobtaining additional research data and knowledge on the rela-tionships between compensable sperm attributes and fertility.However, carrying out field trials with lowered sperm num-bers, i.e. on the dose-response portion of the curve, wouldbe far too ~pensive for the farmers and AI industry andwould not reflect the present commercial Al setting in Nor-way. Consequently, defec~ in compensable sperm attributesmay be masked.

. Toconclude,the relativelyhighnumberof spermatozoaper AI dose may partially be responsible for the lack ofsignifi-cant relationships between some of the sperm parameters andfield fertility. Despite the high number of spermatozoa per Aldose, pre-selection of Al doses and the high field fertility ofthe included freeze codes, the SCSA and TUNEL assays wereassociated with field fertility. This is the first study using thepowerful technique of flow cytometry to assess fIVespermparameters.

Acknowledgments

The authors would like to thank Geno Breeding and AIAssociation for providing the samples for this study.

References

Ahmadi, A., and Ng, S. C. (1999). Fenilizing ability of DNA-damagedspermatozoa.J. Exp. Zool. 284, 696-704. doi:l0.IOO2/(SICI)1097-010X(l9991101)284:6<696::AID-JEZll>3.0.CO:2-E

Amann, R. P. (1989). Can the fertility potential of a seminal sample bepredicted accurately? J. Androl. 10, 89-98.

Amann, R. P. (200S). Weaknesses in ~pons of "fertility" for horsesand other species. Theriogeno(ogy 63, 698-715. doi:l0.1016/I.THERIOOENOLOGY.2004.04.010

Amann, R. P., and Hammerstedt, R. H. (2002). Detection of differencesin fertility. J. Am/rol. 23,317-325.

Anzar, M., He, L., Buhr. M. M., Kroc:tsch, T. G., and Pauls, 1(. P. (2002).Sperm apoptosis in ftesh and cryopreserved bull semen detected byflow cytometry and its relationship with. fertility. Biol. Reprod. 66,354-360. doi: 10.109SIBIOLREPROD66.2.3S4

Arya, M., and Vanha-Perttula, T. (1985). Lectin-binding pattern of bulltestis and epididymis. I Androl. 6, 230-242.

Ballachey, B. E.. Hohcnbolcen, W. D., and Evenson, D. P. (1987).

Heterogeneity of sperm nuclear chromatin structure and its rela-tionship to bull fenility. Bioi. Reprod. 36,915-925. doi:1O.I09S/BIOLREPROD36.4.91S

Ballachey, B. E., Saacla:, R. G., and Evenson, D. P. (1988). Thesperm chromatin structure assay: Relationship with alternate testsof spenn quality and heterospennic performance of bulls. J. A1Idrol.9, 109-11S.

Ericsson, S. A., Garner, D. L., Thomas, C. A., Downing, T. w., andMarshall, C. E. (1993). Interrelationships among fluorometric

788 Reproduction. Fertility and Development

analyses of spermatozoal fimction, classical semen quality parame-ters and the fertility of frozen-thawed bovine spermatozoa. Therio-genology 39, 1009-1024. doi:l0.l0l6/0093-691X(93)90002-M

Evenson, D. P., and Wixon, R. (2006a). Clinical aspects of sperm DNAftagmentation detection and male infertility. Theriogenology 65,979-991. doi:10. 1016/J.THERIOGENOLOGY.2005.09.01I

Evenson. D. R, and Wixon, R. (2006b). Meta-analysis of spcrm DNAfiagmentation using the Spcrm Chromatin Structure Assay. Reprod.Biomed. Online 11,466-472.

Evenson, D. P.,Darzynkiewicz,Z., andMelamed,M. R. (1980). Relationof mammalian sperm chromatin heterogeneity to fertility. ScienceZlO, 1131-1133.

Evenson, D. P., Darzynkiewicz, Z., and Melamccl. M. R. (1982). Simul-taneous measurement by flaw cytometty of sperm cell viabilityand mitochondrial membrane potential related to cell motility.J. Histochem. Cytochem. 30, 279-280.

Evenson, D. P., Jost, L. K.., Marshall, D., Zinaman, M. J., Clegg. E.,Purvis, K.., De Angelis. R, and Claussen, O. R (1999). Utility ofthe spenD chromatin structure assay as a diagnostic and prognos-tic tool in the human fertility clinic. Hum. Reprod. 14, 1039-1049.doi: 10.1 093/HUMREP/14.4.1039

Evenson, D. P., Lanon, K. t., and Jost, L. K. (2002). Sperm chromatinstructure assay: its clinical use for detecting spenD DNA ftagmcn-tation in male infertiJity and comparisons with other techniques.J. A.ndrol. 23, 25-43.

Fatehi, A. N., Bevers,M. M., Schoevers,E., Reelen, B. A. J.,Colenbrander, B., and Gadella, B. M. (2006). DNA damage in bovinesperm does not block fertilization and early cmbtyonic dcvclQJImentbut inducesapoptosisaftertbe firstclcavagcs.J.A.ndrol. 27, 176-188.doi: 10.21641JANDROL.04152

Gadella, B. M., and Harrison, R. A. R (2002). Capacitation inducescyclic adenosine 3',5' -monophosphatc-dependent, but apoptosis-unrelated, exposure of aminophospliolipids at the apical headplasma membrane ofboar sperm cells. Bioi. Reprod. 67, 340-350.doi:l0.l 095IBIOLREPROD67.1.34O

Oavrieli, Y., Sherman, Y~ and Ben Sasson, S. A. (1992). Identificationof programmed cell death in situ via specific labeling of nuclearDNA fiagmentation. J. Cell Bioi. 119, 493-501. doi:l0.1083/JCB.119.3.493

Gillan, L., Evans, G., and Maxwell, W. M. (200S). Flowcytometric evaluation of sperm parametenl in relation tofertility potential. Thel'iogenology 63, 445457. doi:l0.10161J.THERlOGENOLOGY.2004.09.024

Graham, 1.K. (2001). Assessment of sperm quality: a flow cytometricapproach. A.nim. Reprod. Sci. 68, 239-247. doi:l0.1016lS0378~4320(01)00160-9

Harrison, R. A., Ashworth, P.J., and Miller, N. G. (1996). Bicarbonate!CO:z,an effector of capacitation, induces a rapid and reversiblechange in the lipid architc:cturc of boar sperm plasma mem-branes. Mol. Reprod. Dev. 45,378-391. doi:l0.IOO2l(SICI)I098-2795(199611)45:3<378::AID-MRDI6>3.0.CO;2-V

Henkel, R., Hajimohammad,M., StaIf,T., Hoogcndijk,C., Mehnert, C.,Menkveld, R., Gips, H., Schill, W. B., and Kroger, T. F.(2004). Influence of deoxyribonucleic acid damage on fertil-ization and pregnancy. Fe,.,il. Steril. 81, 965-972. doi:l0.1016/!PER~STERT.2003~9~44

K. E. Waterhouse et al.

Hosmer, D. w., and Lcmeshow, S., lr (2ooo).lnterprctation of the fitted

logistic regression model. In. 'Applied Logistic Regression'. (EdsN. A. C. Cressie, N. I. Fisher, I. M. 10hnstone, J. B. Kadane, D. W.

Scott etal.) pp. 47-90. (lohn Wiley & Sons: New York, USA.)Januskauslcas, A., Johannisson,A., and Rodriguez-Martinez, H. (2001).

Assessment of sperm quality through fluorometry and sperm chro.matin structure assay to field fertility of frozen-thawed semenform Swedish AI bulls. Therlogenology 55, 947-961. doi:l0.IOI61S0093-691X(0l )00456-3

1anuskauskas, A., Johannisson, A., and Rodriguez-Martinez, H. (2003).Subtle membrane changes in cryopreserved bull semen in relationwith spenD viability, chromatin structure, and field fertility. Theria-genology 60, 743-758, doi:l0.10161S0093-69IX(03)OOO50-5

Karabinus, D. S., Evenson. D. R, lost, L. K.., Bacr, R. K., andKaproth, M. T. (1990). Comp~son of semen quality in youngand mature Holstein bulls measured by light microscopy and flawcytomeUy..~ Dairy Sci. 73,2364-2371.

Kommisrud, E., Steinc, T., and GIaffer, T. (1996). Comparison offcrtilityrates following insemination with different numbers of spermatozoaper iusemination dose of fi'ozen bovine semen. Reprod. Domest.Anim. 31, 359-362.

Pace, M. M., Sullivan, 1.1.. Elliott, F. I., Graham, E. F., and Coulter, G. H.(1981). Effects oftMwing temperature, nmnberofspcrmatozoa and

spermatozoal quality on fertility of bovine spermatozoa packed in5-mL French StraWs.J. Anim. &i. 53, 693-701.

Ranberg, I. M., Hcringstad, B., Klemetsdal, G., Svendsen,. M., andSteine, T. (2003). Heifer fertility in Norwegian dairy catt1e: variancecomponents and genetic change. J. Dairy Sci. 86,2706-2714.

Saacke, R. G., Dalton, 1. C., Nadir,S., Nebel, R. L., and Bame, 1. H.

(2000). Relationship of seminal traitS and insemination time tofertilization rate and embryo quality. Anim. Reprod. Sci. 60-61,663-677. doi: 10.10 I6IS0378-4320(OO)OO 137-8

Sailer, B. L., Jost. L. K.., and Evenson, D. R (1995). Mammalian

spenD DNA susceptibility to in situ denaturation associated withthe presence of DNA strand breaks as measured by the terminaldeoxynucleotidyl transferase assay. J. A.ndrol. 16, 80-87.

Sakkas, D., Mariethoz, E., Manicardi, G., Bizzaro, D., Bianchi, P. G.,

and Bianchi, U. (1999). Origin of DNA damage in ejaculated humanspermatozoa. Rev. Reprod. 4, 31-37. doi:l0.1 5301R0R.o.o040031

SAS Institute (1999) '8AS 0nIincD0c Version 8.' (SAS Institute: Cary,NC, USA.)

Shen, H. M., Dai, 1., Chis, S. E., Lirn, A., and Ong, C. N. (2002).

Detection of &poptotic alterations in sperm in subfertile patients andtheir correlations with spcnn quality. Hum. Reprod. 17, 1266-1273.doi: 10.1093/HUMREP/17.5.1266

Si1cJca,S. C. (2004). Role of oxidative stress and antioxidants in

andrology and assisted reproductive technology. J. A.ndrol. 25, 5-18.Stokke, T., Solberg, K., De Ange1is, R, and Steen, H. B.

(1998). Propidium iodide quenches the fluoresccncc of TdT-incorporated me-labeled dUTP in apeptonc cells. C)'tametry33, 42~34. doi:l0.IOO2l(SICI)1097-o320(19981201)33:4<428::AJD.CYT06>3.0.CO;2-4

Manuscriptreceived4 April2006;revisedandaccepted10July2006.