BIOLOGY OF REPRODUCTION 69, 469–474 (2003)
Published online before print 2 April 2003.
Serial Analysis of Gene Expression in Turkey Sperm Storage Tubules in the Presence
and Absence of Resident Sperm
Ezhou L. Long,2,3Tad S. Sonstegard,4Julie A. Long,3Curtis P. Van Tassell,4and Kurt A. Zuelke1,3
Biotechnology and Germplasm Laboratory3and Bovine Functional Genomics Laboratory,4Animal and Natural
Resources Institute, USDA-ARS, Beltsville, Maryland 20705
Turkey sperm lose viability within 8–18 h when stored as
liquid semen using current methods and extenders. In contrast,
turkey hens maintain viable, fertile sperm in their sperm storage
tubules (SST) for 45 or more days following a single insemina-
tion. Our long-term objectives are to identify and characterize
differentially expressed genes that may underlie this prolonged
sperm storage and then use this information to develop im-
proved methods for storing liquid turkey semen. We employed
serial analysis of gene expression (SAGE) to compare gene ex-
pression patterns in turkey SST recovered from hens after arti-
ficial insemination (AI) with extended semen (sperm AI) or ex-
tender alone (control AI). We constructed two separate SAGE
libraries with SST RNA obtained from sperm and control AI
hens. We used these libraries to generate 95325 ten-base pair
SAGE tags. These 95325 tags represented 27430 unique genes.
The sperm and control AI libraries contained 47663 and 47662
tags representing 18030 and 19101 putative unique transcripts,
respectively. Approximately 1% of these putative unique genes
were differentially expressed (P ? 0.05) between treatments.
Tentative annotations were ascribed to the SAGE tag nucleotide
sequences by comparing them against publicly available SAGE
tag and cDNA sequence databases. Based on its SAGE tag nu-
cleotide sequence, we cloned a partial turkey avidin cDNA and
confirmed its up-regulation in the sperm AI SST. The bioinfor-
matics and experimental procedures employed to clone turkey
avidin and confirm its differential expression represent a useful
paradigm for analyzing SAGE tag data from relatively unchar-
acterized model systems.
gene regulation, oviduct, sperm, sperm motility and transport
Because of prolonged sperm storage in their distal ovi-
duct, female poultry can produce fertile eggs for several
weeks following a single natural mating or artificial insem-
ination (AI) [1, 2]. Oviductal sperm storage is a feature
common to all avian and a wide range of other species,
including newt , snakes [4, 5], frogs , lizards , and
turtles . Fertile sperm can be maintained in the turkey
hen oviduct from 45 to 112 days following either AI or
1Correspondence: Kurt A. Zuelke, Biotechnology and Germplasm Labo-
ratory, Animal and Natural Resources Institute, USDA-ARS, Bldg 200, Rm
124A, BARC-East, Beltsville, MD 20705. FAX: 301 504 5123;
2Current address: National Cancer Institute, Center for Cancer Research,
Mammary Biology and Tumorigenesis Laboratory, Bldg 10, Rm 5847, Be-
thesda, MD 20892-1402.
Received: 31 December 2002.
First decision: 21 January 2003.
Accepted: 26 March 2003.
? 2003 by the Society for the Study of Reproduction, Inc.
ISSN: 0006-3363. http://www.biolreprod.org
natural mating . Avian sperm are stored in specialized
sperm storage tubules (SST) localized within the luminal
mucosal epithelium of the uterovaginal junction (UVJ) [1,
9]. Turkey SST are comprised primarily of tall columnar,
nonciliated, nonsecretory epithelial cells that appear as bud-
like invaginations of the UVJ mucosal epithelium [10, 11].
In marked contrast to the prolonged sperm storage observed
in SST, current best practices for handling and storing liquid
turkey semen in vitro only maintain sperm viability for ap-
proximately 8–18 h [12, 13].
The spermatozoa are closely apposed to the SST luminal
epithelium, and numerous physiological interactions are
likely both between and within these separate cell types
during prolonged sperm storage in turkeys [1, 9]. Sperm
localized within the SST are typically immotile. Thus, gen-
eral mechanisms that reversibly suppress sperm respiration
and motility, stabilize sperm cell membranes and enzyme
systems, and suppress sperm immunogenicity within the
SST have been proposed to mediate prolonged sperm stor-
age [14–17]. Turkey intraluminal SST epithelial cells con-
tain relatively high levels of carbonic anhydrase, suggesting
that intraluminal SST pH might play a significant role in
modulating sperm motility during oviductal sperm storage
and transport . The presence of progesterone and estro-
gen receptors in SST of laying hens, but not in immature
chicks, suggests that steroid hormones could also play a
role in prolonged sperm storage . Elucidating and char-
acterizing the molecular mechanisms that enable prolonged
sperm storage would assist in developing more efficient
methods for preserving turkey semen in vitro. Because vir-
tually all turkey breeders in the United States use AI, im-
proving sperm viability during short-term liquid semen
handling and storage would increase reproductive manage-
ment options and reduce turkey production costs.
Based on the close apposition between sperm and SST
epithelium combined with the physiological modifications
that occur in the sperm (e.g., decreased motility) within the
SST, we hypothesized that specific and characteristic gene
expression events within the turkey SST enable and regu-
late prolonged sperm storage. An objective of the present
study was to perform comparative serial analysis of gene
expression (SAGE) between turkey SST with or without
resident sperm. A relatively new research strategy, SAGE
enables transcriptome-wide qualitative and quantitative
analysis of gene expression within tissues during discreet
physiological states [20, 21]. We constructed two turkey
SAGE libraries from SST total RNA obtained 48 h after
AI with either extended semen (sperm AI) or extender
alone (no sperm; control AI) and then compared SAGE tag
frequencies between these two libraries to generate a list of
putative differentially expressed genes in the SST epithelial
cells. We then developed an analytical and bioinformatics
paradigm to obtain tentative sequence annotations for the
LONG ET AL.
genes represented by the turkey SAGE tag frequency and
nucleotide sequence data. We identified a differentially
sampled SAGE tag corresponding to turkey avidin and con-
firmed its differential expression via quantitative real-time
reverse transcriptase-polymerase chain reaction (RT-PCR)
to validate this paradigm.
MATERIALS AND METHODS
Large White BUTA (British United Turkeys of America, Lewisburg,
WV) breeder turkeys were maintained under standard management con-
ditions and photostimulated on a daily basis with a 14L:10D photoperiod.
The hens were kept individually in cages, and the toms were kept in
groups of 8–10 in floor pens. Feed and water were provided ad libitum.
Semen was collected by abdominal massage and diluted 1:1 with Beltsville
Poultry Semen Extender (BPSE), and the sperm concentration was deter-
mined . For all hens, AI was performed using 1.5 ? 108sperm/hen.
To generate SST samples for construction of the control library (no
sperm), three randomly selected hens were sham-inseminated using only
BPSE without sperm. Typically, between 65% and 80% of the SST contain
sperm at 48 h after AI under these conditions [11, 22].
Sample Collection and Total RNA Extraction
Hens selected for SST collection within each treatment group had all
laid an egg on the morning of collection. Hens were killed by cervical
dislocation 48 h after insemination to achieve maximal filling of the SST
. The oviduct was immediately removed from the bird and quickly
trimmed free of connective tissue surrounding the vagina and uterus to
reveal the UVJ, and the SST were recovered by scraping the mucosal
epithelium of the UVJ with a scalpel blade . Samples were snap-frozen
in liquid nitrogen and maintained at ?70?C until RNA extraction. Total
RNA was isolated using the Totally RNA Kit (Ambion, Austin, TX) ac-
cording to the manufacturer’s instructions. Briefly, approximately 80 mg
of SST mucosal epithelium were lysed in 800 ?l of denaturing solution
using a Tissuemizer Mark II T25 homogenizer (Tekmar Company, Cin-
cinnati, OH). After phenol/chloroform extraction, total RNA was precip-
itated with an equal volume of isopropanol and washed with 70% (v/v)
ethanol. The RNA pellets were solubilized in 0.1% (v/v)-diethyl pyrocar-
bonate-treated dH2O and stored at ?70?C. Integrity of the total RNA was
assessed by 1% (w/v) agarose-formaldehyde gel electrophoresis.
Construction of SAGE Libraries
Aliquots of total RNA from the SST of three hens (?3.5 ?g RNA/
bird) were pooled within each treatment group before constructing the
SAGE libraries with the I-SAGE kit (Invitrogen, Carlsbad, CA). The pro-
tocol of this kit is based on the original SAGE methodology [20; see also
http://www.sagenet.org]. The anchoring and tagging restriction enzymes
used were NlaIII and BsmFI, respectively. Following PCR amplification,
di-tags (26mers) were released using NlaIII and separated via 12% (w/v)
PAGE. The 26mer di-tags were purified from the polyacrylamide gel, li-
gated to form concatemers, and size fractionated via 8% (w/v) PAGE.
Three size ranges of di-tag concatemers (300–500, 500–800, and ?800
base pairs [bp]) were isolated. The 500- to 800-bp fraction was ligated
into SphI-linearized pZErO-1 vector (Invitrogen) to construct the present
libraries. Ligation products were transformed into One Shot TOP10 Elec-
trocomp cells (Invitrogen) by electroporation and cultured overnight at
37?C on low-salt LB agar plates containing 50 ?g/ml of zeocin. The SAGE
tag inserts were amplified by inoculating individual colonies into separate
wells of a 96-well PCR plate containing 25 ?l of a PCR reaction mix
consisting of 2.5 ?l of 10? PCR Buffer (200 mM Tris-HCl [pH 8.4] and
500 mM KCl), 1.25 ?l of dimethyl sulfoxide, 500 ?M dNTP, 1.7 ?M
MgCl2, 0.2 ?M M13 forward (5?-CCCAGTCACGACGTTGTAAAACG-
3?) and reverse (5?-AGCGGATAACAATTTCACACAGG-3?) primers,
and 1 U of Platinum Taq DNA polymerase (Invitrogen). The thermocy-
cling profile was denaturation at 94?C for 30 sec, annealing at 55?C for
30 sec, and extension at 70?C for 90 sec for 27 cycles. The PCR products
were purified with a 96-well Montage PCR cleanup kit (Millipore, Bed-
ford, MA) and recovered in 50 ?l of nuclease-free water. For sequence
analysis, 2 ?l of the purified PCR product template were transferred to a
384-well plate containing 0.5 ?l of Big-Dye v. 2.0 (Applied Biosystems,
Foster City, CA), 1.5 ?l of Big-Dye extender (Sigma, St. Louis, MO), and
3.2 pmol of SP6 primer (5?-ATTTAGGTGACACTATAG-3?) and were
then amplified using standard thermocycling conditions as recommended
by the manufacturer of Big-Dye. Reaction products were precipitated with
four volumes of 70% (v/v) isopropanol and washed with four volumes of
70% (v/v) ethanol. After drying, reaction products were resuspended in
25 ?l of Hi-Di Formamide (Applied Biosystems), denatured at 95?C for
5 min, and analyzed on an ABI-3700 automated DNA analyzer (Applied
Processing and Analysis of SAGE Tag Sequences
Sequence quality assessment and trimming were performed with phred
v0.980904.e. Vector sequence was identified and trimmed using ‘‘cross
match’’ with the ‘‘-minscore 18’’ and ‘‘-minmatch 12’’ options. Sequence
information in the processed trace files was converted into text files, and
the SAGE tags were extracted and quantified using SAGE 2000 software
version 4.12 (http://www.sagenet.org/Software/software2000.htm). Tag
nucleotide sequences and frequency data were then outputted to MS Ac-
cess (Microsoft, Redmond, WA) database files for subsequent analyses.
For determining differential expression, tag frequencies between both
SAGE libraries were analyzed for significance (P ? 0.05) using tools in
the SAGE 2000 software based on a chi-square analysis combined with
Monte-Carlo simulations .
Annotation of SAGE Tags
Because the NlaIII-recognition sequence CATG lies immediately 5? of
the SAGE tag sequence, CATG was appended 5? to each SAGE tag to
yield 14-bp tags for use in database comparisons . Three primary-
sequence databases were used to ascribe tentative annotation to the turkey-
derived SAGE tags. First, tag sequences were compared with an existing
database of human SAGE tags (human SAGEmap database; National Cen-
ter for Biotechnology Information [NCBI], Bethesda, MD; ftp://
ftp.ncbi.nih.gov/pub/sage/map) . Turkey-derived SAGE tags were then
also compared against cDNA sequence information in the chicken gene
index (GgGI, version 4) available from The Institute of Genome Research
(TIGR; Rockville, MD; http://www.tigr.org/tdb/tgi/gggi)  and a chick-
en expressed sequence tag (EST) database (http://www.chick.umist.ac.uk/
cgi-bin/chicken database.cgi; version 11/06/02) available from the Bio-
technology and Biological Sciences Research Council (BBSRC; Roslin,
U.K.). Database comparisons were performed using BLAST algorithms
 (http://www.ncbi.nlm.nih.gov/BLAST/) and required a perfect 14-
base match for inclusion in the final data set. Tentative annotation of the
turkey SAGE tag representing avidin was also confirmed by BLAST anal-
ysis of this tag with the GenBank nonredundant (nr) database (NCBI; http:
//www.ncbi.nlm.gov/). To increase the likelihood that these tentative
SAGE tag annotations could be ascribed to a defined functional category
(e.g., metabolism, cytoskeletal protein, etc.), initial database matches were
filtered using the following criteria: 1) those that aligned with differentially
expressed (P ? 0.05) SST SAGE tags, 2) those that were represented in
a library five or more times, 3) those that were linked to a Unigene or
TIGR Tentative Consensus (TC) identifier (i.e., tentative contiguous EST
assemblies representing a predicted gene annotation), and 4) those that
were annotated with a gene designation rather than simply an EST clone
identifier (i.e., non-EST annotations). These search criteria yielded the
most consistent gene annotations during preliminary analyses of turkey
SST and swine embryo SAGE library data sets in our laboratory (data not
5?-Rapid Amplification of cDNA Ends
Capture of cDNA sequence information corresponding to a SAGE tag
tentatively identified as turkey avidin was achieved using the 5?-rapid am-
plification of cDNA ends (RACE) System (Invitrogen). First-strand cDNA
synthesis was primed with an aliquot of the sperm-treated SST RNA used
previously for SAGE library construction and the primer (5?-IIIIIGCA-
GCAGCCACATG-3?) consisting of 10 nucleotides complementary to the
tag sequence flanked by CATG (NlaIII-recognition site) and five inosine
nucleotides to aid annealing of primer through nonspecific base-pairing
. First-strand cDNA synthesis products were dC-tailed and amplified
by PCR with an antisense-primer designed from chicken avidin sequence
and complementary to the 14-bp tag sequence (5?-GCAGCAGCCA-
CATGGTCTTC-3?) and the abridged anchor primer provided in the RACE
kit. Products were purified, sequenced, and identified by BLAST analysis
as described above.
SAGE OF TURKEY SPERM STORAGE TUBULES
SAGE tag accrual rates in sperm AI and control AI turkey SST
Real-time PCR was performed using the QuantiTect SYBR Green RT-
PCR kit (Qiagen, Valencia, CA) following the manufacturer’s instruction.
First-strand cDNA was synthesized from total RNA with the oligo-dT(12–
18) primer (Invitrogen) and PCR-amplified in a DNA Engine Opticon Con-
tinuous Fluorescence Detector (MJ Research, Waltham, MA) for as many
as 60 cycles using primers specific to turkey avidin (sense, 5?-
GGCTCCAACATGACCATC-3?; antisense, 5?-GGTGGACTCTGAAAAC-
TTCC-3?). Primers (sense, 5?-CCATGTTTGTGATGGGTGTC-3?; anti-
sense, 5?CTCCACAATGCCAAAGTTGT-3?) specific for turkey glyceral-
dehyde-3-phosphate dehydrogenase (GAPDH) were used as an internal con-
trol for amplification. The GAPDH primers were designed based on the
partial turkey GAPDH sequence (GenBank Accession no. U94327). Spec-
ificity of PCR products was confirmed by melting-curve analysis and gel
electrophoresis. Real-time PCR was repeated in triplicate for each sample
within an individual experiment. The threshold cycle (CT) was the cycle
where increasing fluorescent product was first detectable. Linear standard
curves (CTvs. logarithm of cDNA concentration) were plotted to calculate
the amplification efficiency (AE) of each primer pair by using a logarith-
mic dilution of the cDNA mix reverse transcribed from the control AI
RNA . The AE was calculated using 10(?1/slope of standard curve). The fold
difference in avidin expression between two libraries was calculated as
?CT(Control ? Sperm AI)xindicates the difference in CTof a particular gene X
between two libraries .
? Sperm AI)Avidin)/AE?CT(Control ? Sperm AI)GAPDH), in which
Construction and Comparison Between AI and Control
Insemination SAGE Libraries
Two turkey SST SAGE libraries were constructed and
used to generate a total of 95325 tags. The control and
sperm AI libraries contained 47662 and 47663 tags rep-
resenting 19101 and 18030 putative unique transcripts, re-
spectively. Complete tag sequences and frequencies from
these libraries have been deposited for public access via the
NCBI Gene Expression Omnibus (http://www.ncbi.nlm.
nih.gov/geo; GSM Accession nos. GSM4899 and
GSM4900). Combined, a total of 27430 tags (?29% of
total tags) represented unique putative transcripts present in
these two libraries. Among them 9701 (36%) transcripts
were expressed in both libraries, whereas 9400 (34%) and
8329 (30%) putative transcripts were unique to the control
and sperm insemination libraries, respectively. Statistical
comparison of SAGE tags from each library revealed that
214 potential transcripts (0.78% [214/27430] of total
unique putative genes) were expressed at significantly dif-
ferent levels (P ? 0.05), with 121 putative genes (0.44%
of the total) being up-regulated and 93 (0.34% of the total)
down-regulated between the sperm and control insemina-
tion libraries, respectively. To rule out the possibility that
sperm RNA may have contributed to the gene expression
detected in the sperm AI library, we attempted to extract
RNA from 108turkey spermatozoa for PCR analysis of
putative genes detected by SAGE. No detectable RNA was
isolated from these turkey sperm (data not shown).
To assess the representation of expressed genes relative
to the overall size of each library, the accrual rate of unique
tags identified was plotted as a function of the total tags
sequenced across a series of random subsets of SAGE tags
from within each library (Fig. 1). The accrual rate of new
unique tags decreased steadily and then became more con-
stant as the total number of tags sequenced approached
50000. Unique tag accrual rates and patterns did not differ
between the control and sperm AI SAGE libraries.
Identification and Annotation of Tags Representing
Comparing the differentially expressed (P ? 0.05), SST-
derived SAGE tag sequences against the human SAGEmap
database yielded only six tentative gene annotations. In an
attempt to overcome the lack of poultry-related sequence
information in SAGEmap, turkey SAGE tags obtained from
the sperm and control AI libraries were searched against
the TIGR chicken gene index database. The initial TIGR
chicken gene index database comparison yielded 490432
records. In nearly every case, each SAGE tag corresponded
with multiple tentative annotations. Filtering these data to
include only those tags that were differentially (P ? 0.05)
expressed, that were represented in a library at least five
times, and that were linked to a TIGR TC identifier reduced
the total data set from 490432 to 5923 records. Of the
remaining 5923 records, 3648 records were tentatively an-
notated only as EST clones, leaving just 2275 records that
annotated to a specific, definable gene. Review of these
2275 tentative annotation records for recurring themes (e.g.,
glucose, kinase, tubulin, etc.) enabled ‘‘word-search’’-based
parsing of the data by tentative function, as represented in
Table 1. Comparing the complete SST SAGE tag data set
with the BBSRC Chicken EST database yielded only EST
clone identifiers without more detailed gene annotations
and thus provided no additional information to the TIGR
database search results.
Amplification of Turkey Avidin cDNA Fragment
The SAGE tag sequence (5?-TGGCTGCTGC-3?) ap-
peared approximately 3-fold more frequently in the sperm
AI library and was tentatively annotated as avidin in the
TIGR chicken gene index. The BLAST analysis of this tag
sequence against poultry-annotated sequences in GenBank
(nr and dbEST databases) significantly matched near the
3?-end of an expressed mRNA for chicken avidin (Gen-
Bank accession no. X05343). Sequence alignment indicated
that the turkey avidin SAGE tag sequence mapped to 184
bases 5? of the Poly-A tail at the 3? most distal NlaIII-
recognition (CATG) site of the chicken avidin sequence.
Using 5?-RACE methodology, a 412-bp cDNA fragment
was amplified using the RNA representing the sperm treat-
ment SST sample. The sequence of this turkey cDNA clone
(GenBank accession no. AF545846) shared 92% nucleotide
sequence homology with chicken avidin and thus confirmed
that the SAGE tag (5?-TGGCTGCTGC-3?) represented av-
LONG ET AL.
TABLE 1. Tentative annotations of putative genes represented by differentially expressed SAGE tags in control AI and sperm AI turkey SST SAGE libraries
by comparison with the TIGR chicken gene index (version 4).
SAGE tag counta
1325 TC22536Nicotinic acetylcholine receptor, ?1subunit (Gallus
Precursor polypeptide (amino acids ?24 to 128) (G.
Alpha tubulin (G. gallus)
?-Actin (G. gallus)
?-Actin (G. gallus)
Similar to PIR JC5088 JC5088 pyruvate dehydroge-
Mitochondrial uncoupling protein (G. gallus)
Similar to GP 9652182 gb AAF91430.1 adenosine
Facilitative glucose transporter (G. gallus)
Plasma membrane calcium pump (G. gallus)
N-type calcium channel ?1Bsubunit (G. gallus)
Heat shock protein 90 beta (G. gallus)
Trans-Golgi network protease furin (G. gallus)
138 371 TC22445
aSAGE tag count differences between the libraries were significant (P ? 0.05).
bThe control and sperm AI libraries consisted of 47 662 and 47 663 total tags, respectively.
TABLE 2. Real-time RT-PCR analysis of avidin expression between con-
trol AI and sperm AI SST.a
CT(mean ? SEM)
34.63 ? 0.76
25.88 ? 0.23
34.61 ? 0.33
23.45 ? 0.27
aResults represent three independent experiments. Each PCR reactionwas
performed in quadruplicate within each experiment.
TABLE 3. Standard-curve real-time PCR analysis comparing avidin ex-
pression between control AI and sperm AI SST.
idin mRNA expressed in turkey SST. Another SAGE tag
sequence (5?-CATGGCATCCAAGG-3?) matched the 3?-
end of chicken GAPDH in the TIGR database. This tag was
not differentially expressed between the two SST SAGE
libraries (61 and 78 counts in control and sperm insemi-
nation libraries, respectively). Therefore, GAPDH was used
as a sample reference control for subsequent validation ex-
periments based on real-time RT-PCR assays.
Confirmation of Differential Avidin Gene Expression
To confirm the differential expression of the avidin gene
in SST mucosal epithelium revealed by SAGE analysis,
real-time PCR was performed using SST RNA from the
sperm and control AI treatments. In this assay, differences
in gene expression are reflected by differences in CTwhere-
by the shorter the CT, the higher the concentration of the
initial specific mRNA template in the sample. The CTfor
avidin amplification was significantly shorter (P ? 9.62E-
11) in the sperm-treated versus control SST samples. No
significant difference was found in the CT values for
GAPDH amplification (Table 2). A standard-curve analysis
was then performed to rule out the possibility that differ-
ential expression of avidin between the treatments was an
artifact of the real-time PCR reaction conditions. This anal-
ysis revealed AE values of 1.46 and 1.70 for GAPDH and
avidin, respectively. Using these values and the CTvalues
obtained for both genes, the difference in avidin expression
between the sperm and control insemination samples was
determined to be approximately 2.8-fold (Table 3).
We constructed two turkey SAGE libraries comprised of
95325 total tags to represent and compare differential gene
expression within sperm and control inseminated SST.
These libraries represented 27430 (?29% of total tags)
unique putative transcripts. This level of representation of
putative unique genes is consistent with SAGE analyses
reported for other model systems [21, 29, 30]. It is inter-
esting to note that only 214 of the 27430 (?1%) putative
genes represented between the sperm inseminated and con-
trol SAGE libraries were differentially expressed between
these treatments; the relative frequencies of the vast ma-
jority of tags were beneath the P ? 0.05 statistical thresh-
old. This 1% of differential gene expression in turkey SST
is similar to SAGE results comparing developing mouse
forelimbs and hindlimbs, in which 1% (317/36300) of the
unique transcripts were differentially expressed . The
lack of detectable RNA in turkey sperm rules out the pos-
sibility that sperm RNA may have contributed to the SAGE
results obtained from the sperm inseminated SST. These
data confirm our initial hypothesis that differential gene ex-
pression occurs in the SST mucosal epithelium by 48 h
after insemination with sperm and thereby raises the com-
pelling possibility that the spermatozoa themselves induce
specific gene expression events required for prolonged
sperm storage in the SST.
A complete SAGE analysis of the human transcriptome
SAGE OF TURKEY SPERM STORAGE TUBULES
has been estimated to require sequencing of approximately
650000 tags . However, in practice, accurate and effi-
cient representation of most transcriptomes (?100 mRNA
copies) is obtained after analysis of far fewer tags, most
commonly in the range of 50000 SAGE tags [21, 31, 32].
In silico analyses indicate that 50000 SAGE tags sufficient-
ly account for more than 98% of significant (P ? 0.05) 2-
fold differences at the 0.1% level of expression (i.e., 1 tag
per 1000 tags sequenced) . Therefore, we established
an initial target size of approximately 50000 tags for each
of the SST SAGE libraries. As depicted in Figure 1, the
new tag accrual rates tapered off and reached a ‘‘steady
state’’ after 35000 tags were sequenced in the sperm and
control AI SST libraries. This steady state suggests that new
tags were being sequenced at the same rate as previously
unique tags became redundant. More than 76% of the total
tags were represented at least twice in each final SST li-
brary (data not shown). Thus, the approximately 50000
tags in each of the current turkey SST SAGE libraries prob-
ably represent the most common, along with many of the
rare, unique genes expressed in the SST transcriptome.
It was not surprising that the initial BLAST comparison
of the turkey SST SAGE tag sequences against the human
SAGEmap database yielded so few tentative annotations.
This low yield of tentative annotations probably reflects
existing sequence variation between avian and human
(mammalian) species at the 3?-ends of mRNA transcripts
from which the SAGE tags originated. Comparing the tur-
key SAGE tag sequences against the TIGR chicken gene
index database yielded more results; however, only approx-
imately 0.5% (2275/490432) of these results were associ-
ated with tentative functional annotations. One should view
these initial database comparisons as a baseline for tenta-
tively annotating the genes represented by the SAGE tags
sequenced in the present study. As more chicken genomic
and cDNA sequence data become available in the future,
serial updating of the database search queries established
in the present study will almost certainly yield more exten-
sive and detailed tag-to-gene annotations. It is worth noting
again the results of these BLAST analyses were obtained
by comparing the turkey SAGE sequences to expressed
gene sequences in the TIGR chicken gene index due to the
lack of turkey sequence information. Because of the poten-
tial sequence variation between chickens and turkeys and
the lack of an automated analysis tool to determine SAGE
tag position relative to the matched chicken sequence (i.e.,
3? vs. 5? proximity), these tentative BLAST matches require
further validation to confirm SAGE tag annotation and gene
Bioinformatic analyses of the SST SAGE tag libraries
against both the TIGR chicken gene index and GenBank
nr databases suggested that one of these differentially ex-
pressed SAGE tags represented the avidin gene. Extending
and amplifying this tag with 5?-RACE combined with nu-
cleotide sequence analysis of the resulting cDNA clone
confirmed that this tag indeed represented the turkey avidin
gene. To our knowledge, turkey avidin had not been cloned
previously. Quantitative real-time PCR analysis corroborat-
ed the SAGE analysis data, indicating that avidin mRNA
expression is up-regulated approximately 3-fold in SST fol-
lowing sperm insemination.
The role, if any, that avidin may play in mediating pro-
longed sperm storage within the SST is not known. Avidin
is a major turkey egg-white protein that binds and seques-
ters biotin within the egg . It is possible that increased
avidin expression in sperm-containing SST may provide the
sperm that are resident within those SST with a nutritional
source of biotin or related vitamins. Induction of avidin
expression is commonly employed as a marker of proges-
terone activity in chick oviduct [35, 36]. Thus, increased
avidin expression in sperm SST may also reflect a potential
linkage between prolonged sperm storage and release and
progesterone fluctuations in laying hens. Our laboratory is
currently performing follow-up studies to localize avidin
expression in the SST epithelium and verify its potential
up-regulation in response to sperm at the cellular level.
The expression levels of multiple cytoskeletal protein
genes appeared to be increased in the sperm AI SST (Table
1). The SST epithelial cells possess an intricate F-actin-rich
terminal web that may play a role in mediating contractile
activity within the SST epithelium . It is possible that
the increased expression of cytoskeletal protein genes in the
sperm AI SST may be indicative of increased contractility
to facilitate sperm selection, storage, and then release from
Several strategies have been developed to elucidate mod-
ulated gene expression. Most of these strategies require the
existence of extensively characterized cDNA libraries to
generate EST information [38, 39]. This is especially crit-
ical for microarray hybridization and PCR-based subtrac-
tive cloning. Even though this information greatly facili-
tates the annotation of SAGE tags, we demonstrated that
species-specific EST sequences are not required to perform
the statistical analysis of SAGE data that leads to the iden-
tification of some genes important to a physiological event
of interest. The analytical steps by which the differential
expression of the avidin gene was confirmed represent a
useful paradigm for taking SAGE tag data from a relatively
uncharacterized model system (i.e., turkey SST) to identify
differentially expressed genes, to confirm identity by nu-
cleotide sequence homology and 5?-RACE, and to validate
differential expression through independent qualitative and
quantitative analyses (e.g., real-time RT-PCR). Our labo-
ratory is currently validating the tentative annotation and
expression levels of an extended group of differentially ex-
pressed SST SAGE tags to establish a working list of can-
didate genes to be progressed together in specific follow-
up experiments investigating avian sperm storage mecha-
nisms. We propose that the present SAGE analysis para-
digm could be extremely useful to investigate differential
gene expression on a transcriptome-wide level in a wide
range of otherwise poorly characterized model systems.
The overall aim of the present study was to establish and
characterize the range of genes expressed in turkey SST
following insemination with or without sperm to identify
specific genes that may enable and regulate prolonged
sperm storage. Our SAGE-based strategy achieved this aim
by yielding quantitative assessments of gene expression in-
dependent of original mRNA sequence knowledge and by
providing the capacity for high-throughput and cost-effi-
cient analysis on a transcriptome-wide level. In addition,
these SAGE libraries provided absolute transcript numbers
in a digital format that can be adapted to provide statistical
comparisons of data from multiple laboratories . It must
be noted that mechanisms other than differential gene ex-
pression (e.g., posttranslational modifications, relocation of
proteins in SST epithelial cells, etc.) also likely are impor-
tant factors in SST-mediated sperm storage. Proteomic anal-
yses of SST-mediated sperm storage will be critical to elu-
cidate these additional mechanisms. The SAGE analyses
conducted in the present study are only the beginning of
an iterative process of hypothesis development, testing, and
LONG ET AL.
refinement that will ultimately yield new insights regarding
the physiologic mechanisms that underlie prolonged sperm
storage in the avian oviduct.
The authors thank Jane Garlow for turkey sperm RNA analyses, Tina
Sphon for assistance in DNA sequencing, Larry Shade and Jill Philpot for
assistance in bioinformatics, and Drs. Laura King and Murray Bakst for
assistance in turkey SST anatomy and collection.
1. Bakst MR. Oviductal sperm selection, transport, and storage in poul-
try. Poult Sci Rev 1994; 5:117–143.
2. Christensen VL. Effect of insemination intervals on oviductal sperm
storage in turkeys. Poult Sci 1981; 60:2150–2156.
3. Hardy MP, Dent JN. Transport of sperm within the cloaca of the fe-
male red-spotted newt. J Morphol 1986; 190:259–270.
4. Hoffman LH, Wimsatt WA. Histochemical and electron microscopic
observations on the sperm receptacles in the garter snake oviduct. Am
J Anat 1972; 134:71–96.
5. Sever DM, Ryan TJ. Ultrastructure of the reproductive system of the
black swamp snake (Seminatrix pygaea): part I. Evidence for oviduc-
tal sperm storage. J Morphol 1999; 241:1–18.
6. Sever DM. Female sperm storage in amphibians. J Exp Zool 2002;
7. Bou-Resli MN, Bishay LF, Al-Zaid NS. Observations on the fine
structure of the sperm storage crypts in the lizard Acanthodactylus
scutellatus hardyi. Arch Biol 1981; 92:287–298.
8. Gist DH, Jones JM. Sperm storage within the oviduct of turtles. J
Morphol 1989; 199:379–384.
9. Brillard JP. Sperm storage and transport following natural mating and
artificial insemination. Poult Sci 1993; 72:923–928.
10. Bakst MR. Structure of the avian oviduct with emphasis on sperm
storage in poultry. J Exp Zool 1998; 282:618–626.
11. King LM, Brillard JP, Garrett WM, Bakst MR, Donoghue AM. Seg-
regation of spermatozoa within sperm storage tubules of fowl and
turkey hens. Reproduction 2002; 123:79–86.
12. Christensen L. Diluents, dilution and storage of turkey semen for six
hours. In: Bakst MR, Wishart GH (eds.), First International Sympo-
sium on the Artificial Insemination of Poultry. Savoy, IL: Poultry
Science Association; 1995:90–106.
13. Wishart GJ. Physiological changes in fowl and turkey spermatozoa
during in vitro storage. Br Poult Sci 1989; 22:443–454.
14. Bakst MR. Oviductal sperm storage in poultry: a review. Reprod Fertil
Dev 1993; 5:595–599.
15. Holm L, Wishart GJ. The effect of pH on the motility of spermatozoa
from chicken, turkey and quail. Anim Reprod Sci 1998; 54:45–54.
16. Wishart GJ, Wilson YI. Temperature-dependent inhibition of motility
in spermatozoa from different avian species. Anim Reprod Sci 1999;
17. Holm L, Dkwall H, Wishart GJ, Ridderstra ˚le Y. Localisation of cal-
cium and zinc in the sperm storage tubules from chicken, quail and
turkey using x-ray microanalysis. J Reprod Fertil 2000; 118:331–336.
18. Holm L, Ridderstra ˚le Y. Localization of carbonic anhydrase in the
sperm-storing regions of the turkey and quail oviduct. Histochem J
19. Yoshimura Y, Koike K, Okamoto T. Immunolocalization of proges-
terone and estrogen receptors in the sperm storage tubules of laying
and diethylstilbestrol-injected immature hens. Poult Sci 2000; 79:94–
20. Velculescu VE, Zhang L, Vogelstein B, Kinzler KW. Serial analysis
of gene expression. Science 1995; 270:484–487.
21. Velculescu VE, Vogelstein B, Kinzler KW. Analysing uncharted tran-
scriptomes with SAGE. Trends Genet 2000; 16:423–425.
22. Bakst MR. Fate of fluorescent stained sperm following insemination:
new light on oviductal sperm transport and storage in the turkey. Biol
Reprod 1994; 50:987–992.
23. King LM, Brillard JP, Bakst MR, Donoghue AM. Isolation of sperm
storage tubules from the uterovaginal junction mucosa of the turkey.
Poult Sci 1999; 78:1044–1047.
24. Audic S, Claverie JM. The significance of digital gene expression
profiles. Genome Res 1997; 7:986–995.
25. Lash AE, Tolstoshev CM, Wagner L, Schuler GD, Strausberg RL,
Riggins GJ, Altschul GD. SAGEmap: a public gene expression re-
source. Genome Res 2000; 10:1051–1060.
26. Quackenbush J, Cho J, Lee D, Liang F, Holt I, Karamycheva S, Parv-
izi B, Pertea G, Sultana R, White J. The TIGR gene indices: analysis
of gene transcript sequences in highly sampled eukaryotic species.
Nucleic Acid Res 2001; 29:159–164.
27. Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ. Basic local
alignment search tool. J Mol Biol 1990; 215:403–410.
28. Vanden Berg A, van der Leij J, Poppema S. Serial analysis of gene
expression: rapid RT-PCR analysis of unknown SAGE tags. Nucleic
Acids Res 1999; 27(e17):i–iii.
29. Menssen A, Hermeking H. Characterization of the c-MYC-regulated
transcriptome by SAGE: identification and analysis of c-MYC target
genes. Proc Natl Acad Sci U S A 2002; 99:6274–6279.
30. Margulies EH, Kardia SLR, Innis JW. A comparative molecular anal-
ysis of developing mouse forelimbs and hindlimbs using serial anal-
ysis of gene expression (SAGE). Genome Res 2001; 11:1686–1698.
31. Velculescu VE, Madden SL, Zhang L, Lash AE, Yu J, Rago C, Lal
A, Wang CJ, Beaudry GA, Ciriello KM, Cook BP, Dufault MR, Fer-
guson AT, Gao Y, He TC, Hermeking H, Hiraldo SK, Hwang PM,
Lopez MA, Luderer HF, Mathews B, Petroziello JM, Polyak K, Zawel
L, Kinzler KW. Analysis of human transcriptomes. Nat Genet 1999;
32. Anisimov SV, Tarasov KV, Tweedie D, Stern MD, Wobus AM, Bohler
KR. SAGE identification of gene transcripts with profiles unique to
pluripotent mouse R1 embryonic stem cells. Genomics 2002; 79:169–
33. Man MZ, Wang X, Wang Y. Power SAGE: comparing statistical tests
of SAGE experiments. Bioinformatics 2000; 16:953–959.
34. White HB III. Biotin-binding proteins and biotin transport to oocytes.
Ann N Y Acad Sci 1985; 447:202–211.
35. Niemela ¨ AO, Elo HA. Effects of oestradiol-17? and diethylstilboestrol
on progesterone-induced protein (avidin) production in chick oviduct:
evidence for differences in the actions of steroidal and non-steroidal
oestrogens. J Endocrinol 1983; 96:465–469.
36. Hora J, Gosse B, Rasmussen K, Spelsberg TC. Estrogen regulation of
the biological activity of the avian oviduct progesterone receptor and
its ability to induce avidin. Endocrinology 1986; 119:1118–1125.
37. Freedman S, Akuffo V, Bakst MR. New evidence for the innervation
of the sperm storage tubules in the turkey (Meleagris gallopavo). Re-
production 2001; 121:809–814.
38. Marra MA, Hillier L, Waterston RH. Expressed sequence tags—Es-
tablishing bridges between genomes. Trends Genet 1998; 14:4–7.
39. Kozian DH, Kirschbaum BJ. Comparative gene-expression analysis.
TIBTECH 1999; 17:73–78.