Theor Appl Genet (2002) 104:804–812

DOI 10.1007/s00122-001-0795-y

D. Roche · J. A. Conner · M. A. Budiman · D. Frisch

R. Wing · W. W. Hanna · P. Ozias-Akins

Construction of BAC libraries from two apomictic grasses to study

the microcolinearity of their apospory-specific genomic regions

Received: 25 March 2001 / Accepted: 20 June 2001 / Published online: 6 February 2002
© Springer-Verlag 2002

Abstract We have constructed bacterial artificial chro- Keywords Apomixis · Apospory · BAC library ·
mosome (BAC) libraries from two grass species that Colinearity · Pennisetum
reproduce by apospory, a form of gametophytic apomixis.
The library of an apomictic polyhaploid genotype (line
MS228-20, with a 2C genome size of approximately Introduction
4,500 Mbp) derived from a cross between the obligate
apomict, Pennisetum squamulatum, and pearl millet Apomixis is a naturally occurring mode of reproduction
(P. glaucum) comprises 118,272 clones with an average in which plants, regardless of their level of heterozygosity,
insert size of 82 kb. The library of buffelgrass (Cenchrus clonally propagate themselves through seeds (Asker and
ciliaris, apomictic line B-12-9, with a 2C genome size of Jerling 1992). The understanding as well as the control
approximately 3,000 Mbp) contains 68,736 clones with of the mechanisms underlying apomixis could have a
an average insert size of 109 kb. Based on the genome remarkable impact on crop breeding and seed production
sizes of these two lines and correcting for the number for (Hanna 1995). We have concentrated our research efforts
false-positive and organellar clones, library coverages on one form of gametophytic apomixis, pseudogamous
were found to be 3.7 and 4.8 haploid genome equivalents apospory, in which unreduced embryo sacs originate
for MS 228-20 and B12-9, respectively. Both libraries from somatic nucellar cells (Nogler 1984). The 2n egg
were screened by hybridization with six SCARs cell in each embryo sac is not fertilized by a sperm cell
(sequence-characterized amplified regions), whose tight but nevertheless develops parthenogenetically to form an
linkage in a single apospory-specific genomic region had embryo whose genotype is identical to that of the seed-
been previously demonstrated in both species. Analysis bearing plant. We have recently shown that a single apo-
of these BAC clones indicated that some of the SCAR spory-specific genomic region (ASGR) is sufficient for
markers are actually amplifying duplicated regions the expression of apospory in two grasses, Pennisetum
linked in coupling in both genomes and that restriction squamulatum Fresen and Cenchrus ciliaris L. [syn.
enzyme mapping will be necessary to sort out the dupli- P. ciliare (L.) Link; buffelgrass], both of African origin
cations. (Ozias-Akins et al. 1998; Roche et al. 1999). ASGR-
linked molecular markers are conserved between both
Communicated by D. Hoisington species (Lubbers et al. 1994; Roche et al. 1999). Unfor-
tunately, recombination between these markers has not
D. Roche · W.W. Hanna been detected, thereby hampering the prospect of finer
Crop Genetics and Breeding, USDA-ARS, genetic mapping in order to clone the gene(s) for apo-
Coastal Plain Experiment Station, Tifton, GA 31793, USA
spory using positional information. Given the conserva-
J.A. Conner · P. Ozias-Akins (✉) tion of markers within the ASGR of both species, we
Department of Horticulture, University of Georgia – decided to analyze the microcolinearity of this region in
Tifton Campus, Tifton GA 31793-0748, USA
e-mail: Ozias@tifton.cpes.peachnet.edu order to identify DNA regions highly conserved between
Fax: +1-229-3863356 the two species. Analysis of gene content within the con-
M.A. Budiman · D. Frisch · R. Wing
served regions of the ASGR from both species would
Clemson University Genomics Institute, Jordan Hall, Clemson, provide candidate gene targets. A necessary prerequisite
SC 29634, USA for high-resolution comparative analysis at the structural
Present address: level is the construction of large-insert genomic libraries.
D. Roche, Plants, Soils and Biometerology Department, For all organisms, bacterial artificial chromosome
UMC 4820, Utah State University, Logan, UT 84322, USA (BAC) libraries have become the most efficient tool by
 805

which to clone whole genomes since the demonstration stirred on ice for 15–20 min. The homogenate was successively
by Shizuya et al. (1992) that a bacterial plasmid could be filtered through three nylon sieves of 295, 105 and 60 µm into
250-ml centrifuge bottles and spun at 1,800 g for 10 min. The
used to clone large DNA inserts. Insert sizes often are in supernatant was discarded, and the pellet was gently resuspended
the average range of 80–150 kb, thus three to four times with a small paintbrush in 30 ml of homogenization buffer, then
larger than those of cosmid libraries (25–45 kb) (Sambrook filtered through Miracloth (CN Biosciences, Darmstadt, Germany)
et al. 1989). Although BAC inserts are significantly when necessary to remove clumps of nuclei. A light spin at 60 g
for 2 min was used to further remove nuclei clumps. The supernatant
smaller than those in yeast artificial chromosomes (YAC was spun at 2,000 rpm for 5 min to collect the nuclei, and the
libraries; average of 500 kb) (Burke et al. 1987), the pro- pellet was gently resuspended in 15 ml of cold homogenization
duction of BAC libraries, as well as their utilization, buffer. Three further nuclei washes were conducted with the same
present many advantages over YAC libraries, including conditions for centrifugation and resuspension. All filtrations and
the ease of DNA isolation, the low frequency of chime- centrifugations described above were carried out at 0 °–4 °C. The
final nuclei pellet was warmed at 45 °C and resuspended at a final
ras, and insert stability (Cai et al. 1995). We document concentration of 107 nuclei per milliliter with an equal volume of
here the construction of BAC libraries from an apomictic 1.6% Incert Agarose (FMC, Rockland, Md.) in homogenization
polyhaploid line derived from an F1 hybrid between buffer prewarmed at 45 °C. Nuclei-agarose plugs of 100 µl each
pearl millet [Pennisetum glaucum (L.) R. Br.] and were made on ice in a plastic mold.
P. squamulatum (Dujardin and Hanna 1986) and an apo-
mictic buffelgrass (Cenchrus ciliaris). Preparation of cloning vector
The single-copy BAC vector, pBeloBAC11, was prepared as
described by Woo et al. (1994) omitting purification on a cesium
Materials and methods chloride density gradient.

Plant material
Size-selection and cloning of megabase DNA
Apomictic line MS228-20 was germinated from seeds of an open-
pollinated apomictic polyhaploid F1 line derived from a cross Nuclei in agarose plugs were lysed for 48 h, and then plugs were
between Pennisetum glaucum (pearl millet) and P. squamulatum washed and equilibrated in buffer as described by Zhang et al.
(Dujardin and Hanna 1986), hereafter referred to as simply “poly- (1995). Agarose plugs were divided into thirds, and then each of
haploid”. The presence of the ASGR in the polyhaploid was con- the approximately 30-µl segments was finely chopped and equili-
firmed by the polymerase chain reaction (PCR) using sequence- brated on ice for 45 min in HindIII reaction buffer. Megabase
characterized amplified regions (SCARs) as outlined in Ozias- DNA was digested by adding 2.5–5 U HindIII restriction enzyme
Akins et al. (1998). Other sexual and apomictic F1 individuals (NEB, Beverly, Mass.) in a total volume of 50 µl, incubating the
(290-105, 290-124) from the mapping population (Ozias-Akins et complete reaction mix on ice for 45 min, followed by incubation
al. 1998) were used during restriction fragment length polymor- for 20 min at 37 °C. Partially digested DNA was electrophoretically
phism (RFLP) analysis. Apomictic line B-12-9 of bufflegrass separated on 0.9% agarose CHEF gels in 0.5×TBE at 12 °–15 °C
(Cenchrus ciliaris) was provided by R.T. Sherwood and D.L. Gus- (first size selection: 17 h, 5.0 V/cm, constant switch at 75 s;
tine, USDA-ARS, University Park, Penn. The pedigree of this line second size selection: 15 h, 5.0 V/cm, constant switch at 5 s). Gel
can be found in Sherwood et al. (1994). A sexual buffelgrass gen- slices containing DNA of different molecular weights were elec-
otype, B-2s, received from the same source, had been previously troeluted in 1×TAE at 4 °C for 2 h in dialysis membranes
used as the sexual parent in segregating crosses to map the apo- (12,000–14,000 mwco). Following gel quantification of an ethidium
mixis locus (Gustine et al. 1997; Roche et al. 1999). Plants from bromide-stained aliquot of the electroeluted DNA, 80–150 ng of
species and hybrids were propagated vegetatively for several plant DNA was mixed with 20 ng of HindIII-cut and dephospho-
years. DNA isolations were conducted as previously described rylated BAC vector and ligated at 16 °C overnight in a total
(Ozias-Akins et al. 1998). volume of 100 µl in the presence of 10 U T4 DNA ligase and under
the manufacturer’s reaction buffer conditions (Promega, Madison,
Wis.). Following a desalting incubation of 2 h on ice in a 1.5-ml
Nuclear DNA content analysis of different germplasm lines microcentrifuge tube containing 1.0 ml 100 mM glucose and 1%
LE agarose (FMC), 1.0 µl from each ligation was used to trans-
A piece of young leaf was chopped with a razor blade in buffer form 20 µl of competent cells (electrocompetent DH10B, Life
according to Otto (1994). The chopped tissue was diluted with Technologies, Rockville, Md.). Electroporation was carried out
3 ml of 0.4 M Na2 HPO4 containing the DNA-specific fluoro- using an EC600 apparatus (BTX Corp, San Diego, Calif.) at
chrome DAPI (0.2 mg/100 ml) and filtered through a 40-µm sieve. 1.5 kV, 129 ohms, a pulse of 4.85–4.95 ms, and 1-mm gap cuvettes.
Suspended nuclei (10,000 per sample) were analyzed on a PAS-III Electroporated cells were immediately diluted in 1 ml of SOC
flow cytometer (Partec, Munster, Germany) equipped with a medium (Sambrook et al. 1989) and incubated at 200 rpm at 37 °C
100-W high-pressure mercury lamp. for 1 h. Dimethylsulfoxide (DMSO) was added to the cells (7%
v/v final concentration) before freezing at –80 °C. The addition of
DMSO maintained the viability of the transformed cells for several
BAC library construction months (data not shown). Hence, loss of titer was conveniently
prevented during clone sizing and characterization of each liga-
Isolation of nuclei for preparation of megabase DNA tion. Transformed cells were plated on 200 ml of selective medi-
um (LB; Luria-Bertani medium with 15 µg/ml chloramphenicol,
Plants were etiolated for 72 h, after which fresh leaves and stems 0.55 mM IPTG and 80 µg/l of X-gal) poured into Q-trays (Genetix,
were harvested to isolate nuclei for megabase DNA preparation. Queensway, UK). Following two 24-h incubations in the dark
Ten to twenty grams of tissue was ground in liquid nitrogen in (first 24-h period at 37 °C, second at room temperature), white
2- to 3-g aliquots. Liquid nitrogen was never added to partially recombinant colonies were picked robotically (Q-Bot, Genetix)
ground tissue to prevent nuclei damage. A single homogenization and stored in 384-well microtiter plates (Genetix) filled with 65 µl
buffer (Zhang et al. 1995) containing 0.5% triton X-100 was used of freeze broth (Woo et al. 1994) per well. Recombinant colonies
throughout the nuclei preparation. The powdered tissue was mixed which were avoided by the robot because they were either too
with homogenization buffer (10 ml per gram of tissue) and slowly close to other colonies or to the tray edge were picked manually
806
Table 1 Nuclear DNA contents
estimated from DAPI-stained 2C DNA content (Mbp) Number of haploid
nuclei genome equivalents
Present estimate Previous estimates in BAC libraries

Oryza sativa 980c

a Marie and Brown (1993) Pennisetum glaucum (2×) 3,900 2,150a; 4,600b; 4,700–5,200c
b Martel et al. (1997) Pennisetum glaucum (4×) 7,800
c C-value database at Pennisetum squamulatum 10,300 9,400c
www.rbgkew.org.uk F1 290-124 9,300
d Extrapolated from the
Polyhaploid 4,500 4,700d 3.6–3.8
literature values for P. glaucumb Cenchrus ciliaris 3,000 2,600c 4.8–5.5
and P. squamulatumc

with sterile disposable pipette tips. Microtiter plates were incubated

overnight at 37 °C and robotically (Q-Bot) duplicated for storage Results
in different freezers.
Analysis of genome sizes
Characterization of BAC clones
The genome size of Pennisetum glaucum [syn. P. ameri-
Individual BAC clones were grown overnight at 37 °C in a total canum (L.) Leeke and P. typhoides (Burm.) Stapf et
volume of 3 ml LB medium containing 15 µg/ml chlorampheni- Hubb.] has been estimated previously to range from
col. Isolation of the plasmid was carried out using a standard 2,100 Mbp to 5,200 Mbp (Marie and Brown 1993;
alkaline-lysis method (Sambrook et al. 1989). The final pellet was Bennett and Leitch 1997; Martel et al. 1997). Because of
dissolved in a total of 15 µl of TE buffer, from which 5 µl was
used for each restriction enzyme digestion prior to electrophoretic the variability reported in the literature, we chose to carry
analysis. out our own estimates of DNA content using flow cytom-
etry of DAPI-stained nuclei with rice (Oryza sativa cv.
Lemont) as an internal standard for diploid pearl millet.
Screening of BAC libraries
Table 1 shows the DNA contents of all materials estimated
High-density colony filters were prepared with the Genetix Q-Bot in this study based on their 2C peaks of relative fluores-
as described in Tomkins et al. (1999a). Six sets of six and four cence intensity. DNA contents of both buffelgrass geno-
filters each were made for the polyhaploid and buffelgrass libraries, types (B-2s and B-12-9) were the same, as estimated
respectively. Radioactive decay (not stripping) was used between
two successive hybridizations of the same filter. Radiolabeling of using diploid pearl millet as the standard for comparison.
probes was done by PCR-incorporation of [32 P] using SCAR Genome sizes of the polyhaploid and P. squamulatum
primers and SCAR clones as template DNAs (Ozias-Akins et al. were estimated using tetraploid pearl millet as the stan-
1998). Hybridization and identification of addresses for positive dard. Tetraploid pearl millet showed exactly twice the
clones were performed as indicated at www.genome.clemson.edu/ DNA content of diploid pearl millet when both were run
groups/bac/protocols.
together. Since the genome size of P. squamulatum is
approximately 10 billion bp it would have been a major
Chloroplast and mitochondrial DNA probes undertaking to obtain a representative BAC library of this
species. Thus, we chose a 21-chromosome polyhaploid
A segment of chloroplast (cp) DNA was amplified by PCR from
total DNA of P. squamulatum and buffelgrass with the primer line that was spontaneously derived from an apomictic F1
combination trnH and trnK (Demesure et al. 1995). Both species individual (Dujardin and Hanna 1986) and was shown to
yielded similar 1,900-bp products that were radiolabeled. Total contain all markers for the ASGR from P. squamulatum.
mitochondrial (mt) DNA from P. glaucum was isolated from a cell This line contains seven pearl millet (1×) and 14
culture line (Ozias-Akins et al. 1987) and labeled with [32 P] by
random priming. P. squamulatum (1.5×) chromosomes, one of which
represents the apospory linkage group. Upon screening of
the reproductive phenotype and analysis of ASGR-linked
Fingerprinting analysis and contig assembly molecular markers (data not shown), the polyhaploid line
A subset of the clones identified in the library screens were finger-
was found to display similar results to any other apomictic
printed at CUGI (Clemson University Genomics Institute). From F1 individual produced by the cross between tetraploid
the polyhaploid library this included: 3-A14M clones (all ASGR- pearl millet and P. squamulatum (Ozias-Akins et al.
linked), 8-Q8M clones (all ASGR-linked; 5 PCR-positive, 3 PCR- 1998). Thus, in the polyhaploid, one copy of the ASGR is
negative), 7-UGT197 clones (7 ASGR-linked), and 9-O7M clones represented in only half the genome size contained in any
(6 ASGR-linked, 3 not ASGR-linked). From the buffelgrass libra-
ry this included: 7-Q8M clones (all ASGR-linked; 4 PCR-positive, F1 individual, and the number of BAC clones that are
3 PCR-negative), 7-UGT197 clones (all ASGR-linked), and 6-O7M required for 1×genome coverage is reduced by half. Even
clones (none ASGR-linked). BAC clones were digested with though the ASGR is simplex in these plants, other loci in
HindIII, run on an agarose gel, and imaged at CUGI. Band calling the polyhaploid or buffelgrass libraries are likely to be
and contig building were accomplished using the Image (version
3.10, Sanger Centre, UK) and FPC (version 4.6, CUGI) software represented by multiple alleles; thus, for the calculation of
with Linux. Parameters used in the FPC analysis were a fixed genome coverage we followed the standard of 1×coverage
tolerance value of 7 and a cutoff score of 10–12. being equivalent to a 1C value.
 807
Fig. 1 Analysis of BAC clones
by PFG electrophoresis.
Ethidium bromide-stained
CHEF gels of randomly picked
recombinant BAC clones from
the buffelgrass (A) and poly-
haploid (B) libraries digested
with NotI. The first and last
lanes of both gels contain a
lambda concatemer marker
(NEB). Fragment sizes (in
kilobases) are indicated on the
right-hand side of each panel

Construction and characterization of the libraries positives, chloroplast, and mitochondrial clones, as well
as present and previous estimates of genome sizes, the
The polyhaploid library contains 118,272 clones with an coverage of the polyhaploid genome is 3.6–3.8 haploid
average insert-size of 82 kb (n=l00), and 12% of these genome equivalents and the buffelgrass genome is
clones are false-positives (clones with no insert). The 4.8–5.5 haploid genome equivalents. Random samples of
buffelgrass BAC library includes 68,736 clones with an BAC clones from both species are shown in Fig. 1.
average insert size of 109 kb (n=l05) and a small fraction The construction of libraries from these two grasses
(3%) of false-positives. To estimate the representation of was challenging. For the polyhaploid line, a relatively
cpDNA in both libraries we hybridized one high-density non-vigorous, vegetatively propagated plant, we per-
filter from each library (18,432 clones) with a 1,900-bp formed more than 200 ligations with 20–30 different
fragment amplified between the trnH and trnK primers size selections of megabase DNA and were never
(Demesure et al. 1995). Correcting for the insert size of successful at exceeding 92 kb as the average insert-size.
the libraries and for an approximate chloroplast genome A majority of the ligations resulted in few total colonies
size in P. glaucum of 120 kb (Smith et al. 1987), an esti- with a high representation of white, false-positive
mated 0.1% and 0.7% of clones are of chloroplast origin clones containing no inserts (up to l00%, variable with
in the polyhaploid and buffelgrass libraries, respectively. different cloning attempts). Analysis of some false-
The proportion of mitochondrial clones were estimated positive clones with digestion by restriction enzymes
at 0.2% and 0.5% in polyhaploid and buffelgrass and gel electrophoresis revealed partial degradation of
libraries, respectively. Taking into consideration false- the cloning vector (Fig. 2). Deletions of 1–2 kb of the
808

In the case of the polyhaploid, all six markers had

been mapped in the total F1 population as PCR products,
therefore, PCR alone was initially used to assign clones
to the ASGR. Out of 58 cross-hybridizing clones from the
polyhaploid library, 25 (43%) showed amplification with
the respective SCAR primers (Table 2). For Q8M, however,
only 5 out of 12 clones showed PCR amplification of the
SCAR marker even though Q8M was previously shown
by RFLP analysis to be hemizygous; i.e., to hybridize at
high stringency only to apomictic F1 s (3 DraI fragments)
and not to any fragments in sexual F1 s (Ozias-Akins et
al. 1998). This prior result suggested that there might be
RFLP markers that did not correspond to PCR-amplified
SCAR markers but were nevertheless linked with apospory.
Fig. 2 Degradation of HindIII-cut and dephosphorylated pBelo- By RFLP analysis of BACs, we were able to assign three
BAC11. Lanes: 1 PstI-digested lambda DNA, 2 intact pBelo- additional polyhaploid, Q8M-hybridizing clones to the
BAC11 vector (isolated from a blue colony) cut with SalI and
yielding the predicted size fragments of 6,384 bp, 843 bp, and 280 ASGR based on their content of an apospory-linked, but
bp (arrows), 3–7 individual false-positive clones digested with PCR-negative RFLP fragment.
SalI; the expected 280-bp fragment that contains the HindIII cloning In buffelgrass, only two types of BAC clones
site was absent in these clones, while the adjacent 843-bp fragment (UGT197, Q8M) were further characterized with SCAR
was present in only lanes 3 and 6. Deletions of 1–2 kb of the BAC
vector were apparent in lanes 4, 5 and 7–9. Lane 10 undigested primers since the remaining SCAR markers (A10H, C4,
pBeloBAC11 C16, O7M) were not apomict-specific and had been
previously mapped as RFLPs (Roche et al. 1999). BACs
isolated with these latter four markers required an analysis
total BAC vector were apparent in four out of seven of RFLP content in which the polymorphic RFLP frag-
clones. ments previously mapped were used to determine which
BAC clones could be assigned to the ASGR. For example,
a total of ten BAC clones from the buffelgrass library
Identification of ASGR-linked BAC clones were isolated with the SCAR C4 probe; however, only
two of these hybridized to the same restriction fragments
Both BAC libraries were screened with six [32 P]-labeled that cosegregated with the trait of apospory in buffel-
ASGR-linked SCAR probes. However, only five out of grass (Fig. 3). The larger of the two cosegregating RFLP
six SCARs were used on both libraries. SCAR A14M was fragments was also present in P. squamulatum, the poly-
utilized as a probe onto the polyhaploid library but not on haploid line and the apomictic F1 (P. glaucum ×
the apomictic buffelgrass in which this molecular marker P. squamulatum) 290-124, but not in the sexual F1 290-105
is absent (unpublished results). C16, an ASGR-linked (Fig. 3). Out of 83 buffelgrass clones, 28 (33%) were
RFLP marker in buffelgrass, could not be mapped in an F1 assigned to the ASGR based on their content of mapped
(P. glaucum × P.squamulatum) population segregating for markers. Similar to what we observed in the polyhaploid
mode of reproduction (unpublished results), thus it was with Q8M BACs, one group of BACs from buffelgrass
not used to screen the polyhaploid library. The SCARs was not PCR-positive but could be assigned to the
were able to identify 58 and 83 weakly to strongly cross- ASGR based on its content of one of two RFLP fragments
hybridizing clones in polyhaploid and buffelgrass cosegregating with apospory. In both species, apospory-
libraries, respectively (Table 2). All BAC clones were linked RFLPs for Q8M could be separated into distinct
characterized further by screening for the presence of the BAC clones, one of which was SCAR PCR-positive and
respective marker shown to be linked with apospory. one of which was SCAR PCR-negative.

Table 2 Screening of BAC

libraries from two apomictic Polyhaploid BAC library Buffelgrass BAC Library
grasses with probes mapped to
the apospory-specific genomic ASGR-specific Total BAC ASGR ASGR Total BAC ASGR ASGR
region (ASGR). In all cases marker clones PCR-positive RPLF-positive clones PCR-positive RFLP-positive
labeled probes were generated
by PCR using [32 P]-dCTP, A10H 5 2 – 12 – 1
SCAR primers, and respective A14M 3 3 – N/A N/A N/A
SCAR plasmid as template DNA. C4 10 1 – 10 – 2
Analysis and picking of posi- C16 N/A N/A N/A 17 – 0
tively hybridizing clones were O7M 19 6 – 22 – 7
done following two washes of Q8M 12 5 3 15 5 6
30 min each at 65 °C in 0.5× UGT197 9 8 – 7 7 N/A
and 0.1× SSPE plus 1% SDS, Total 58 25 3 83 12 16
respectively (N/A not available)
 809

Fig. 3 Analysis of the RFLP content of C4-BAC clones relative to

C4-hybridizing restriction fragments in genomic DNA. Southern-
hybridization with [32 P]-labeled C4 SCAR to DraI-cut genomic
DNA of B-12-9 [apomictic (A)], B-2s [sexual (S)], an apomictic
and sexual F1 (290-124 and 290-105, respectively), polyhaploid
(poly), P. squamulatum (Ps) and ten DraI-cut C4 BAC clones
(lanes 1–10) isolated from the buffelgrass library. Aliquots of
12.5 µg of total genomic DNA and 0.5 µg of each BAC clone were
used per lane. Two polymorphic DNA fragments (arrowheads)
linked to apospory in buffelgrass are found in the BAC clones of
lanes 2 and 6

Polyhaploid and buffelgrass clones from Table 2 were

grouped into SCAR-specific pools and further tested
with SCAR markers that could not be used in filter Fig. 4 Southern analysis of RFLP and homology conservation of
hybridization due to repetitive elements within the Q8M PCR-negative BAC clones from buffelgrass and polyhaploid
SCAR. Six SCARs (P16R, R13, U12H, V4, W10M, libraries. Plasmid extracts from six Q8M PCR-negative BAC
clones from buffelgrass and polyhaploid libraries (prefixed C and P,
X18R) (Ozias-Akins et al. 1998) were used to screen the respectively) were digested with EcoRI and separated by regular
polyhaploid pools, while four SCARs (P16R, U12H, V4, agarose gel electrophoresis (left). Southern analysis was carried
X18R) were screened against the buffelgrass pools since out with a [32 P]-labeled C2 BAC (right). The first lane contains a
only these four had been mapped to the ASGR in buffel- l-kb ladder (NEB)
grass (Roche et al. 1999). No amplification of these
SCARs was observed on the BAC-clone pools (data not
shown). From this initial screening in both species, no limit false associations of clones. A single contig was
two SCARs were found in any one BAC clone, which created for the three polyhaploid A14M clones. The
suggests that the ASGR is likely to be larger than 1 Mbp seven polyhaploid UGT197 clones created two contigs
in size. of two clones each and three singletons. The seven
buffelgrass UGT197 clones created two contigs of three
and four clones. The six ASGR-linked polyhaploid
Fingerprint analysis of BAC clones O7M clones created two contigs of two and three clones
and one singleton. Interestingly, the Q8M clones from
In order to determine relationships between clones the two libraries were contained within three contigs
identified with various probes, a subset of ASGR-linked with one singleton. Six Q8M PCR-negative clones
and unlinked clones were fingerprinted. Twelve contigs (three from each species) were grouped together in the
were assembled with seven clones remaining singletons. same contig, suggesting that this region is highly
The contigs contained from two to six clones each. No conserved between the two species. These BAC clones
contig assembled contained BAC clones from different were digested with EcoRI and probed with total labeled
SCAR markers, indicating that there was no substantial C2 BAC, which demonstrated that most EcoRI sites and
overlap between clones from different markers. The homology were conserved between the two species
nine clones not shown to be linked with the ASGR fell (Fig. 4). The other two Q8M contigs contained
into two contigs with one singleton. The separation the PCR-positive clones from the two libraries. Again,
of ASGR-linked and-unlinked clones verified that one of these contigs contained clones from both
the conditions for contig assembly were strict enough to libraries.
810

clones has been previously attributed to the storage

conditions of the dephosphorylated vector (Danesh et al.
1998) or to electroporation conditions (Lin et al. 1999).
Since we used the same batch of frozen vector for suc-
cessful and unsuccessful ligations and kept electropora-
tion conditions constant across experiments, it is likely
that additional factors played a role in the generation of
false-positive clones. At 3%, the percentage of false-
Fig. 5 Southern analysis of UGT197 PCR-positive BAC clones positive clones in the buffelgrass library was similar to
from buffelgrass and polyhaploid libraries. Plasmid extracts from that reported in soybean (Tomkins et al. 1999a), sugarcane
six UGT197 PCR-positive BAC clones from each of the buffelgrass
and polyhaploid libraries (prefixed C and P, respectively) were
(Tomkins et al. 1999b), and tomato (Budiman et al.
digested with NotI and separated by pulsed-field electrophoresis. 2000) libraries. The 12% false-positive clones in the
Southern analysis was carried out with a [32 P]-labeled UGT197 polyhaploid library fell in the range of the 7–16% reported
SCAR. Two distinct classes of BACs were recovered from the for rice (Wang et al. 1995), potato (Song et al. 2000),
buffelgrass library and three classes from the polyhaploid library. and Medicago truncatula (Nam et al. 1999). The BAC
Sizes, in kilobases are indicated on the right-hand side
inserts from both buffelgrass and polyhaploid libraries
typically produced more than one NotI restriction frag-
Duplication of ASGR markers in polyhaploid ment, which is consistent with previous reports of a
and buffelgrass higher frequency of NotI sites in GC-rich grasses (Woo
et al. 1994; Wang et al. 1995; Moullet et al. 1999;
The identification of multiple contigs for individual Tomkins et al. 1999b) compared with most dicots (Frijters
SCAR markers was verified for the UGT197 clones et al. 1997; Tomkins et al. 1999a; Budiman et al. 2000).
through the use of RFLP data. Six buffelgrass and six Over 50 BAC clones that could be assigned to the
polyhaploid ASGR-linked clones were digested with ASGR were isolated from the two libraries after hybrid-
NotI and probed with the SCAR marker. Distinct NotI ization of the high-density filters with six low-copy
patterns were identified for both species (Fig. 5). None SCAR clones. BAC clones were assigned to the ASGR
of the SCAR-hybridizing NotI fragments represented an in the polyhaploid library (25 or 5.0 per probe from an
insert-vector junction-fragment; therefore, the different approximately 4× genome coverage after excluding the
size classes could be explained by linked duplications. A14M clones that had no counterpart from buffelgrass)
This result demonstrates that there are at least two and in the buffelgrass library (28 or 5.6 per probe from
distinct regions of DNA in buffelgrass that correspond to an approximately 5× genome coverage). The representa-
the two FPC contigs and at least two, maybe more, tion of a probe in our libraries did not differ substantially
distinct regions for the polyhaploid. Analysis with FPC from the two to eight clones per probe recovered from
software allowed the positioning of P2 and P3 within the two soybean libraries, each with 4.7 haploid genome
same contig. P1 was not fingerprinted. Upon FPC analysis, equivalents (Meksem et al. 2000). The larger total number
P4 and P5 were not placed in the same contig, although of clones from buffelgrass (83 vs. 58 from polyhaploid)
their NotI restriction patterns were similar. could be due to the slightly greater genome coverage of
the library, but also to the larger number of unmapped,
multiplex RFLP fragments detected in genomic DNA
Discussion that could represent allelic, linked, or unlinked sequences
(Ozias-Akins et al. 1998; Roche et al. 1999).
We constructed BAC libraries from two grass species Here we provided examples (for SCARs – C4,
that reproduce by apospory. Both of these libraries have UGT197, Q8M) of the feasibility of using BAC clones to
a low (0.3–1.2%) representation of organelle DNA study the microcolinearity of the ASGR in both species.
among the clones. This low frequency of c1ones with BAC clones conserved partially or totally between the
cpDNA and mtDNA may be attributed to the use of two species may harbor the gene(s) of interest for this
nuclei as the megabase DNA source rather than protoplasts mode of reproduction. However, at the onset of library
and also to the incorporation of 0.5% triton X-100 in construction, and based on our molecular marker mapping
successive steps throughout the preparation of nuclei. in both species, we did not expect to find as many classes
Usually fewer than 2% of the clones resulting from of BAC clones and separate DNA contigs for each
nuclei preparations contain cpDNA (Wang et al. 1995; SCAR. It seems that the ASGR in both species comprises
Marek and Shoemaker 1997; Tomkins et al. 1999a, b; several duplicated regions. We have previously docu-
Nam et al. 1999; Budiman et al. 2000). In the construc- mented that the ASGR is partially hemizygous in nature
tion of the polyhaploid BAC library we encountered high and lacks genetic recombination (Ozias-Akins et al.
levels of false-positive clones found to have no inserts 1998). Furthermore, its nucleotide divergence for non-
upon plasmid isolation and analysis. In the study reported coding nuclear DNA regions between the two species is
here, we showed that partial degradation of the BAC similar to that observed between non-coding regions of
vector was responsible for the “false-positive” nature of the chloroplast genome in the same two species (Roche
the transformed cells. The occurrence of false-positive et al. 1999). Some of the ASGR-linked markers have
 811

also been found in other Pennisetum species (Lubbers et Budiman MA, Mao L, Wood TC, Wing RA (2000) A deep-
al. 1994 and unpublished). It is possible that the ASGR coverage tomato BAC library and prospects toward development
of an STC framework for genome sequencing. Genome Res
was organized prior to speciation within the Pennisetum/ 10:129–136
Cenchrus complex. Considering that apomixis may Burke DT, Carle GF, Olson M (1987) Cloning of large fragments of
significantly increase the fitness of progenies issued exogenous DNA into yeasts by means of artificial chromosome
from interspecific crosses (Harlan and de Wet 1963), the vectors. Science 236:806–812
Cai L, Taylor JF, Wing RA, Gallagher DS, Woo SS, Davis SK
ASGR may have moved laterally through different grasses (1995) Construction and characterization of a bovine bacterial
of sub-Saharan Africa, the geographical region from artificial chromosome library. Genomics 29:413–425
which Pennisetum squamulatum and Cenchrus ciliaris Danesh D, Penuela S, Mudge J, Denny RL, Nordstrom H, Martinez
originated. JP, Young ND (1998) A bacterial artificial chromosome library
To determine if any of the remaining mapped SCAR for soybean and identification of clones near a major cyst
nematode resistance gene. Theor Appl Genet 96:96–202
markers could be located within any of the isolated BAC Demesure B, Sodzi N, Petit RJ (1995) A set of universal primers
clones, each pool of BAC clones was tested by PCR with for amplification of polymorphic non-coding regions of
the other SCARs. None of these PCR analyses was posi- mitochondrial and chloroplast DNA in plants. Mol Ecol 4:
tive, indicating that none of the isolated clones contained 129–131
Dujardin M, Hanna WW (1986) An apomictic polyhaploid obtained
more than one SCAR marker. Before these analyses we from a pearl millet × Pennisetum squamulatum apomictic
considered the possibility that the ASGR may be a rela- interspecific hybrid. Theor Appl Genet 72:33–36
tively small, non-recombining genomic region. In this Frijters ACJ, Zhang Z, van Damme M, Wang GL, Ronald PC,
case, a bulked-segregant analysis in extremely heterozy- Michelmore RW (1997) Construction of a bacterial artificial
chromosome library containing large EcoRI and HindIII
gous F1 progenies may have efficiently saturated the genomic fragments of lettuce. Theor Appl Genet 94:390–399
region with molecular markers (Ozias-Akins et al. 1998). Gustine DL, Sherwood RT, Huff DR (1997) Apospory-linked
However, our current findings in the polyhaploid line molecular markers in buffelgrass. Crop Sci 37:947–951
indicate that at least 28 ASGR-assigned BAC clones do Hanna WW (1995) Use of apomixis in cultivar development. Adv
not include more than one SCAR marker. The amount of Agron 54:333–350
Harlan JR, de Wet JMJ (1963) Role of apomixis in the evolution
overlap among these 31 clones is not finalized as the end- of the Bothriochloa-Dichanthium complex. Crop Sci 3:
clone analysis is in progress and end clones are being 314–316
used for walking. However, with an average insert six of Jackson SA, Wang ML, Goodman HM, Jiang J (1998) Application
82 kb it is likely that the ASGR, as previously defined of fiber-FISH in physical mapping of Arabidopsis thaliana.
Genome 41:566–572
with 12 independent SCARs (Ozias-Akins et al. 1998), Lin Y-R, Zhu L, Ren S, Yang J, Schertz KF, Paterson AH (1999)
may be at least several hundred kilobase pairs in size. A Sorghum propinquum BAC library, suitable for cloning
Physical mapping of the ASGR in buffelgrass and genes associated with loss-of-function mutations during crop
P. squamulatum will be expanded by application of the domestication. Mol Breed 5:511–520
Lubbers EL, Arthur L, Hanna WW, Ozias-Akins P (1994) Molecular
BAC resource using a combination of DNA contig markers shared by diverse apomictic Pennisetum species.
assembly and fluorescence in situ hybridization (FISH). Theor Appl Genet 89:636–642
The latter technique has proved to be essential for Marek LF, Shoemaker RC (1997) BAC contig development by
resolving physical distances in complex organisms fingerprint analysis in soybean. Genome 40:420–427
where gaps in DNA contigs frequently occur (Jackson et Marie D, Brown SC (1993) A cytometric exercise in plant DNA
histograms, with 2C values for 70 species. Biol Cell 78:41–51
al. 1998) or may be obscured by duplications (Stupar et Martel E, De Nay D, Siljak-Yakovlev S, Brown S, Sarr A (1997)
al. 2001). The two BAC libraries we constructed will be Genome size variation and basic chromosome number in pearl
invaluable tools to decipher the organization and eventually millet and fourteen related Pennisetum species. J Hered 88:
the gene content of the ASGR in both species. 139–143
Meksem K, Zobrist K, Ruben E, Hyten D, Quanzhou T, Zhang
H-B, Lightfoot DA (2000) Two large-insert soybean genomic
Acknowledgements We are grateful for the technical assistance libraries constructed in a binary vector: applications in
of Anne Bell, Evelyn Perry, Jacolyn Merriman (USDA-ARS), and chromosome walking and genome wide physical mapping.
Sue Dove at the Coastal Plain Experiment Station, and Michael Theor Appl Genet 101:747–755
Atkins, John Bishop, Barbara Blackmon, and Scheen Thurmond at Moullet O, Zhang H-B, Lagudah ES (1999) Construction and
the Clemson University Genomics Institute. We acknowledge characterization of a large DNA insert library from the D
support of D. Roche by Pioneer Hybrid International and Limagrain genome of wheat. Theor Appl Genet 99:305–313
SA through a cooperative research and development agreement, of Nam Y-W, Penmetsa RV, Endre G, Uribe P, Kim D, Cook DR
JA Conner by a DOE-EnergyBiosciences research fellowship, as (1999) Construction of a bacterial artificial chromosome library
well as comprehensive support of our work by the USDANRI of Medicago truncatula and identification of clones containing
Plant Genome Program, award nos. 93-37304-9363 and 99-35300- ethylene-response genes. Theor Appl Genet 98:638–646
7691. Experiments have been conducted in compliance with Nogler GA (1984) Gametophytic apomixis. In: Johri BM (ed)
current US laws. Embryology of angiosperms. Springer, Berlin Heidelberg New
York, pp 475–518
Otto FJ (1994) High-resolution analysis of nuclear DNA
References employing the fluorochrome DAPI. Methods Cell Biol 41:
211–217
Asker S, Jerling L (1992) Apomixis in plants. CRC Press, Boca Ozias-Akins P, Pring DR, Vasil IK (1987) Rearrangements in the
Raton mitochondrial genome of hybrid cell lines of Pennisetum
Bennett MD, Leitch IJ (1997) Nuclear DNA amounts in angio- americanum (L.) K. Schum. + Panicum maximum Jacq. Theor
sperms – 583 new estimates. Ann Bot 80:169–196 Appl Genet 74:15–20
812
Ozias-Akins P, Roche D, Hanna WW (1998) Tight clustering and Stupar RM, Lilly JW, Town CD, Cheng Z, Kaul S, Buell CR,
hemizygosity of apomixis-linked markers in Pennisetum Jiang J (2001) Complex mtDNA constitutes an approximate
squamulatum implies genetic control of apospory by a divergent 620-kb insertion on Arabidopsis thaliana chromosome 2:
locus that may have no allelic form in sexual genotypes. Proc Implication of potential sequencing errors caused by large-unit
Natl Acad Sci USA 95:5127–5132 repeats. Proc Natl Acad Sci USA 98:5099–5103
Roche D, Cong P, Chen Z, Hanna WW, Gustine DL, Sherwood Tomkins JP, Mahalingam R, Smith H, Goicoechea JL, Knap HT,
RT, Ozias-Akins P (1999) An apospory-specific genomic Wing RA (1999a) A bacterial artificial chromosome library
region is conserved between buffelgrass (Cenchrus ciliaris L.) for soybean PI 437654 and identification of clones associated
and Pennisetum squamulatum Fresen. Plant J 19:203–208 with cyst nematode resistance. Plant Mol Biol 41:25–32
Sambrook J, Fritsch E, Maniatis T (1989) Molecular cloning: Tomkins JP, Yu Y, Miller-Smith H, Frisch DA, Woo SS, Wing RA
a laboratory manual, 2nd edn. Cold Spring Harbor Laboratory (1999b) A bacterial artificial chromosome library for sugarcane.
Press, Cold Spring Harbor, N.Y. Theor Appl Genet 99:419–424
Sherwood RT, Berg CC, Young BA (1994) Inheritance of apospory Wang G-L, Holsten TE, Song W-Y, Wang H-P, Ronald PC (1995)
in buffelgrass. Crop Sci 36:1490–1494 Construction of a rice bacterial artificial chromosome library and
Shizuya H, Birren B, Kim U, Mancino V, Slepak T, Tachiiri U, identification of clones linked to the Xa-21 disease resistance
Simon M (1992) Cloning and stable maintenance of 300-kilo- locus. Plant J 7:525–533
basepair fragments of human DNA in Escherichia coli using an Woo S-S, Jiang J, Gill BS, Paterson AH, Wing RA (1994)
F-factor-based vector. Proc Natl Acad Sci USA 89:8794–8797 Construction and characterization of a bacterial artificial
Smith RL, Chowdhury MKU, Pring DR (1987) Mitochondrial DNA library of Sorghum bicolor. Nucleic Acids Res 22:4922–4931
rearrangements in Pennisetum associated with reversion from Zhang HB, Zhao XP, Ding XL, Paterson AH, Wing RA (1995)
cytoplasmic male sterility to fertility. Plant Mol Biol 9:277–286 Preparation of megabase-size DNA from plant nuclei. Plant J
Song J, Dong F, Jiang J (2000) Construction of a bacterial artificial 7:175–184
chromosome (BAC) library for potato molecular cytogenetics
research. Genome 43:199–204