Point mutation impairs centromeric CENH3 loading andinduces haploid plantsRaheleh Karimi-Ashtiyania,1, Takayoshi Ishiia,1, Markus Niessenb, Nils Steina, Stefan Heckmanna,2, Maia Gurushidzea,Ali Mohammad Banaei-Moghaddama,3, Jörg Fuchsa, Veit Schuberta, Kerstin Kochb, Oda Weissa, Dmitri Demidova,Klaus Schmidtb, Jochen Kumlehna, and Andreas Houbena,4

aLeibniz Institute of Plant Genetics and Crop Plant Research Gatersleben, 06466 Stadt Seeland, Germany; and bKWS SAAT SE, 37555 Einbeck, Germany

Edited by James A. Birchler, University of Missouri-Columbia, Columbia, MO, and approved July 10, 2015 (received for review March 3, 2015)

The chromosomal position of the centromere-specific histone H3variant CENH3 (also called “CENP-A”) is the assembly site for thekinetochore complex of active centromeres. Any error in transcrip-tion, translation, modification, or incorporation can affect the abilityto assemble intact CENH3 chromatin and can cause centromereinactivation [Allshire RC, Karpen GH (2008) Nat Rev Genet 9(12):923–937]. Here we show that a single-point amino acid ex-change in the centromere-targeting domain of CENH3 leads to re-duced centromere loading of CENH3 in barley, sugar beet, andArabidopsis thaliana. Haploids were obtained after cenh3 L130F-complemented cenh3-null mutant plants were crossed with wild-type A. thaliana. In contrast, in a noncompeting situation (i.e.,centromeres possessing only mutated or only wild-type CENH3),no uniparental chromosome elimination occurs during early em-bryogenesis. The high degree of evolutionary conservation of theidentified mutation site offers promising opportunities for appli-cation in a wide range of crop species in which haploid technologyis of interest.

haploid induction | CENH3 mutant | plant breeding | CENH3 loading |chromosome elimination

The generation and use of haploids is one of the most powerfulbiotechnological means to accelerate the breeding process of

cultivated plants. The advantage of haploid plants for breeders isthat homozygosity can be achieved at all loci in a single gener-ation via whole-genome duplication, without the need of selfingor backcrossing over many generations as conventionally re-quired to obtain true-breeding lines. Haploids can be obtainedin vitro or in vivo, although many species and genotypes arerecalcitrant to these processes (reviewed in ref. 1). Alternatively,substantial changes to centromeric histone H3 (CENH3), such asreplacing the hypervariable N-terminal tail of CENH3 with thetail of conventional histone H3 and fusing it to GFP (producing“tailswap-cenh3”), or complementing the cenh3.2-null mutantwith homologs from the mustard family CENH3s creates haploidinducer lines in the model plant Arabidopsis thaliana (2–4).Haploidization occurred only when such a haploid inducer wascrossed with a wild-type plant. The haploid inducer line provedto be stable upon selfing, suggesting that competition betweenmodified and wild-type centromeres in the developing hybridembryo results in the inactivation of the centromeres from theinducer parent. Consequently, chromosomes from the inducerparent are lost, and progeny can be recovered that retain onlythe haploid chromosome set of the wild-type parent.Because CENH3 is almost universal in eukaryotes, this method

has the potential to produce haploids in any plant species. Toelucidate whether, in addition to the severe conformational changeusing the CENH3-tailswap (2, 3), nontransgenic-induced mini-mal mutations in endogenous CENH3 also could affect the cen-tromere function for haploid induction, we screened a populationof barley (Hordeum vulgare) produced by ethyl methanesulfo-nate-induced targeting of local lesions in genomes (TILLING)(5); this diploid species has two functional variants of CENH3,αCENH3 and βCENH3 (6, 7). Assuming that either CENH3

variant can compensate for the absence of the other, viableoffspring should be observed in the presence of a loss-of-function allele in one of the two CENH3 variants. A total of 25TILLING mutants were identified for both barley CENH3 genes(Table S1). The functionality of mutated CENH3s of homozy-gous M2 individuals of nine TILLING lines carrying non-synonymous mutations was determined by immunostaining thecentromeres with barley CENH3 variant-specific antibodies.αCENH3 and βCENH3 signals at centromeres were revealed inall but one mutant TILLING genotype 4528 (called “Hvßcenh3L92F”) carrying a homozygous leucine-to-phenylalanine sub-stitution at amino acid 92. It showed no centromeric βCENH3signals in mitotic, meiotic, or interphase cells, and only minorβCENH3 signals were observed in the nucleoplasm outside thecentromeres (Fig. 1). Because no obvious differences in thetranscription levels of both CENH3 genes were found in wild-type CENH3 and Hvßcenh3 L92F (Fig. S1), the centromericloading of the mutated βCENH3 seems to be impaired. Cen-tromeres without βCENH3 are sufficient for mitotic centromerefunction, because no obvious chromosome segregation defects,such as anaphase bridges or changes in endopolyploidy, could befound (Fig. S2).The potential ofHvßcenh3 L92F to act as a haploid inducer was

tested. The analysis of 577 F1 plantlets obtained from crosses

Significance

The generation of haploids is the most powerful means to ac-celerate the plant-breeding process. We elucidated whetherpoint mutations in the centromere-specific histone H3 variantCENH3 could be harnessed for the induction of haploids. Weidentified plants with impaired centromere loading caused by amutation in the centromere-targeting domain (CATD). The samemutation results in reduced loading of CENH3 in transgenicArabidopsis and sugar beet. Arabidopsis plants carrying thissingle point mutation in wild-type CENH3 were used as haploidinducers. Because the identified mutation site is highly conservedand because point mutations can be generated by mutagenesisor genome editing, the described method offers opportunitiesfor application in a wide range of crop species.

Author contributions: K.S., J.K., and A.H. designed research; R.K.-A., T.I., M.N., N.S., S.H.,M.G., A.M.B.-M., J.F., V.S., K.K., O.W., and D.D. performed research; R.K.-A., T.I., and A.H.analyzed data; and A.H., R.K.-A., and T.I. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Freely available online through the PNAS open access option.1R.K.-A. and T.I. contributed equally to this work.2Present address: School of Biosciences, University of Birmingham, Birmingham, B15 2TT,United Kingdom.

3Present address: Department of Plant Sciences, School of Biology, College of Science,University of Tehran, Tehran, 1417614411 Iran.

4To whom correspondence should be addressed. Email: houben@ipk-gatersleben.de.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1504333112/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1504333112 PNAS | September 8, 2015 | vol. 112 | no. 36 | 11211–11216

AGRICU

LTURA

LSC

IENCE

S

involving the Hvßcenh3 L92F mutant as the maternal (35 spikes)or paternal (22 spikes) partner for wild-type barley spikes did notreveal any haploid or otherwise hypoploid plants, whereas all 18plantlets derived from nine wild-type barley spikes pollinatedwith Hordeum bulbosum (as a positive control for the procedureused to induce uniparental genome elimination) proved to beinvariably haploid. This result indicates that the Hvßcenh3 L92Fmutation in the presence of native αCENH3 is not sufficient forchromosome elimination during early zygotic embryogenesis.The Hvßcenh3 L92F mutation is located in the CENH3 cen-

tromere-targeting domain (CATD), defined by loop 1 and the α2helix of the histone fold domain (Fig. S3). This domain has beenshown to be required for centromere loading of CENH3 by Scm3/HJURP chaperons in nonplant species (8, 9). The correspondingposition in human CENH3, L91, is essential for CENH3 locali-zation into the centromeres after binding with chaperones (8).CENH3 chaperons are highly variable among different organisms(10), and no analog for plants has been identified yet. However,the possibility that this domain in plants also mediates interactionwith a chaperon cannot be excluded.

To determine whether the CATD mutation caused the ob-served impaired centromere loading, YFP was N-terminallyfused to the coding sequence of A. thaliana CENH3 with an L→Ior L→F exchange (L130I or L130F) at the corresponding posi-tion in A. thaliana CENH3. Double immunolabeling of trans-genic A. thaliana with anti–wild-type CENH3 and anti-GFPrevealed significantly reduced centromere targeting, especially ofthe L-to-F mutated CENH3s compared with wild-type CENH3fused to YFP (Fig. 2A). In addition, the functional significance ofthe identified mutation was assayed in sugar beet (Beta vulgaris).RFP reporter constructs containing the cDNA of sugar beet CENH3with an L106I or L106F exchange (corresponding to amino acidposition 92 of barley; Fig. S3) were generated and used for stabletransformation of sugar beet cells. Again, centromere targetingof the two mutated CENH3s was reduced compared with theendogenous CENH3 fused with RFP in callus and leaf nuclei ofregenerated plants (Fig. 2B). In both species, the L-to-F ex-change resulted in a stronger effect than the L-to-I exchange,likely because of improper function or folding of the protein asthe result of steric challenges imposed by the additional aromatic

Fig. 1. Centromeres of barley TILLING line 4528 lost βCENH3. Chromosomes of wild-type and homozygous TILLING line 4528 after immunostaining withantibodies specific for αCENH3 (green) and βCENH3 (red). Note the absence of βCENH3-specific signals in line 4528.

11212 | www.pnas.org/cgi/doi/10.1073/pnas.1504333112 Karimi-Ashtiyani et al.

group provided by phenylalanine (11). These results indicateimpaired centromere targeting of CENH3 by a point mutation inthe conserved amino acid sequence of monocot and eudicot species.To test for haploid inducer ability in A. thaliana, a genomic

CENH3 construct previously used for functional complementa-tion of cenh3.1 homozygous knockout plants (2) was L130Fmutated and used to transform heterozygous cenh3.1 knockoutplants (2). Lines with one, two, or three Atcenh3 L130F transgeneinsertions were selected (Fig. S4A). Consistent with the resultobtained with the reporter cenh3 L/F construct in A. thaliana, andsugar beet, a high proportion of Atcenh3 L130F-complementedcenh3.1-null mutants displayed limited centromeric anti-CENH3signals in diploid leaf and haploid sperm interphase nuclei (Fig. 2C and D). The single insertion line Atcenh3 L130F-1 revealed thehighest proportion of nuclei with impaired CENH3 distribution.Unlike the tailswap-CENH3 haploid inducer (2, 3), the pheno-type, meiosis, and seed setting of homozygous Atcenh3 L130Flines were almost unaffected (Fig. S4 B and C). Thus, despite adiminished centromere loading of Atcenh3 L130F, mitosis andmeiosis work sufficiently well to produce diploid offspring uponselfing. When Atcenh3 L130F-1 plants were pollinated with wild-type A. thaliana, 4.8% of total F1 seedlings were haploid (24.1%and 2.8% haploids of F1 seedlings grown from shriveled andnormal seeds, respectively) and possessed only the chromosomesof the wild-type parent (Tables 1 and 2 and Figs. S5 and S6). Inaddition, plants derived from shriveled seeds were more oftenaneuploid or mixoploid. The same combination, including theother Atcenh3 L130F lines, resulted in diploids and aneuploidsonly (Tables 1 and 2). The reciprocal cross did not generatehaploid plants.To test whether the efficiency of the haploid inducer correlates

with the total amount of CENH3, a comparative Western blotanalysis was conducted. Indeed, lines L130F-1 and -2 possessedless total CENH3 than wild-type and L130F-3 plants (Fig. S7).The reduced amount of nuclear CENH3 in both lines could becaused either by efficient proteasome-mediated degradation thatprevents promiscuous misincorporation of CENH3 throughoutchromatin (12) or reduced nuclear transport of modified CENH3.Our results suggest that, depending on the amount of CENH3, asingle point mutation in CENH3 is able to generate a haploidinducer in species carrying no more than one CENH3 variant.We propose a possible model of how the process of uniparental

chromosome elimination works in inducer cenh3 × wild-typeCENH3 hybrid embryos (Fig. 3). It is likely that egg cells derivedfrom haploid inducers contain less CENH3 than eggs derivedfrom wild-type plants or contain a reduced amount of an un-known “CENH3-transgeneration required signature.” Accordingto studies performed with a wild-type CENH3-GFP reporter,A. thaliana sperm nuclei but not egg cells are marked withCENH3 (13). However, it is possible that residual amounts ofmaternal CENH3 generating a “centromeric imprinting” aretransmitted to the progeny. Within a few hours after fertilizationpaternal wild-type CENH3 also is actively removed from thezygote nucleus, and centromeric reloading of CENH3 in thezygote occurs at the 16-nuclei stage of endosperm developmentin A. thaliana (13). In embryos undergoing haploidization, cen-tromeric reloading of the maternal chromosomes is impaired ordelayed, causing chromosomes to lag at anaphase because ofcentromere inactivity. Subsequently, micronucleated haploid in-ducer chromosomes will degrade, and a haploid embryo willdevelop, as demonstrated for unstable interspecific hybrid em-bryos (14). In contrast, in a noncompeting situation in whichcentromeres possess only mutated or wild-type CENH3, thetiming of centromere assembly is similar during early embryo-genesis, and no chromosome elimination occurs.In summary, the CATD of the CENH3 gene was mutated

in barley by a single-nucleotide exchange leading to impairedcentromere loading of βCENH3. The effect was reproduced in

Fig. 2. Characterization of the CENH3 point mutation in A. thaliana andB. vulgare. (A) Reduced centromere targeting of L130-mutated CENH3 inA. thaliana. (Upper) Quantification of centromere colocalization patterns(complete, partial, and no) in flower bud nuclei of wild-type A. thaliana (502nuclei from three plants), cenh3 L130I (543 nuclei from four plants), andcenh3 L130F (1,105 nuclei from seven plants). (Lower) Representative doubleimmunostaining patterns. Construct and endogenous wild-type CENH3 weredetected with anti-GFP and anti-CENH3 antibodies, respectively. (B) Reducedcentromere targeting of L106-mutated CENH3 in B. vulgaris. (Upper) Quan-tification of centromere colocalization patterns (complete, partial, and no)in callus and leaves of plants transformed with RFP reporter of wild-typeCENH3 (230 nuclei from five plants) and Bvcenh3 L106I (160 nuclei from twoplants) or Bvcenh3 L106F (350 nuclei from five plants) constructs. (Lower)Representative centromere patterns. (C and D, Upper) Distribution of CENH3in Atcenh3 L130F-complemented cenh3.1-null mutants. Quantification ofcentromere colocalization patterns (strong, weak, or dispersed) in leaf nucleiof wild-type (1,071 nuclei from three plants), cenh3 L130F-1 (2,666 nucleifrom six plants), cenh3 L130F-2 (1,157 nuclei from three plants), and cenh3L130F-3 (1,656 nuclei from four plants) (C) and in sperm nuclei of wild-type(1,121 nuclei from three plants), cenh3 L130F-1 (1,918 nuclei from fiveplants), cenh3 L130F-2 (1,321 nuclei from three plants), and cenh3 L130F-3(1,201 nuclei from three plants). (C and D, Lower) Representative anti-CENH3distribution patterns. Double asterisks indicate 1% significance versusCENH3 wild type; single asterisks indicate 5% significance versus CENH3 wildtype. (E) Flow histograms of diploid and haploid A. thaliana plants togetherwith representative nuclei after FISH using an A. thaliana centromere-specificprobe (in red).

Karimi-Ashtiyani et al. PNAS | September 8, 2015 | vol. 112 | no. 36 | 11213

AGRICU

LTURA

LSC

IENCE

S

transgenic A. thaliana and sugar beet exhibiting the same CENH3mutation. Thus, genotypes carrying the described L130F mutationor a different amino acid switch at this position or at anotherpositon within CATD could be developed into a general in-strument for haploid induction in a wide range of eudicot andmonocot species. Finally, because single amino acid mutationscan be generated by chemical mutagenesis, the entire processof haploidization via application of a haploid inducer line isnontransgenic. Alternatively, haploidy inducers could be gen-erated by genome editing (15) without any modification of thegenetic background.

Materials and MethodsHordeum vulgare.Mutagenesis of barley αCENH3 and βCENH3 by TILLING. We screened a TILLING pop-ulation of 7,979 ethyl methanesulfonate-treated (EMS) diploid barley (H. vulgare)plants of cv. Barke (5) to identify mutant alleles of αCENH3 and βCENH3. Four andthree primer combinations (Table S2) were used to amplify all exons and part ofthe corresponding introns of the αCENH3 and βCENH3 variants respectively, byusing PCR with a heteroduplex step as described earlier (5). PCR productswere digested with the dsDNA Cleavage Kit and analyzed using MutationDiscovery and Gel-dsDNA reagent kits on the AdvanCETM FS96 systemaccording to the manufacturer’s guidelines (Advanced Analytical).RNA extraction, PCR, and quantitative real-time RT-PCR. Total RNA was isolatedfrom roots and leaves using the TRIzol method (16) and from anthers (mi-croscopically staged between meiosis and development of mature pollen),carpel, endosperm, and embryo by a Picopure RNA isolation kit (Arcturus)according to the manufacturer’s instructions. Transcript levels of each genewere normalized to GAPDH by the following formula: R = 2̂ (-(CtGOI–CtH))*100

(17), where R = relative changes, GOI = the gene of interest, and H =housekeeping (GAPDH). The specificity and efficiency of both primers weredetermined by quantitative RT-PCR (qRT-PCR) using a dilution series ofplasmids of cloned full-length barley αCENH3 and βCENH3 genes. A similarCt value (the PCR cycle at which the fluorescent signal of the reporter dyeexceeds background level) for equal amount of plasmid indicates that bothprimers can amplify specific transcripts with the same efficiency.

Indirect immunostaining. Indirect immunostaining of nuclei and chromosomeswas carried out as described in ref. 6. CENH3 of barley was detected withguinea pig anti-αCENH3– and rabbit anti-βCENH3–specific antibodies. Arabbit HTR12-specific antibody (ab72001; Abcam) was used for the detectionof A. thaliana CENH3 (AtCENH3).

A. thaliana.Cloning and generation of CENH3 transgenes. To generate CENH3 genomicfragments carrying mutations, resulting in phenylalanine 130 (F130) andisoleucine 130 (I130) instead of wild-type leucine 130 (L130), a genomicCENH3 fragment in the pCAMBIA1300 vector used to complement cenh3-1/cenh3-1 (the cenh3-null mutant) (2, 18) was subcloned via the unique HindIIIand BamHI sites into pBlueScript II KS (Strategene; www.genomics.agilent.com/). Mutations of CENH3 (L130I or L130F) were generated in pBlueScript IIKS using a Phusion Site-Directed Mutagenesis Kit (Finnzymes; diagnostics.finnzymes.fi/reagents_index.html) according to the manufacturer’s in-structions with minor changes as described in ref. 19. The following 5′-phosphorylated primers were used for mutagenesis: CH3A+L130_F_for-ward, CENH3L130_I_forward, and CENH3L130_F+I_reverse (Table S2).Mutated CENH3 genomic fragments were subcloned via the unique HindIIIand BamHI sites into the initial pCAMBIA1300 vector (18) containing ahygromycin resistance marker. All constructs were verified by sequencing. Togenerate p35S::eYFP-CENH3 fusion constructs containing mutations withinthe CENH3 coding DNA sequence, resulting in L130I or L130F, a plasmid(p35S-BM; www.dna-cloning.com) containing a p35S::eYFP-CENH3 expres-sion cassette (20) was used as template for the Phusion Site-Directed Mu-tagenesis Kit (Finnzymes; diagnostics.finnzymes.fi/reagents_index.html). Theprimers (Table S2) and strategies used to introduce the desired mutationswere the same as above. The resulting expression cassettes [35Spro, eYFP-(mutated)CENH3, and NOS terminator] were subcloned via unique SfiI restric-tion sites into the pLH7000 vector containing a phosphinotricine resistancemarker (www.dna-cloning.com) and were verified by sequencing.Plant transformation, culture conditions, and cross-pollination. A. thaliana wild-type and cenh3-1/CENH3 heterozygotes plants (both accession Columbia-0)were transformed by the floral dip method (21). Transgenic progenies wereselected on MS (22) solid medium containing the corresponding antibiotic.Plants were germinated on Petri dishes under long-day conditions (20 °C,

Table 1. Seed analysis of offspring derived from the reciprocal cross Atcenh3 L130F × wild-type A. thaliana

Shading emphasizes the different crossing direction.

Table 2. Ploidy analysis of offspring derived from the reciprocal cross Atcenh3 L130F × wild-type A. thaliana

Seedlings were derived from seeds with a normal or shriveled phenotype. Shading emphasizes the different crossing direction.*Two hundred ten seedlings out of 307 germinated seeds were analyzed.†One hundred seedlings out of 129 germinated seeds were analyzed.

11214 | www.pnas.org/cgi/doi/10.1073/pnas.1504333112 Karimi-Ashtiyani et al.

16 h light/18 °C, 8 h dark), grown for 4 wk under short-day conditions (20 °C8 h light/18 °C 16 h dark), and then shifted to long-day conditions again. Forcrossing, closed buds of the A. thaliana cenh3 mutant were emasculated byremoving the immature anthers. The stigmas of emasculated buds were pol-linated using the yellowish pollen from mature anthers of freshly opened,wild-type A. thaliana flowers.DNA extraction and genotyping of A. thaliana. Genomic DNA preparations andPCR-based genotyping were performed using standard methods. DNA wasextracted according to ref. 23. Plants were genotyped for cenh3-1 in a de-rived cleaved amplified polymorphic sequence (dCAPS) genotyping reaction.The dCAPS primers cenh3-1_mut_forward and cenh3-1_mut_reverse wereused to amplify CENH3. Amplicons were digested with EcoRV and resolvedon a gel. The cenh3-1 mutant allele is not fragmented (215 bp), whereas thewild-type CENH3 allele is cut (191 and 24 bp). (For primers see Table S2.) Togenotype the endogenous CENH3 locus for cenh3-1 in the offspring ofcenh3-1/CENH3 plants transformed with the CENH3 genomic locus (un-tagged CENH3 transgene with L130, L130I, or L130F), an initial PCR wasperformed with one primer outside the transgene CENH3 locus, allowingspecific amplification of the endogenous but not the transgenic CENH3 locus.

Primers used were cenh3-1_mut_for and cenh3-1_mut2429r. Ampliconswere purified and used as template for a second dCAPS PCR genotypingreaction as described above for cenh3-1 plants. The presence of the trans-gene was verified by sequencing of the RT-PCR product using the primer setof A.tha-CENH3-F and A.ar-CENH3CDNA-R (Table S2).Western blot analysis. Nuclear proteins were isolated (24) and separated in10% polyacrylamide gels according to ref. 25. Samples were electrotransferredonto Immobilon PVDF membranes (Millipore). Membranes were simulta-neously incubated for 12 h at 4 °C in PBS and 5% low-fat milk with histone H3mouse monoclonal (1:5,000; Sino Biological Inc., catalog no.100005-MM01-50) and A. thaliana CENH3 (HTR12) rabbit polyclonal (1:5,000; ab72001;Abcam) primary antibodies. Anti-mouse IRDye 680RD (1:5,000; LI-COR) andanti-rabbit IRDye 800CW (1:5,000; LI-COR) were used as secondary anti-bodies. Signals were detected by the Odyssey CLx Imager (LI-COR). HistoneH3 signals were used to equalize the amounts of loaded proteins.Flow cytometric analysis of A. thaliana plants. For flow cytometric ploidy analysesof plants, equal amounts of leaf material from 5–10 individuals werechopped together with a sharp razor blade in nuclei isolation buffer (26)supplemented with DNase-free RNase (50 μg/mL) and propidium iodide(50 μg/mL). The nuclei suspensions were filtered through a 35-μm cell-strainer cap into 5-mL polystyrene tubes (BD Biosciences) and measured on aFACStarPLUS cell sorter (BD Biosciences) equipped with an argon ion laserINNOVA 90C (Coherent). Approximately 10,000 nuclei were measured andanalyzed using the software CELL Quest ver. 3.3 (BD Biosciences). Theresulting histograms were compared with a reference histogram of a diploidwild-type plant. If an additional peak indicated haploidy, single plants weremeasured again to identify the haploid individuals.Barley test crosses. The barley TILLING line 4528 was first made homozygousfor the leucine-to-phenylalanine substitution at amino acid 92 of βCENH3 viagenerational segregation. To reduce the mutation background, it then wascrossed with wild-type barley cv. Golden Promise. As expected, all F1 hybridscarried the ßcenh3 mutation in a heterozygous state. Upon selfing of F1plants and sequencing βCENH3 of F2 individuals, 14 selected homozygousβcenh3 segregants derived from seven independent F1 hybrids were usedfor reciprocal crosses with wild-type barley. To this end, the spikes used asmaternal material were emasculated and bagged to prevent any un-intended pollination. Mutant plants were used as the maternal parent in 36crosses and as pollinator in 22 crosses. As a positive control for uniparentalgenome elimination, emasculated spikes of Golden Promise were pollinatedby H. bulbosum (accessions HbPB1 and HbFBB of the Leibniz Institute ofPlant Genetics and Crop Plant Research (IPK) GenBank). Cross-pollinatedplants were transferred to the glasshouse with day/night temperatures of20/18 °C, respectively (27). To stimulate caryopsis and embryo development,despite potential chromosome loss, 100 mg/L 2,4-D or dicamba was appliedby injection into the uppermost internode of the spike or by dropping intothe individual florets 1–4 d after pollination (28). Embryos were excisedaseptically from the caryopses 18–21 d post pollination and placed with thescutellum facing down on B5 medium (29) to regenerate plantlets as de-scribed (30). Genome size measurements of plantlets established in soil wereconducted using a Ploidy Analyzer I (Partec).

B. vulgaris.Plant transformation and culture conditions. Stable transformation of B. vulgariscallus was performed as described (31) by selection using kanamycin. After∼2 mo [24 °C, 16 h light (55 μmol·m−2·s−1)/8 h dark], callus cells were micro-scopically analyzed.Cloning and generation of CENH3 transgenes. To generate the 35S::RFP-CENH3fusion construct, CENH3 was amplified from sugar beet cDNA (BvCENH3-cds1 and BvCENH3-cds2) and cloned into a vector containing a 35Spro, RFP,and 35S-terminator expression cassette. For constructs containing mutationswithin the CENH3 coding sequence, resulting in F106 and I106 instead ofL106, the above-mentioned plasmid containing the 35S::RFP-CENH3 ex-pression cassette was used as template for primer-based mutagenesis. ThePstI site close to the position of the desired mutation was used to splitCENH3 into two parts. The desired mutations were integrated via mutationsin the primers (BvCENH3_mut_Fw, BvCENH3_L->F_Rv, BvCENH3_L->I_Rv;Table S2). The resulting expression cassettes [35Spro, RFP-(mutated)CENH3and 35S-terminator] were verified by sequencing. For further details see SIMaterials and Methods.

ACKNOWLEDGMENTS. We thank the late Simon Chan (University ofCalifornia, Davis) for sharing the A. thaliana cloned CENH3 genomic frag-ment and A. thaliana cenh3-null mutant plants; Inna Lermontova [LeibnizInstitute of Plant Genetics and Crop Plant Research (IPK)] for providing thep35S::eYFP-CENH3 fusion construct; Twan Rutten (IPK) for scanning electron

Fig. 3. Model explaining the uniparental chromosome elimination in hap-loid inducer CENH3 × wild-type CENH3 intraspecific hybrid embryos (Left)compared with embryos originating from A. thaliana wild-type crossings(Right). (A) Diploid homozygous haploid inducer and wild-type CENH3 par-ents produce haploid gametes during meiosis. (B) Egg cells derived from thehaploid inducer likely contain either less CENH3 than eggs derived from wildtype or a reduced amount of an unknown CENH3-transgeneration requiredsignature. (C) After fertilization, the paternal wild-type CENH3 is activelyremoved from the zygote nucleus. (D) Centromeric reloading of wild-typeCENH3 in the zygote occurs at the 16-nuclei stage of endosperm develop-ment in A. thaliana. (E and F) In embryos undergoing haploidization, cen-tromeric reloading of the maternal chromosomes is impaired or delayedcausing (i) lagging chromosomes and (ii) subsequent micronuclei formationbecause of centromere inactivity (E). Subsequently, micronucleated haploid-inducer chromosomes will degrade, and a haploid embryo will develop (F).Embryos contain paternal-derived chromosomes in the background of ma-ternal-derived cytoplasm.

Karimi-Ashtiyani et al. PNAS | September 8, 2015 | vol. 112 | no. 36 | 11215

AGRICU

LTURA

LSC

IENCE

S

microscopy; Ingo Schubert and Florian Mette (IPK) for stimulating discussions;and Katrin Kumke, Karla Meier, Heike Büchner, Petra Hoffmeister, JacquelinePohl, and Marzena Kurowska for excellent technical assistance. This work was

supported by the German Federal Ministry of Education and Research Plant2030 Project HAPLOIDS FKZ 0315965, FKZ0313123C, and FKZ0315052B andthe German Research Foundation (DFG) Collaborative Research Center 648.

1. Dunwell JM (2010) Haploids in flowering plants: Origins and exploitation. PlantBiotechnol J 8(4):377–424.

2. Ravi M, Chan SW (2010) Haploid plants produced by centromere-mediated genomeelimination. Nature 464(7288):615–618.

3. Ravi M, et al. (2014) A haploid genetics toolbox for Arabidopsis thaliana. NatCommun 5:5334.

4. Maheshwari S, et al. (2015) Naturally occurring differences in CENH3 affect chromo-some segregation in zygotic mitosis of hybrids. PLoS Genet 11(1):e1004970.

5. Gottwald S, Bauer P, Komatsuda T, Lundqvist U, Stein N (2009) TILLING in the two-rowed barley cultivar ‘Barke’ reveals preferred sites of functional diversity in the geneHvHox1. BMC Res Notes 2:258.

6. Sanei M, Pickering R, Kumke K, Nasuda S, Houben A (2011) Loss of centromeric his-tone H3 (CENH3) from centromeres precedes uniparental chromosome elimination ininterspecific barley hybrids. Proc Natl Acad Sci USA 108(33):E498–E505.

7. Ishii T, et al. (2015) The differential loading of two barley CENH3 variants into distinctcentromeric substructures is cell type- and development-specific. Chromosome Res23(2):277–284, 10.1007/s10577-015-9466-8.

8. Bassett EA, et al. (2012) HJURP uses distinct CENP-A surfaces to recognize and tostabilize CENP-A/histone H4 for centromere assembly. Dev Cell 22(4):749–762.

9. Foltz DR, et al. (2009) Centromere-specific assembly of CENP-a nucleosomes is medi-ated by HJURP. Cell 137(3):472–484.

10. Bai Y, Zhou Z, Feng H, Zhou BR (2011) Recognition of centromeric histone variant CenH3sby their chaperones: Structurally conserved or not. Cell Cycle 10(19):3217–3218.

11. Thomas PJ, Qu BH, Pedersen PL (1995) Defective protein folding as a basis of humandisease. Trends Biochem Sci 20(11):456–459.

12. Moreno-Moreno O, Medina-Giró S, Torras-Llort M, Azorín F (2011) The F box proteinpartner of paired regulates stability of Drosophila centromeric histone H3, CenH3(CID).Curr Biol 21(17):1488–1493.

13. Ingouff M, et al. (2010) Zygotic resetting of the HISTONE 3 variant repertoire par-ticipates in epigenetic reprogramming in Arabidopsis. Curr Biol 20(23):2137–2143.

14. Gernand D, et al. (2005) Uniparental chromosome elimination at mitosis and interphase inwheat and pearl millet crosses involves micronucleus formation, progressive hetero-chromatinization, and DNA fragmentation. Plant Cell 17(9):2431–2438.

15. Puchta H, Fauser F (2014) Synthetic nucleases for genome engineering in plants:Prospects for a bright future. Plant J 78(5):727–741.

16. Chomczynski P, Sacchi N (1987) Single-step method of RNA isolation by acid guani-dinium thiocyanate-phenol-chloroform extraction. Anal Biochem 162(1):156–159.

17. Schmittgen TD, Livak KJ (2008) Analyzing real-time PCR data by the comparative C(T)

method. Nat Protoc 3(6):1101–1108.18. Ravi M, et al. (2010) The rapidly evolving centromere-specific histone has stringent

functional requirements in Arabidopsis thaliana. Genetics 186(2):461–471.19. Heckmann S, et al. (2011) The E2F transcription factor family regulates CENH3 ex-

pression in Arabidopsis thaliana. Plant J 68(4):646–656.20. Lermontova I, et al. (2006) Loading of Arabidopsis centromeric histone CENH3 occurs

mainly during G2 and requires the presence of the histone fold domain. Plant Cell

18(10):2443–2451.21. Clough SJ, Bent AF (1998) Floral dip: A simplified method for Agrobacterium-mediated

transformation of Arabidopsis thaliana. Plant J 16(6):735–743.22. Murashige T, Skoog F (1962) A revised medium for rapid growth and bio assays with

tobacco tissue cultures. Physiol Plant 15(3):473–497.23. Edwards K, Johnstone C, Thompson C (1991) A simple and rapid method for the

preparation of plant genomic DNA for PCR analysis. Nucleic Acids Res 19(6):1349.24. Gendrel AV, Lippman Z, Martienssen R, Colot V (2005) Profiling histone modification

patterns in plants using genomic tiling microarrays. Nat Methods 2(3):213–218.25. Schägger H, von Jagow G (1987) Tricine-sodium dodecyl sulfate-polyacrylamide gel

electrophoresis for the separation of proteins in the range from 1 to 100 kDa. Anal

Biochem 166(2):368–379.26. Galbraith DW, et al. (1983) Rapid flow cytometric analysis of the cell cycle in intact

plant tissues. Science 220(4601):1049–1051.27. Pickering RA, Morgan PW (1985) The influence of temperature on chromosome

elimination during embryo development in crosses involving Hordeum spp., wheat

(Triticum aestivum L.) and rye (Secale cereale L.). Theor Appl Genet 70(2):199–206.28. Matzk F, Mahn A (1994) Improved techniques for haploid production in wheat using

chromosome elimination. Plant Breed 113(2):125–129.29. Gamborg OL, Miller RA, Ojima K (1968) Nutrient requirements of suspension cultures

of soybean root cells. Exp Cell Res 50(1):151–158.30. Houben A, Sanei M, Pickering R (2011) Barley doubled-haploid production by uni-

parental chromosome elimination. Plant Cell Tissue Organ Cult 104(3):321–327.31. Kishchenko EM, Komarnitskii IK, Kuchuk NV (2005) Production of transgenetic sug-

arbeet (Beta vulgaris L.) plants resistant to phosphinothricin. Cell Biol Int 29(1):15–19.32. Barow M, Meister A (2003) Endopolyploidy in seed plants is differently correlated to

systematics, organ, life strategy and genome size. Plant Cell Environ 26(4):571–584.

11216 | www.pnas.org/cgi/doi/10.1073/pnas.1504333112 Karimi-Ashtiyani et al.