The dsc1-R mutation arose from a Mu transposon line (Scanlon et al., 1994) and was introgressed into a B73 background for at least six generations before harvesting kernels used in phenotypic and gene expression analyses. The dsc1-C06, dsc1-H02, and dsc1-B09 alleles were identified after screening the trait utility system for corn (TUSC), a Mu transposon-mutagenized population (Meeley and Briggs, 1995). Primers for this screen can be found in Table A1 in Appendix.

For histological analyses, wild type and discolored mutant kernels were harvested 6 DAP to 20 DAP and fixed overnight in FAA (37% formaldehyde: ethanol: glacial acetic acid: water at 10:50:5:35). The kernels were dehydrated in an ethanol/tert-butyl alcohol series, embedded in paraplast, and 10 μm thick sections were stained with either safranin O-fast green or safranin O-orange G as previously described (Ruzin, 1999). For in situ hybridizations, wild type and discolored mutant kernels were harvested 6 DAP to 20 DAP, fixed in FAA, dehydrated, embedded in paraplast, sectioned, and hybridized with gene specific probes as previously described (Jackson, 1991). Primers used to make probes can be found in Table A1 in Appendix. All samples were imaged using the Zeiss Axio Imager Z1-Apotome microscope (Thornwood, New York) and Zeiss Axiovision release 4.6 software.

The Invitrogen Superscript III One Step RT-PCR Platinum Taq HiFi kit was used to clone dsc1. Briefly, tissue segments including the vegetative SAM were cut from 10 day old B73 seedlings. Total RNA was isolated using the RNeasy Plant Mini Kit (Qiagen) and poly(A) RNA was isolated by the Oligotex mRNA mini kit (Qiagen). Primers, which can be found in Table A1 in Appendix, were designed in the 5′ and 3′ ends of the predicted full length transcript and RT-PCR was performed to generate the full length dsc1 transcript. For analysis of gene expression using RT-PCR and quantitative RT-PCR, total RNA was isolated from harvested wild type kernels (6 DAP, 8 DAP, 12 DAP, 14 DAP, 16 DAP, 18 DAP) and discolored mutant kernels (16 DAP). Dissected embryo and endosperm tissue was flash frozen in liquid nitrogen and ground in SDS extraction buffer as previously described (Prescott and Martin, 1986) with some modifications. Following the five minute incubation on ice with chloroform/isoamyl alcohol (24:1), samples were centrifuged for 10 min at 4°C. After transferring the aqueous phase to a new tube, 1 mL of TRIzol (Invitrogen) was used to extract RNA following the manufacturer’s protocol. Total RNA was extracted from whole 14 day old seedlings and from the upper third of the emerging leaf blade from 14 day old seedlings grown on soil, and roots from 14 day old seedlings grown on 0.02% agar using TRIzol (Invitrogen) according to the manufacturer’s protocol. Superscript III (Invitrogen) was used to synthesize cDNA from 1 μg of RNA treated with DNaseI (Invitrogen). SYBR-green (Quanta) methodology combined with gene specific primers (Table A1 in Appendix) as described in (Zhang et al., 2007) was used in the quantitative RT-PCR analysis. Relative gene expression normalized to 18 s rRNA was determined using the 2^−ΔΔCT method as described in Livak and Schmittgen (2001).

The Gateway Recombination Cloning System (Invitrogen) was used to clone the DSC1 ORF into the pEarleyGate104 destination vector as described (Earley et al., 2006). Details about the organelle marker constructs are previously described (Kohler et al., 1997; Boevink et al., 1998; Reisen and Hanson, 2007; Geldner et al., 2009). Electroporation was used to transform agrobacterium strain C58C1 with the N-terminal 35S-YFP<DSC1 fusion construct, the control construct (empty pEarleyGate104 vector), and the organelle marker constructs (35S-ERD2<GFP, 35S35SAMV-COXIV<GFP, 35S-DsRed<CAT, UBQ10-mCherry<VTI12, UBQ10-mCherry<Got1p; Kohler et al., 1997; Boevink et al., 1998; Reisen and Hanson, 2007; Geldner et al., 2009). Transformants from individual construct lines were grown overnight at 28°C in 2 mL of LB medium containing 50 μg/mL kanamycin and 5 μg/mL tetracycline. After centrifugation, the cells were resuspended in 10 mM MgCl₂ to an OD of 0.5 and incubated at room temperature for 2–4 h. For localization of YFP-tagged DSC1 and the control construct 35S-YFP, N. benthamiana leaves were infiltrated as described in Goodin et al. (2002) and observed using epifluorescence microscopy between 48 and 72 h later. Images were obtained on the Zeiss Axio Imager Z1-Apotome microscope (Thornwood, New York) and Zeiss Axiovision release 4.6 software. For co-localization assays, equal parts of the suspensions transformed with each one of the five organelle markers were individually mixed with the suspension transformed with YFP-tagged DSC1 prior to incubation at room temperature. N. benthamiana leaves were infiltrated as described in Goodin et al. (2002) and observed using confocal microscopy between 48 and 72 h later. Imaging of fluorescent proteins was performed using a Leica TCS-SP5 confocal microscope (Leica Microsystems, Exton, PA, USA) using either 10× or 40× objectives (NA 0.4 or 0.85, respectively). Images were obtained sequentially to separate signal from the two channels and were later superimposed. Time lapse series were collected non-sequentially. All images were taken using either a blue argon ion laser (Ar) or a diode pumped solid state laser (DPSS). Excitation and emission parameters are presented in Table A2 in Appendix. Leica LAS-AF software (version 1.8.2) was used to process all images.

Plants heterozygous for dsc1-R were crossed with ZmPIN1a∼YFP transgenic individuals (Gallavotti et al., 2008). The resulting progeny were planted, screened for the dsc1-R allele and YFP, grown to maturity, and self-pollinated. Ears were harvested the same day kernels were removed for live imaging. Embryos were dissected and put on culture media as described in Scanlon et al. (1997). A BFA (Sigma-Aldrich) stock solution was diluted in DMSO and added to liquid culture media to make a final concentration of 100 μM BFA. Mock treatments were made by adding the same amount of DMSO (minus BFA) to liquid culture media. Harvested embryos were incubated in culture media containing BFA or the mock treatment for at least 4 h before confocal microscopy image analysis. Images were collected non-sequentially as described above.

Self-pollinated plants heterozygous for the dsc1-Rmutation segregate mutant kernels with aberrations in both embryo and endosperm development (Scanlon et al., 1994; Scanlon and Myers, 1998). Wild type kernels are yellow at 12 DAP, where upon discolored mutant kernels are white and smaller than wild type siblings (Figure 1A). Embryo structures are not discernible in mutant kernels dissected after 18 DAP, and the reduced endosperm development fails to fill the kernel space (Figures 1B,C; Scanlon et al., 1994; Scanlon and Myers, 1998). At maturity, all discolored mutant kernels are brown, misshapen, and embryo lethal (Figure 1D).

Detailed phenotypic analyses of discolored mutant kernels harvested at different time-points following pollination show that development of both the embryo and endosperm is normal, albeit delayed, before eventual kernel abortion and tissue disintegration. For example, whereas late proembryo-staged embryos are harvested from wild type embryos at 8 DAP (Figures 2A,B), mutant embryos harvested from the same ear comprise far fewer cells than wild type siblings and are still encased within the ESR (Figures 2G,H). Wild type embryos harvested at 12 DAP are in stage L2, having already elaborated the scutellum, the coleoptile, and the first foliar leaf (Figures 2C,D). In contrast, 12 DAP discolored mutant sibling embryos are stalled at the early transition stage (Figures 2I,J). Three foliar leaf primordia (i.e., stage L3) are initiated in wild type embryos harvested at 16 DAP (Figures 2E,F), where upon discolored mutant embryos exhibit a variable range of phenotypes, including embryos retarded at stage L1(Figures 2K,L) and others in which identification of embryo developmental stage is impossible owing to tissue degeneration (Figures 2M–P). In agreement with the mutant embryo phenotypes described above, molecular markers for shoot meristem maintenance [knotted1(kn1)], vasculature development [ran binding protein2(ranbp2)], and scutellum initiation [Zea mays yabby14(zyb14)] showed normal transcript accumulation in transition-staged discolored mutant embryos (Figures 3A–F; Smith et al., 1995; Juarez et al., 2004). Likewise, degenerated later-staged discolored mutant embryos do not accumulate kn1or zyb14 transcripts after 16 DAP (Figures 3G,H).

Similar to the development discolored mutant embryos, the differentiation of aleurone, BETL, and starchy endosperm cell types is also delayed and eventually aborted in discolored mutant kernels. For example, all these endosperm-specific structures are fully differentiated in 8 DAP wild type kernels. Aleurone cells assume their distinctive cuboidal shape within a single layer surrounding the perimeter of the starchy endosperm (Figure 4A), and three layers of highly extended BETL cells develop in the base of the endosperm, immediately juxtaposed to the maternally derived placenta (Figure 4B). In contrast, differentiated aleurone cells are not identified in the discolored endosperm at 8 DAP (Figure 4E), and just a single layer of BETL cells form at the base of the mutant endosperm (Figure 4F). By 16 DAP however, both discolored mutant and wild type sibling kernels have fully differentiated endosperm structures, including an anatomically distinct aleurone layer surrounding the enlarged, vacuolated cells of the starchy endosperm, and a BETL comprising three cell layers (Figures 4C,D,G,H). Finally, discolored mutant endosperm development is again aberrant by 20 DAP, wherein undifferentiated cells are observed within the mutant aleurone and BETL layers and other cells are degenerated and of non-descript identity (Figures 4I–L).

Previously, a 3,808 bp Mu1-inserted 5′ genomic fragment of the dsc1 locus on chromosome 4S was identified by transposon-tagging analysis of the dsc1-R mutation (Scanlon and Myers, 1998). Three additional Mu-insertion alleles of dsc1 (dsc1-H02, dsc1-C06, and dsc1-B09) obtained from the TUSC (Meeley and Briggs, 1995) failed to complement the dsc1-R mutation (Figure 5A), and thus provided further confirmation that the dsc1 locus has been cloned. Alignment of the ∼2.4 KB genomic DNA fragment of the dsc1 clone to the sequenced maize genome (MaizeSequence.org Release 5b.60; Schnable et al., 2009) identified the predicted full length gene, the 2.472 KB full length transcript, the boundaries of the 17 predicted introns, and the 823 amino acid sequence of the predicted dsc1 gene product. The sequence of the dsc1 cDNA was confirmed by RT-PCR using primers anchored within the dsc1-R transposon-inserted genomic DNA fragment described above (Figures 5A,B). PCR analyses utilizing dsc1gene-specific primers (Table A1 in Appendix) and a primer targeted to the Mu transposon termini (MuTIR; Table A1 in Appendix) identified the Mu-insertion sites of the four dsc1 mutant alleles. As indicated in Figure 5A, the dsc1-R and dsc1-H02 alleles harbor 5′ Mu transposon insertions located 233 bp and 283 bp upstream of the dsc1start codon, respectively. In contrast, the dsc1-C06 and dsc1-B09 alleles contain Mu-insertions within the first intron, 284 bp and 250 bp downstream from the DSC1 start codon. Notably, no exon insertions were identified among our dsc1 mutant alleles. The predicted 823 amino acids of the DSC1 protein encode a putative ARF-GAP protein comprising a BIN-amphiphysin-RVS (BAR) domain, a pleckstrin homology motif, and two ankyrin repeats, in addition to an ARF-GAP domain (Figure 5B). Both the BAR and pleckstrin homology domains are implicated in membrane interactions (Hurley, 2006), whereas ankyrin repeats are known to function in protein-protein interactions (Inoue and Randazzo, 2007). ARF-GAPs comprise a highly conserved group of proteins within the eukaryotes, and function during the regulation of vesicle trafficking (Vernoud et al., 2003; Jiang and Ramachandran, 2006; Inoue and Randazzo, 2007). Taken together, the DSC1 protein is predicted to function during endomembrane trafficking in maize.

A total of 43 predicted maize genes encode an ARF-GAP domain, including dsc1 and a close paralog (95% nucleotide identity/98% amino acid identity) located on chromosome 1 (GRMZM5G872204, designated here as dsc2). In adherence to the mammalian classification system, nine predicted maize ARF-GAP proteins (including DSC1 and DSC2) belong to the ACAP subgroup, comprising ARF-GAPs that function during post-Golgi transport (Jackson et al., 2000; Miura et al., 2002; Nie et al., 2003; Randazzo and Hirsch, 2004). The ARF-GAP domain of DSC1 is homologous (66% identity) to that of the Arabidopsis protein VAN3 (Koizumi et al., 2005; Sieburth et al., 2006), which regulates endomembrane trafficking of the auxin efflux protein PIN1, although yet another maize ARF-GAP (GRMZM2G059225) is more similar to VAN3 (69% identity). ClustalW alignments of the ARF-GAP domains of DSC1, DSC2, VAN3, and a human ACAP-type ARF-GAP (ACAP2) are shown in Figure 5C.

RT-PCR identified dsc1 transcript accumulation during multiple stages of kernel development (6DAP, 8 DAP, 12 DAP, 14 DAP, 16 DAP, and 18 DAP) and in 14-day-old seedling shoots and roots, but not in the fully differentiated distal tips of emerged seedling leaves (Figure 6A). Notably, dsc1 transcript accumulation in 12 DAP, early transition-staged discolored mutant kernels is reduced to less than 0.3 times the level found in wild-type transition-staged seed harvested at 8 DAP (Figure 6B). Furthermore, transcripts of the kernel patterning genes dek1 and cr4 both accumulate to less than 0.4 fold the level of wild type in transition-staged discolored mutants, although accumulation of sal1 is not significantly altered by the dsc1-R mutation (Figure 6B).

Yellow fluorescent protein (YFP)-tagged DSC1 constructs were generated to determine the subcellular localization of DSC1 in the N. benthamiana leaf (Earley et al., 2006). Epifluorescence and confocal microscopic imaging of infiltrated transgenic leaf sectors reveals35S-YFP<DSC1 accumulation within distinct intracellular compartments, and as foci at the plasma membrane (Figures 7A,D,G,J,M; Figures A1A,C and A2A,D,G in Appendix). 35S-YFP<DSC1 labeled bodies also actively move around the cell, as evidenced by time series taken of cells to illustrate the intracellular movement of 35S-YFP-tagged DSC1 (Movie S1 in Supplementary Material). In contrast, N. benthamiana leaves expressing control 35S-YFP construct that lacked the DSC1 open reading frame show fluorescent signal within the nucleus, or dispersed within the cytoplasm (Figures A1E,G in Appendix).

The lipophilic, styryl dye FM4-64 (Invitrogen) fluoresces in hydrophobic environments, and is commonly utilized to trace endocytotic activities in cells (Bolte et al., 2004). Once internalized FM4-64 labeled membranes successively co-localize with the trans-Golgi network/early endosomes, prevacuolar compartments, and ultimately with the tonoplast (vacuolar membrane; Bolte et al., 2004; Geldner et al., 2009). Co-localization assays with several organelle-specific markers and the endocytic tracer FM4-64 were performed to identify more precisely the intracellular localization of YFP-tagged DSC1 protein. These studies confirmed that 35S-YFP<DSC1does not co-localize with the mitochondrial marker 35S35SAMV-COXIV<GFP, the peroxisome marker 35S-DsRed<catalase, auto-fluorescent chloroplasts, the cis-Golgi marker 35S-ERD2<GFP; or with the Golgi stacks marker UBQ10-mCherry<Got1p (Figures A2A–I in Appendix; Figures 7A–F; Kohler et al., 1997; Boevink et al., 1998; Reisen and Hanson, 2007; Geldner et al., 2009). However, in some instances35S-YFP<DSC1 labeled bodies localize next to the cis-Golgi and Golgi stacks markers (Figures 7A–F). Recurrently, 35S-YFP<DSC1 labeled bodies co-localize with the trans-Golgi network/early endosome marker UBQ10-mCherry<VTI12 (Figures 7G–I), in addition toFM4-64 labeled intracellular compartments and the plasma membrane (Figures 7J–O; Geldner et al., 2009). Taken together, the cellular mobility of YFP-tagged DSC1 labeled compartments and their co-localization with both FM4-64 and a marker for the trans-Golgi network/early endosome illustrate that DSC1 functions in endomembrane trafficking.

In Arabidopsis, the ARF-GAP VAN3 functions during the transport of PIN1 from the plasma membrane to the recycling endosome (Sieburth et al., 2006). YFP-tagged ZmPIN1a localization was observed in wild type and discolored mutant embryos to determine if the endomembrane transport of ZmPIN1a is disrupted by the dsc1-R mutation. ZmPIN1a<YFP localizes to the plasma membrane in both mutant and non-mutant 14 DAP embryos (Figures 8A–D). After treatment with Brefeldin A, which blocks vesicle cycling from the endosome to the plasma membrane (Steinmann et al., 1999; Geldner et al., 2001), ZmPIN1a<YFP accumulates in endosomal compartments in both wild type and discolored mutant embryos and no differences in size or number of ZmPIN1a<YFP-tagged endosomal compartments are observed (Figures 8E,F).