2.1. Protein sequence alignments

Using the AtBAM2 sequence as a reference (NP_191958.3), a BLAST search was conducted using the NCBI RefSeq database to identify BAM2 and BAM7 genes in other annotated land plant genomes. Similarly, the sequence of AtBAM1 (NP_189034.1) was used to find BAM1 genes in land plants for comparison with the BAM2 and BAM7 genes. Species names and accession codes are listed in Supplementary Table S1. The FASTA-formatted protein sequences were then downloaded from NCBI and aligned using Clustal Omega (Madeira et al., 2019). The full alignment is available as Supplementary File S2.

2.2. Protein start-site predictions

Initial identification of the alternate start site came from expressed sequence tag (EST) and sequenced cDNA information on the UCSC Genome Browser site for corn BAM7 using B73 RefGen_v3 at chr7:62,317,954–62,321,954 (https://genome.ucsc.edu; Kent et al., 2002). Nucleotide sequences of BAM7 genes were manually analyzed for alternative in-frame translational start sites (ATG) in the intron between the BZR1 and BAM domains. After in silico translation, cellular locations of predicted full-length BAM7 proteins (BAM7-L) and shorter proteins lacking the BZR1 domain and starting from in-frame Met residues (BAM7-S) were conducted using LOCALIZER (Sperschneider et al., 2017).

2.3. Expression vector construction

DNA sequences for AtBAM2 (NM_116273.5) and ZmBAM7 (NM_001350702.1) were obtained from NCBI. The sequence of ZmBAM7-S was determined based on its in-frame start-codon prediction 42 bases upstream of exon 2. The lengths of the predicted chloroplast transit peptides were determined using TargetP-2.0 (Almagro Armenteros et al., 2019). Synthesis of the AtBAM2 and ZmBAM7-S coding sequences lacking the predicted 55- and 66-residue chloroplast transit peptides, respectively, was carried out by GenScript (Piscataway, New Jersey, USA) using codon optimization for expression in E. coli (sequence available in Supplementary Fig. S2). The cDNAs were then cloned into pET-15b such that the expressed proteins would contain an N-terminal His tag. Transformation of competent E. coli DH5α cells (New England BioLabs, Beverly, Massachusetts, USA) with the plasmid DNA was carried out using the manufacturer's protocol. Plasmids were isolated by miniprep and confirmed after digestion with BamHI and NdeI. Transformation of E. coli BL21 cells with each plasmid DNA was carried out using the rapid colony transformation procedure (Micklos & Freyer, 1990). The BAM1 cDNA was a gift from Heike Reinhold and has been described previously (Monroe et al., 2014).

2.4. Protein expression and purification

Cultures of E. coli BL21 cells lacking any plasmid (control) or containing one of the previously described recombinant plasmids were grown to an optical density of 0.7 at 600 nm in Luria–Bertani medium with 100 µg ml⁻¹ carbenicillin (AtBAM2 and ZmBAM7-S) or in 2×YT medium (RPI) along with 1 µl ml⁻¹ kanamycin (AtBAM1) at 37°C with shaking at 250 rev min⁻¹. Isopropyl β-D-1-thiogalactopyranoside (IPTG) was added to a final concentration of 0.8 mM and the flasks were shaken at 250 rev min⁻¹ at 20°C overnight. The cells were pelleted by centrifugation and then frozen at −80°C for at least 10 min. The cell pellets were thawed and resuspended in 50 mM NaH₂PO₄, 0.5 M NaCl, 0.2 mM tris(2-carboxyethyl)phosphine (TCEP), 2 mM imidazole pH 8.0 with EDTA-free protease-inhibitor tablets (Pierce A32965). Cell lysis was completed by sonication in an ice bath (Misonix S-4000; Microtip). After centrifugation of the cell lysate, the supernatants of AtBAM2 and ZmBAM7-S were separately loaded onto a TALON cobalt-affinity column using an ÄKTA start system (Cytiva Life Science, Marlborough, Massachusetts, USA), while AtBAM1 was loaded onto a GE Healthcare nickel-affinity column using the same ÄKTA start system. The bound proteins were eluted from the affinity columns by the stepwise addition of a second buffer consisting of 50 mM NaH₂PO₄, 0.5 M NaCl, 0.2 mM TCEP, 200 mM imidazole pH 8.0. The four 10 ml elution steps contained 12.5, 50, 125 or 200 mM imidazole mixed using the ÄKTA start system. Fractions were analyzed for purity by SDS–PAGE.

Pure ZmBAM7-S and AtBAM2 were dialyzed overnight using SpectraPor tubing (Spectrum, New Brunswick, New Jersey, USA) with a molecular-weight cutoff of 6–8 kDa. The dialyzed proteins were concentrated in an Amicon Ultra-15 concentrator with a molecular-weight cutoff of 10 kDa at intervals of 30 min at 5000g and 4°C until the desired volume was reached. Protein concentrations were determined using the Bio-Rad Protein Assay Kit with BSA as the standard. AtBAM1 was not dialyzed, but was immediately concentrated in a Spin-X UF concentrator with a PES filter of molecular-weight cutoff 5 kDa for 30 min intervals at 4200 rev min⁻¹ (3215g) until the desired volume of 1 ml was reached.

IbBAM5 from Sigma–Aldrich was resuspended at 7 mg ml⁻¹ in 20 mM HEPES pH 7.3, 150 mM NaCl, 0.2 mM TCEP and separated using a HiLoad 16/60 Superdex 200 column equilibrated with SEC buffer (20 mM HEPES pH 7.3, 150 mM NaCl, 0.2 mM TCEP).

2.5. Size-exclusion chromatography (SEC)

Concentrated ZmBAM7-S and AtBAM1 were further purified using a HiLoad 16/60 Superdex 200 column (Cytiva). Pure ZmBAM7-S and AtBAM1, as confirmed by SDS–PAGE, were concentrated as before, distributed into the plate for SAXS and then flash-frozen in liquid nitrogen. The concentration of ZmBAM7-S was determined by the absorbance at 280 nm using an extinction coefficient of 101 760 M⁻¹ cm⁻¹. This value was calculated from the recombinant protein sequence including the His tag using ProtParam (Gasteiger et al., 2005). Similarly, the concentration of AtBAM1 was determined at 280 nm using a Synergy H4 Hybrid Reader (BioTek) on a Take3 plate; the path length was 0.05 cm and the extinction coefficient of 59 511 M⁻¹ cm⁻¹ was calculated from the sequence using ProtParam.

The size-exclusion chromatography data for ZmBAM7-S were used to predict the molecular weight and quaternary structure of this protein. The predicted molecular weight of a ZmBAM7 monomer was calculated using ProtParam. Experimental molecular weights were calculated using calibration standards from the Gel Filtration Molecular Weight Markers Kit for molecular weights 12–200 kDa (Sigma) and the Gel Filtration Calibration Kit for molecular weights 43–669 kDa (Cytiva). These standards were run on the HiLoad 16/60 Superdex 200 column to determine their respective void and elution volumes. The equation of the calibration curve used for molecular-weight calculations was y = −3.62x + 10.37 using a void volume of 40.93 ml. The data were analyzed using Microsoft Excel (version 16.46), R using the tidyverse package (version 4.0.3; 10/10/2020; Wickham et al., 2019) and RStudio (version 1.3.1093; R Core Team; https://www.r-project.org/).

2.6. ZmBAM7-S homology-model construction and quaternary-structure prediction

A homology model of a ZmBAM7-S monomer including the purification tag was produced using AlphaFold2 and ColabFold (Mirdita et al., 2021; Jumper et al., 2021). The sequence alignment was generated using MMseqs2 (Steinegger & Söding, 2018; Mirdita et al., 2019). The best model based on the pLDDT score was used without further processing for data fitting in SASREF and FoXS. SASREF was used through ATSAS online to predict the structure of ZmBAM7-S oligomers (Petoukhov & Svergun, 2005; Petoukhov et al., 2012). A single SAXS curve and a single PDB file of the ZmBAM7-S monomer were used for prediction, with the number of subunits in the final complex controlled by setting the overall symmetry input. No weighting, constraints or contact conditions were added during the fitting process. For comparative fitting of all of the SASREF models, the PDB files from SASREF were combined into a single zip file and uploaded onto the FoXS website (Schneidman-Duhovny et al., 2016). The same SAXS data as used for SASREF prediction were provided for data fitting.

2.7. Small-angle X-ray scattering data collection

Full-length ZmBAM7-S was prepared for SAXS by diluting the SEC-purified protein using SEC buffer to five different concentrations (1.76, 2.64, 3.53, 5.29 and 6.17 mg ml⁻¹) in 35 µl. Samples in a 96-well sample plate were flash-frozen with liquid nitrogen. Three protein-free controls consisting of SEC buffer alone were included with the samples. Purified IbBAM5 was diluted to concentrations of between 1 and 10 mg ml⁻¹ in 20 mM HEPES pH 7.3, 150 mM NaCl, 0.2 mM TCEP and flash-frozen in the plate using liquid nitrogen. The sample plate was shipped overnight on dry ice to the Advanced Light Source at Lawrence Berkeley National Laboratory. Prior to data collection, the plate was spun at 3700 rev min⁻¹ for 10 min by the beamline staff. Scattering data were collected from the samples and controls every 0.3 s for a total of 10 s, resulting in 33 frames of data per sample. The beam energy was 11 keV and the detector was 2 m from the sample holder. Samples were kept at 10°C during data collection. Data from the protein-free buffer were collected before and after ZmBAM7-S data collection to ensure there was no difference in scattering due to contamination of the sample cell.

For AtBAM1, data were collected using size-exclusion chromatography coupled to small-angle X-ray scattering. Samples were shipped overnight at 4°C to the SIBYLS beamline at the Advanced Light Source. The sample buffer was 50 mM MES pH 6.7, 100 mM NaCl, 1 mM DTT. The sample was injected into an Agilent 1260 series HPLC with a Shodex KW-802.5 analytical column at a flow rate of 0.5 ml min⁻¹. Small-angle X-ray scattering (SAXS) data were collected from the eluent as it came off the column. The incident light wavelength was 1.03 Å at a sample-to-detector distance of 1.5 m.

Buffer scattering was subtracted from sample scattering in RAW (version 2.0.3; Hopkins et al., 2017). We used SAXS FrameSlice (version 1.4.13) to identify SAXS data frames that lacked discernable aggregation or radiation-induced degradation of the protein. This setup results in scattering vectors, q, ranging from 0.013 to 0.5 Å⁻¹, where the scattering vector is defined as q = 4π;sinθ/λ and 2θ is the measured scattering angle.

2.8. BAM solution-structure comparison

We calculated the radius of gyration and molecular weight from SAXS data for ZmBAM7-S, our previous SAXS data for AtBAM2, the reference data for I. batatas BAM from SASBDB entry SASDA62 and data collected for I. batatas BAM from Sigma and AtBAM1 using RAW. Molecular weights were determined through Bayesian inference. The R_g value was determined using the Guinier fit function. We then plotted the pair distance distribution function for comparison of the shape and size of all of the proteins. The homology model of ZmBAM7-S was fitted to the intensity data using FoXS (Schneidman-Duhovny et al., 2016). All SAXS data sets have been deposited in the SASBDB as SASDM26 (IbBAM5), SASBMX5 (ZmBAM7-S) and SASDMZ5 (AtBAM1).

2.9. Enzyme assays

Amylase activity assays were conducted using purified protein samples in 0.5 ml containing 50 mM MES pH 6, 100 mM KCl and various concentrations of soluble starch (catalog No. AC424495000, Acros Organics, Morris Plains, New Jersey, USA). After 20 min at 25°C, the reaction tubes were immersed in boiling water for 3 min to stop the reaction. The reducing sugars in each reaction were then measured using the Somogyi–Nelson assay (Nelson, 1944). Data were analyzed using Microsoft Excel (version 16.46) and fitted to the Michaelis–Menten equation (1), which accounts for cooperativity, using Solver,

3.1. Functional residue alignment of BAM2 and BAM7

A. thaliana BAM7 (AtBAM7) has an N-terminal, BZR1-like DNA-binding domain that contains a bipartite nuclear localization signal (NSL; Reinhold et al., 2011). Although AtBAM7 is reported to be catalytically inactive, the BAM domain of AtBAM7 is necessary for specific DNA binding (Soyk et al., 2014). The `active site' contains four mutations among the 15 residues that form hydrogen bonds to the four glucose residues at the nonreducing end of starch (see Fig. 2b in Monroe & Storm, 2018). These mutations are likely to contribute to the catalytic inactivity of AtBAM7 that was found under certain conditions (Reinhold et al., 2011). In the course of analyzing the BAM proteins from sequenced plant genomes, we noticed that predicted BAM7 genes from several plants contained BAM domains that seemed to be more similar to AtBAM2 than to AtBAM7. We identified 12 BAM7 genes from basal angiosperms, monocots and basal eudicots that lacked a BAM2 gene and compared them with the BAM2 and BAM7 protein sequences from 14 eudicot genomes that contain separate BAM2 and BAM7 genes. We also included BAM1 proteins from the same genomes in our analysis. We used Clustal Omega (Madeira et al., 2019) to align the amino-acid sequences (Supplementary File S2) and we then identified residues that play a role in the specific functions of either BAM7 (Reinhold et al., 2011; Soyk et al., 2014) or BAM2 (Monroe et al., 2017, 2018; Fig. 1). A putative bipartite NLS was found in the BZR1-like domain of each BAM7 gene, with only minor differences. In five of the DF-BAM7 proteins the distance between the two regions of positively charged residues was 13 or 15 residues, as opposed to 14 in all of the BAM7 proteins, while in half of the BAM7 and DF-BAM7 proteins His in the second positive region was substituted with Gln (Fig. 1a; Supplementary Table S2). In addition, a Glu residue that was confirmed to be essential for DNA binding (Glu87 in AtBAM7; Soyk et al., 2014) is perfectly conserved in all BAM7 and DF-BAM7 proteins (data not shown).

Figure 1 Conservation of key residues in BAM2 and BAM7 from flowering plants and how they compare with the corresponding residues in DF-BAM and BAM1, and a map of the corn BAM7 gene with its two predicted transcripts. (a) WebLogo illustrating the conservation of residues predicted to be part of the nuclear localization signal in the DNA-binding domain as identified by mutagenesis in A. thaliana BAM7 (Reinhold et al., 2011). The alignment included 14 species of flowering plants that contain BAM7 that were selected to represent a diversity of orders and 12 species that contain DF-BAM7 (see Supplementary Table S1 for species and accession numbers). (b) WebLogo illustrating the conservation of 15 residues in the same species as in (a) identified as forming hydrogen bonds to starch in the active site of soybean (Glycine max) BAM5 (Laederach et al., 1999). (c) WebLogo illustrating the conservation of residues as in (a) compared with the three residues identified in BAM2 as being involved in allosteric regulation of activity in BAM2 (Ser464) or starch binding to a surface binding site for starch (Gly335, Gly446 and Trp449) (Monroe et al., 2017, 2018). (d) WebLogo illustrating the conservation of three residues in the same species as in (a) that were identified by mutagenesis as being important for tetramer stabilization: Trp456 and Asp590 in interface `A' and Phe238 in interface `B' (Monroe et al., 2018). (e) Predicted dual-function ZmBAM7 gene model. Coding regions of exons are colored black, with the exception of a region unique to the N-terminus of ZmBAM7-S that is colored red. The locations of the two putative transcriptional start sites (TSS1 and TSS2) and their respective translational start sites (AUG) in the two predicted transcripts are indicated with arrows. The 5′ and 3′ untranslated regions (UTR) of both transcripts are colored white.

The active-site starch-binding residues (Laederach et al., 1999) are perfectly conserved in all but one of the BAM2 proteins and in all of the BAM1 and DF-BAM7 proteins, suggesting that these enzymes are all likely to be catalytically active (Fig. 1b). In contrast, all but one of the BAM7 sequences from eudicots we analyzed that also contained a separate BAM2 gene had mutations among the active-site residues (Fig. 1b; Supplementary Table S2).

AtBAM2 is unusual among characterized BAMs in having a sigmoidal substrate-saturation curve, in being tetrameric and in having a secondary starch-binding site (SBS) in a groove between the monomers of each dimer (Monroe et al., 2017, 2018; Chandrasekharan et al., 2020). We next looked for key amino acids within the BAM7 and DF-BAM7 sequences that had previously been identified as functioning in each of these unique aspects of BAM2 (Supplementary Fig. S3). Ser464, Gly335, Gly446 and Trp449, which were previously identified as being associated with the SBS and sigmoidal kinetics in BAM2 (Monroe et al., 2017, 2018), are all perfectly conserved in the BAM2 and DF-BAM7 proteins, and in ten of the 12 BAM7 proteins (Fig. 1c, Supplementary Fig. S3). Residues that altered the oligomerization of BAM2 when mutated include Phe238, Trp456 and Asp490 (Monroe et al., 2018). With the exception of two BAM2 proteins in which one of these residues differed, they were conserved in BAM2, BAM7 and DF-BAM7 proteins and were not conserved in the monomeric BAM1 (Fig. 1; Sparla et al., 2006). Together, these results suggest that the DF-BAM7 proteins share most, if not all, of the key residues identified as being important for BAM7 and BAM2, and thus we hypothesize that they may serve both functions.

3.2. Dual-function BAM7 gene structure

Based on the above sequence analysis, we hypothesize that the BAM7 genes in corn and other land plants that lack a separate BAM2 gene have alternative transcription start sites (TSSs) so that the first start site leads to a longer transcript that encodes a BAM7-like protein and contains an NLS. The shorter transcript would be initiated at the second TSS, which we predict is located in the first intron of the DF-BAM7 gene and encodes a BAM2-like protein with an N-terminal chloroplast transit peptide (cTP). We are currently using 5′-RACE to test this prediction, but the hypothesis is supported by known ESTs and cap analysis of gene expression (CAGE) data (Mejía-Guerra et al., 2015). Using the annotated corn BAM7 gene, we created a proposed gene-structure model showing both putative TSSs and their respective translational start sites (Fig. 1e). The longer transcript (BAM7-L) includes the first exon of the coding region which contains the DNA-binding domain and predicted NLS. This gene product would be transcribed from the first TSS, contain all ten exons and function as BAM7. The shorter transcript lacks the first exon and would encode BAM7-S, which we predict is initiated from a cryptic start codon at the N-terminus of a 66-amino-acid putative cTP (Fig. 1e). RNA encoding the N-terminal 14 amino acids of this cTP would be spliced out of the BAM7-L transcript, so it is specific to the BAM7-S transcript. Both transcripts have a common BAM domain with nine exons and a common translational termination site (Fig. 1e). Upon closer inspection of the 12 DF-BAM7 genes that we identified, an in-frame start codon was identified in a similar position within the first intron of each gene. This cryptic translational start site is also predicted to be part of the cTP. Therefore, the shorter versions of the DF-BAM7 genes might be expressed and targeted to the chloro­plast if they were translated from a shorter mRNA transcript. In contrast, only four of the 14 BAM7 genes from genomes that also contained a separate BAM2 gene had an in-frame ATG codon in intron 1, and none of these four was predicted to initiate a cTP (data not shown). Importantly, the longer ZmBAM7-L protein would lack a functional cTP because it is not located at the N-terminus, and the shorter ZmBAM7-S protein would lack the NLS because it lacks the DNA-binding domain which contains the NLS.

3.3. Recombinant protein purification

To test the hypothesis that the BAM7 gene encodes both BAM7-like and BAM2-like proteins in some plants that lack a BAM2 gene, we expressed and purified ZmBAM7-S and AtBAM2 in E. coli. Both ZmBAM7-S and AtBAM2 were expected to have a molecular weight of 58 kDa including the His tag. The absence of a prominent 58 kDa band in the sonicated supernatant of E. coli BL21 cells was used to determine the expression and solubility of the recombinant proteins (Fig. 2a). ZmBAM7-S was expressed (Fig. 2a) and found to be soluble after sonication and centrifugation based on the 58 kDa band in the sonicated supernatant (Fig. 2a).

Figure 2 Purification of recombinant ZmBAM7-S. (a) Pure protein was eluted in 10 ml fractions at increasing imidazole concentrations during affinity chromatography; only one elution fraction is shown for ZmBAM7-S (lane 6). A wash fraction with less than 12.5 mM imidazole is also shown (lane 5). Size-exclusion chromatography (SEC) was also performed in preparation for small-angle X-ray scattering analysis (lane 7). Markers are present in lanes 1 and 8. (b) SEC of ZmBAM7-S. Absorbance data were normalized to the largest value. The peak elution volume for ZmBAM7-S was 50.0 ml. (c) SEC molecular-weight calibration curve. The black line with the equation y = −3.62x + 10.37 was created from nine different calibration standards (gray points). The expected ZmBAM7-S tetrameric molecular weight (black) calculated from the ZmBAM7-S sequence (232 kDa) is shown. (d) Disorder predictions from IntFOLD and IUPred2A for ZmBAM7-S. Using a disorder probability score cutoff of 0.5 (pink line), the probability of being disordered was predicted for each residue in the ZmBAM7-S homology model (black line) or from the sequence of ZmBAM7-S using IUPred2A (blue line).

We then purified ZmBAM7-S using TALON cobalt-affinity chromatography followed by size-exclusion chromatography (SEC; Fig. 2a). The pure protein had a molecular weight of about 58 kDa on SDS–PAGE. Following SEC, the molecular weight of ZmBAM7-S using the trend-line equation of the calibration curve was calculated to be 384.8 kDa (Figs. 2b and 2c). The molecular weight of ZmBAM7-S from its sequence should be 58 kDa, suggesting that ZmBAM7-S forms oligomers of up to six subunits in solution.

3.4. ZmBAM7-S homology model

Since there are no experimentally determined structures of ZmBAM7-S, we produced homology models of the recombinant ZmBAM7-S protein sequence using AlphaFold2 and ColabFold (Jumper et al., 2021; Mirdita et al., 2021). The resulting model showed an extended N-terminal region of 77 amino acids followed by a TIM-barrel domain. This latter folded region is consistent with the structure of other BAM enzymes (Monroe & Storm, 2018). To support the model, we then predicted the disorder of ZmBAM7-S, finding that residues 1–90 of the model had the highest probability of being disordered (Fig. 2d; McGuffin, 2008; Mészáros et al., 2018; Erdős & Dosztányi, 2020).

3.5. Small-angle X-ray scattering

We next acquired small-angle X-ray scattering (SAXS) data for ZmBAM7-S using the SIBYLS beamline. The SAXS data-collection parameters and analysis software used are outlined in Table 1. We did not observe a trend in the radius of gyration (R_g) for ZmBAM7-S with concentration, indicating that ZmBAM7-S forms a consistent oligomer (Fig. 3a). Because there is limited solution structure information for any BAM, we further collected SAXS data for A. thaliana BAM1 (AtBAM1) and I. batatas BAM5 (IbBAM5) for comparison with the ZmBAM7-S data. We calculated the R_g for AtBAM1 to be 25.8 ± 0.2 Å with a molecular weight of 53.1 kDa (95% confidence interval) and that for IbBAM5 to be 43.8 ± 0.1 Å with a molecular weight of 185.8 kDa (95% confidence interval) (Table 1). IbBAM5 is generally accepted to be a tetramer in solution, while BAM1 is thought to be monomeric, and these data are consistent with these previous proposals (Cheong et al., 1995; Sparla et al., 2006). When we compared the pair distance distribution function (PDDF) for ZmBAM7-S with the data for AtBAM2, AtBAM1 and IbBAM5, we observed that the data for ZmBAM7-S showed a similar peak distance value to the data for AtBAM2. However, there was a long tail towards D_max suggestive of a structurally extended region (Fig. 4c). We then used SASREF to fit the ZmBAM7-S homology model to the data and construct oligomers of the homology model using the SAXS data as a guide. Matching the SEC data, SASREF showed that oligomers of ZmBAM7-S fitted the data better than a monomer (χ² of 2.25–2.83 for a tetramer versus 1123.26 for a monomer) (Fig. 4d). A tetramer with P222 symmetry fitted better than a tetramer with P4 symmetry (Figs. 4d and 4e). Collectively, these data support a core structure for ZmBAM7-S that is similar in shape and construction to those of AtBAM2 and IbBAM5.

Figure 3 SAXS data for ZmBAM7-S and truncated ZmBAM7-S. (a) Log of intensity versus momentum transfer for ZmBAM7-S (black) and truncated ZmBAM7-S (gray) The inset plot shows the radius of gyration (Rg) versus ZmBAM7-S concentration (mg ml−1). The Rg values of ZmBAM-S at four different concentrations submitted to SAXS analysis were calculated from the Guinier plot (red data points) and from the P(r) plot (black data points). (b) Kratky plot of ZmBAM7-S (black) and truncated ZmBAM7-S (gray). (c) Pair distance distribution function plot comparing ZmBAM7-S with other BAMs. ZmBAM7-S is shown as a solid black line, truncated ZmBAM7-S is shown as a solid gray line, AtBAM1 is shown as a dotted blue line, IbBAM5 is shown as a dotted gray line and AtBAM2 is shown as a dotted black line. (d) SASREF fits of ZmBAM7-S monomer and oligomers to the SAXS data. SASREF models were combined into a single file and fitted to the SAXS data using FoXS. The χ2 values were 1123.26 for P1, 258.1 for P2, 2.83 for P4, 2.25 for P222, 11.81 for P6, 7.32 for P32, 20.16 for P42 and 1150.59 for P52. (e) FoXS fit of the P222 model from SASREF to the SAXS data.

Figure 4 Activity of ZmBAM7-S. (a) Gel showing the purity and amount of AtBAM2 and ZmBAM7-S used in activity assays. (b) Effect of substrate concentration on ZmBAM7-S (gray) and AtBAM2 (black) activity on a per milligram of protein basis. Points are shown for each of three replicate assays. Data were fitted to the Michaelis–Menten equation for cooperative enzymes. For AtBAM2 the Km was 59.4 mg ml−1 (95% confidence interval 58–61 mg ml−1) and for ZmBAM7-S the Km was 86.9 mg ml−1 (95% confidence interval 83–91 mg ml−1). The fitted data and all kinetic values are shown in Supplementary Fig. S4.

3.6. Enzyme-activity assays

If ZmBAM7-S functions as a BAM2-like protein, we predict that it would be catalytically active and exhibit sigmoidal kinetics (Monroe et al., 2017). In order to compare the catalytic activity of ZmBAM7-S with that of AtBAM2, both proteins were purified to near-homogeneity (Fig. 4a). The purified corn protein was catalytically active, and this activity did exhibit sigmoidal kinetics, but because soluble starch at or above 100 mg ml⁻¹ rapidly retrogrades at 25°C, we were unable to conduct assays at higher levels of substrate to reach saturation (Fig. 4b). From fitting of the data to a cooperative Michaelis–Menten equation, we calculate the maximum activity of ZmBAM7-S to be 160.7 U mg⁻¹ (95% confidence interval 137–185 U mg⁻¹), while the value for AtBAM2 was 367 U mg⁻¹ (95% confidence interval 353–381 U mg⁻¹). ZmBAM7-S also shows an apparently weaker affinity for soluble starch, with a K_m of 86.9 mg ml⁻¹ (95% confidence interval 83–91 mg ml⁻¹) compared with a value of 59.4 mg ml⁻¹ (95% confidence interval 58–61 mg ml⁻¹) for AtBAM2 (Supplementary Fig. S4). These data show that ZmBAM7-S is apparently a less efficient enzyme; however, the inability to saturate the enzyme with substrate means that these values are estimates and should be taken as a preliminary study of ZmBAM7-S activity.