Amino acid sequences of NF-YC homologs were analyzed using the CLUSTALW software from the Kyoto Encyclopedia of Genes and Genomes (KEGG) (https://www.genome.jp/tools-bin/clustalw), with standard parameters (i.e., Gap Open Penalty: 10, and Gap Extension Penalty: 0.05). The neighbor-joining tree was constructed using MEGA version X based on the JTT+G+F model (Kumar et al., 2018). All positions containing gaps and missing data were eliminated for analysis. Bootstrap values were calculated from 1000 trees. The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test are shown next to the branches.

A plus mating type of the C. reinhardtii wild-type (WT) 21gr strain were used as controls. The 21gr (mt+) WT strains and the QQS transgenic lines were grown photoautotrophically in 200-mL photobioreactors containing a minimal growth medium consisting of 10 mM urea, 1.22 mM K₂HPO₄, 0.76 mM KH₂PO₄, 0.405 mM MgSO₄·7H₂O, 0.34 mM CaCl₂·2H₂O, 1 mL/L of Hutner′s trace elements, and buffered with 20 mM MOPS titrated with Tris to pH 7.3 (Xie et al., 2014). Cultures were aerated with a mixture of 5% CO₂ in air and continuously illuminated with 6500 K fluorescent lamps at an intensity of 300 μmol photons·m⁻²·s⁻¹ PAR under the continuous illumination. Photobioreactor cultures were initiated at a cell density of 1 x 10⁵ cells/mL, maintained at 25°C, and grown to stationary phase (6 days). Growth was monitored by measuring both optical density at 750 nm and by counting cell density using a Z1 Coulter Particle Counter (Beckman Coulter Inc. USA). Samples were collected at 2-, 4-, and 6-day intervals for dry weight, protein, and starch analysis.

For QQS expression in C. reinhardtii, the AtQQS ORF (Open Reading Frame) was PCR-amplified from QQSpB2GW7 (Li et al., 2015) with specific primers ( Supplementary Table S1 ) that introduced an NdeI site overhanging the translational ATG start codon at the 5′ end, and an EcoRI site after the stop codon at the 3′ end. The amplified DNA was digested by EcoRI and NdeI and ligated into the corresponding sites of the pGenD plasmid (Fischer and Rochaix, 2001), which placed the QQS ORF between the constitutive PsaD promoter and PsaD terminator. A HindIII/KpnI DNA fragment containing the AphVIII selection marker from pSI103-delta (Sizova et al., 2001) was inserted downstream of the PsaD terminator to complete the final R15-QQS plasmid. This plasmid was linearized and used to transform C. reinhardtii, strain 21gr (CC-1690, Chlamydomonas Resource Center, USA).

WT 21gr cells grown mixotrophically in liquid TAP medium were harvested at the early log phase (0.5-1 X 10⁶ cells/mL) and directly used for electroporation without the autolysin-mediated removal of the cell wall. Electroporation was performed as described (Shimogawara et al., 1998) with a Gene Pulser Xcell electroporator (BioRad, USA; the condition was set as 650 V, 25 µF, 0 resistance with an exponential pulse). After electroporation, cells were resuspended in 15-mL TAP medium, and recovered for 24 h with gentle agitation (150 rpm) before plating on TAP plates supplemented with 15 µg/mL paromomycin for selection. Colonies grown under selective conditions were transferred to fresh TAP plates for further gene expression analysis. First, the presence of QQS sequence integrated into the genome in the putative transformants was detected by colony PCR with a specific primer complementary to the PsaD promoter sequence and a primer complementary to the QQS coding sequence ( Supplementary Table S1 ). To detect QQS mRNA in the transformants bearing QQS transgene, each transformed cell line was grown in a liquid TAP medium, and RNA was isolated as previously described (Li et al., 2015), using TRIzol reagent (Invitrogen, USA) and purified with RNeasy-mini-Qiagen kit (Qiagen, USA), and treated with DNase I. Complementary DNA was synthesized using M-MLV Reverse Transcriptase (New England BioLabs, USA). Subsequently, RT-PCR was performed for QQS transcript analysis. CBLP (encoding guanine nucleotide-binding protein beta subunit-like protein) gene was used as the internal control (Fang et al., 2012). Assays were conducted on at least three biological replicates. The primers used in these analyses are listed in Supplementary Table S1 .

Dry weight was determined after centrifuging 10 mL of cells at 4800 x g for 10 min, decanting the supernatant, washing cells once with distilled water, and transferring cells to a pre-weighed aluminum tray to be placed in a 75°C oven for 48 h. Starch content was determined as previously described in Fan et al. (2012). Protein content was measured by mixing 0.1 M NaOH with culture at a ratio of 3:1, heating at 95°C for 30 min. After centrifugation at 22,000g for 1 min, the supernatant was recovered and protein concentration was determined by using the Pierce BCA Protein Assay Kit (Fisher Scientific, USA) (Bigelow et al., 2014).

WT yeast strains were cultured in YPD (1% yeast extract, 2% peptone, 2% dextrose from Sigma-Aldrich, USA) media as described (Treco and Lundblad, 2001). Strains carrying kanMX disrupted mutant alleles were selected using YPD media containing 200 mg/L G418. Synthetic defined (SD) media was used to culture and select yeast strains containing plasmids that complimented auxotrophic mutant alleles (ura3 or his3). The media was composed of 0.2% drop-out mix (minus uracil and/or histidine), 0.17% yeast N base (minus amino acids) and 2% dextrose (Sigma-Aldrich, USA). All strains were cultured at 30°C in liquid media or on 2% agar plates.

S. cerevisiae haploid strain BY4741, mating type a (MATa), and haploid strain BY4742 mating type α (MATα), carrying hap2, hap3, hap4, or hap5 mutant alleles were obtained from the American Tissue Culture Collection (www.atcc.org), and these strains were the products of Saccharomyces Genome Deletion Project (www.sequence.stanford.edu/group/yeast_deletion_project/deletions3.html). Mutant strains carrying multiple Knock-out (KO) alleles of the HAP complex were generated by mating and sporulating single hap KO strains, as previously described (Giaever and Nislow, 2014). MATa and MATα strains were grown on YPD-agar plates for 24 to 48 h at 30°C. A single colony of each mating type was mated on a YPD plate and grown at 30°C for 4 h. A patch of cells was then transferred from the mating plate to an SD (Lys- Met-) plate and grown for 24 h at 30°C. The cells were transferred to GNA pre-sporulation plates (5% D-glucose, 3% Difco nutrient broth, 1% Difco yeast extract) and grown at 30°C for 24 h. The cells were transferred to a new GNA plate and grown for 24 h at 30°C. A patch of cells was then transferred to 1-mL sporulation media (1% potassium acetate, 0.006% zinc acetate, 0.2% drop-out mix [met- lys-]) in sterile tubes. Sporulation tubes were placed in a room temperature shaker for 5 days, then transferred to a 30°C shaker for 3 days. Twenty-μL of 1 mg/mL Zymolyase was added to 100 μL suspension of sporulated cells in a 1.5-mL Eppendorf tube. Tubes were incubated at 37°C for 5-10 min and were then placed on ice. Twenty-μL suspension of the treated cells was pipetted onto the top half of a YPD plate. Tetrads were visualized using a dissection microscope. A total of 8–10 sets of tetrad spores were separated from each other and placed on the empty half of the YPD plate. The plate was placed in a 30°C incubator for 24-48 h, and the resulting yeast colonies were saved for testing of mating type.

The mating type of each colony was tested by the use of a “halo effect” induced by the use of the Δbar1 strain (Manney, 1983). Patches of each colony were transferred to a YPD plate previously spread with a lawn of Δbar1 colonies. Plates were incubated in a 30°C incubator for 24 h. Strains of MATa type displayed a halo effect; Δbar1 yeast did not grow on top of the patches of MATa yeast, resulting in a circle of no growth around the patches. MATα strains can be determined by observing lawn growth of Δbar1 on top of the MATα patches of yeast.

The hap2, hap3, hap4, and hap5 KO strains were obtained as MATa type, and thus were mated with BY4742 to generate single KO strains with the MATα mating type. Single KOs were mated with each other and sporulated to generate tetrads containing double KOs of the HAP complex. Double KOs were then mated with each other and sporulated to generate tetrads containing the triple and quadruple KOs of the HAP complex. All spores were tested using PCR and gel electrophoresis analysis of the amplified DNA product to confirm the KOs of hap2, hap3, hap4 and/or hap5.

The Arabidopsis QQS or yeast HAP5 ORFs were cloned from entry vectors (pDONR221) into pAG destination vectors (Alberti et al., 2007) using the LR reaction as described by the product manual of Gateway^™ LR Clonase^™ II Enzyme mix (Fisher Scientific, USA). The destination vectors were isolated from the bacterial cells using the PureLink^™ Quick Plasmid Miniprep Kit (Fisher Scientific, USA). The pAG destination vectors were pAG426 (high copy vector, carrying the URA3 auxotrophic marker), pAG416 (low copy vector, carrying the URA3 auxotrophic marker), pAG423 (high copy vector, carrying the HIS3 auxotrophic marker), and pAG423 (low copy vector, carrying the HIS3 auxotrophic marker). Expression from these episomal vectors was constitutive in the presence of glucose (GPD). The vectors also contain an ampicillin resistance gene for selection in E. coli.

The pAG plasmids containing QQS or HAP5 ORFs were transformed into yeast cells using the lithium acetate/single-stranded carrier DNA/PEG rapid transformation method (Gietz and Stenback, 1998). The resulting cells were grown on SD media with either ura- and/or his- selection depending on the pAG plasmid that was used.

The presence of yeast mutant alleles and the presence of QQS and/or HAP5 genes carried by pAG vectors were confirmed by PCR and gel electrophoresis analysis of the amplified DNA product. Primer sequences for these PCR-based genotyping experiments are listed in Supplementary Table S1 .

All PCR reactions were performed using a mix of 12.5 μL GoTaq^® Hot Start Green Master Mix (Promega, USA), 0.25 μL of each primer at 10 μM concentration, 2 μL of DNA template and 10 μL ddH₂O per reaction. The 25 μL reactions were performed in 8-well strips or 96-well plates. Aliquots of 5–15 μL of PCR product were analyzed by gel electrophoresis. The genotype for each strain was determined by comparing the observed fragment size to the expected size of the PCR products.

Overnight liquid cultures of each yeast strain were inoculated with a single colony into 5 mL of the rich YPD (Yeast Peptone Dextrose) media as specified previously (Rose et al., 1990; Guthrie and Gerald, 1991; Campbell et al., 2019). Once the cultures reached a sufficient optical density (OD₆₀₀ > 0.5), they were diluted 1: 20 with fresh media, and 100 μL of the diluted culture was pipetted into a 96-well plate. The plates were then placed in a BioTek Eon^™ spectrophotometer (BioTek Instruments, USA), with constant (double orbital) shaking until each strain reached stationary phase of growth (achieved by 36 h of growth). The OD₆₀₀ for each strain was collected every 30 min. Doubling time was calculated to quantify the growth rate for each strain.

Growth rates were also determined in shake-flasks. Three independent colonies from three confirmed strains were inoculated into 5-mL starter cultures and grown overnight (OD₆₀₀ > 1.5). The cultures were then diluted to OD₆₀₀ = 0.1, and 50 mL of culture was grown in 250-mL Erlenmeyer flasks in a shaker at 30°C, and a 0.5-mL aliquot was removed every hour for the first 12 h and then every 3 h, to determine the OD₆₀₀ for each culture. These data were used to determine the time required to reach the stationary phase. These data, in conjunction with the 96-well plate reader data, were used to identify the appropriate sample collection times for biochemical analyses.

Forty-eight unique yeast genotypes ( Figure 1C ) were subjected to biochemical analyses. Three biological replicates were analyzed per strain. Strains were grown in sets of 40, and the strain composition of each set was randomized. Culture samples from strains with a faster growth phenotype were collected after 15 h, while culture samples from strains with a slower growth phenotype were collected after 24 h of growth. This was done to ensure that all samples were collected at the early stationary phase. The culture samples from each flask were aliquoted into separate tubes depending on the type of metabolite extraction. The 50-mL culture was used for all of the metabolite extraction and quantification experiments.

At each chosen time-point in the early stationary phase of the yeast cultures, two 10-mL culture aliquots were removed and transferred to individual pre-weighed 15-mL Falcon tubes. Cells were pelleted by centrifugation at 4,000 x g for 5 min, the supernatant was removed, and pellets were flash frozen in liquid nitrogen, and stored at -80°C. Prior to extraction, samples were dried by lyophilization, and the dry weight of each cell pellet was determined.

Fatty acids were extracted using a sulfuric acid/methanol method (Dietrich et al., 2005; Perera et al., 2010; Quanbeck et al., 2012), and the order of these extractions was randomized. The dried pellets were transferred to screw-capped glass tubes, and each sample was spiked with 10 µg of nonadecanoic acid, which served as the internal standard for quantification purposes. Following the addition of glass beads and 1 mL of 5% sulfuric acid in methanol, the tubes were vortexed for 5 min and sonicated for 10 min. The screw-capped tubes were incubated at 80°C for an hour, and then allowed to cool to room temperature. The fatty acid methyl esters (FAMEs) were recovered by the addition of 1 mL of hexane:chloroform (4:1) and 1 mL of 0.9% sodium chloride solution. The tubes were vortexed for 5 min, and then centrifuged at 700 x g for 5 min to separate the aqueous and organic phases. The upper organic layer was collected and transferred to a GC vial, and the lower aqueous layer was re-extracted with two 1-mL aliquots of hexane:chloroform (4:1). The recovered organic phases were pooled and concentrated by evaporation, and the dried extracts were stored at -80°C. Prior to gas chromatographic (GC) analysis, the dried samples were resuspended in 300 µL of hexane:chloroform (4:1), and 1 µL of sample was injected into a GC equipped with a flame ionization detector (FID). A few samples were analyzed by GC-mass spectrometry (MS) in order to determine the chemical identity of the FAME peaks. Quantification was accomplished by comparing the peak areas of each FAME, with the peak area of internal standard FAME, and then summing the total amount of fatty acid detected for the C₁₂-C₁₈ FAMEs.

Two-mL aliquots of culture samples were collected in pre-weighed tubes at each chosen time-point, and cells were pelleted by centrifugation at 4000 x g for 5 min. After removing the supernatant, the tubes containing the cell pellets were flash frozen by submerging in liquid nitrogen, weighed, and stored at -80°C until extraction.

Protein extracts were prepared in sets of 80 samples. In order to reduce the variance introduced by extracting in separate sets, the 144 total samples were extracted in random order. Protein extraction was conducted using a DEST52 kit (Invitrogen Life Technologies, USA) following instructions provided by the manufacturer. The cell lysis buffer was modified by incorporating 2% (v/v) 2-mercaptoethanol and 0.1% (v/v) Triton-X100. The cell pellets were washed with 500 µL of lysis buffer and re-pelleted by centrifugation for 5 min at 1500 x g and 4°C. The supernatant was removed, cells were resuspended in 100 µL of lysis buffer, and an equal volume of acid-washed glass beads was added to the samples. Samples were vigorously vortexed for three 30-s periods, interrupted by a 30-s rest on ice and were centrifuged at 15,000 x g for 10 min. The supernatant was recovered into a new tube, and stored at -20°C.

Protein concentrations in each sample were measured using the Bradford colorimetric method (Bradford, 1976). These determinations were conducted in sets of 80 individual samples, and the order of samples was randomized. Twenty µl of each protein extract was pipetted into separate wells of a 96-well plate, and 200 µL of Bradford reagent was added to each well. Following a 15-min incubation at room temperature, the A₅₉₅ of each well was determined using the BioTek Eon^™ spectrophotometer (USA). Protein concentrations were calculated from a parallel generated standard curve using bovine serum albumin as the standard.

At each chosen time-point in the early stationary phase of the culture, two 5-mL aliquots of each culture were collected and transferred to a pre-weighed 15-mL Falcon tube. Following centrifugation at 4,000 x g for 5 min, the supernatant was removed, and the collected cell pellets were flash frozen in liquid nitrogen and stored at -80°C prior to extraction. Prior to extraction, the cell pellets were dried for 24 h by lyophilization, and the dry weight of the pellet was recorded.

To reduce the potential of introducing variance associated with the extraction protocol, glycogen extracts were prepared in sets of 40 samples, selected at random from the 144 total samples. Glycogen was extracted using an optimized hot alkali method (Gunja-Smith et al, 1977; Seo et al, 2002). Initially, cell pellets were washed twice with 1-mL aliquots of H₂O to remove glucose that carried over from the culture medium. The washed pellets were suspended in 1 mL of 20% (w/v) potassium hydroxide solution and placed in a boiling water bath for 1 h. After cooling, the samples were adjusted to pH 6-7 using 5 N hydrochloric acid. Two volumes of 100% ethanol were added to each sample, and the precipitates were collected by centrifugation for 5 min at 4,000 x g at 4°C. The pellets were washed twice with 67% (v/v) ethanol, and the tubes were inverted over paper towels for at least 5 min to remove excess ethanol. The pellet was resuspended in 2 mL H₂O by heating and vortexing each sample. Any undissolved material was pelleted by centrifugation for 5 min at 4,000 x g at 4°C, and the supernatant was collected to fresh 15-mL Falcon tube. Pellet was resuspended and washed with another aliquot of H₂O, and following centrifugation, the supernatant was recovered and pooled. Polysaccharides in the pooled supernatants were precipitated by the addition of 1.5 volumes of 100% ethanol, and following incubation, on ice for 20 min, the precipitate was collected by centrifugation for 10 min at 4,000 x g at 4°C. The pellets were resuspended in 1 mL of H₂O and flash-frozen in liquid nitrogen, and stored at -80°C. Once all 144 glycogen extracts were prepared, the order of samples for glycogen analysis was again randomized, and these analyses were conducted in sets of 80 samples.

Thawed glycogen extracts were adjusted to 10 mM sodium hydroxide and sonicated for 10 s. The glycogen in each sample (0.2-mL aliquots) was hydrolyzed in individual wells of a 96-well plate, with the addition of 30 µL of 100 mM sodium acetate, pH 5.0, containing 0.012 U of amyloglucosidase. The plates were incubated at 37°C for 2 h, and the concentration of the released glucose was quantified using the GAHK-20 Glucose Oxidase Assay Kit (Sigma-Aldrich, USA). At the start of these assays, a 66-µL aliquot was withdrawn from each assay and transferred to a new 96-well plate and the initial A₃₄₀ of the solution was determined. Thirty-three µL of glucose oxidase reagent containing 1.5 mM NAD⁺, 1.0 mM ATP, 1.0 U/mL hexokinase, and 1.0 U/mL glucose-6-phosphate dehydrogenase was added to each well. The plate was placed in a spectrophotometer, and A₃₄₀ of each assay was measured for a period of 1 h. A glucose standard curve was generated in parallel with each 96-well plate. Glucose content of each glycogen extract was quantified by relating the change in A₃₄₀ to the change observed with the glucose standards.

A minimum of three biological determinations from each independent transgenic line and each control were used for qualitative and quantitative analyses of composition. For compositional analyses, data are presented as mean ± SE (Standard Error). Statistical significance relative to the control was calculated with Student’s t-test (one tailed, not paired), or by analysis of variance (ANOVA).

The plant transcription factor database, PlantRegMap/PlantTFDB v5.0 (Tian et al., 2020) was used to identify and analyze NF-YA, NF-YB, and NF-YC homologs in plants. As compared to mammals and yeast genomes, which encode single NF-Y subunit genes (Hooft et al., 1990; Nardini et al., 2013; Oldfield et al., 2019), higher plant genomes contain a large number of genes that encode these subunits ( Figure 2 ). This is typified by the situation in the Arabidopsis genome, which encodes 21 NF-YA, 27 NF-YB, and 21 NF-YC gene models. The genome organization of these genes is simplified, but also more intriguing in the green algae clades, such as the unicellular C. reinhardtii genome, which encodes three NF-YB homologs and three NF-YC homologs, but there is no identifiable NF-YA homolog ( Supplementary Tables S2 , S3 ). Interestingly, single NF-YA homologs are recognizable in other green algae, such as Ostreococcus lucimarinus, Micromonas pusilla, and Micromonas sp. RC299, while there are no sequence homologs of NF-YA in the multicellular green alga, Volvox carteri.

Initially, we explored the phylogenetic relationships among the NF-YC proteins that bind to QQS. In its natural host, Arabidopsis, QQS interacts with NF-YC4 (Li et al., 2015). We therefore constructed a neighbor-joining phylogenetic tree using sequences of QQS-binding domain (residues 73-162) of the NF-YC4 protein (Li et al., 2015) from seven species (i.e., S. cerevisiae, Homo sapiens, C. reinhardtii, A. thaliana, Oryza sativa, Glycine max, and Zea mays) ( Supplementary Figure S1 , Supplementary Table S4 ). These proteins segregate into four major clades, and the clade that includes AtNF-YC4 also contains the single-copy NF-YC protein of S. cerevisiae and H. sapiens and one of the three NF-YC isoforms of C. reinhardtii (Cre12.g556400). The S. cerevisiae and H. sapiens NF-YC sequences are located in adjoining branches, reflecting the fact that these homologs share 53% amino acid identity. The three Arabidopsis isoforms, AtNF-YC9, AtNF-YC3, and AtNF-YC4, are also located close to each other, consistent with the finding that these three proteins are functionally redundant (Kumimoto et al., 2010). It is interesting to note that unlike most of the rice and maize NF-YC isoforms that are located in a distinct clade, away from the AtNF-YC4 isoform, the rice (LOC_Os3g14669) and maize NF-YC4 isoforms (GrmZm2g089812 in genome Version 2, now ZmCA5P11: Zm00001eb010770 in genome Version 5) that bind to QQS (Li et al., 2015) are located in a clade adjoining the AtNF-YC4 homolog. Based on this correlation, one may expect that the C. reinhardtii NF-YC4 homolog (Cre12.g556400), which is adjoining to AtNF-YC4 may bind to QQS ( Supplementary Figure S1 ).

This hypothesis was supported by additional analyses of the amino acid sequences of these homologous proteins. The CLUSTALW multisequence alignment shown in Supplementary Figure S2 illustrates that Cre12.g556400 (CrNF-YC4) has very high homology with the AtNF-YC4, specifically encompassing the QQS-binding domain; namely 98% similarity and 73% identity between residues 73-162 of AtNF-YC4. This hypothesis was therefore evaluated by the expression of QQS in C. reinhardtii.

Five transgenic events were recovered from the transformation of C.reinhardtii with the QQS transgene. These recovered QQS-E strains were confirmed by PCR-based genotyping, and RT-PCR was used to confirm the expression of the QQS transcript ( Supplementary Figure S3 ). The impact of QQS expression on the growth of C. reinhardtii was evaluated by monitoring the optical density ( Figure 3A ), cell density ( Supplementary Figure S4 ) and dry weight biomass yield of cultures ( Figure 3B ) for all mutant strains as compared to WT. Collectively these analyses indicate that transgenic expression of QQS significantly affected the cell density as the cultures approached the stationary phase of growth (i.e., only after 144 h), while it did not have a statistically significant impact on the growth of C. reinhardtii (at the P level of 0.05) as compared to the WT strain of 21gr⁺.

Parallel analyses of the protein and starch content showed that nearly all QQS-E strains had a statistically significant increase in protein levels as compared to the WT strain, the exception being QQS-E #8 (in mg/mL, interestingly this correlates with its QQS expression level in Figures S3 ) ( Figures 3C, D ). These changes in protein content depended on the stage of the cultures and ranged in increases of 2-8% at 48 h to 9-19% at 144 h (P < 0.01 for at least one time-point per QQS-E strain from the 48-h time-point onward, Figure 3C ). Juxtaposed to this increase in protein content, starch content was decreased in the QQS-E strains, decreasing by up to 13-24% at 144 h (P < 0.01 for at least one time-point per QQS-E strain from the 96-h time-point onward; Figures 3E, F ). Hence, these data establish that the transgenic expression of QQS affects C and N partitioning in C. reinhardtii, and this is accomplished even though this organism does not encode a recognizable NF-YA homolog.

To further evaluate the effect of QQS on C and N allocation, we used another genetically simpler, unicellular organism (i.e., S. cerevisiae) as the host for similar transgenesis experiments. In S. cerevisiae, the NF-Y complex is known as the Heme-Activated Protein (HAP) complex, which is composed of the three structural subunits, HAP2, HAP3, and HAP5 (Olesen et al., 1987; Hahn and Guarente, 1988; Forsburg and Guarente, 1989; McNabb et al., 1995), homologs of NF-YA, NF-YB and NF-YC, respectively ( Figure 1A ). In addition, in yeast a fourth protein (HAP4) binds to the HAP2/HAP3/HAP5 DNA-binding complex, and stimulates transcription (McNabb and Pinto, 2005). In contrast, plant genomes do not encode a HAP4 homolog. Using this system therefore, we evaluated the ability of QQS to interact with HAP5, the NF-YC homolog. The multisequence alignment shown in Supplementary Figure S2 identifies the homology between HAP5 and the Arabidopsis, C.reinhardtii and human NF-YC homologs, illustrating the conservation that is shared among these proteins in the QQS binding domain (Li et al., 2015); the yeast HAP5 protein shares 67% identity and 83% similarity with the QQS binding domain of AtNF-YC4.

We genetically evaluated if QQS interacts with the yeast HAP complex by expressing it or overexpressing HAP5 subunit in different S. cerevisiae backgrounds that carried the knockout (KO) mutations in the HAP subunit genes, and compared the resulting strains to the control recipient strains ( Figure 1 ). Mutant strains carrying KO mutant alleles in the four HAP subunit genes were obtained in both MATa and MATα mating types and these were all viable. Using an intermating strategy illustrated in Figure 1B for generating the hap2,3 double mutant, we generated haploid strains that carried all possible combinations of the double mutant, triple mutant, and quadruple mutant strains ( Figure 1C ). The resulting collection of 15 mutant strains and the WT strain were subsequently recipients for three expression vectors. One of these was the pAG426 plasmid, which served as the control in all subsequent experiments. The other two vectors that were used in these experiments expressed either the QQS gene (i.e., QQS-E strains) or the HAP5 gene (i.e., HAP5-OE strains), both being under the transcriptional regulation of the GPD-promoter that was carried by the pAG426 base-vector.

The resulting 48 strains ( Figure 1C ) were evaluated to ascertain the effect of these genetic manipulations on growth. Specifically, we monitored the rate of change in the optical density of cultures of each strain, from which doubling time was calculated. Supplementary Figure S5A plots the inverse of doubling times for the control and hap2, hap3, hap4 and hap5 single, double, triple and quadruple combination mutant strains; note all these strains carried the pAG426 empty vector plasmid specifically for the purpose of serving as the control strains for the QQS-E and HAP5-OE strains. The growth of all these mutant strains was indistinguishable from the WT (i.e., P > 0.1).

We also determined the effect on cell growth when QQS was expressed ( Supplementary Figure S5B ) or HAP5 overexpressed ( Supplementary Figure S5C ) in each of these mutant strains. The growth of the majority of these strains was unaffected by the expression or overexpression of either QQS or HAP5. The exceptions were the hap2,4 double mutant strains, which showed a ~12% increase in growth in response to QQS expression (P = 0.03), whereas the hap3 mutant strain showed a ~8% reduction in growth rate (P = 0.049). Although growth of additional mutant strains was statistically significantly affected by the overexpression of HAP5 (namely the WT (P = 0.042), hap2 (P = 0.041), hap5 (P = 0.03), hap2,4 (P = 0.02), hap4,5 (P = 0.041), and hap2,4,5 (P = 0.050)), in all cases, these effects were minor with only an 8 to 20% increase in growth rate.

In yeast, the major forms of C storage are glycogen and fatty acids, which together can account for approximately 10-20% of the wet weight of the cells (Quain and Tubb, 1983; Suutari et al., 1990; Cahill et al., 2000). To obtain an understanding of the role of the HAP complex in determining C and N allocation, we ascertained protein, glycogen and fatty acid content in yeast strains that carried hap2, hap3, hap4 and/or hap5 KO mutant alleles ( Figure 4 ). The data shown in Figures 4A, B indicate that the hap5 mutation has the most statistically consistent effect on protein and glycogen levels, increasing glycogen content by ~25% (P = 0.019) and decreasing protein content by a similar amount (~20%; P = 0.027). Parallel analyses of all potential double, triple and quadruple mutants reinforce the importance of the HAP5 gene in regulating protein and glycogen levels ( Figures 4D, E ). Specifically, any of the double, triple and quadruple mutants that included the hap5 mutation showed increased glycogen levels (10-20% increase) ( Figure 4E ), while some of these mutants also expressed decreased protein levels (e.g., hap2,5 and hap3,5) ( Figure 4D ). The parallel analysis of the fatty acid content in these KO mutants indicates that these HAP subunits do not appear to have statistically significant impacts in total fatty acid content (P > 0.05) ( Figure 4C ).

The QQS-E and HAP5-OE strains developed in the WT background expressed a ~40% (P = 0.045) and ~50% (P = 0.095) increase in protein content, respectively ( Figures 5A, B ), and reduced glycogen content by similar amounts ( Figures 6A, B ). In contrast, these genetic manipulations had no significant impact on fatty acid content ( Supplementary Figure S6A, B ). These effects were further dissected by evaluating protein ( Figure 5 ), glycogen ( Figure 6 ), and fatty acid ( Supplementary Figure S6 ) content in the QQS-E and HAP5-OE strains developed in genetic backgrounds that carried hap2, hap3, hap4, and hap5 single, double, triple and quadruple KO mutant combinations. The effect of expressing QQS or HAP5 on the fatty acid content of most of these mutant strains was statistically indistinguishable from the recipient strains (P > 0.1) ( Supplementary Figure S6 ). The two exceptions were the ~15% increase in the fatty acid content of the QQS-E strain in the hap3 mutant background (P = 0.062) and of the HAP5-OE strain in the hap5 mutant background (P = 0.024).

However, the QQS-E and HAP5-OE strains in these mutant backgrounds showed distinct changes in protein content, indicating a role for HAP2 and HAP3, and possibly also for the HAP4 subunit. Specifically, protein content was increased by the expression of QQS or HAP5 in the WT strain, and this modulation in protein content also occurs in the strains carrying the hap4 KO mutant allele [P = 0.069 for QQS-E ( Figure 5A ), and P = 0.083 for the HAP5-OE strains ( Figure 5B )]. However, because of somewhat larger variation in the data, we cannot exclude the possibility that a functional HAP4 allele may be required to increase protein content by the expression of either QQS or HAP5. These increases in protein content did not occur in the QQS-E (P > 0.1) and HAP5-OE (P > 0.1) strains that carry either hap2 or hap3 KO mutant allele. Because prior studies had established that QQS mediates its effect by interacting with the NF-YC4 (HAP5) subunit (Li et al., 2015), it was expected that QQS-E could not increase protein content in the hap5 KO mutant allele. However, the QQS-E and HAP5-OE in hap2 or hap3 KO mutants affected protein content, similar to in the hap5 KO mutant. Therefore, these results suggest that in addition to the need for HAP5, the QQS-effect on protein content may also involve the HAP2 or HAP3 or potentially the HAP4 subunits.

The parallel analyses of glycogen content in these QQS-E ( Figure 6A ) and HAP5-OE ( Figure 6B ) strains indicate that the effect is more complex. Specifically, expressing QQS in any of the four hap single mutants decreased glycogen content by 38% in hap2 (P = 0. 038), 51% in hap3 (P = 0. 053), 49% in hap4 (P = 0. 08), and 64% in hap5 (P = 0. 001) ( Figure 6A ). However, when QQS was expressed in any combinations of the hap2 or hap3 KO alleles, where a WT HAP5 allele was still active, glycogen levels were decreased; specifically, by 44% in hap2,3 (P = 0.027), 43% in hap2,4 (P = 0.002), 41% in hap3,4 (P = 0.003) double mutants, and 50% in hap2,3,4 (P = 0.003) triple mutant. Moreover, there was no statistically significant change in glycogen levels in QQS-E strains that combined the hap5 KO mutant allele with any combination of hap2 or hap3 KO alleles; specifically, the hap2,5 (P = 0.192), or hap3,5 (P = 0.075) double mutants, or the hap2,3,5 (P = 0.14) triple mutant. However, because of some larger variation in the data, we cannot exclude the possibility that a functional HAP4 allele may be required to affect glycogen content. Collectively therefore, these results suggest that when HAP5 is absent, HAP2, HAP3 and probably HAP4 subunits may have redundant functions to enable QQS to affect glycogen levels. This is consistent with the finding that when QQS was expressed in the hap2,3,4,5 quadruple mutant, the glycogen level was only reduced by ~10% (P = 0.046). The small change in glycogen level displayed by this quadruple mutant strain may be indicative of another unknown mechanism that enables QQS to affect the glycogen content in the absence of any HAP subunits.

Finally, overexpression of HAP5 significantly decreased glycogen levels by between 30% and 70% in most strains (P < 0.05), except the hap3,4 and hap3,5 double mutants (P < 0.1), and these effects are consistent among all the different hap mutants that were evaluated, similar to the effect of overexpressing HAP5 in WT yeast ( Figure 6B ). These effects of HAP5 overexpression on glycogen content in yeast parallel our initial characterizations of QQS affecting starch content in plants, by activating the plant homolog of HAP5, namely AtNF-YC4 (Li et al., 2015).