The Dunaliella strains used in this study (D. salina CCAP 19/18, D. viridis dumsii, D. tertiolecta CCMP 364, D. primolecta UTEX LB1000) were obtained from the Culture Collection of Algae and Protozoa (CCAP¹), National Center for Marine Algae and Microbiota (NCMA²), or UTEX³.

Strains were grown in 125 ml Erlenmeyer flasks containing modified Ben-Amotz media (mBA, pH 7.5) (Ben-Amotz et al., 1989; Srirangan et al., 2015) and maintained in exponential phase prior to inoculation of growth experiments. All cultures (except those used in uptake assays, see below) were grown at 21°C under continuous cool white fluorescent light with an intensity of 150 μmol photons m^-2 s^-1 at the culture surface.

Nitrogen source growth experiments were carried out in batch culture using filter-sterilized mBA lacking nitrogen (mBA -N) medium in 4 ml volumes within 12-well polystyrene tissue culture plates. The cell density of Dunaliella cultures was quantified using a TC10 Automated Cell Counter (Bio-Rad). For each biological replicate, 8 × 10⁷ cells were harvested by centrifugation at 3,441 × g for 2 min. The supernatant was discarded and cultures were washed in mBA -N. These cell suspensions were again centrifuged, the supernatant was discarded, and cells were re-suspended in 1 ml of mBA -N. An inoculum of 4.0 × 10⁶ cells in a volume of 50 μl was used to seed each well containing media. For cultures of D. salina, only 2 × 10⁶ cells were used. Cells were mixed and the density of each well was recorded. Plates were then wrapped with parafilm and grown for 144 h under the conditions described above. At 0 and 144 h post-inoculation, cultures were mixed and the cell densities and diameter were determined. There were four biological replicates used for each Dunaliella strain.

High affinity urea transport in plants is carried out by homologs of the yeast DUR3 urea transporter (Kojima et al., 2006). To find Dunaliella DUR3 homologs, yeast DUR3 (yeastgenome.org: YHL016C) and Arabidopsis thaliana AtDUR3 (TAIR: AT5G45380) protein sequences were used as BLAST queries against the D. salina CCAP 19/18 genome⁴ and against the assembled transcriptomes of D. viridis strain dumsii (Srirangan et al., 2015), D. tertiolecta CCMP 364, and D. primolecta UTEX 1000 (Matasci et al., 2014). Blast hits that also contained the conserved domain PF00474 were considered to be DUR3 homologs (Finn et al., 2016).

The metabolism of urea to ammonia in organisms can be accomplished via two different routes. The first consists of a nickel dependent urease producing ammonia and carbon dioxide. The second method uses urea carboxylase and allophanate hydrolase to convert urea to ammonium and bicarbonate through an allophanate intermediate (Fan et al., 2012). The urease protein sequence from A. thaliana (TAIR: AT1G67550) and the urea carboxylase (Phytozome_v11: Cre08.g360050.t1.1) and allophanate hydrolase (Phytozome_v11: Cre08.g360100.t1.2) protein sequences from Chlamydomonas reinhardtii (Merchant et al., 2007) were used as BLAST queries against the D. salina CCAP 19/18 genome⁵ and against the assembled transcriptomes of D. viridis strain dumsii (Srirangan et al., 2015), D. tertiolecta CCMP 364, and D. primolecta UTEX 1000 (Matasci et al., 2014).

Free NH₄⁺ in growth medium was quantified using the phenate method amended for use with mBA medium in a 96 well plate (Eppley et al., 1969). NH₄⁺ standards of 5 to 2,000 μg ml^-1 were made using NH₄Cl in mBA -N. Each sample or NH₄⁺ standard was added in triplicate 50 μl aliquots to 96 well microplate. To each aliquot, 40 μl of 20% (w/v) crystalline phenol in EtOH and 40 μl 0.5% (w/v) sodium nitroprusside in distilled H₂0 were added and mixed. A 10 ml solution of 20% (w/v) sodium citrate and 2% NaOH in distilled H₂O was combined with 3.5 ml of commercial bleach (7.25% sodium hypochlorite) and 70 μl of the resulting solution was added to each well and immediately mixed 10 times. Plates were developed for 1 h in darkness. Free NH₄⁺ was quantified by measuring absorption at 630 nm.

Amino acid uptake assays for Dunaliella spp. were amended from a yeast protocol (Su et al., 2004). Briefly, 50 ml Dunaliella cultures in mBA were maintained in a Conviron growth chamber (model number ATC60) under long-day condition (16 h light/8 h dark, 22°C/18°C, respectively) and 135 μmol photons m^-2 s^-1 in exponential phase prior to inoculation of growth medium containing different nitrogen sources. From each growth medium, either 5 × 10⁷ (Figure 5A) or 2 × 10⁸ (Figure 5B) cells were harvested by centrifugation and suspended in 1 ml of mBA -N. Each cell suspension was dispensed into 100 μl aliquots. Aliquots were maintained on ice until 10 μl of 1 M glucose was added to each suspension, after which cells were mixed and incubated for 5 min under cool white fluorescent light at 25°C (40 μmol photons m^-2 s^-1). Uptake solutions were prepared by mixing non-labeled amino acid (final concentration of 1 mM) with two microliters of the following labeled amino acids at mCi ml^-1: L-cystine [3,3′-¹⁴C], L-histidine [2,5-³H], and L-glutamine [3,4-³H(N)]. Uptake solution was mixed with incubated cell aliquots, yielding a final specific activity of ∼91 nCi μl^-1. A portion of each sample (50 μl) was taken at several time points and added to 5 ml of 1 M NaCl on a DHI Filtration Manifold (CAT# EQU-FM-10X20-SET) fitted with 24 mm glass filters (Whatman CAT# 1822-024). The solution was removed by application of vacuum, and the washing was repeated two more times with 5 ml 1 M NaCl. The filters containing the cells were loaded into scintillation vials, 500 μl of commercial bleach (5% sodium hypochlorite) was added to each scintillation vial. Radioactivity in each vial was measured using a Beckman-Coulter scintillation counter (LS6500 Multi-Purpose Scintillation Counter).

The amino acid composition of D. viridis cells was determined using 8 ml of culture grown for 144 h. Samples were centrifuged for 10 min at 16,000 × g. Supernatant was removed and cells were lysed in 400 μl of distilled water. Cell lysates were stored at -20°C until derivatization. For total amino acid content analysis, 50 μl of frozen cell lysates were thawed and hydrolyzed using 1 ml vacuum hydrolysis tubes (Life Technologies) containing 400 μl of 6 N HCl with 0.02% β-mercaptoethanol. Hydrolysis was carried out under vacuum and 105°C for 20 h and terminated using 400 μl of 6 N NaOH.

Cell lysates and hydrolyzed cell lysates were separately derivatized with AccQ – Tag (Waters). Derivatized samples were run on a Waters UPLC system in reverse phase mode. Standard curves for each amino acid and ammonium were generated using amino acid standards from Sigma Aldrich (Catalog no.: AA-S-18) for a working range of 0 to 50 mM.

For all data presented, at least three biological replicates were used. Statistical analysis was performed in Microsoft Excel. A Student’s t-test was used for all comparisons between two means. Tukey’s HSD was used to establish significant groupings when performing multiple comparisons.

A screen was conducted to determine whether various organic and inorganic nitrogen substrates could support growth of D. viridis (Figure 2). Two surprising results of this nutrient screen were that (i) urea did not support growth of D. viridis and that (ii) NH₄⁺ in excess of 1 mM was generally inhibitory or toxic to all Dunaliella species tested (Figure 3 and Table 1). Other Dunaliella species were able to grow to high density when supplemented with urea, and also possessed gene sequence encoding likely transporters and enzymes necessary for urea utilization (Table 1). In contrast, D. viridis lacked any available sequence that might code for a urea transporter, and also does not appear to possess a sufficient set of enzymes needed for urea conversion (Table 1). NH₄⁺ above 2 mM was inhibitory regardless of the species tested, a situation previously observed only in D. tertiolecta (Figure 3; Chen et al., 2011). Ribonucleoside and nucleobase monomers were also evaluated as a nitrogen source, but did not rescue growth of nitrogen starved D. viridis (Supplementary Figure S1).

Our screen revealed that compared to a nitrogen starvation control, cultures of D. viridis could reach high cell densities when supplemented with the amino acids Gln, His, Cys, and Trp (Figures 2, 4 and Table 2). The other 16 proteinogenic amino acids either had no observable effect on growth, or caused a small decrease in final cell density. However, glycine was an exception under one condition. When glycine (5 mM) was mixed with KNO₃ (5 mM) cultures had a final cell density and neutral lipid productivity nearly 50% greater than cultures grown with KNO₃ alone (Supplementary Figure S2). This initial nitrogen source screen was expanded to include three other species of Dunaliella (D. salina, D. tertiolecta, D. primolecta). We found that all four species consistently reached higher cell densities than nitrogen starvation controls when supplied with Gln, His, Cys, Trp (Table 2). Just as for D. viridis, the other 16 proteinogenic amino acids had no observable effect on growth.

There were differences in the level of metabolites accumulated by cells of each Dunaliella species, which was significant both when comparing between species and when comparing nitrogen supplementation treatments (Figure 4). While nitrogen starvation was generally associated with lower cell densities and decreased chlorophyll content, nitrogen-starved D. salina and D. viridis accumulated different levels of carbon storage metabolites and protein (Figure 4). Nitrogen-starved cultures of both D. salina and D. viridis accumulated higher amounts of neutral lipids on a cell basis, but D. viridis cells accumulated about five times more neutral lipids than KNO3– fed controls. Cells of D. salina accumulated about three times more carbohydrates under nitrogen starvation, while carbohydrate levels were unchanged in cells of D. viridis subjected to the same treatment (Figure 4). Cell protein abundance was significantly lower during nitrogen starvation in D. salina, but not in D. viridis.

Gln, His, Cys and Trp supplementation altered the metabolite profiles of Dunaliella in a species specific manner (Figure 4). While supplementation with each of four amino acids recovered the growth of Dunaliella, the color of these cultures was often different compared to KNO₃ controls (Supplementary Figure S3). This color change was obvious even when comparing cultures of equivalent cell density. Consequently, each amino acid altered the metabolite profile of species in ways that were unique from both controls and other amino acids.

Gln supplementation yielded final cell densities and cell protein content that were nearly as high as a KNO₃ control (Figure 4). However, for both D. viridis and D. salina the chlorophyll content in Gln supplemented cultures tended to be lower, and Gln supplemented cultures D. viridis were visibly yellow (Supplementary Figure S3). Compared to KNO₃ controls, Gln supplemented cultures had almost double cellular carbohydrate content and consequently produced at least 50% more carbohydrates in total. Gln was also associated with a twofold higher level of cellular triacylglycerol production in D. viridis (Figure 4).

Although cultures supplemented with Cys did not grow as well as those supplemented with Gln, they accumulated similar levels of metabolites (Figure 4). Like Gln cultures, Cys supplemented cultures were also visibly yellow (Supplementary Figure S3). Neutral lipids, carbohydrates, and chlorophyll were present at equivalent levels in both treatments in D. viridis. However, Cys supplementation produced higher cell protein content than Gln or KNO₃ cultures (Figure 4).

The growth rate of Dunaliella cultures supplemented with His remained low for all tested species (Table 2). Protein levels of His-fed cultures were only marginally higher than nitrogen-starved controls (Figure 4). As was the case for Gln and Cys supplementation, His-supplemented cultures had low chlorophyll content, giving rise to a yellowish appearance (Supplementary Figure S3).

Trp added to growth media produced a vibrant orange color after incubation for several days and exposure to light (Supplementary Figure S3). This effect also occurred in sterile media. Trp-supplemented cultures were often contaminated with bacteria [data not shown], which was never seen for the other amino acids. Data could only be obtained for cultures that had undergone a rigorous selection procedure to isolate cultures free of contaminating bacteria (Supplemental Methods). Cultures supplemented with Trp were unique in that both protein and chlorophyll levels tended to be highest in these treatments (Figure 4). While Trp supplementation marginally increased the carbohydrate content of D. salina, it reduced the neutral lipid content of D. viridis cells to 31% of KNO₃ fed cells. Trp fed cultures produced one fifth the volume of neutral lipids compared to KNO₃ controls. This effect was not observed in other species, but occurred even if D. viridis was simultaneously cultured with KNO₃ and Trp (Supplementary Figure S2).

The mechanism by which Dunaliella spp. are able to obtain nitrogen from the amino acids His, Gln, Trp, and Cys was investigated using the ecotype D. viridis as a model. Routes supported by prior literature include transport of amino acids across the cell membrane and recovery of NH₄⁺ separated from amino acids either by abiotic oxidation chemistry or the activity of a deaminating enzyme (Kirk and Kirk, 1978a; Abraham and Podell, 1981; Palenik and Morel, 1990, 1991; Gracanin et al., 2009).

The uptake of radiolabeled Gln, cystine, and His by D. viridis was measured over a period of 120 min (Figure 5A). Cystine was used as a proxy for Cys because of the relative instability of Cys, which rapidly oxidizes to cystine in aqueous solution (Kendall and Nord, 1926). No uptake of cystine or Gln was detected, only His was significantly taken up. This was true even if the culture used for the uptake assay was nitrogen-starved for 72 h prior to the assay (Data not shown). The uptake for radiolabeled His was approximately linear over a period of 20 min (Figure 5B).

The degree of NH₄⁺ production by sterile His, Gln, Glu, Cys, and Trp-containing media was assessed using the same light and temperature conditions that are used for growth of Dunaliella cultures. Gln, Cys, and Trp media generated increasing NH₄⁺ concentrations over time (Figure 6). In contrast, His-containing media gave an initially high signal, but the apparent NH₄⁺ concentration of this media did not significantly change over time. Glu was used as a control and did not produce any detectable amount of free NH₄⁺.

The NMR spectra of media was used to ascertain the fate of Gln in culture. Spent media from D. viridis cultivated with mBA -N containing Gln revealed that peaks corresponding to Gln were present at all time points, and that the absolute value of these peaks did not change significantly (Figure 7). However, a set of peaks corresponding to pyroglutamate appear prominently after 24 h, and the strength of this signal increased at 48 h (Figure 7). The estimated rate of pyroglutamate formation in this experiment was considerably higher than that obtained for the abiotic production of NH₄⁺ from Gln-containing media (Figures 6, 7).

The amino acid content of D. viridis was obtained in order to ascertain what proportion of the biomass could be recycled using Gln, His, Trp, and Cys as fertilizer supplementation. There was little variation in the overall amino acid content of D. viridis even between nitrogen-depleted and KNO3– supplemented cultures (Table 3). The sum of relative Gln, His, Trp, and Cys fractions was at most 16% of the total amino acid fraction.

Dunaliella viridis could not use urea as a nitrogen source (Figure 2 and Table 1). This is in contrast to other Dunaliella spp., which recover growth when supplemented with urea (Table 1; Hellio and Le Gal 1998, 1999). The use of urea as a nitrogen source requires at least two components: urea uptake via a membrane transport protein and subsequent enzymatic degradation into HCO₃^-/ CO₂ and NH₄⁺ (Fan et al., 2012). NH₄⁺ released by these two activities can then be assimilated via the GS-GOGAT pathway. We therefore probed the genomes of each Dunaliella species for the presence of sequence encoding putative urea transporters, urease, urea carboxylase, and allophanate hydrolase. No predicted proteins similar to urea transporters could be detected in the available genomic resources for D. viridis (Table 1). However, sequences putatively coding for these transporters could be identified in the other species of Dunaliella. The surveyed Dunaliella species do not contain any coding sequences for urease (Table 1). D. salina, D. tertiolecta, and D. primolecta contain sequences for urea carboxylase and allophanate hydrolase, while D. viridis only contains a sequence for urea carboxylase (Table 1). D. salina, D. tertiolecta, and D. primolecta grow when supplemented with urea as a sole nitrogen source and possess three necessary proteins that enable urea acquisition and metabolism. In contrast, this particular ecotype of D. viridis likely cannot utilize urea because it does not possess a transporter to carry urea across the cell membrane for subsequent assimilation. Even if D. viridis had an unknown protein capable of urea transport, it could convert urea to allophanate, but could not complete the conversion to ammonium and bicarbonate unless it possesses an unknown or alternate enzyme rescuing the activity of allophanate hydrolase.

One caveat is that the genomic resources available for D. viridis are derived entirely from RNA sequencing. Urea transporter and allophanate hydrolase mRNAs may not have been captured in the samples used for sequencing. We probed for the presence of urea transporter and allophanate hydrolase genes in the genomic DNA of D. viridis using consensus primers derived from the sequence of other Dunaliella spp. No putative urea transporter or allophanate hydrolase genes were amplified from D. viridis genomic DNA using D. salina, D. tertiolecta, and D. primolecta genomic DNA as positive controls (Data not shown). It is therefore highly unlikely that D. viridis possesses sequence coding for these two genes.

Urea carboxylase, allophanate hydrolase, and a urea transporter are grouped sequentially together within the D. salina genome⁸. This gene arrangement is also reflected on chromosome 8 of the C. reinhardtii genome (Merchant et al., 2007) and scaffold 3 of Volvox carteri (Prochnik et al., 2010), both species within the Chlamydomonadales order along with Dunaliella. If this syntenic relationship is conserved within the Chlamydomonadales order, this block of genes may have been deleted in D. viridis.

Although NH₄⁺ is a useful source of nitrogen for photosynthetic organisms, it is also toxic at high concentrations. NH₄⁺ toxicity is due to the decoupling effect of NH₄⁺/ NH₃ equilibrium, which can disturb membrane proton and charge gradients (Krogmann et al., 1959). Surprisingly, NH₄⁺ supplementation of the Dunaliella cultures used in this study is toxic at concentrations less than previously reported to be suitable (5 mM) (Figure 3; Giordano et al., 1994; Giordano, 1997). Instead, we found growth maxima for some Dunaliella spp. at 1 mM NH₄Cl, with higher concentrations causing inhibition of growth or cell death. It is therefore necessary to understand what factors contribute to the ability of some Dunaliella spp. to tolerate this nutrient better than others. However, there are several key differences in the methods used to cultivate Dunaliella spp. between our study and those prior that may explain this discrepancy.

Acclimation of Dunaliella spp. to the presence of NH₄⁺ may affect the ability of these organisms to resist gradient decoupling effects. Dunaliella spp. have a long life cycle (0.47 – 1.22 divisions per 24 h) (Oren, 2005), and therefore may require substantial time to shift genetic and protein machinery when first exposed to NH₄⁺. Prior studies involving higher amounts of NH₄⁺ were materially different than ours in that the Dunaliella cultures used were acclimated to NH₄Cl as a sole nitrogen source prior to the start of experimentation (Giordano et al., 1994; Giordano, 1997). In contrast, the cultures used in our study were grown using KNO₃ as a sole nitrogen source prior to experimentation.

Another important factor affecting NH₄⁺ toxicity is the amount of headspace and airflow in the container used for algal cultivation. NH₄⁺/ NH₃ equilibrium increasingly favors the production of NH₃, a volatile gas, under more basic conditions. The algal cultivation media used in this study begins neutral (pH 7.5), but becomes basic (pH 10) after 72 h. It is likely that as Dunaliella growth experiments progress, NH₃ leaves the liquid phase of the cultures and lowers the effective NH₄⁺ concentration in the media. Additionally, this process should proceed to an equilibrium given a limited space for gas expansion. As NH₃ volatilizes out of liquid over time and occupies the headspace of its container, NH₄⁺/ NH₃ equilibrium will be further driven toward NH₃ production, likely lowering the effective NH₄⁺ concentration. This means that the headspace of its cultivation vessel likely has a large effect on the NH₄⁺/ NH₃ equilibrium and therefore the effective NH₄⁺ concentration. Compared to other studies using NH₄⁺ supplementation in Dunaliella growth experiments, the 12 well tissue culture plates used in this study have less headspace. It is therefore reasonable to assume that the reason our cultures were more susceptible to NH₄⁺ toxicity was that they were subjected to higher effective levels of NH₄⁺.

Gln and Cys support growth of Dunaliella spp., but also increase the amount of carbon storage metabolites of cells compared to KNO₃ fertilizer controls. It is unlikely to be the case that Dunaliella spp. can access or assimilate the carbon from Gln and Cys, as we found no evidence that these amino acids cross the cell membrane. Therefore, one likely reason for increased carbon storage is that Gln and Cys actually supply NH₄⁺ via their oxidative degradation, which is a more efficient source of nitrogen than KNO₃. Cells must convert KNO₃ to NH₄⁺ before nitrogen can be fixed to Glu via the GS-GOGAT cycle. This conversion process requires eight electrons, which are ultimately derived by light capture during photosynthesis. When NH₄⁺ is assimilated, the eight electrons that would have otherwise been needed for nitrate assimilation can be either utilized elsewhere, i.e., for carbon fixation, or the cell can reduce its rate of electron capture from water oxidation by lowering light harvesting activity and chlorophyll content. This is consistent with the observed levels of chlorophyll in Gln and Cys-supplemented cultures of both D. salina and D. viridis, which possessed significantly lower cellular chlorophyll levels than KNO₃ fertilizer controls (Figure 4 and Supplementary Figure S3). Surprisingly, Gln and Cys-supplemented cultures produced higher cell concentrations of carbohydrates and neutral lipids while retaining similar amounts of protein compared to KNO₃ controls. NH₄⁺ utilization is consistent with these metabolic effects, as this scenario allows that less electron capture could exist simultaneously with increased carbon capture.

Although Trp releases NH₄⁺ in solution like Cys and Gln, Trp supplementation resulted in uniquely elevated chlorophyll and protein levels as well as diminished neutral lipid content in D. viridis. These cultures were also commonly susceptible to contamination by bacteria, unlike other nutrient regimes we tested. Trp is also unique because it degrades into a variety of products due to light catalyzed oxidation of the pyrrole ring (Gracanin et al., 2009). This process is fast enough to be visually apparent, as cultures were visibly orange after as little as 24 h. Therefore, in addition to NH₄⁺, there remains the possibility that one of many Trp degradation products or Trp itself is sensed or acquired by Dunaliella spp.

Extracellular deamination of amino acids by an enzyme is known to occur in microalgae (Munoz-Blanco et al., 1990; Palenik and Morel, 1990, 1991; Piedras et al., 1992; Vallon et al., 1993). Although there was no prior evidence for their existence in Dunaliella spp., we did not rule out the possibility that at least one of those enzymes was present and active. Dunaliella spp. are in the same Order as C. reinhardtii, which possesses a non-specific L-amino acid oxidase (LAO1) located in the periplasmic space that deaminates a wide range of amino acids, releasing NH₄⁺ and corresponding keto-acids (Munoz-Blanco et al., 1990; Piedras et al., 1992; Vallon et al., 1993). We probed the genomes of four species of Dunaliella and could not find any protein-coding genes of significant similarity to LAO1 (Data not shown). The fact that Dunaliella cells grew only in the presence of amino acids subject to abiotic deamination does not support the hypothesis of the existence of an amino acid oxidase. Additionally, our NMR data reveal that cultures of glutamine-fed D. viridis form pyroglutamate outside the cell (Figure 7). If activity analogous to LAO1 existed we expect to instead find a signal corresponding to 2-keto-glutaramic acid.

Uptake of His has now been demonstrated in two species of Dunaliella and appears to support growth in each of the species tested (Table 2 and Figure 5; Hellio and Le Gal 1998, 1999). We therefore speculate that His uptake and utilization may be a general feature of this genus. Although His is transported across the cell membrane, His supplementation conferred little improvement in growth or accumulation of major carbon or nitrogen metabolites over cultures that were starved for nitrogen. This is surprising, as transported His should functionally supply nitrogen and carbon via conversion by histidine ammonia lyase. However, the kinetics of His uptake, the identity of the transporting protein, and the mechanism of transport are unknown. This situation is intriguing because a parallel situation exists in the model alga C. reinhardtii. C. reinhardtii efficiently transports only one amino acid, arginine, which like histidine is positively charged, and the identity and mechanism of transport and assimilation of arginine are unknown. This similarity might yield an evolutionary insight if the respective transport mechanisms could be identified.

It is unlikely that His is able to supply nitrogen via enzymatic or abiotic degradation outside the cell, because light catalyzed degradation of His does not yield free NH₄⁺, but aspartate and urea (Tomita et al., 1969). However, aspartate and urea do not support growth of D. viridis. Therefore, only His itself could supply nitrogen to cells of D. viridis. Additionally, His supplementation only marginally improved growth even in cultures that can use urea as a nitrogen source.