Biosynthesis and Genetic Tuning of Guanine Crystals via Metabolic Engineering of Yeast

ABSTRACT

Purine crystals, such as guanine, generate diverse optical phenomena in animals and function as important nitrogen‐storage reservoirs in microalgae. Here, we demonstrate the formation of guanine crystals in Brewer's yeast. Whilst guanine crystals do not naturally contribute to the physiology of this yeast species, we show how crystallization may be artificially induced and then use genetic manipulation to control crystal properties. Specifically, we use genetic knockouts of purine salvage enzymes to modify crystal morphology, size and crystalline order. Deletion of Hpt1 and Gud1, which convert guanine to guanosine monophosphate and xanthine, respectively, transforms spherical guanine nanoparticles in wild‐type cells into crystalline dumbbells and nanoflowers. While we are lacking a definitive mechanism to rationalize this transformation, we hypothesize that changes in crystal morphology are likely driven by genetically induced shifts in the local supersaturation and metabolite composition within the yeast vacuole. Leveraging yeast's genetic programmability and ubiquitous industrial usage, our findings provide a powerful biotechnological strategy for the biosynthesis of optically functional guanine crystals, as environmentally friendly alternatives to toxic pigments in a range of industries.

We demonstrate how yeast cells can be harnessed as cellular factories for the synthesis of optically functional guanine crystals, where genetic editing is used to tune crystal size, order and optical properties. This demonstrates a powerful new biosynthetic approach for the development of sustainable and biocompatible optical pigments, which utilizes the programming capabilities of cells.

Highly reflective guanine crystals generate diverse optical phenomena in animals [1, 2, 3] and are also used by microalgae as high‐capacity nitrogen (N) storage reservoirs [4, 5, 6]. Microalgae synthesize guanine crystals de novo from inorganic N‐nutrients (e.g., nitrate, ammonium) [4, 6] or by direct uptake of N‐heterocycles from their environment [5]. During periods of nutrient depletion, the cells dissolve the crystals, releasing organic N for cell growth [4]. Microbial crystallization [7, 8] represents an unexplored frontier for organic biomineralization, opening up exciting opportunities in biotechnology. Recently, we showed that dinoflagellate cells may be harnessed as “factories” for the biosynthesis of difficult‐to‐crystallize molecular materials [5], which hold promise as sustainable optical pigments in paints, pharmaceuticals and cosmetics [9]. Here, we explore the formation of guanine crystals in Brewer's yeast, Saccharomyces cerevisiae. Yeast offers key advantages over dinoflagellates in the biosynthesis of crystalline materials due to its repertoire of genetic manipulation tools and use in industrial fermentation [10]. We show how deletion of purine salvage pathway genes can be used to modulate the size, morphologies and crystallinity of crystalline guanine spherulites in yeast. Previous work showed how the scattering properties of molecular crystalline spherulites [11] depend on their size [12], birefringence [9, 13], morphologies [14], crystallinity and chemical composition [15]. Thus, this work provides a proof‐of‐concept for utilizing genetic editing to tailor the synthesis and optical properties of molecular materials.

To investigate guanine crystal formation in yeast, we grew two strains (W1588 and BY4741) of S. cerevisiae (Figures 1A and S1) to the exponential phase in YPD medium before replacing the medium with a 16 µM guanine solution in 4% glucose. After medium replacement, birefringent spherical particles (Figure 1A(i, ii)) and occasionally dumbbells were observed rapidly rotating in the central lumen of the cell vacuole (Figure S1, Movie S1). The median diameter of birefringent particles from light microscopy at 4.5 h was 497 ± 126 nm (Figure 1B). In vivo Raman spectroscopy confirmed these particles are composed of β‐guanine crystals, which exhibit significant disorder—evidenced by broad lattice mode Raman peaks (Figure S1) [16, 17]. The number of crystal‐positive cells reached a maximum of 65% at 1.5 h before dropping (Figure 1C), as accumulated guanine particles are dissolved and metabolized and/or exported out of the cell. Uptake experiments with other N‐heterocycles (e.g., xanthine, hypoxanthine, uric acid, adenine, cytosine, 5‐flurocytosine, theobromine and caffeine) did not result in crystal formation, indicating a guanine‐specific crystallization mechanism in S. cerevisiae. This contrasts with the early work of Roush [18, 19] and Svihla [20] who described the accumulation and crystallization of several purines in Candida utilis yeast. These observations are consistent with C. utilis being able to assimilate a much broader array of carbon and nitrogen sources than S. cerevisiae [21].

FIGURE 1.

Guanine crystal formation in S. cerevisiae. (A) Polarized light micrographs of yeast 4.5 h after medium replacement with guanine. (i) Control (no guanine), (ii) WT, insets; without (top) and with (bottom) polarizer, (iii) fcy2Δ, (iv) hpt1Δ, insets; Maltese cross extinction pattern (top) and dumbbell (bottom) and (v) gud1Δ. (B) Crystal size (the box indicates the interquartile range (25th–75th percentiles), the horizontal line marks the median, and the whiskers extend to the most extreme values within 1.5 × IQR; n is indicated below each box; Wilcoxon test, **** = p < 0.0001) and (C) crystal‐positive cells (bars represent the mean ± SD; % of n > 100 indicated for each time point; Student´s t‐test, * = p < 0.05) in the WT (16 µM guanine solution), as a function of time after medium replacement (measured by light microscopy) and a comparison with hpt1Δ and gud1Δ (2 µM guanine solution). (D) Normalized UV–vis spectra of the growth medium as a function of time after medium replacement. In the WT, a red‐shift in λ ₁ (from 246 to 250 nm), a blue‐shift in λ ₂ (from 274 to 268 nm) and a decrease in the λ ₁:λ ₂ ratio occurs in the medium over time. Mean of n = 3 is indicated. (E) In vivo Raman spectra of birefringent deposits in WT, hpt1Δ and gud1Δ yeast 1.5 h after feeding show the presence of β‐guanine crystals [16]. Spectral intensity was normalized against the intensity of the guanine ring‐breathing mode at 648 cm⁻¹. Dashed lines; the π‐stacking mode (106 cm⁻¹), H‐bonding mode (396 cm⁻¹) of β‐guanine and the ring‐breathing mode of guanine (648 cm⁻¹) [16].

To ascertain whether accumulated guanine crystals facilitate cell growth (as in microalgae) [4, 5], we compared the growth of S. cerevisiae in glucose solutions with and without guanine. The presence of guanine crystals did not allow for higher cell proliferation at any glucose concentration (Figure S2). Furthermore, only glucose, but not galactose, lactose or sucrose, enabled crystal formation (Figure S2). Also, in contrast to dinoflagellates [4, 5], crystallization only occurred in the exponential growth regime and not in the stationary phase (Figure S2). S. cerevisiae is a Crabtree‐positive yeast species which specializes in the rapid uptake of sugars, preferentially glucose, for fermentation [22]. Our results indicate that guanine crystallization is linked to glucose specific metabolism and that guanine uptake may occur as an incidental (non‐functional) by‐product of the high energy demands of cells during rapid cell growth. Another hypothesis is that guanine crystallization may occur as a stress response [23], where guanine crystal formation could prevent the buildup of high concentrations of cytotoxic metabolites such as guanosine monophosphate (GMP) [24].

Making use of S. cerevisiae as a genetic model, we then investigated how silencing purine transporter and salvage pathway genes affects crystallization. For this, knock‐out (KO) strains of; (i) FCY2, encoding the transporter purine‐cytosine permease [25], (ii) HPT1, encoding hypoxanthine‐guanine phosphoribosyltransferase, which metabolizes guanine to GMP [26], and (iii) GUD1, encoding guanine deaminase, which catabolizes guanine to xanthine and ammonia [6, 27], were generated by gene replacement with a NAT cassette. WT and KO cells were compared using a 2 µM guanine solution to increase the sensitivity of the phenotype screening.

Deletion of FCY2 resulted in complete inhibition of crystallization, showing that purine‐cytosine permease is crucial for guanine transport into the cells (Figure 1A(iii)) [25]. This deletion also severely affected cell morphology and size (Figure S3) [28]. Deletion of HPT1 resulted in the formation of significantly bigger crystals than in the WT (Figure 1A(iv),B, median diameters in WT and hpt1Δ; 485 and 761 nm at 4.5 h, respectively; see also Table S1), while the number of crystal‐positive cells remained unchanged (Figure 1C). Cell size and morphology was moderately affected (Figure S3) although the cells remained viable for at least 1 week following guanine feeding [28]. To investigate if the deletion of HPT1 can increase crystal size beyond the physiological boundaries of the WT at higher guanine concentrations, we incubated WT and hpt1Δ cells with 8 and 24 µM guanine for 1.5 and 4.5 h. In contrast to dinoflagellates (where guanine concentration and incubation time affect crystal size) [5], neither increases in time or guanine concentration allowed the WT to grow crystals as big as that observed after HPT1 deletion (Figure S3). Finally, deletion of GUD1 resulted in no detectable change in the crystal size (according to light microscopy screening) in the first 12 h compared to the WT (Figure 1A(v),B; see also Table S1). However, whereas guanine crystals are depleted in the WT at 12 h, gud1Δ cells retained the maximum quantity of crystals after 12 h (Figure 1C). Crystals were even observed 1 week after transfer to the guanine medium (Figure S3). These late‐stage phenotypic effects are consistent with Gud1 being active in stationary‐phase cultures [6, 27]. No change in cell size or morphology were observed (Figure S3).

To provide further insights into the dynamics of guanine accumulation and degradation, UV–vis spectroscopy was used to track changes in guanine in the growth medium supernatant over time (Figure 1D). The UV–vis spectrum of guanine is characterized by absorption maxima at 246 (λ ₁) and 274 nm (λ ₂), originating from π → π* electronic transitions [20, 29]. In the WT, the slight decrease in guanine concentration in the first 1.5 h shows that only a small fraction of guanine is converted into intracellular crystals and/or that guanine is exported back into the growth medium after uptake. From 4.5 to 48 h, during crystal degradation (Figure 1C), shifts in the positions and relative intensities of λ ₁ and λ ₂ (Figure 1D) may suggest that accumulated guanine is converted enzymatically to different metabolites (e.g., GMP by Hpt1 and xanthine by Gud1), which are expelled back into solution. In hpt1Δ, few changes in the UV–vis spectra were observed even at late time points, as the conversion of guanine (via GMP) to other metabolites is inhibited and the medium is maintained as a relatively pure guanine solution. In gud1Δ, a steady decrease in the intensity of the λ ₁ and λ ₂ peaks occurred in the first 24 h. The absence of shifts in λ ₁ and λ ₂ suggests guanine remains as the dominant metabolite in the medium. Raman spectroscopy (Figures 1E and S1) confirms that the birefringent inclusions in both hpt1Δ and gud1Δ are composed of crystalline guanine.

FIGURE 2.

Guanine crystal morphologies in WT, hpt1Δ, and gud1Δ cells. (A) SEM and (B) TEM micrographs of extracted semi‐crystalline spheres from WT cells. (C) SEM and (D) TEM micrographs of extracted crystalline dumbbells from hpt1Δ cells, inset: SAED shows intense (100) reflections along the long axis of the dumbbell, corresponding to periodic ordering along the π‐stacking axis of the guanine crystallites. (E) SEM and (F) TEM micrographs of extracted crystalline nanoflowers from gud1Δ cells.

SEM (Figures 2A and S4), TEM (Figures 2B and S5), and electron diffraction (Figure S5 insets) revealed that guanine spheres in WT yeast have a granulated texture and are poorly crystalline, consistent with the observation of weak lattice mode Raman peaks (Figures 1E and S1) [16]. Deletion of HPT1 and GUD1 lead to a dramatic change in guanine crystal morphology. In hpt1Δ cells, dumbbell crystals are formed. The dumbbells are composed of many faceted guanine crystals elongated along the π‐stacking axis and displaying the characteristic (100) face twinning of β‐guanine [30, 31] (Figures 2C,D and S4). In polarizing light microscopy, the Maltese‐cross extinction pattern (Figure 1A(iv) insets) indicates that within these dumbbells, the guanine molecules are arranged as spherulites with radial symmetry [32]. Selected area electron diffraction patterns (Figure 2D inset) exhibit strong {100} reflections oriented along the long dumbbell axis, demonstrating the π‐stacking axis of the guanine crystallites is preferentially oriented in this direction (Figure S6). Intense Bragg reflections demonstrate the dumbbells are significantly more ordered than the spherical particles in the WT, consistent with sharper low‐frequency Raman peaks and a higher intensity of the π‐stacking mode at 106 cm⁻¹ compared to the ring‐breathing mode at 648 cm⁻¹ (ratio of respective peak areas: 0.22 in WT and 0.75 in hpt1Δ, Figures 1E and S1) [16]. Polarized light microscopy shows that in gud1Δ, the guanine deposits exhibit significantly higher light scattering than in the WT (Figure 1A(v) inset). SEM and TEM of extracted crystals (Figures 2E,F and S4 and S7) show the gene deletion produces a nanoflower morphology, composed of aggregated guanine plates. The gud1Δ nanoflowers appear to have higher crystallinity with respect to the WT spheres as indicated by the π‐stacking: ring‐breathing mode peak area ratio (0.32, Figure 1E). According to high‐resolution TEM, crystal size increased in gud1Δ and hpt1Δ with respect to the WT (Table S1).

We recently showed how the morphologies of biosynthesized molecular crystals can be tuned within microalgae by modification of culture conditions and by the addition of crystallization inhibitors [5]. Here, we extend the “cellular factories” concept, showing the utility of genetic editing as a tool for controlling crystal “phenotype”—i.e., crystal size, morphology, and crystallinity. We are lacking a definitive mechanism for the crystal morphology change in yeast. One hypothesis is that vacuolar crystallization may be regulated by changes in pH [33]. However, pH tracking using LysoTracker dye did not produce a vacuole‐specific signal (Figure S2). Since we also observed (Figure S1, Movie S1) that the crystals are located within the central lumen of the vacuole (rather than at the membrane interface), crystal morphology is also not dictated by the shape and size of the confining membrane. Thus, we suggest that guanine morphology is influenced by changes in the local composition of guanine metabolites within the vacuole—determined by the activities of Hpt1 and Gud1.

Our hypothesis for morphology regulation draws on the resemblance of the yeast crystals to “isotropic” and “sheaf” spherulites encountered in polymer chemistry. Isotropic spherulites, comprising spokes radiating from a single nucleation point, are formed at high supersaturation and/or in the presence of impurities, when the rate of secondary nucleation is ≫ anisotropic growth [34]. Within the yeast vacuole, guanine supersaturation and metabolite composition are determined by the activities of purine transporters and salvage enzymes. Editing these enzymes provides a means to modulate these parameters and alter crystal shape—a biological analogue of the approach Li et al. who modified guanine morphology by controlling its local supersaturation in double emulsion compartments [35]. In WT yeast, conversion of guanine to GMP by Hpt1 (or to xanthine by Gud1) would create a heterogeneous mixture of metabolites in the vacuole during crystallization, affecting molecular diffusion and/or the composition of the crystals. Compositional heterogeneity likely perturbs anisotropic growth of ordered guanine crystals, resulting in high secondary nucleation rates, producing spherulites that are isotropic and poorly crystalline due to numerous nanometer‐sized grains (Figure 2A,B).

The dumbbells in hpt1Δ, resemble sheaf spherulites [34, 36, 37], whose formation is associated with lower supersaturation or more homogeneous compositions, where anisotropic growth > the secondary nucleation rate in early growth stages. In hpt1Δ, conversion of guanine to GMP is inhibited, and the solute is thus enriched in guanine. Additionally, xanthine is unlikely to be a major component of the solute, because the xanthine‐producing Gud1 was reported to be active only after exponential growth [6, 27]—that is, when the dumbbells are already formed. Compositional homogeneity likely favors the growth of larger, more ordered, prismatic guanine crystals [31, 38]. However, the chevron morphology, and sheaf‐like branching of the crystallites show that the crystallization kinetics and thus secondary nucleation rate is still relatively high [31]. At later growth stages, additional branching and “splaying” leads to the two lobes of the dumbbell (Figures 2C,D and S4 and S6).

In gud1Δ, the nanoflower spherulites are isotropic, but have a more “open” morphology compared to the WT and less ordered crystallites compared to hpt1Δ. This suggests an intermediate crystallization driving force between the WT (high) and hpt1Δ (low). Since Gud1 is active after exponential growth, early stages of crystallization are likely to be similar to the WT, where the presence of GMP disrupts anisotropic crystal growth, resulting in highly branched isotropic spherulites. However, at later stages, after guanine uptake, inhibition of guanine deaminase prevents guanine degradation, allowing the guanine spheres to “mature” and reorganize into a more ordered, nanoflower morphology.

In addition to affecting crystal morphology, Hpt1 and Gud1 likely also drive guanine crystal dissolution (and export) from WT cells, degrading free guanine in the vacuole to downstream metabolites—driving further dissolution of the crystals. Future time‐dependent imaging studies and growth simulations will be used to test these hypotheses of crystal formation and dissolution.

Recently, Deis et al. showed how guanine morphologies in zebrafish could be controlled by genetic editing of the purine nucleoside phosphorylase enzyme, which determines the hypoxanthine content of the crystals [39]. In zebrafish, guanine crystallization is strictly controlled with guanine crystals being formed in dedicated pigment organelles [39]. In contrast, crystal formation does not appear to be functional in S. cerevisiae. However, our results show how purine transporters and biosynthetic enzymes, intended for other essential biological functions, can be genetically manipulated to influence crystal properties. Given that HGPT (an ortholog of HPT1) is omnipresent in eukaryotes [8], our genetic editing strategy may be generalized to a range of biological systems, including microalgae, where crystallization may be first induced and then genetically modified.

In animals, plate‐like guanine crystals, preferentially expressing the high refractive (100) face, are formed [40]. In contrast, the spherulite guanine nanoparticles in yeast more closely resemble isoxanthopterin [9, 14, 41], leucopterin [15], and uric acid [42] nanospheres, which produce structural colors in animals. The ∼500 nm diameter of the WT and gud1Δ particles makes them ideal for light scattering in the Mie regime, and their spherical morphology facilitates colloidal self‐assembly into photonic superstructures. Given that the spectral response (e.g., color hue, saturation) of photonic assemblies depends on the size, refractive index, crystallinity, morphology, and polydispersity of the component particles [9, 12], genetic editing may be applied for tuning the properties of optically functional molecular crystals. Combining its ease of genetic manipulation and use in industry, yeast provides a powerful biotechnological platform for the sustainable synthesis of biocompatible optical alternatives to toxic or environmentally harmful inorganic optical materials [43]. This work further highlights the utility of harnessing the programmability of living [44, 45] and synthetic [46, 47] cells for constructing advanced materials.

Author Contributions

Jerome N. Janssen: conceptualization, investigation, writing – original draft, methodology, visualization, writing – review and editing, and formal analysis. Yahel Fishman: investigation, writing – original draft, methodology, visualization, writing – review and editing, and formal analysis. Yi Li: conceptualization, investigation, and writing – review and editing. Avital Wagner: investigation, conceptualization, and writing – review and editing. Mariela J. Pavan: investigation, methodology, writing – review and editing, and formal analysis. Zohar Paleiov: investigation. Tali Lemcoff: investigation. Rotem Lichtenstein‐Wolfheim: investigation. Alexander Upcher: investigation and visualization. Einat Nativ‐Roth: investigation and visualization. Amir Aharoni: conceptualization, methodology, writing – review and editing, and supervision. Benjamin A. Palmer: conceptualization, funding acquisition, writing – original draft, writing – review and editing, project administration, resources, and supervision.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File 1: anie74163‐sup‐0001‐SuppMat.docx.

Supporting File 2: anie74163‐sup‐0002‐MovieS1.mp4.

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.