Yeast collection

We first conducted a literature review on OCFA-producing yeasts, to guide the selection of species for this study. The final selection, comprising a total of 19 strains, was based on the following criteria: (i) documented capacity to produce OCFAs in sugar-based media, (ii) capacity to grow on and produce OCFAs from PA or combinations of different SCFAs (synthetic mixes or real SCFAs effluents), and (iii) oleaginous yeast species that had not been tested before for criteria (i) or (ii) (Table 2). Five strains (L. tetrasporus, W. anomalus, Z. florentina, B. adeninivorans, R. mucilaginosa) were selected due to reports on traces of OCFAs in their fatty acid profiles upon growth in sugar-based media. Five strains (Y. lipolytica, T. delbrueckii, R. toruloides, C. cutaneum, C. oleaginosus) because they had previously been cultivated in PA or SCFAs synthetic mixtures or effluents for production of OCFAs. Finally, the remaining nine strains were chosen because of their status as oleaginous yeast (reviewed in [32]), and to ensure representation of a diverse array of families within both the Ascomycota and Basidiomycota phyla. To the best of our knowledge, these latter strains have not yet been explored with respect to their abilities to produce OCFAs.

Table 2 List of reasons of selection of the yeast species used in this studyGrowth profiles in glucose based medium

Initially, we assessed the growth capabilities of all strains in glucose minimal medium at molar C/N ratio 9 and 50 in 96-well microtiter plates and determined the maximum specific growth rates and the duration of the lag phases.

At a molar C/N ratio of 9, all strains exhibited short lag phases (≤ 14 h) and reached maximum specific growth rates between 0.15 and 0.30 h⁻¹ (Suppl. Figure S1). By comparison, growth profiles at a C/N ratio of 50 showed clear strain differences (Fig. 2). B. californica, M. pulcherrima, R. mucilaginosa, S. silvae and W. anomalus displayed similar growth profiles in C/N ratio 50 as in C/N ratio 9, but with accentuated deceleration phases, reaching the same or almost the same final ODeq as in C/N ratio 9 only a few hours later. In contrast, strains such as B. adeninivorans, P. occidentalis, R. toruloides, S. occidentalis, S. bombicola, T. delbrueckii and Y. lipolytica displayed growth profiles in C/N ratio 50 that were clearly different from those in C/N ratio 9, with pronounced biphasic growth, lower maximum specific growth rates and longer time needed to reach the same final ODeq as in C/N ratio 9. C. cutaneum, C. oleaginosus, D. hansenii, L. tetrasporus, and Z. florentina primarily exhibited non-biphasic growth profiles, with low maximum specific growth rates (0.02–0.09 h⁻¹). These strains also failed to achieve the final ODeq observed under C/N ratio of 9 within the 5-days experimental timeframe. We can conclude that the growth characteristics were clearly influenced by the C/N ratio in a species-dependent manner.

Fig. 2

Comparison of growth profiles in glucose-based media with high and low C/N ratios. Yeast strain growth was analyzed in 20 g/L glucose at molar C/N ratios of 9 and 50. The graphs display data obtained from a Growth Profiler in a 96-well plate system. Data are shown as mean ± standard deviation for biological triplicates. The y-axis shows the biomass formed during cultivation on a logarithmic scale using green values, considered equivalent to OD, plotted against time (days) on the x-axis. A) strains with similar growth profiles at C/N ratios of 9 and 50. B) strains with distinct biphasic growth at C/N ratio 50, reaching the same or almost the same final ODeq as in C/N ratio 9. C) strains with non-biphasic growth at C/N ratio 50, lower maximum specific growth rates, and lower final ODeqs than those observed at C/N ratio 9

Fatty acids profiles resulting after cultivation in glucose-based medium

Next, we determined the yeasts’ capacities to produce fatty acids in minimal medium containing 20 g/L glucose at molar C/N ratio 50. At the end of cultivation (115 h), when all the strains had reached stationary phase, we sampled and determined cell dry weight (g/L), residual glucose (g/L) and fatty acid profiles. Unfortunately, C. cutaneum, D. hansenii, L. tetrasporus, and Z. florentina did not form enough biomass in glucose-based medium to enable all measurements (including technical replicates for fatty acid profile analysis). Consequently, the analyses were performed on the 15 yeasts that produced high enough cell mass concentrations.

Analysis revealed that ECFAs are the predominant fatty acids produced under glucose-based conditions, accounting for 95.0-99.5% of the total fatty acids across all studied strains (Fig. 3A). The percentage of specific fatty acids on total fatty acids differed between the different yeasts, although palmitic acid (C16:0), palmitoleic acid (C16:1), stearic acid (C18:0), oleic acid (C18:1) andlinoleic acid (C18:2) were among the most abundant fatty acids for all strains. S. occidentalis, P. occidentalis, S. silvae, W. anomalus and B. californica produced also α-linolenic acid (C18:3), with levels ranging between 1.9% and 12.7%. Most yeasts produced more C16:0 than C16:1, except in T. delbrueckii and S. bombicola where C16:1 was more abundant than C16:0. On the contrary, the unsaturated fatty acids C18:1 and C18:2 were generally the preferred ones compared to saturated C18:0. Very long chain fatty acids (VLCFAs) such as behenic acid (C22:0) were produced by R. toruloides (1.3%), and lignoceric acid (C24:0) was mainly produced by R. fluvialis, R. toruloides, R. mucilaginosa and Y. lipolytica (1.0, 4.1, 2.2 and 3.9% respectively). Notably, R. toruloides also produced a small amount (1.2%) of cerotic acid (C26:0), highlighting the capability of this yeast to produce VLCFAs [46].

In terms of OCFAs, pentadecanoic acid (C15:0), heptadecanoic acid (C17:0) and heptadecanoic acid (C17:1) were the main fatty acids produced. M. pulcherrima produced the highest percentage of OCFAs, almost 5% of the total fatty acids, followed by the two Blastobotrys strains B. raffinosifermentans and B. adeninivorans (3.4%), which exhibited very similar fatty acid profiles overall, and B. californica (3.1%). Whereas all strains displayed relatively low OCFA yields, the yields of ECFAs and yeast cell biomass on consumed substrate varied substantially between strains (Fig. 3B). C. oleaginosus exhibited the highest ECFAs yield on consumed substrate (0.13 g/g), followed by R. fluvialis, R. toruloides, Y. lipolytica and R. mucilaginosa (0.07, 0.07, 0.06, 0.05 g/g, respectively). B. raffinosifermentans and B. adeninivorans on the other hand exhibited the highest yeast biomass titers (6.95 and 6.85 g/L, respectively) and a CDW yield of 0.47 g/g, but lower ECFAs yields (0.03 g/g for both). This suggests that these yeasts prioritized carbon allocation toward biomass production rather than lipid synthesis under the tested condition. All other analyzed strains reached ECFAs yields of less than 0.02 g/g.

Overall, the titers indicated that R. fluvialis, and R. toruloides, C. oleaginosus and Y. lipolytica were amongst the best fatty acid producers (1.09–1.31 g/L ECFAs) (Fig. 3C). In contrast, R. mucilaginosa, B. raffinosifermentans, and B. adeninivorans produced an intermediate amount of ECFAs (0.43 to 0.50 g/L), while the remaining yeast strains showed a poor production of ECFAs (0.05 to 0.24 g/L). The OCFA titers for all strains were negligible, although some differences were observed, within the range of 1–16 mg/L, and where the top four OCFA-producing yeasts were R. toruloides, Y. lipolytica, B. raffinosifermentans, and B. adeninivorans (Fig. 3D).

Fig. 3

Results from fatty acids analysis represented with different metrics. The 15 strains that produced high enough cell mass titers when cultivated in 20 g/L glucose at molar C/N ratio 50 are displayed. All plotted data are ordered by decreasing ECFAs g/L metric. A) Relative percentages ratio (%) of single fatty acids on total fatty acids profile; B) Histograms representation of yields on consumed substrate (S) in g/g of final biomass (CDW/S), final odd chain fatty acids (OCFAs/S) and final even chain fatty acids (ECFAs/S). Overlapping scatter point of final residual substrate (g/L) and final CDW (g/L) relative to each strain; C) titer in g/L of the final total amount of ECFAs; D) titer in mg/L of the final total amount of OCFAs

Growth in increasing concentrations of propionic acid

We assessed the growth capabilities of all 19 yeast strains in minimal medium with six different concentrations of PA (g/L): 5, 10, 15, 19, 24, and 29, and buffered to pH 6.8. Only those strains that displayed growth at lower concentrations of PA were included in the trials with higher concentrations. It is important to note that PA has dual roles, functioning both as both a carbon and energy source while also acting as a weak acid that imposes cellular stress. This complexity makes interpreting growth patterns more challenging. With increasing PA concentrations, cells have access to more carbon for biomass formation, but the accompanying stress can lead to longer lag phases, reduced specific growth rates, and/or lower final optical density (ODeq).

Eight strains could clearly grow on PA concentrations up to 24 g/L: B. adeninivorans, B. raffinosifermentans, B. californica, C. cutaneum, C. oleaginosus, R. toruloides, W. anomalus and Y. lipolytica (Fig. 4). At 24 g/L and 29 g/L of PA, Y. lipolytica, B. californica, and C. oleaginosus exhibited higher specific growth rates than the other strains. Y. lipolytica displayed the most robust growth characteristics on PA with short lag phases and high final ODeq, while B. californica and C. oleaginosus displayed extended lag phases, although they eventually reached the same final ODeq as in 19 g/L PA. The other five yeasts showed the highest ODeq in 15 g/L PA, though some strains did not appear to reach stationary phase within the experimental timeframe.

Conversely, all 19 strains grew when PA (5–10 g/L) was supplemented with 2 g/L of glucose (Suppl. Fig. S2). However, L. tetrasporus, M. pulcherrima, S. occidentalis, S. bombicola, W. anomalus, and Z. florentina did not reach the same final ODeq as the control with 2 g/L of glucose alone. This suggests that these strains experience greater cellular stress in response to PA compared to the other strains.

Fig. 4

Comparison of growth profiles in increasing concentrations of propionic acid. Yeast strain tolerance to PA were analyzed in 5, 10, 15, 19, 24, and 29 g/L of PA at molar C/N ratio of 9 following the growth using a Growth Profiler in a 96-well plate system. Data are shown as means of three biological replicates. The y-axis shows the biomass formed during cultivation on a logarithmic scale using green values, considered equivalent to OD, plotted against time (days) on the x-axis

Fatty acids profiles in propionic acid-based medium

The eight strains that could grow on PA at C/N ratio 9 were next cultivated in 15 g/L of PA and at C/N ratio 50 to promote fatty acids accumulation. All eight strains displayed growth profiles similar to those in C/N ratio 9 (Suppl. Fig. S3), in contrast to the varying profiles observed for most strains in glucose-based cultivation at C/N ratios 9 and 50 (Fig. 1). The microtiter growth cultivations revealed that 188 h was an appropriate harvesting time, as all strains had ceased to grow at this time (Suppl. Fig. S3). We cultivated the eight PA tolerant yeasts in shake flasks with medium containing 15 g/L of PA at molar C/N ratio 50, and samples were collected and used to determine the cell dry weight (g/L), the residual substrate (g/L) and fatty acid profile. Unfortunately, W. anomalus did not grow to high enough cell mass titers in shake flasks, despite its evident growth in the microtiter plate, and was therefore left behind.

All seven yeasts showed fatty acid profiles rich in OCFAs, but the ratio of OCFAs on total fatty acids varied significantly (Fig. 5A). R. torulodies and B. californica produced close to 90% of their total fatty acids as OCFAs, while C. oleaginosus produced about 80%. Y. lipolytica and the two Blastobotrys strains produced 55% OCFAs, and C. cutaneum 37%. Moreover, the percentage of individual fatty acids on total fatty acids varied between yeasts, although all accumulated preferentially C15:0, C17:0 and C17:1 (similar to the OCFA-profiles when using glucose as substrate). R. toruloides, C. oleaginosus and C. cutaneum also produced measurable amounts of nonadecanoic acid (C19:0), and R. toruloides generated 1.44% of its total fatty acids as tricosilyc acid (C23:0). The primary ECFAs accumulated were C16:0, C18:0, C18:1 and C18:2.

We also identified additional OCFAs and ECFAs peaks, not present in our standards. These are assigned as “other OCFAs” or “other ECFAs” in Fig. 5A. All yeast species except Y. lipolytica produced nonadecenoic acid (10-C19:1). R. toruloides also produced a substantial relative amount of heneicosanoic acid (C21:0, 1.86%) and pentacosylic acid (C25:0, 6.83%) (Fig. 5A) which, together with C23:0, underline the capability of this yeast to produce VLCFAs also on PA. We also detected the “other ECFAs” hexadecenoic acid (7-C16:1), octadecadienoic acid (9,12-C18:2) in all strains except C. cutaneum.

C. oleaginosus and R. toruloides displayed the highest OCFA yields on consumed PA (0.07 and 0.05 g/g respectively), while the other five yeasts displayed more modest yields (0.01–0.02 g/g) (Fig. 5B). Moreover, C. oleaginosus, Y. lipolytica and B. californica used most of the available carbons, whereas the other strains used substantially less (2.9–7.6 g/L of the initial 15 g/L). C. oleaginosus and C. cutaneum exhibited the highest ECFA yields based on consumed substrate (0.02 and 0.04 g/g respectively). The three strains that demonstrated the most efficient PA conversion also reached the highest fatty acid titers. C. oleaginosus produced 1.22 g/L of total fatty acids, with OCFAs accounting for 0.94 g/L. B. californica reached 0.29 g/L of total fatty acids, of which 0.26 g/L were OCFAs, and Y. lipolytica produced 0.36 g/L of total fatty acids, with 0.20 g/L as OCFAs (Fig. 5C).

In conclusion, all the yeast species cultivated in 15 g/L of PA showed profiles enriched in OCFAs compared to their fatty acid profiles from glucose-based cultivation, although there were clear differences between strains in terms of PA uptake and conversion and the accumulated fatty acid profiles.

Fig. 5

Fatty acids profiles of seven strains cultivated on 15 g/L propionic acid at molar C/N ratio 50. A) relative % fatty acid profiles. Other ECFAs: 7-C16:1, 9,12-C18:2. Other OCFAs: 10-C19:1, C21:0, C25:0; B) yields on substrate (S) of biomass (CDW/S), odd chain fatty acids (OCFAs/S) and even chain fatty acids (ECFAs/S), overlapping illustrations of residual substrate (g/L) and CDW (g/L) to show the full potential of every analyzed yeast; C) titer (g/L) of the total amount of odd chain and even chain fatty acids

Comparison of fatty acid profiles in glucose and propionic acid

Finally, we compared the yeasts’ capacities to produce fatty acids on glucose and PA by evaluating the percentage of fatty acids relative to dry yeast biomass, as well as the fatty acid profiles and abundance of individual fatty acids produced.

In PA-based medium, all yeast strains exhibit a more diverse fatty acid profile compared to those in glucose-based medium, as indicated by the presence of a mix of both OCFAs and ECFAs (Fig. 6). Interestingly, C. cutaneum demonstrated poor growth in glucose-based medium at a C/N ratio of 50 (unpublished data), which prevented us to proceed with downstream measurements, compared to PA-based medium, where we obtained 1.75 g/L of CDW (Fig. 5A), indicating a potential preference for PA as a substrate. In glucose-based medium, R. toruloides, C. oleaginosus and Y. lipolytica showed high levels of C16:0, C18:0 and C18:1, while the other strains contained mainly C16:0 and C18:1. In PA-based medium, all strains exhibited strain-dependent production of ECFAs, yet all continued to produce substantial amounts of C16:0 and C18:1. The consistency in the production of these ECFAs in presence of different carbon sources likely reflect the importance of these fatty acids for membrane structures [47, 48].

R. toruloides and C. oleaginosus demonstrated superior oleaginous capabilities in the presence of both carbon sources, accumulating 33.6% (w/w) and 30.9% (w/w) of fatty acids on dry biomass in glucose, and 20.4% (w/w) and 29.4% (w/w) in PA, respectively (Table 3). Notably, C. oleaginosus accumulated similar amounts of total fatty acids per g of dry biomass in both conditions (Fig. 6), although it produced more total fatty acids per g of carbon consumed in glucose-based media (0.32 g/g-C) compared to PA media (0.17 g/g-C) (Table 3). Y. lipolytica, which demonstrated superior growth on PA (Fig. 4), produced comparatively fewer grams of fatty acids from this carbon source compared to glucose (0.05  vs. 0.14 g/g-C) (Table 3). Interestingly, B. californica was the only yeast that produced more total fatty acids per g of consumed carbon on PA compared to glucose-based media (0.05 vs. 0.01 g/g-C). This suggests that PA is a more favorable substrate than glucose for promoting oleaginicity in B. californica, as it appears to channel more carbon towards lipid accumulation under PA growth.

Fig. 6

Comparison between the fatty acid profiles in 15 g/L of PA and 20 g/L of glucose at molar C/N ratio 50. The y-axis shows the µg fatty acids produced per mg of biomass (CDW). The x-axis shows the labels of the single fatty acid present in the profile. The orange and white bars represent the PA and glucose conditions, respectively

Table 3 Comparison of metrics between glucose and propionic acid-based cultivations