Uncoupling protein production from growth: different strategies for intracellular and secreted proteins in yeast

Strains and strain construction

Strains ScNTi001, ScNTp001 and ScNTp002 were used in this study (Table 1). All strains were grown in appropriate media until late exponential phase and stored as glycerol stocks (30% v/v) at -80 °C. ScNTi001 was derived from CEN.PK113-7D [20], by integration of the dual expression cassette (P_HSP12-mRuby2-tt_ADH1 - P_TEF1-ymNeongreen-tt_TDH1) in the X-2 locus [21]. ScNTp001 and ScNTp002 are derived from CEN.PK113-5D [20] by transforming a multicopy (2µ) plasmid carrying the klURA3 gene and a ymNeongreen secretion cassette driven by the TEF1 or HSP12 promoter. All plasmids and primers used are shown in Supplementary Tables 1 and 2, respectively.

To incorporate the promoter region, the 1 kb upstream sequence of the HSP12 start codon was synthesized (IDT, USA), with overhangs for integration into the Yeast Tool Kit [22]. The HSP12 promoter fragment was cloned into the entry plasmid (YTK001) via Golden Gate cloning, followed by E. coli transformation and purification. The yeast-optimized mNeongreen fluorescent protein coding sequence with specific overhangs was amplified from a ymNeongreen plasmid (addgene #125704) from [23] using PCR with primers designed to incorporate the desired overhang sequences. The PCR product was assembled into the entry plasmid (pYTK001) using Golden Gate cloning to generate a pYTK-ymNeongreen part plasmid. For protein secretion, we fused the Ksh1 secretion signal sequence [24] and ymNeongreen coding sequence. The Ksh1 secretion signal peptide (SP_Ksh1) and ymNeongreen coding sequences were amplified separately, while ensuring seamless fusion between them and compatibility with the entry part plasmid (YTK001). The SP_Ksh1 coding sequence was ordered as two long complementary primers (IDT, USA), while ymNeongreen was amplified from the pYTK-ymNeongreen plasmid. The annealed oligonucleotide duplex and PCR product were assembled into the entry plasmid (YTK001) resulting in pYTK-SP_Ksh1-ymNeongreen.

The intracellular fluorescence protein-expressing strain ScNti001 was constructed from the prototrophic strain S. cerevisiae CEN.PK113-7D. Using Golden Gate cloning, a reporter plasmid was assembled containing two transcription units: P_HSP12-mRuby2-T_ADH1, and P_TEF1-ymNeonGreen-T_TDH1. This plasmid was used as to PCR amplify the integration cassette consisting of both transcription units. Integration into the X-2 region was facilitated by a single CRISPR-Cas9 plasmid encoding the spCas9 gene and a gRNA targeting the locus on chromosome X. This plasmid was constructed following the approach of Juergens et al., based on plasmid pUDP002 (Addgene #103872) [25, 26]. Successful integration was verified via colony PCR. The CRISPR-Cas9 plasmid was cured by culturing the strains in YPD without selective pressure.

For protein-secretion, S. cerevisiae strain CEN.PK113-5D was transformed with plasmid pP_HSP12-SP_Ksh1-ymNeongreen or pP_TEF1-SP_Ksh1-ymNeongreen. Multicopy plasmids with the URA3 selection marker were constructed from part plasmids using Golden Gate cloning [22]. Plasmids were validated by restriction analyses and sequencing.

Yeast transformations were carried out using the lithium acetate method [27]. YPD media with 200 µg/mL hygromycin were used as selective pressure for integration, synthetic medium [28] lacking uracil was used for selection of cells carrying the plasmid.

Fed-batch cultivations

Shake-flask precultures for fed-batch experiments were grown in 500 mL shake flasks containing 100 mL of 6.7 g/L YNB medium (Y0626 Sigma Aldrich), set to pH 6.0 with 2 M KOH before autoclaving and supplemented with 20 g/L glucose. These shake-flask cultures were inoculated with 2 mL of frozen stock culture and incubated in an orbital shaker at 200 rpm and 30 °C.

Fed-batch cultivations were performed in 3-liter bioreactors (Getinge/Applikon, Delft, the Netherlands). The operational parameters were set as follows: the temperature was maintained at 30 °C, the constant stirring speed was 1000 rpm, and sterile air was sparged at 900 mL/min. pH was controlled at 5.0 for intracellular protein experiments and 6.5 for secretion experiments by automatic addition of 2 M potassium hydroxide. The cultures for the fed-batch were grown in 6.7 g/L YNB with 3 g/L ammonium sulphate, 1.94 g/L yeast synthetic dropout supplement without uracil (Y1501 Sigma Aldrich), 0.02 g/L BSA, 1 mL/L antifoam B (Sigma-Aldrich) and 17 g/L glucose for the batch phase. The feed medium consisted of 6.7 g/L YNB with 3 g/L ammonium sulfate, 3.88 g/L yeast synthetic dropout supplement without uracil, 0.02 g/L BSA, 1 mL/L antifoam B (Sigma-Aldrich) and 70 g/L glucose. Feeding started once extracellular carbon sources were exhausted during the batch phase, with a constant feed rate of 30 mL/h.

Determination of substrate and biomass concentrations and off gas composition

The culture dry weight was determined by filtering precisely 10 mL of an appropriately diluted culture broth through pre-dried and pre-weighed membrane filters (Gelman Science). Samples were diluted with demineralised water when biomass concentration was higher than 15 g/L, prior to culture dry weight assays. The filters were then washed with demineralized water, dried for 20 min at 350 W in a microwave oven, and reweighed [30]​. Supernatants for sugar analysis were prepared by centrifuging culture samples for 5 min at 16,000×g. These supernatants and media were analysed in a YSI Glucose Analyzer (2950 Biochemistry Analyzer, Yellow Spring Instruments). OD was measured at 600 nm in determined time intervals using a Hach Lange DR600 spectrophotometer. The exhaust gas from the fed batch cultures was cooled using a condenser set at 4˚C and subsequently dried. The dried gas was then analysed for carbon dioxide, oxygen and ethanol concentrations using a mass spectrometer (Prima δB, Thermo Fisher).

Determination of intracellular relative fluorescent protein levels and viability

Aliquots of 1 mL culture broth were diluted to 1 × 10⁷ with PBS buffer pH 7.5 (VWR). Fluorescence intensity was measured using a Cytoflex flow cytometer (Beckman Coulter, the Netherlands), excited with a 488 nm laser. ymNeongreen fluorescence was measured with a 525/40 nm filter, while mRuby2 fluorescence was measured with a 610/20 nm filter. In addition, culture viability was assayed by staining 1 × 10⁷ cells with 4 µg/mL propidium iodide (PI) as described previously [31]. After 15 min incubation in the dark, fluorescence of the samples was measured with a Cytoflex flow cytometer (Beckman Coulter, the Netherlands), excited with 488 nm, and measured with a 690/50 nm filter. A total of 10,000 cells were counted for each sample. Prior to sample acquisition, the cytometer was calibrated using fluorescent calibration beads to ensure consistent detector performance across experiments. A threshold was based on forward scatter (FSC-A) to exclude small particles and background noise. Cells were gated based on FSC-A versus side scatter (SSC-A) to exclude debris. PI-positive and PI-negative populations were distinguished based on 690/50 nm fluorescence intensity versus FSC-A plots. Flow cytometry data were analysed using FlowJo v10.8.1 software (BD Biosciences, Franklin Lakes, NJ, USA).

Determination of secreted fluorescence protein intensity

1 mL aliquots of culture broth were centrifuged at 5000 g for 5 min. 200 µL of supernatant was placed into a black 96 well plate (Nunc, Thermo Fisher) and analysed for green fluorescence. Measurements were performed in a Synergy H1 plate reader (Biotek) with monochromators set to the excitation wavelength of 488 nm and emission at 507 nm.

Calculation of fed-batch growth kinetics

Titers of intracellular and extracellular fluorescent protein were calculated using Eqs. (1) and (2), respectively. In Eq. (1), T_IP denotes intracellular protein titer (a.u./L), fluorescence intensity is the mean fluorescent intensity detected using flow cytometry (a.u.) corrected for mean cell size (based on FSC-A, a.u.), c_X is the biomass concentration (g/L). In Eq. (2), T_EP denotes extracellular protein titer (a.u./L), Fluorescence intensity is the measured fluorescence intensity of the supernatant (a.u./L).

1

2

Specific growth rates (µ in h^{− 1}) were calculated as previously described [7]. Specific protein production rates (q_P in a.u./g/h) were calculated using Eq. (3). In this equation r_P denotes the protein production rate at the time t (h), M_X is the biomass amount (g) in the reactor at the time t. Biomass amounts were calculated based on Eq. (4). Protein production rates were calculated by determining the slope of protein amount (i.e., Titer x Volume) versus time. For first-order kinetics, the rate corresponds to the linear slope of protein amount versus cultivation time. For second-order behaviour, the rate was derived from the time-dependent change in the slope, calculated as the first derivative of the fitted curve, where time was the independent variable.

3

4

Protein yield on substrate (Y_PS) at the time point t was calculated with Eq. (5), where q_S is the specific substrate consumption rate (g/g/h) estimated with a previously described model [7].

5

All normalizations were performed against the value at the start of the feeding phase. For strains ScNTp001 and ScNTp002 the average value of both strains at the start of the feeding phase was used for normalisation. As a result, all normalized values represent relative amounts and are unit-less.

Statistical analyses

All statistical analyses were performed using GraphPad Prism (version 10.4.1, Dotmatics). Cell size comparisons between strains were evaluated using unpaired t-tests and simple linear regression. Correlation of specific protein production rates (q_P) or secretion rates (q_p^ex) with growth rate (µ) were analyzed using Pearson correlation to assess linear relationships, with significance set at p-value < 0.05. The relationship between secretion rates (q_p^ex) and intracellular fluorescence intensity was evaluated using Spearman correlation. Biomass yield on substrate (Y_XS) comparisons between strains were analyzed using a paired t-test. All statistical tests were conducted with a significance threshold of p-value < 0.05.

Effect of intracellular protein expression on physiology

Fed-batch cultivation is a semi-continuous method in which feed is continuously added without removing any culture. This approach enables high cell densities and product titers in a controlled environment, making it popular in industrial settings [32, 33]. Different feeding strategies can lead to varying growth characteristics in fed-batch cultures. We maintained a constant feed flow rate to gradually decrease the specific growth rate over time. This allowed us to study how productivity varies with growth rate when employing stress-responsive (P_HSP12) or constitutive (P_TEF1) promoters.

During the feeding phase, the initial specific growth rate (µ) was 0.081 ± 0.003 h⁻¹, which gradually decreased to 0.015 ± 0.001 h⁻¹ over 42 h (Fig. 1A). At the end of the cultivation, the biomass yield reached 0.32 ± 0.01 g/g. Using Pirt’s model we estimated the distribution of consumed glucose (q_S) between growth (µ), protein formation (q_p), and non-growth associated processes (m_S). The non-growth associated glucose consumption was estimated at 0.017 ± 0.001 g/g/h, corresponding to 37.0 ± 0.0% of the glucose consumed at the end of the fed-batch culture. As the product accumulated intracellularly this value includes both the cellular maintenance requirement and part of the glucose allocated to product formation under non-growing conditions. Cell viability remained above 98% (Supplementary Fig. 1A), while the average cell size increased significantly during the cultivation, leading to app. 1.5-fold larger cells (t-test p-value = 0.0041, Supplementary Fig. 1B).

Expression driven by TEF1 promoter shows a positive correlation with the specific growth rate

Cellular fluorescence intensity caused by fluorescent protein expression was measured using flow cytometry (Fig. 2 and Supplementary Fig. 2). The cellular mRuby2 fluorescence intensity driven by the HSP12 promoter (P_HSP12) and ymNeongreen fluorescence intensity controlled by the TEF1 promoter (P_TEF1), showed opposite trends over time (Figs. 1A and 2). At the start of the feeding phase, all cells show medium to high green fluorescence (mean intensity = 60.4 ± 0.2 kAU), gradually decreasing to lower intensities (mean intensity = 17.0 ± 0.4 kAU). This implies that P_TEF1 activity is correlated with the specific growth rate. Indeed, specific product formation rates (q_P) show a positive, linear correlation with specific growth rate (µ) (Fig. 1B) (p < 0.0001, Pearson r = 0.9998). Extrapolation to a growth-rate of zero, predicts a significant growth-rate independent specific production rate (m_P, p < 0.0001).

While ymNeongreen protein levels per cell decreased, titers remained almost constant during the fed-batch cultures due to the increase in biomass concentration (Fig. 1C, Supplementary Fig. 1). In addition to biomass concentration, the culture volume increased, and the total amount of ymNeongreen protein increased over time. This indicates that ymNeongreen production continued, even though glucose availability was limited. This is also illustrated by a growth-rate-independent protein yield (Fig. 1D).

Expression driven by P_HSP12 enables strong increase in intracellular protein titers

While the P_TEF1 mediated protein production resulted in average cellular fluorescence intensities that gradually decreased, P_HSP12 mediated protein production resulted in increased cellular protein levels with time and decreasing growth rate (Figs. 1A and 2). At the start of the feeding phase, only 12.6 ± 3.6% of the cells were highly fluorescent (i.e., with fluorescence intensities above 10kAU). After already 4 h of feeding (µ = 0.058 h^− 1), fluorescence intensity increased dramatically with 81.7 ± 2.3% of cells exhibiting high fluorescence. By the end of the cultivation, this proportion increased further to 96.6 ± 1.0% (Fig. 2B and Supplementary Fig. 2). Combined with increasing biomass concentrations, these increased cellular levels represent an increase in mRuby2 titer (Fig. 1C). E.g., after 42 h of constant feeding, the mRuby2 titer had increased 10-fold.

Also, for mRuby2 specific production rates (q_P) were calculated based on the change of intracellular protein levels. These still showed a positive correlation within the range of growth rates tested here (Fig. 1B). Yet, the correlation parameters between production and glucose consumption differed, which resulted in an increase in product yield at lower growth rates (Fig. 1D). Linear regression furthermore revealed a growth-rate independent specific production rate (m_P) that was 2.67 times higher than for the P_TEF1.

Impact of promoter on correlation between protein secretion and specific growth rate

Based on the strong increase in intracellular protein titers at decreasing specific growth rates when using P_HSP12, we explored if this also translated to higher secreted protein titers. Protein secretion provides several advantages for industrial-scale production, including more straightforward downstream processing and reduced costs. The same fed-batch strategy was applied for strains equipped with a multicopy plasmid harbouring the ymNeongreen coding sequence, including the Ksh1 signal peptide sequence [24] with either P_TEF1 (ScNTp001) or P_HSP12 (ScNTp002). As reporter protein, only ymNeongreen was used to avoid interfering effects from differences in protein trafficking that could occur when comparing it with mRuby2. Ksh1 is a high-level surface display protein [34] and its secretion signal has been used before to secrete fluorescent proteins in S. cerevisiae efficiently [24]. For secreted proteins, the pH of the medium impacts the fluorescence stability. Notably, within 2 h, ymNeongreen loses over 50% of its fluorescence at pH 5.0, whereas at pH 6.5 it retains 90% of its fluorescence [23]. Therefore, to increase the stability of extracellular ymNeongreen fluorescence, the pH was maintained at 6.5 instead of 6, to be well above the in vivo pKa of 5.42 [23]. Furthermore, extracellular proteases can contribute to the degradation of secreted recombinant proteins, potentially reducing overall yield. To mitigate this, we supplemented the media with bovine serum albumin (BSA), a known stabilising agent that has been demonstrated to prevent proteolytic degradation in previous studies [35, 36]. During the feeding phase, the specific growth rate decreased from 0.14 ± 0.004 h⁻¹ to 0.019 ± 0.001 h⁻¹ for strain ScNTp001 (P_TEF1), while for strain ScNTp002 (P_HSP12) this was comparable from 0.11 ± 0.005 h⁻¹ to 0.018 ± 0.001 h⁻¹ (Fig. 3A).

During the first 25 h, the extracellular titer of ymNeongreen increased 2.6-fold for P_HSP12 controlled protein expression. However, after 25 h, titers did not increase further, but stabilised (Fig. 3A). This contrasted with the intracellular production observed in the ScNti001 strain, in which accumulation continued for the P_HSP12-driven protein (Fig. 1C). On the other hand, the P_TEF1 initially resulted in lower extracellular titers, but extracellular ymNeongreen titers did not level off in the 2-day timespan, and final yields were 1.5-fold higher compared to the P_HSP12 results (Fig. 3A, Supplementary Table 3).

When we checked the specific secretion rates (q_p^ex), the q_p^ex of strain ScNTp002 (P_HSP12) showed a positive linear correlation with growth rate (Fig. 3B, p-value < 0.0001, Pearson r = 0.9996). Although a positive linear correlation was also observed for intracellular production of P_HSP12 driven mRuby2 (Fig. 1B), the correlation for secretion differed notably: cells appeared to stop secreting the fluorescent protein at growth rates below 0.02 h^− 1 (Fig. 3B). This is in line with the plateauing titers (Fig. 3A). Surprisingly, for ScNTp001 (P_TEF1), the correlation between q_p^ex and µ was inverted, suggesting that the specific product secretion rate increases with a decreasing growth rate (Fig. 3B). This correlation clearly differed from the one observed for intracellular production in the integrated strain (Fig. 1B).

To understand the differences between protein titers and q_P -µ correlations between the intracellular accumulating strain (Fig. 1) and secreting strains (Fig. 3), we also monitored intracellular protein levels in the secreting strains. The intracellular ymNeongreen levels of ScNTp002 (P_HSP12) were higher than of ScNTp001 (P_TEF1) throughout the entire cultivation (Fig. 3C). However, intracellular ymNeongreen levels slightly decreased in ScNTp002 (P_HSP12) after 25 h, which was not observed for ScNTp001 (P_TEF1). Specific production rates (q_p) based on these intracellular levels showed a positive correlation with the growth rate in both strains (Fig. 3D, p-value = 0.019, Pearson r = 0.8331 for ScNTp001 (P_TEF1), p-value < 0.0001, Pearson r = 0.9856 for ScNTp002 (P_HSP12), in line with observations for the intracellularly accumulating strain (Fig. 1B).

These results suggest that a low, but constant expression results in higher extracellular titers at lower growth rates, in contrast to high expression. This implies a bottleneck in the secretory pathway. Comparison of specific secretion rates (q_p^ex) versus specific intracellular production rates showed clear differences in the secretion efficiencies between both strains (Fig. 4A). While for ScNTp002 (P_HSP12) the secretion efficiency (ratio q_P^ex/q_P) decreased on average by 40%, for the P_TEF1 it increased over 3-fold. Intracellular protein levels can also indicate a secretory bottleneck, as efficient secretion should result in lower intracellular accumulation. A positive trend between intracellular levels and specific secretion rates was observed when the TEF1-promoter controlled expression (p-value = 0.018, Spearman r = 0.71), indicating that higher intracellular accumulation drove secretion. The P_HSP12 led to an opposite correlation, where high intracellular levels corresponded to lower specific secretion rates (p-value = 0.02, Spearman r = -0.70) (Fig. 4B).

Effect of protein expression and secretion on physiology

Final biomass concentrations, and especially biomass yields on glucose (Y_XS) were higher for ScNTp001 (P_TEF1) (0.42 ± 0.03 g/g) compared to for ScNTp002 (P_HSP12) (0.34 ± 0.050 g/g) (Supplementary Table 3, Supplementary Fig. 1C. This is in line with a higher non-growth associated glucose consumption by ScNTp002 (P_HSP12) (0.018 ± 0.007 g/g/h) compared to ScNTp001 (P_TEF1) (0.008 ± 0.003 g/g/h). This means that expression by P_HSP12 leads to the allocation of 36.2 ± 10% of the available glucose to these processes, which include cellular maintenance and product formation, at the end of the fed-batch culture, while for P_TEF1 mediated protein production this is only 17.6 ± 5.1%. To further assess the impact of the multicopy plasmid and protein secretion, we monitored viability. Viability remained above 99% for ScNTp001 (P_TEF1), while P_HSP12 driven expression resulted in a decline in viability, especially after 25 h of cultivation to 89.0 ± 3.1% (Supplementary Fig. 1A).

For both strains, two subpopulations with regards to cellular fluorescence intensity were observed at all time points (Supplementary Fig. 3). This separation into two populations is likely caused by the presence or absence of the plasmid, where cells containing the plasmid express ymNeongreen resulting in higher fluorescence intensities. However, 70% of the cells showed above background, green fluorescence at all the time points, indicating fairly good plasmid stability under the given conditions. P_HSP12-based expression resulted in higher average intensities (Fig. 3C and Supplementary Figs. 3B, 5D), but also more variation in fluorescent intensity among cells was observed, than for P_TEF1-based expression (Supplementary Fig. 3A, C).

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.