A highly optimized human in vitro translation system

Summary

In vitro translation is an important method for studying fundamental aspects of co- and post-translational gene regulation, as well as for protein expression in the laboratory and on an industrial scale. Here, by re-examining and improving a human in vitro translation system (HITS), we were able to develop a minimal system where only four components are needed to supplement human cell lysates. Functional characterization of our improved HITS revealed the synergistic effect of mRNA capping and polyadenylation. Furthermore, we found that mRNAs are translated with an efficiency equal to or higher than existing state-of-the-art mammalian in vitro translation systems. Lastly, we present an easy preparation procedure for cytoplasmic extracts from cultured HeLa cells, which can be performed in any cell culture laboratory. These methodological advances will allow HITSs to become a widespread tool in basic molecular biology research.

Graphical abstract

Highlights

• •Programmable translation with high yields from purified mRNA
• •A minimal translation buffer using only 4 essential components
• •Cap- and poly(A) tail-dependent regulation of translation initiation in vitro
• •Simple protocol for cheap preparation of HeLa cell extract for in vitro translation

Motivation

Human in vitro translation systems are an ideal model for studying human translation and related processes. Such systems also enable fast and scalable production of recombinant human proteins. However, human in vitro translation systems are difficult to produce, generally provide low translation efficiency, and are limited in their ability to faithfully reproduce regulated translation from human cells. Through a stepwise optimization, we were able to develop a human in vitro translation system that requires a minimal number of components while exceeding the translation efficiency of currently available systems.

Bothe and Ban optimize a human cell-based in vitro translation system, which can be used to produce proteins in a cell-free way from purified mRNA in a test tube. Through systematic screening of chemicals used to supplement cell lysate, they have found a minimal composition that enables highly efficient translation.

Introduction

Translation of messenger RNA (mRNA) is strictly regulated in eukaryotes to modulate protein synthesis during cellular homeostasis, development, and cellular stress response. Mammalian in vitro translation systems have made key contributions to our biochemical and structural understanding of the general translation mechanism.^1,2,3 Lysate-based in vitro reactions that approximate conditions in living cells are powerful resources to study co- and post-translational gene regulation. Systems based on rabbit reticulocyte lysate (RRL) have been used widely for decades, whereas the use of analogous systems based on human cell extracts was hampered because they are more difficult to produce, are expensive, and offer lower yields.

Although RRL-based in vitro translation systems have some advantages, such as inexpensive preparation and high translation efficiency, they are limited as a model system for several reasons. Due to its specialized cell type, reticulocyte lysate is enriched with globin protein and mRNA, and therefore translation of exogenous mRNA is hampered by competition with endogenous mRNAs.⁴ To enable efficient translation from exogeneous transcripts, RRL is often treated with staphylococcal nuclease to degrade intrinsic mRNAs,⁴ creating an artificially mRNA-depleted system; however, nuclease treatment of RRL was reported to lead to dysregulated translation. In particular, the translation-enhancing effects of the mRNA 5′ cap and 3′ polyadenosine (poly(A)) tail were reported to be additive rather than synergistic^5,6 in nuclease-treated RRL. Another reason why RRL may not be an ideal model system is the fact that truncated mRNAs can be used to stall translating ribosomes in RRL,^7,8 implying that the system is deficient in ribosome quality control pathways, which usually resolve stalled translation states.⁹ Together, these observations indicate that RRL does not properly recapitulate translation regulation of typical mammalian cells. Due to these limitations of RRL, human in vitro translation systems (HITSs) are better suited for studying such regulatory pathways. For example, HITSs have been reported to readily reproduce mRNA cap/poly(A) tail synergism.^10,11 Efficient HITSs promise to expand our knowledge of translation regulation and therefore human gene expression and protein homeostasis.

Since the first demonstration of translationally active cell-free translation systems based on human cell extracts,¹² several studies have developed HITSs. Major methodological advances have been achieved regarding the use of translation-enhancing proteins,^13,14 as well as the recognition of viral mRNA sequences as a means to enhance translation initiation for mammalian cell-free protein synthesis.^15,16,17 However, despite these developments, the essential components used in translation buffers remained unaltered for several decades. Surprisingly, our review of available literature revealed that the composition of translation buffers differs widely between protocols both in number of optional components and, by up to two orders of magnitude, in the concentrations of individual components (Table S1).^{12,14,18,19,20,21,22,23} Given these large variations, we were interested to understand the functional significance of translation buffer components for human in vitro translation.

In the present study, we developed a minimal HITS through systematic optimization of translation buffer components, identifying several dispensable constituents. Our improved system offers a translation performance exceeding that of previous systems and commercial in vitro translation kits. Additionally, we show that translation-competent extracts can be reliably prepared from cultured human cells using a simple and fast protocol. The high translation efficiency combined with the low costs of production demonstrated in this study allow easy integration of the HITS into the toolbox of laboratories studying eukaryotic translation.

Results

A minimal HITS

Published protocols for preparing HITSs widely differ in the number of components added to the translation reaction. As a starting point for a minimal system, we selected a recent protocol containing eight components in the translation buffer.²² To investigate the role of individual components, we prepared HITSs from cytoplasmic HeLa cell extracts and varied the concentration of selected components in the translation reactions. We monitored the translation efficiency by expressing Renilla luciferase (RLuc) from a capped mRNA, followed by quantification of RLuc enzyme activity with luminescence assays.

Our systematic screening of individual components showed that RNase inhibitor, amino acids, and purified creatine kinase are all dispensable components for HITSs (Figures 1A–1C). These components were iteratively removed from the translation buffer and not used in subsequent experiments.

Figure 1.

Minimizing the translation supplement

Translation efficiency of human in vitro translation system (HITS) was measured using capped RLuc reporter mRNA (c = 5 nM). Green bars indicate the common concentration range used in published protocols for human in vitro translation. Average and error-propagated standard deviations (error bars) of at least three independent translation reactions are shown.

(A) Translation efficiency of HITS with recombinant RNase inhibitor (+RI) compared with water as a control. No signal was detected when mRNA was omitted from the reaction (−mRNA). Values are normalized to water control (N = 6).

(B) Translation efficiency of HITS with various concentrations of equimolar amino acid mixture. Values are normalized to water control (N = 6).

(C) Translation efficiency of HITS supplemented with creatine kinase at various concentrations normalized to water control (N = 3).

(D) Translation efficiency of HITS with various proportions of acetate and chloride normalized to the value with 100% acetate (N = 4).

(E) Components of the translation supplement before and after the simplification.

See also Table S1.

The cations Mg²⁺ and K⁺ are crucial for ribosome structure and function^24,25 and need to be supplemented for any in vitro translation system. Curiously, the choice of counter ions for these cations in published protocols varies, though most compositions use acetate or chloride in varying proportions. When preparing HITSs with different ratios of acetate to chloride while keeping the total concentration of Mg²⁺ and K⁺ constant, we found acetate to be detrimental for in vitro translation (Figure 1D). A similar, albeit weaker, inhibitory effect of acetate at similar concentrations of K⁺ was previously reported for L cell and HeLa cell extracts.²⁶ Based on this result, we decided to exclusively use chloride as the anion in HITSs. This left us with a simplified system where only 4, instead of 8, components are used to supplement HeLa cytoplasmic extract (HCE) (Figure 1E).

Optimized conditions enhance cell-free translation

Our experiments demonstrated that several established components of the translation buffer are not required for efficient translation in HITSs. In the next step, we set out to determine the optimal concentrations for the remaining components of this minimal system. We determined the concentration optima for MgCl₂ and KCl simultaneously with a two-dimensional screening approach (Figure 2A). Our results demonstrate that the system is very sensitive to small deviations in magnesium concentration, with an optimum between 0.6 and 0.9 mM MgCl₂, largely independent of the concentration of KCl. In contrast, the optimal range for KCl was much broader, centering around 90 mM KCl. Interestingly, the omission of MgCl₂ in the translation supplement could be compensated somewhat by using increased the concentration of KCl, whereas KCl is essential. These results were obtained with a capped mRNA. When we repeated the assay with an uncapped mRNA containing the encephalomyocarditis virus internal ribosome entry site (EMCV IRES), we found a similar MgCl₂ optimum; however, the KCl optimum was shifted to much higher concentrations (Figure S1). Nevertheless, we chose to avoid such unphysiologically high concentrations of KCl in favor of conditions that allow efficient cap-dependent translation.

Figure 2.

Optimizing salt and buffer concentrations for human in vitro translation

Translation efficiency of HITS programmed with a capped RLuc reporter mRNA (c = 5 nM).

(A) Translation efficiency of HITS as a function of MgCl₂ and KCl concentration of the translation supplement (interpolated graph from 87 independent data points).

(B) Translation efficiency of HITS as a function of HEPES pH and buffer concentration (interpolated graph from 21 averaged data points, N = 3).

(C) Translation efficiency of HITS as a function of creatine phosphate and MgCl₂ concentration (interpolated graph from 80 independent data points).

See also Figure S1 and S2.

Translation supplements are usually buffered at a pH range from 7.4 to 7.6 using HEPES. We expanded our screening around those pH values to include the entire buffering range of HEPES and determined the translation efficiency for several concentrations of HEPES at each pH (Figure 2B). Surprisingly, the optimum was pH 7.0, where translation is roughly 60% more efficient compared to pH 7.6. Additionally, higher concentrations of HEPES reduced translation efficiency regardless of the pH.

Finally, we optimized the concentration of creatine phosphate (CrP) used in the translation reaction. Omission of CrP resulted in no product formation, proving its essential role (Figure 2C). Translation efficiency increased as a function of CrP concentration up to 20 mM CrP and was inhibited at very high concentrations above 60 mM. We observed a weak correlation of the MgCl₂ optimum with the concentration of CrP between 8 and 60 mM. Since we do not observe a translation enhancement above 20 mM CrP, we decided to use a concentration of 20 mM in the final system.

It was previously reported that a truncated form of human protein phosphatase 1 regulatory subunit 15A (PPP1R15A, also referred to as GADD34) can be added to HITSs to enhance translation,²⁷ which acts by dephosphorylating and thereby activating eukaryotic initiation factor 2 (eIF2).²⁸ This counteracts cellular pathways that phosphorylate eIF2 to inhibit translation during the integrated stress response (ISR).²⁹ Addition of purified GADD34^Δ1–240 to our optimized system improved the protein yield up to 4-fold in a concentration-dependent manner (Figure S2).

Based on the optimization outlined above, we established the following final concentrations of components: 75% (v/v) HCE, 15 mM HEPES (pH 7.0), 0.9 mM MgCl₂, 90 mM KCl, and 20 mM CrP. In total, we were able to improve the translation efficiency of our minimal HITS by more than one order of magnitude compared to the initial composition (not considering the additional effect of GADD34).

Characteristics of translation in the improved HITS

In mammalian cells, translation is especially regulated during the initiation stage, which includes mRNA binding, scanning, and start codon selection. How well a specific mRNA is translated depends on many factors, such as the composition of its 5′ and 3′ untranslated regions (UTRs) and the quality of the translation initiation site (TIS).^30,31 In particular, the presence of a 7-methylguanosine cap at the 5′ end and a 3′ poly(A) tail are mRNA features that promote initiation.^32,33 By in-vitro-translating reporter mRNAs with or without a cap and poly(A) tail, we could confirm the synergistic roles of the 5′ cap and 3′ poly(A) tail to promote efficient translation (Figure 3A), confirming that our system enables efficient canonical cap-dependent initiation.

Figure 3.

Characterizing performance of the optimized human in vitro translation system

For all graphs, average and standard deviations (error bars) of independent translation reactions (N = 3) are shown. For normalized data, the error bars correspond to the error-propagated standard deviations.

(A) Translation efficiency of RLuc reporter mRNA (c = 5 nM) with and without a 5′ cap and 3′ poly(A) tail. Values are normalized to uncapped mRNA without a poly(A) tail.

(B) Time course of translation in HITS for capped and polyadenylated RLuc mRNA (c = 100 nM).

(C) Translation efficiency of HITS for RLuc mRNAs (c = 100 nM) bearing different 5′ untranslated regions (5′ UTRs).

(D) Translation yield of HITS as a function of mRNA concentration for a reporter encoding a 3×FLAG-tagged RLuc. The translation product was quantified via a luminescence assay and qualitatively confirmed by western blotting. Luminescence data were normalized to the lowest mRNA concentration.

(E) Western blot showing the protein products from two reporter mRNAs differing only in the sequence surrounding the first of two in-frame start codons (constructs are shown schematically below the blot).

See also Figures S3 and S4.

To assess the kinetics of protein production, we recorded a time course during translation of luciferase mRNA. Active RLuc was first detected 5 min after the start of the reaction, and translation continued at a high rate for up to 1 h, after which translation drastically slowed down and eventually ceased (Figure 3B). In similar experiments performed with lower concentrations of CrP, we observed a net breakdown of protein after prolonged incubation, suggesting a switch to protein catabolism under conditions of energy scarcity (Figure S3). No such degradation was observed for our optimized system within several hours. Therefore, we conclude that the cease of translation after 1 h is not a consequence of energy depletion in the lysate.

Given that the translation reaction in our system is most productive during the first hour of incubation, it is important to maximize the yield of protein expression during this time. In contrast to nuclease-treated RRL, which is devoid of mRNA, our system retains cytoplasmic mRNAs. Therefore, the exogenous template mRNA competes with endogenous mRNA for translational resources, especially at low concentrations. However, altering the 5′ UTR of mRNAs may support preferential translation initiation. To demonstrate the regulatory role of the 5′ UTR, we programmed HITS reactions with mRNAs encoding the same RLuc protein while containing distinct 5′ UTRs. In particular, viral IRES sequences in the 5′ UTR allow translation initiation in a cap-independent manner, thereby bypassing the complex and rate-limiting canonical initiation pathway. This is confirmed by the fact that capping had only a minimal effect on the translation efficiency of mRNA containing a hepatitis C virus IRES (Figure 3C). We observed that the EMCV IRES sequence conveyed a 6-fold higher translation efficiency compared to capped mRNA harboring the 5′ UTR of human lamin B1, demonstrating that the choice of 5′ UTR influences the protein yield in HITSs (Figure 3C).

Protein synthesis in HITSs is most affected by the concentration of the template mRNA that is added to the reaction. To maximize protein yield in HITSs, we determined the mRNA-concentration dependence for a capped reporter mRNA encoding a fused RLuc-β-globin protein using luminescence measurements and western blot (Figure 3D). Our results show that the yield of active luciferase is a linear function of mRNA concentration in the range between 5 and 100 nM. Above a concentration of 100 nM, the linear relationship is lost, indicating that the system is saturated with mRNA. In similar experiments but using mRNA encoding for thioredoxin or green fluorescent protein, the in vitro translation reaction could be further stimulated even above an mRNA concentration of 100 nM (Figure S4). A possible explanation is that the length of the open reading frame determines the number of ribosomes that can simultaneously translate an mRNA, which is why shorter mRNAs may require higher mRNA concentrations to enable high levels of translation. Additionally, mRNAs with weak TISs have a competitive disadvantage compared to the endogenous mRNAs in the lysate and could require increased mRNA concentrations to yield a high amount of protein.

The TIS quality depends on the type of start codon and the sequence surrounding it (Kozak sequence). To demonstrate the effect of different Kozak sequences on the efficiency of translation initiation in HITSs, we used mRNA reporters that contained two in-frame start codons encoding for RLuc with an optional N-terminal (SGGGG)₁₄ tail. We translated two mRNA reporters in our optimized HITS that differed only in the Kozak sequence of the first start codon (shown schematically in Figure 3E). With a weak first Kozak sequence (UUUAUGU), a considerable proportion of ribosomes fail to initiate on the first TIS and initiate translation on the second start codon, producing a shorter protein that can be detected as a shift in a western blot (“1” in Figure 3E). In contrast, we found that a strong first Kozak sequence³⁴ (ACCAUGG) led to initiation mainly on the first start codon, producing a longer protein (“2” in Figure 3E). In this case, almost no signal was detected for the product initiating from the second TIS since ribosomes scan the mRNA in the 5′-3′ direction starting from the cap and a strong first TIS captures most scanning ribosomes. These results show that our system can reproduce start site selection in a Kozak-sequence-dependent manner. Importantly, this has the practical consequence that, for capped mRNAs, the choice of Kozak sequence influences the protein yield, and a strong Kozak sequence such as ACCAUGG should be employed for efficient translation initiation and a high protein yield.

Comparing mammalian in vitro translation systems

Currently, commercial mammalian in vitro translation kits are mostly based on RRL, whereas to our knowledge, only one human cell-based system is currently available on the market. The latter system is based on HeLa cell extracts and is optimized to translate mRNAs containing an EMCV IRES 5′ UTR. Since we optimized our HITS using capped mRNAs, we were interested in how it would compare to the 1-Step Human Coupled IVT kit (Thermo Scientific). To this end, we produced luciferase from either uncapped EMCV-RLuc or capped pCMV-3×FLAG-RLuc-β-globin mRNA in both systems. In comparison to the commercial kit, our improved HITS yielded a higher luciferase signal by a factor of 2–12 for IRES- and cap-driven translation, respectively (Figure 4A). The superior protein yield of our HITS from capped mRNAs was confirmed by western blot (Figure 4B).

Figure 4.

Benchmarking translation yield among mammalian in vitro translation systems

(A) Using reporter mRNA (c = 100 nM) encoding RLuc and containing either the 5′ UTR of encephalomyocarditis virus (EMCV) or the mammalian expression vector pCMV-3FLAG (pCMV), we quantified the luciferase yield after translation in our HITS and using 1-Step Human Coupled IVT kit (Thermo Scientific, “TS”). Average and standard deviations (error bars, N = 3) are shown.

(B and C) Comparison of translation yield for different proteins expressed from capped mRNAs (c = 400 nM). Protein products were detected using western blot with α-FLAG antibody. (B) Comparison between our minimal HITS and the 1-Step Human Coupled IVT kit (TS). (C) Comparison between our system and the Flexi RRL kit (Promega, “RRL”).

See also Figure S5.

We also compared our system to the Flexi Rabbit Reticulocyte Lysate kit (Promega) and found that the protein expression levels were similar in both systems (Figure 4C).

In addition to relative comparison between mammalian in vitro translation systems, we were interested in quantifying the absolute protein yield from our improved HITS. Using a luminescence standard curve prepared with purified recombinant RLuc protein, we estimate that our HITS incubated with 200 nM uncapped EMCV-hRLuc mRNA yielded 515 ± 24 nM RLuc protein in 1 h, which is equivalent to 18.8 ± 0.9 μg mL⁻¹ (Figure S5).

Preparation of human cell lysate for in vitro translation

Commercial HCE, which we used for the above experiments, does not have many vendors and can be expensive or impossible to modify. To decrease costs and allow control over lysis conditions, we prepared HCE from cell cultures grown in house. We modified a protocol³⁵ based on hypotonic swelling and homogenization with a glass pestle (Figure 5A). Compared to detergent-based lysis, this method is milder and assures compatibility with downstream applications such as cryoelectron microscopy for structure determination.

Figure 5.

Preparation of HITSs from cultured HeLa cells

(A) Flow diagram of the HITS preparation procedure starting from a suspension culture of HeLa cells.

(B) Translation efficiency of HITS as a function of HeLa cell culture density at time of harvest. Average and standard deviations (error bars, N = 3) of independent translation reactions are shown. Horizontal error bars represent 10% of the measured cell density. For comparison, the translation efficiency of a HITS prepared from commercial HeLa cytoplasmic extract is also shown.

(C) Relative translation enhancement due to addition of GADD34 for four different batches of HeLa extract. Translation efficiency was normalized to the buffer control. Average and error-propagated standard deviations (error bars, N = 3) of independent translation reactions are shown.

See also Figure S6.

All our HCE preparations gave rise to highly efficient HITSs. We observed that the extracts we obtained from cultures harvested at cell densities below 10⁶ cells/mL were more active (Figure 5B). The total protein concentration in our HCE preparations was variable (∼6 to ∼13 mg mL⁻¹ compared to ∼17 mg mL⁻¹ for commercial HCE); however, we found no correlation between protein concentration and activity of HITSs prepared from these extracts. We also successfully used a fast lysis protocol, which can be performed in as little as 45 min, and observed no reduction in translation efficiency compared to the standard protocol (Figure 5B, “fast lysis”). We would like to note that HCE is very stable over several freeze-thaw cycles when thawed on ice, which is of major importance for reproducibility and handling. In contrast, thawing the lysate at 30°C damages the system (Figure S6).

Remarkably, all HITSs prepared with homemade HCE were more efficient than those from commercial HCE, possibly a consequence of growing the cells at lower culture densities or the omission of DTT in our lysis buffer. The above experiment was done without adding purified GADD34. Interestingly, when we added GADD34 to HITS from homemade and commercial HCE, we found that the translation enhancement conveyed by GADD34 was much weaker for homemade extracts (Figure 5C), implying that the cellular stress response was not induced as strongly under our lysis conditions.

Discussion

Understanding the regulation of translation in human cells has been an active area of biochemical research for many years. The ability to study concerted biochemical mechanisms in a controlled, cell-free environment, while preserving the characteristics of translation that are present inside of living cells, can accelerate research. In particular, in vitro translation systems are compatible with the demands of structural biology experiments (e.g., cryoelectron microscopy) since fragile complexes can be purified directly. Additionally, lysate-based in vitro translation systems are a potential alternative to current in vivo systems of protein production, enabling fast synthesis of protein within hours while retaining proper folding and incorporation of correct post-translational modifications.³⁶ However, thus far, HITSs have still been occupying a niche, and published protocols for producing HITSs differ widely in their complexity and efficiency. HITSs are still considered difficult to prepare and expensive, which, so far, has prevented their adoption as a mainstream tool in molecular biology. In the current study, we developed a minimal in vitro translation system based on HeLa cell extracts, which reproducibly enables highly efficient translation.

We were able to show during our optimization that several components can be removed from the HITS without affecting translation efficiency. To our knowledge, creatine kinase and amino acids have been ubiquitously used in protocols for HITSs so far. The fact that these components are dispensable suggests that metabolic pathways remain functional in vitro, allowing the recycling of amino acids and the regeneration of ATP and GTP from CrP. Looking at the final composition of our minimal HITS, three out of four components serve to maintain salt and pH conditions that are similar inside living cells, while CrP provides energy for protein synthesis. Our approach of minimizing the translation buffer contrasts with a recent study where 18 different components were added to HEK cell lysate for in vitro translation, including three metabolic enzymes.²³ While such metabolically engineered systems can be useful for efficient cell-free protein synthesis, the main potential of HITSs lies in their ability to sustain complex regulatory networks for studies outside of living cells.

In our benchmarking assays, we were able to show that the protein yield of our improved HITS exceeds that of a commercial state-of-the-art human in vitro translation kit. Given the fact that the composition of our final HITS, which we used for these benchmarking experiments, is suboptimal for translation of EMCV-IRES-containing mRNAs, it is remarkable that our HITS translates EMCV-RLuc more efficiently than the commercial kit, which is optimized specifically for the expression of mRNAs containing an EMCV IRES sequence. This comparison demonstrates the leap in translation efficiency that we were able to realize through systematic optimization. Our results suggest that using a higher concentration of potassium chloride could further enhance the translation efficiency for EMCV-IRES-bearing mRNAs by almost 2-fold.

In addition, we found that the protein yields obtained from our HITS are very similar to a commercial RRL-based system. Since the RRL kit we used for comparison is treated with nuclease to remove endogenous mRNAs, it is noteworthy that our system can translate mRNAs with comparable efficiency despite intrinsic mRNAs competing with exogenously supplied mRNA for ribosomes. Moreover, apart from an improved efficiency of our HITS, we estimate that our system is less expensive per reaction volume by a factor of 20–80 relative to RRL and human in vitro translation kits, respectively.

The use of phosphorylated molecules, including CrP, as secondary energy sources for in vitro translation leads to the accumulation of inorganic phosphate over time, which is inhibitory to translation by perturbing the concentration of free magnesium in the system.^37,38 Conceptually, phosphate accumulation might be overcome if fed-batch or dialysis systems were established to replenish critical components such as magnesium or CrP during the reaction, which might allow translation for a longer time. However, such a setup would also dilute intrinsic components of the lysate and therefore is probably difficult to control and optimize. Given that we describe an inexpensive way to prepare HITSs, scaling up the reaction volume is probably the easiest way to increase the total yield for recombinant protein production.

In addition to the thorough optimization, we demonstrated that we could prepare highly efficient HITSs using homemade HeLa cell extracts. We found that even shortening the lysis protocol by a factor of two did not affect the translation efficiency, which implies that the precise timing is not critical. The robustness of the lysis protocol allows a reliable production of HITSs from cultured human cells in all laboratories equipped with a basic cell culture facility. Similar lysis methods can be adopted to a wide range of human cell lines, including HeLa cell lines where certain proteins were knocked down²⁰ or knocked out. Different cell lines may be more appropriate for studying certain biochemical pathways, and the choice of cell line can also influence the patterns of post-translational protein modifications that can be incorporated. In addition to a fast and simple lysis protocol, the small number of translation buffer components in our protocol can accelerate the optimization for other cell lysates, keeping in mind that the expression levels of creatine kinase might be too low depending on the cell line or physiological state, which might require addition of purified enzyme. Furthermore, using HITSs obtained from homemade HeLa cell extracts that seem to be less dependent on GADD34 may be useful for studying the ISR. The fact that GADD34-directed eIF2 dephosphorylation can be applied on more stressed HITSs implies that the regulatory pathways underlying the ISR are (at least partially) maintained, though future studies need to confirm whether protein kinases and stress sensing pathways remain active in vitro.

In conclusion, we present a highly efficient, minimal HITS that can be prepared with little effort from cultured human cells. This tool will accelerate studies that require in vitro protein production and deepen our understanding of the mechanisms and the regulation of translation. Additionally, HITSs can be used to rapidly express recombinant human proteins. Beyond translation, such an in vitro system may recapitulate many cytoplasmic pathways of protein homeostasis, including protein folding, assembly of protein complexes, and protein degradation.

Limitations of the study

One limitation of HITS might manifest when intrinsic membrane proteins or glycosylated, secreted proteins are synthesized since the lysis procedure removes parts of the intracellular membrane system. The presence of small microsomal vesicles in the lysate might not be able to process a high protein load, resulting in partial translocation and modification. However, the addition of purified microsomal membranes to the HITS could likely alleviate such restrictions.^39,40

If protein synthesis is the main goal, a coupled in vitro transcription-translation system could be set up by additionally supplementing an RNA polymerase and nucleotides, as was previously described for human cell-free systems.¹⁷ This allows the direct use of linear DNA templates and circumvents the need for a separate in vitro transcription step to produce mRNAs. However, coupled systems have little capping ability, which is why they generally utilize cap-independent IRES-driven translation to allow highly efficient protein synthesis (e.g., the cricket paralysis virus IRES).¹⁶

For structural studies on ribosomal complexes, RRL has long been the system of choice when preparing mammalian ribosome-nascent chain complexes because translation can be stalled using 3′ truncated mRNA. When using HITSs, this stalling method might be less suitable since ribosome rescue pathways can, in principle, resolve such stalled ribosome states. A useful alternative for stalling ribosomes is using an mRNA-encoded arrest peptide such as the engineered human Xbp1u sequence that stalls translation through specific nascent chain interactions inside the polypeptide exit tunnel of the ribosome.^41,42 Even though HITSs may require updated stalling approaches for translating ribosomes, we believe that an efficient HITS can functionally replace RRL for structural studies of translation-related complexes.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
FLAG (host: mouse)	Sigma Aldrich	Cat# A8592; RRID:AB_439702
Renilla luciferase (host: rabbit)	Abcam	Cat# ab185926; RRID:AB_3083537
Rabbit IgG (host: goat)	Santa Cruz Biotechnology	Cat# sc-2004; RRID:AB_631746
Rabbit IgG (host: mouse)	Santa Cruz Biotechnology	Cat# sc-2357; RRID: AB_628497
Actin (host: mouse)	Thermo Fisher Scientific	Cat# MA1-744; RRID:AB_2223496
Mouse IgG (host: goat)	Thermo Fisher Scientific	Cat# 62–6520; RRID:AB_2533947
Bacterial and virus strains
Escherichia coli MACH-1	Thermo Fisher Scientific	Cat# C862003
Escherichia coli BL21(DE3)pLysS	Sigma Aldrich	Cat# 69451-3
Biological samples
HeLa cytoplasmic extract	IPRACELL	Cat# CC-01-40-50
Chemicals, peptides, and recombinant proteins
T7 polymerase	This study	N/A
SUMO protease	This study	N/A
ScriptCap m7G capping kit	Cellscript	Cat# C-SCCE0625
1-Step Human Coupled IVT kit	Thermo Fisher Scientific	Cat# 88881
Flexi® Rabbit Reticulocyte Lysate System	Promega	Cat# L4540
EX-CELL® HeLa Serum-Free Medium	SAFC	Cat# 14591C
Critical commercial assays
Renilla-Glo® Luciferase Assay System	Promega	Cat# E2710
Amersham ECL Prime Western Blotting Detection Reagent	Cytiva	Cat# RPN2232
Experimental models: Cell lines
HeLa S3	Sigma Aldrich	Cat# 87110901; RRID:CVCL_0058
Oligonucleotides
PCR Primers for creation of linear templates for in vitro transcription from plasmids (see Table S2)	Microsynth	N/A
Recombinant DNA
pRSET-A-His6-GADD34Δ1-240	Gurzeler et al.22	N/A
pUC19-Lamin_B1-hRLuc	GenScript	N/A
pUC19-HCV_IRES-hRLuc	This study	N/A
pT7CFE1-CHis	Thermo Fisher Scientific	Cat# 88860
EMCV_IRES-hRLuc	This study	N/A
pET21(+)-RLuc-Kozak-bad-good	Twist Bioscience	N/A
pET21(+)-RLuc-Kozak-good-good	Twist Bioscience	N/A
pET-28a(+)-His6-SUMO-hRLuc	Twist Bioscience	N/A
pCMV-3×FLAG-RLuc-β-glo	Udy and Bradley43	Addgene plasmid #184393
FLAG-HA-GFP	Sowa et al.44	Addgene plasmid #22612
FLAG-Trx1	Liu and Min45	Addgene plasmid #21283
Software and algorithms
Origin Pro	OriginLab	Version 2023
PGFPlots	Christian Feuersänger	https://pgfplots.sourceforge.net

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Nenad Ban (ban@mol.biol.ethz.ch).

Materials availability

Plasmids generated in this study are available on request. Their full nucleotide sequences are provided as a supplementary .xlsx file with this paper (Table S3).

Data and code availability

• •Raw data for luminescence measurements are provided as a supplementary .xlsx file with this paper (Table S4).
• •This paper does not report original code.
• •Any additional information required to reanalyse the data reported in this paper is available from the lead contact upon request.

Experimental model and study participant details

Eukaryotic cell culture

Lysates for in vitro translation were prepared from cultured HeLa cells (female). They were grown in suspension in EX-CELL HeLa serum-free medium (SAFC) at 37°C, 110 rpm, 4.5% CO₂ in glass bottles. The cells were passaged every other day to a cell density of 0.3–0.5 × 10⁶ cells ml⁻¹.

Bacterial cell culture

Cells of Escherichia coli (strain BL21(DE3)pLysS) were transformed with plasmids for recombinant expression of protein. Details about the culture media used and incubation conditions are described in the Method details below.

Cells of Escherichia coli (strain MACH-1) were used for amplification of plasmids. They were grown in LB medium supplemented with appropriate antibiotic for selection, and incubated at 37°C, 180 rpm.

Method details

Plasmids

A plasmid containing a T7 promotor, parts of the human Lamin B1 5′-untranslated region and a Renilla luciferase open reading frame optimized for expression in human cells (hRLuc), followed by a poly(A) tail, was synthesized by GenScript (pUC19-Lamin_B1-hRLuc). A variant of this plasmid containing the Hepatitis C virus internal ribosome entry site instead of the Lamin B1 5′-UTR was generated by E. Kummerant (pUC19-HCV_IRES-hRLuc). A plasmid containing the encephalomyocarditis virus internal ribosome entry site, a luciferase open reading frame, and a poly(A) tail (EMCV_IRES-hRLuc) was generated by amplification of the open reading frame from pUC19-Lamin_B1-hRLuc and ligation into the linearized vector pT7CFE1-CHis (Thermo Scientific) using NEBuilder HiFi DNA Assembly Master Mix (NEB). The plasmids pET21(+)-RLuc-Kozak-bad-good and pET21(+)-RLuc-Kozak-good-good were synthesized by Twist Bioscience, and contain two in-frame start codons separated by an open reading frame encoding an unstructured (SGGGG)₁₄ linker, fused to a 3′ hRLuc open reading frame. The two plasmids differ only in the sequence surrounding the first start codon (Kozak sequence).

A plasmid encoding the fusion protein 3×FLAG-RLuc-β-glo (pCMV-3×FLAG-RLuc-β-glo) was a gift from Robert Bradley (Addgene plasmid #184393).⁴³ A plasmid encoding FLAG-HA-GFP was a gift from Wade Harper (Addgene plasmid #22612).⁴⁴ A plasmid encoding FLAG-Trx1 was a gift from Wang Min (Addgene plasmid #21283).⁴⁵

A plasmid encoding Renilla luciferase with an N-terminal His₆ tag separated by the cleavage site for SUMO protease (pET-28a(+)-His₆-SUMO-hRLuc) was synthesized by Twist Bioscience and used for recombinant expression of Renilla luciferase in Escherichia coli.

A plasmid encoding His₆-GADD34^Δ1-240 (pRSET-A-His₆-GADD34^Δ1-240) was kindly shared by O. Mühlemann (University of Bern) and used for recombinant expression in E. coli.

The sequences of all plasmids generated for this study are provided (Table S3, .xlsx file).

Preparation of reporter mRNAs

Plasmids were linearized by PCR using the Q5 DNA polymerase (NEB) with primers surrounding the template region for transcription (spanning the region from T7 promoter to poly(A) tail). For plasmids in which the poly(A) tail was not encoded, an A₃₀ tail was added through an overhang of the reverse primer. The DNA primers used for PCR are listed in Table S2.

After PCR, plasmid DNA was digested with DpnI (NEB) and the linear PCR product was purified with the QiaQuick PCR cleanup kit (Qiagen). The presence of a single band as template for in vitro transcription was affirmed by agarose gel electrophoresis (1% (w/v) agarose in TAE buffer).

Transcription was carried out in microcentrifuge tubes using 5–20 nM linear DNA template (PCR product) and purified recombinant T7 polymerase (1.15 mg mL⁻¹) in transcription buffer (40 mM Tris/HCl pH 7.6, 30 mM MgCl₂, 1 mM DTT, 2 mM spermidine, 10 mM each of ATP, GTP, CTP and UTP, and 0.1 U/μL RNaseOUT). After 2 h of incubation at 37°C the mRNA product was purified by adding 3/4 volumes of cold 7 M LiCl and precipitated 1 h at −20°C. Precipitated mRNA was pelleted by centrifugation (15 min, 13,000 rpm, 4°C), and the supernatant was discarded. The pellet was washed with 500 μL 70% ethanol and pelleted once more. After discarding the supernatant, the mRNA pellet was air-dried and resuspended in pure water. The concentration was calculated from the A₂₆₀ absorbance using sequence-based molar extinction coefficients, and the presence of a single mRNA product was confirmed by agarose-bleach gel electrophoresis (1% (w/v) agarose in TAE buffer with 0.06% (v/v) sodium hypochlorite).

Purified mRNAs were capped in vitro using the ScriptCap m⁷G capping system (Cellscript) according to the manufacturer’s instructions except that up to 300 pmol mRNA were capped per 100 μL reaction. Capped mRNA was purified using LiCl precipitation and its integrity was analyzed using agarose-bleach gel electrophoresis.

Purification of recombinant proteins

Purification of GADD34

For expression, Escherichia coli BL21(DE3)pLysS cells were transformed with the plasmid pRSET-A-His₆-GADD34^Δ1-240 by electroporation and grown at 37°C, 200 rpm in TB medium supplemented with 100 μg mL⁻¹ Ampicillin until OD₆₀₀ = 5.0. Expression was induced with 1 mM isopropyl β-D-1-thiogalactopyranoside (IPTG), and cells were further incubated for 6 h before harvest. The cell pellet was frozen in liquid nitrogen and stored at −80°C.

Cells were resuspended in lysis buffer (50 mM HEPES pH 7.6, 500 mM KCl, 5 mM MgCl₂, 10 mM imidazole pH 7.6, 10% (w/v) glycerol, 2 mM TCEP, 100 μM PMSF, 2 μM E−64, 10 μM Leupeptin, 10 μM Bestatin, 1 μM Pepstatin A, 10 μg mL⁻¹ Aprotinin, pinch DNaseI, pinch lysozyme) and lysed by sonication. The lysate was cleared by centrifugation in an SS-34 rotor (Sorvall) at 20,000 rpm for 1 h at 4°C. Protein was bound to a HisTrap HP column (Cytiva), washed with buffer A (50 mM HEPES pH 7.6, 150 mM KCl, 10% (w/v) glycerol, 40 mM imidazole, 1 mM TCEP) and eluted with a linear gradient of buffer B (50 mM HEPES pH 7.6, 100 mM KCl, 10% (w/v) glycerol, 500 mM imidazole, 1 mM TCEP) over 10 column volumes. The pooled elution fractions were separated from contaminating nucleic acids (high A₂₆₀ absorption) using a HiTrap Q HP column (Cytiva) equilibrated in low salt buffer (50 mM HEPES pH 7.6, 150 mM KCl, 10% (w/v) glycerol, 1 mM TCEP). Protein was applied to the column, washed with low salt buffer and eluted with a linear gradient of high salt buffer (50 mM HEPES pH 7.6, 1 M KCl, 10% (w/v) glycerol, 1 mM TCEP) over 10 column volumes. The first peak was pooled and concentrated to a volume of 1.5 mL using an Amicon Ultra-4 30K centrifugal filter (Merck). The protein was further purified using preparative size exclusion chromatography by applying the sample to a HiLoad 16/60 Superdex 75 prep grade column (Cytiva) equilibrated in final sample buffer (10 mM HEPES pH 7.6, 30 mM KCl) and eluting with final sample buffer at a flow rate of 1 mL min⁻¹. The void peak was collected, concentrated using a centrifugal filter, flash-frozen in aliquots in liquid nitrogen and stored at −80°C.

Purification of Renilla luciferase

For expression, E. coli BL21(DE3)pLysS were transformed with the plasmid pET-28a(+)-His₆-SUMO-hRLuc by electroporation and grown at 37°C, 120 rpm in LB medium supplied with 50 μg mL⁻¹ Kanamycin. When the culture reached an OD₆₀₀ of 0.65, expression was induced by adding 0.5 mM IPTG, and the culture was further incubated at 18°C overnight at 120 rpm. The cell pellet was flash-frozen in liquid nitrogen and stored at −80°C until purification.

Cells were resuspended in lysis buffer (see above) and lysed by sonication. The lysate was cleared in an SS-34 rotor (Sorvall), at 20,000 rpm, 1 h, 4°C. Protein was applied to a HisTrap FF column (Cytiva) equilibrated in lysis buffer. The column was washed with buffer A (see above) and eluted with a linear gradient of buffer B (see above) over 10 column volumes. The pooled fractions from the elution peak were concentrated in an Amicon Ultra-15 30K MWCO centrifugal filter (Merck) and then diluted 10-fold with Dilution buffer (50 mM HEPES pH 7.6, 150 mM KCl, 10% (w/v) glycerol, 1 mM TCEP). The His₆-SUMO tag was cleaved by adding 30 μL of 11.6 mg mL⁻¹ purified SUMO protease and incubating at 37°C for 90 min while shaking at 400 rpm in a heat block. Precipitated protein was removed by passing the solution through a 0.2 μm syringe filter (Filtropur S, Sarstedt) and the filtrate was loaded onto a HisTrap FF column. The flowthrough was collected and contained the cleaved Renilla luciferase protein. It was subsequently concentrated and buffer-exchanged to final sample buffer (50 mM HEPES pH 7.6, 100 mM KCl, 10% (w/v) glycerol) using a centrifugal filter. The purified protein was flash-frozen and stored at −80°C.

Human cell lysate preparation

HeLa S3 cells were grown in suspension in EX-CELL HeLa serum-free medium (SAFC) at 37°C, 110 rpm, 4.5% CO₂, in glass bottles. Cells were passaged every other day to a cell density of 0.3–0.5 × 10⁶ cells/ml. Cells were harvested at various cell densities (indicated in Figure 5B) from 200 mL culture by centrifugation in an SLA-1500 rotor at 1800 g, 4°C for 20 min. After the supernatant was discarded, the pellet was resuspended in PBS and transferred to a 15 mL polypropylene tube and centrifuged at 1800 g, 4°C for 10 min. The supernatant was discarded, and the packed cell volume (pcv) was estimated from the graduation on the tube. The cell pellet was washed once more by resuspending it in 5 pcv PBS and centrifuged (1800 g, 4°C, 10 min). The pellet was quickly resuspended in 5 pcv hypotonic buffer (10 mM HEPES pH 7.6, 0.5 mM MgAc₂, 10 mM KCl) and centrifuged (1800 g, 4°C, 5 min). The supernatant was discarded. At this point the cells will have swollen by up to 50% of their pcv. The pellet was resuspended with hypotonic buffer in a final volume of 3 pcv and cells were swollen for 10 min on ice. Cells were disrupted using a KIMBLE KONTES Dounce Tissue Grinder (7 mL) in combination with the pestle A, which has a clearance of 71–119 μm. No force was applied to the pestle other than gravity for the downward strokes, and the upward strokes were done roughly with the same speed. At this point, the combination of osmotic and mechanical forces will have disrupted the cell membrane, while the large clearance of the pestle keeps nuclei intact. The cell homogenate was cleared by centrifugation (3220 g, 4°C, 15 min) to remove (swollen) nuclei and cell debris, and the supernatant was flash-frozen in liquid nitrogen in small aliquots and stored at −80°C. We found that for 1 volume of cells, the yield is typically 2 volumes of extract (supernatant). The whole procedure for lysate preparation is represented schematically in Figure 5A.

Commercial HeLa cytoplasmic extract (IPRACELL) was thawed on ice, cleared at 10,000 g and 4°C for 20 min using an SS-34 rotor, flash-frozen in liquid nitrogen in aliquots and stored at −80°C.

In vitro translation reactions

Home-made or commercial cleared HeLa cytoplasmic extract (HCE) was thawed on ice (unless stated otherwise: Figure S6) and supplemented with RNaseOUT (Invitrogen), an equimolar concentration of 20 L-amino acids (Sigma), purified creatine kinase from rabbit muscle (Roche), potassium acetate (Sigma), potassium chloride (Supelco), magnesium acetate tetrahydrate (Sigma), magnesium chloride hexahydrate (Sigma), HEPES (Sigma) and creatine phosphate dibasic tetrahydrate (Sigma), unless stated otherwise. Supplemented HCE was flash-frozen in liquid nitrogen and stored at −80°C.

In vitro translation was carried out in 1.5 mL microcentrifuge tube or flat-bottom 96-well plates as follows: supplemented HCE was thawed on ice. Purified mRNA was diluted in the volume required, preincubated at 65°C for 5 min and immediately placed on ice to unfold unspecific secondary structures. mRNA was added to supplemented HCE and incubated at 33°C in a waterbath. The translation reaction was stopped by placing the tube on ice. The incubation times were: 50 min (Figures 1, 2, 3A, 3D, 5B, 5C, S1, S2, and S6), 60 min (Figures 3C, 3E, and S5), 90 min (Figure 4), unless stated otherwise (Figures 3B, S3, and S4).

In vitro translation using commercial systems (Figure 4) was performed according to the instructions provided by the manufacturers except that the reaction volume was set to 20 μL and the incubation time to 90 min.

In our initial experiments, the concentrations in the translation reaction were as follows²²: 75% (v/v) HCE, 1 U/μL RNaseOUT, 40 μM amino acids, 15 mM HEPES pH 7.6, 0.3 mM magnesium chloride, 24 mM potassium chloride, 28 mM potassium acetate, 6 mM creatine phosphate, 102 ng μL⁻¹ creatine kinase and 5 nM capped mRNA. Single components were supplied at various concentrations or removed during our initial optimization (Figure 1). After removing four components and optimizing the composition (Figure 2), final concentrations in the translation reaction were as follows: 75% (v/v) HCE, 15 mM HEPES pH 7.0, 90 mM KCl, 0.9 mM MgCl₂, 20 mM creatine phosphate.

For the titration of GADD34 (Figure S2), purified His₆-GADD34^Δ1-240 was included in the translation reaction at the concentrations indicated. For testing the susceptibility of home-made lysates to GADD34 (Figure 5C), 0.14 mg mL⁻¹ GADD34 or final sample buffer was included in the translation reaction. In any other case where GADD34 was included (Figures 3, 4, S4, and S5), the system was prepared as follows: HCE was thawed on ice, incubated 5 min with 0.15 mg mL⁻¹ GADD34 at 33°C in a waterbath and placed back on ice. It was flash-frozen after supplementation of HEPES, KCl, MgCl₂ and creatine phosphate and used for translation as above.

Luciferase assay

After in vitro translation, the translation reaction (20 μL) was transferred to a flat-bottom white 96-well plate (Falcon). An equal volume of luciferase substrate was added, using the Renilla-Glo Luciferase Assay System (Promega). Luminescence was measured after 10 min with a Synergy 2 plate reader (Biotek) in endpoint mode with 5 s integration time. The sensitivity was usually set to 200 and reduced in case the signal was too high. For time course measurements, a sample (5 μL) was taken from a batch translation reaction (60 μL or more) at each respective time point, diluted on ice with 15 μL water in a 96-well plate, and luminescence was quantified as written above.

To estimate the absolute yield of translation reactions, HITS reactions without mRNA were supplemented with purified Renilla luciferase and used to setup a standard curve. HITS reactions incubated with RLuc-encoding mRNA were diluted such that their luminescence values were in the linear range of the standard curve.

Western blot

Samples for gel electrophoresis were taken from the translation reaction, mixed with an equal volume of 2×SDS loading dye (4% (w/v) SDS, 10% (v/v) β-mercaptoethanol, 20% (w/v) glycerol, 172 mM DTT, 125 mM Tris/HCl pH 6.8, 0.006% (w/v) bromphenol blue) and incubated at 95°C for 5 min. For comparison of different in vitro translation systems (Figure 4), diluted gel samples from one system incubated with mRNA were mixed with an equal volume of the other system incubated with water and mixed with an equal volume of 2×SDS loading dye. This way the protein background could be normalized between lanes, even though systems with different protein composition were compared.

Proteins were separated on mPAGE 4–20% Bis-Tris gradient gels (Millipore) using a Mini-PROTEAN chamber (BioRad) and Tris-MOPS-SDS running buffer (GenScript), for 75 min at 130 V constant voltage.

After rinsing the gel in water to remove excess SDS, proteins were blotted onto a nitrocellulose membrane using the eBlot L1 Protein Transfer System (GenScript). Following transfer, the membrane was incubated on an orbital shaker for 1 h at room temperature in 50 mL 5% (w/v) milk in TBS, followed by 1 h incubation in 14 mL primary antibody solution (antibody diluted in 1% (w/v) milk in TBS; α-Renilla luciferase: 1:10,000; α-FLAG: 1:1,000) at room temperature. The membrane was washed for 15 min with 50 mL TBS-T and twice for 15 min with 50 mL TBS. In case of an HRP-conjugated primary antibody, the western blot was developed by applying 2 mL of Amersham ECL Prime Western Blotting Detection Reagent (Cytiva) to the membrane, incubation for 2 min at room temperature and recording the chemiluminescence. In case of non-conjugated primary antibody, the membrane was incubated for 1 h in 14 mL secondary antibody solution (antibody diluted in 1% (w/v) milk in TBS; goat α-rabbit IgG: 1:5,000; mouse α-rabbit IgG: 1:1,000). The membrane was washed as above (15 min TBS-T, 2 × 15 min TBS) and developed as above.

After recording the chemiluminescence, the membrane was briefly rinsed in TBS and then incubated twice for 10 min each in freshly prepared stripping buffer (0.1% (w/v) SDS, 1% (v/v) Tween 20, 200 mM glycine, adjusted to pH 2.2 with HCl) to remove bound antibodies. The membrane was then washed twice for 10 min in TBS and twice for 10 min in TBS-T. Following stripping, blocking was done as described and the membrane was incubated with antibodies as above to detect the loading control (α-Actin: 1:10,000 with secondary antibody goat α-mouse IgG: 1:3,000) and imaged as above.

The key resource table lists all antibodies that were used in the present study.

Quantification and statistical analysis

Luminescence measurements were performed on independent translation reactions that were incubated at the same time. The number of biological replicates N for each condition is indicated in each respective figure legend. All graphs show the individual measurements of replicates as small circles along with the arithmetic mean as squares and error bars corresponding to standard deviations, except for Figure 2B where due to the three-dimensional nature of the plot only the mean values of triplicates are plotted.

Where applicable, data was normalized to a specific experimental condition such that the normalized value $a’$ was calculated by $a^{\prime }=a/b$ where $b$ is the mean value of the reference used for normalization. The error of a′ was estimated by error propagation according to the following formula:

$$\Delta a^{\prime }=\sqrt{{\left(\frac{\partial a^{\prime }}{\partial a}\Delta a\right)}^2+{\left(\frac{\partial a^{\prime }}{\partial b}\Delta b\right)}^2}=\sqrt{{\left(\frac{\Delta a}{b}\right)}^2+{\left(-\frac{a}{b^2}\Delta b\right)}^2}$$

where $a$ and $b$ are mean values with standard deviations $\Delta a$ and $\Delta b$.

It should be noted that $b^{\prime }=b/b=1$; upon applying the partial derivative in the above formula, the error $\Delta b^{\prime }$ of $b^{\prime }$ is equal to 0 (i.e., the reference used for normalization has no error by definition).

For the absolute quantification of Renilla luciferase (RLuc), the luminescence y for the standard curve of purified RLuc was fitted with a linear equation $y=m\times c_{\text{RLuc}}+b$. The Regression tool in Microsoft Excel was used to calculate the standard errors of fit parameters, $\Delta m$ and $\Delta b$. The concentration of recombinant RLuc produced in HITS was calculated according to $c_{\text{RLuc}}=\left(y-b\right)/m$ where $y$ was the average luminescence of three independent translation reactions with standard deviation $\Delta y$. The error in $c_{\text{RLuc}}$ was calculated according to the following formula:

$$\Delta c_{\text{RLuc}}=\sqrt{{\left(\frac{\partial c_{\text{RLuc}}}{\partial m}\Delta m\right)}^2+{\left(\frac{\partial c_{\text{RLuc}}}{\partial b}\Delta b\right)}^2+{\left(\frac{\partial c_{\text{RLuc}}}{\partial y}\Delta y\right)}^2}=\sqrt{{\left(\frac{\left(y-b\right)}{m^2}\Delta m\right)}^2+{\left(\frac{\Delta b}{m}\right)}^2+{\left(\frac{\Delta y}{m}\right)}^2}$$

Published: April 11, 2024

Supplemental information

Table S4. Raw data for luminescence measurements, related to STAR Methods

Data is organised as one sheet per panel for all Figures from the main text and supplement.