This describes a novel cell-free system in which endogenous ribosomes from the reticulocyte lysate are replaced by ribosomes from any tissues, organs or cell types. This enables to control the overall ribosomal concentration and to study the translation of specialized genes.
Cell free protein synthesis systems (CFPS) have been used for decades to express proteins from exogenously added nucleic acids that could be either mRNAs or cDNAs . They can be used to produce proteins with a pharmaceutical interest such as vaccine components and cytokines [2–5] and are routinely exploited to perform ribosome display  and high throughput production of protein libraries . In vitro translational assays represent a good alternative for directly sampling and screening molecules due to the absence of cell wall. Above all, CFPS have been used in many ground breaking experiments such as the discovery of the genetic code  and have been instrumental in deciphering many biological pathways involved in mRNA translation, stability, processing and miRNA regulation .
In theory, it should be easy to prepare translation competent cell free extracts from any cell type. However, this is not the case and only a few cell-free active systems have been developed in the last three decades, with lysates made from Escherichia coli, wheat germ (WG), insect cells and rabbit reticulocyte being the most commonly used . Whereas the E.coli, WG and insect cells can produce large amounts of a given protein, they are not adequate for studies on the pathway of mammalian translational control.
This is the reason why most discoveries have instead been conducted in the rabbit reticulocyte lysate which was initially designed by Hunt and Jackson in 1974  and later optimized for protein production by disposing of the endogenous mRNAs with the calcium-activated S7 micrococcal nuclease . Since then, both the untreated (uRRL) and the S7 nuclease-treated rabbit reticulocyte lysate (RRL) have been successfully commercialized and widely used by the scientific community to produce labelled or unlabelled proteins for biochemical purposes such as protein characterization or protein–protein, protein–RNA and protein–DNA interactions. This is the reason why the RRL has been instrumental in deciphering many of the biochemical pathways involved in the regulation of translation . However, a major concern with the RRL is that it does not recapitulate some important translation characteristics that are found in a cellular environment. One of them concerns the lack of cap- and poly(A)-dependence which is a critical determinant in translational control [13–15]. Advances in the field of RNA silencing has further stressed the importance of the poly(A) tail as a key player to control both translational efficiency and mRNA stability . Likewise, we have shown that the uRRL could recreate cap/poly(A) synergy which means that addition of the two elements has a combined effect greater than the sum of individual ones. This cap/poly(A) synergy is better revealed when exogenous mRNAs have to compete with other mRNAs, probably due to limitation of translation factors availability . In addition, the physiological relevance of the RRL has often been criticized for the study of human genes or viral mRNAs that infect humans as it may not contain all factors required for their expression. Along this line, it is noteworthy that many IRES (internal ribosome entry site)-driven mRNAs of cellular and viral origins are not, or poorly, translated in the rabbit reticulocyte lysate [17–20]. For this reason, supplementation of the RRL with cellular extracts has been instrumental in studies focused on IRES-driven translation in order to implement the translation reaction with cellular factors that are absent or missing from the RRL [17,21]. This has opened the way to the development of other in vitro translation systems based on mammalian cells extracts [22–25]. However, whereas all these systems can faithfully to recreate a competitive cellular environment, they are not easy to make and very inefficient compared with the RRL. Therefore, we have designed an in vitro system which relies on mixing components obtained from the rabbit reticulocyte lysate with a ribosomal fraction that can be isolated from cultured cells or even from tissues and organs. Such a reconstituted in vitro lysate retains the high efficiency of the parental uRRL and recapitulates translational characteristics observed in cells from which the ribosomes have been isolated.
MATERIALS AND METHODS
The p0–glo (β-globin)–renilla and p0–PV (poliovirus)–renilla coding for renilla luciferase and driven by the 5′-UTR of globin or PV IRES were described previously . The p0-bicistronic construct was cloned inserting the encephalomyocartditis virus (EMCV)–renilla fragment previously amplified by PCR from p0–EMCV–renilla into the p0–glo-firefly vector  previously digested by EcoRI restriction enzyme.
cDNA for polyadenylate binding protein interacting protein 1 (PAIP1) and eukaryote initiation factor (eIF)3g were obtained by reverse transcriptase-PCR (RT-PCR) using HeLa cells total RNAs as template and inserted into the EcoRI/NotI sites of the pCIneo–HA vector, as previously described [26,27].
cDNA for polymerase basic protein 1 of influenza (PB1), polymerase acidic protein of influenza (PA) and NS1 (influenza A non-structural protein 1) were obtained by PCR using PHW2000-PB1, PHW2000-PA, PHW2000-NS from A/Puerto Rico/8/1934(H1N1) strain  and inserted into p0–gloRenilla previously digested by BamHI and EcoRV restriction enzymes respectively, as previously described .
In vitro transcription
mRNAs were transcribed using the T7 RNA polymerase from templates linearized either at the AflII for polyadenylated mRNAs or at the EcoRV sites for non-polyadenylated mRNAs. Uncapped mRNAs were obtained by using 1 μg of linear DNA template, 20 units of T7 RNA polymerase (Promega), 40 units of RNAsin (Promega), 1.6 mM of each ribonucleotide triphosphate, 3 mM DTT in transcription buffer [40 mM Tris/HCl (pH 7.9), 6 mM MgCl2, 2 mM spermidine and 10 mM NaCl]. For capped mRNAs, the rGTP concentration was reduced to 0.32 mM and 1.28 mM of m7GpppG cap analogue (New England Biolabs) was added. The transcription reaction was carried out at 37°C for 2 h, the mixture was treated with DNAse and the mRNAs were precipitated with ammonium acetate at 2.5 M final concentration. The mRNA pellet was then resuspended in 30 μl of RNAse free water and mRNA concentration was determined by absorbance using Nanodrop technology. mRNA integrity was checked by electrophoresis on non-denaturing agarose gel.
HeLa, baby hamster kidney (BHK), A549, HuH7.3 and Jurkat cells were obtained originally from A.T.T.C. collection. Mouse stem cells were kindly donated by D. Aubert (IGF-Lyon, France) . HeLa, A549, HuH7.3 and BHK cells were typically grown in Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal calf serum (FCS) supplemented with 50 units/ml of penicillin, 50 μg/ml of streptomycin (PS). Jurkat cells were grown in Roswell Park Memorial Institute (RPMI) medium containing 10% FCSC supplemented with 50 units/ml of penicillin, 50 μg/ml of streptomycin, 10 mM Hepes (pH 7.2–7.5), 2 mM L-glutamine and 1 mM pyruvate. RML12 cells were expanded as described previously . Mouse stem cells were grown in Glasgow minimum essential medium (GMEM) containing 10% FCS supplemented with 50 units/ml of penicillin, 50 μg/ml of streptomycin, 1% modified Eagle's medium (MEM) non-essential amino acids (Gibco), 1 mM sodium pyruvate, 2 mM L-glutamine, 40 μM 2-mercaptoethanol and 400 μl leukaemia inhibitory factor (LIF; Chemicon). All the cells were grown under a humidified atmosphere containing 5% CO2 at 37°C, except for avian cells which were incubated at 42°C.
Cell lysate preparation (S10)
All following steps were performed at 4°C.
From cell lines
The pellet of 108 cells was diluted in an isovolume of lysis buffer [buffer R: Hepes 10 mM, pH 7.5, CH3CO2K 10 mM, (CH3CO2)2Mg 1 mM, DTT 1 mM]. The cell suspension was homogenized in a Potter homogenizer to obtain complete membrane dissociation and stopped before any degradation of the nuclei; this phase was monitored by microscopic observation, then the cell suspension was centrifuged at 16000 g for 10 min to yield the S10 supernatant extract.
From fresh organs
Mice were killed by cervical dislocation on day 15. Zebrafish fishes were kept under standard conditions and killed following protocols in accordance with regulations from the French Ministry of Agriculture and the European Union (agreement number A693870602) and organs were extracted and immediately washed in lysis buffer, then finely chopped in a new isovolume of clean lysis buffer and finally homogenized using a Dounce homogenizer and centrifuged at 16000 g for 10 min to yield the S10 supernatant extract. Wheat germ extract (WGE) was directly purchased from Promega.
Functional cell lysate preparation
For uRRL preparation, the method used was identical with that described  and can be briefly summarized as follow: 1 ml of uRRL purchased from Promega was supplemented with 25 μM Hemin (Fluka), 25 μg of creatine phosphokinase (Sigma–Aldrich), 5 mg of creatine phosphate (Fluka), 50 μg of bovine liver tRNAs (Sigma–Aldrich) and 3 mM of D-glucose (Sigma–Aldrich). 1 ml of functional HeLa lysate was constituted with HeLa S10 supernatant extract supplemented with 25 μM haemin (Fluka), 25 μg of creatine phosphokinase (Sigma–Aldrich), 5 mg of creatine phosphate (Fluka), 50 μg of bovine liver tRNAs (Sigma–Aldrich) and 3 mM of D-glucose (Sigma–Aldrich).
Reconstituted system and hybrid system preparation
Three-hundred microlitres of S10 preparation was centrifuged through a 1 ml of sucrose cushion (1 M sucrose in buffer R) for 2 h and 15 min at 240000 g. After removal of the sucrose solution, the resulting pellet is gently rinsed in buffer (buffer R2: 20 mM Hepes, pH 7.5, 10 mM NaCl, 25 mM KCl, 1.1 mM MgCl2, 7 mM 2-mercaptoethanol) and resuspended in 30 μl of buffer R2 and stored at −80°C. Ribosomal concentration is determined by Bradford or absorbance using Nanodrop technology. We obtain the equivalence: 1 μg of ribosomal fraction by Bradford=6–9 D260 ribosomes.
Fractionation of the reticulocyte lysate
After centrifugation for 2 h and 15 min at 240000 g of 1 ml of uRRL, 900 μl of post-ribosomal supernatant is collected [and called uRRL supernatant (Su)], frozen and stored at −80°C. The ribosomal pellet is rinsed in buffer R2 and resuspended in 100 μl of buffer R2 as above to constitute uRRL 10× ribosomal fraction (Ru) stock.
Five microlitres of S100 supernatant isolated from uRRL (Su; 50% v/v of translation mixture) was mixed with 1 μg of ribosomal fraction (R). These fractions are derived from uRRL ribosomes (Ru; Figure 1A) to give a reconstituted in vitro system or derived from cultured cells (HeLa, Jurkat, A549, HuH7.3, mouse stem cells etc.) or organs (heart, lung, liver, brain and muscle) to give a hybrid in vitro system. In vitro transcribed mRNAs were translated at 2.7 nM (corresponding to 10 ng of glo–renilla mRNA) unless specified in the figure legend in a final volume of 10 μl of hybrid lysate supplemented with 75 mM KCl, 0.75 mM MgCl2, 20 μM amino acids mixture. The translation reaction was left incubated for 30 min (unless stated) at 30°C before the reaction is stopped by addition of renilla lysis buffer (Promega).
Fractionation and reconstitution of the rabbit reticulocyte lysate
RNA extraction and quantitative RT-PCR
Ten microlitres of translation mixture was supplemented with 90 μl of pure water, then 1 ml of TRIzol Reagent (Life Technologies) was added and RNAs were extracted as indicated by the manufacturer. Three-hundred nanograms of these RNAs were treated with RQ1 DNase (Promega) and reversed using the high-capacity RNA-to-cDNA Master Mix (Life Technologies). Quantitative PCR (qPCR) was then performed as described  on endogenous glyceraldehyde 3-phosphate dehydrogenase (GAPDH) mRNA or the renilla luciferase mRNA. Relative copy numbers of GAPDH or renilla luciferase cDNAs were determined by their Ct value. List of the primer used for qRT-PCR:
qPCR S Renilla: TGGAGAATAACTTCTTCGTGAAAAC
qPCR AS Renilla: GCTGCAAATTCTTCTGGTTCTAA
qPCR S GAPDH: AGCCACATCGCTCAGACAC
qPCR AS GAPDH: GCCCAATACGACCAAATCC
Endogenous and neosynthetized proteins detection
Samples were resolved on a SDS/PAGE (12% gel), transferred to PVDF membrane and blotted using anti-eIF3e, anti-α-tubulin, anti-GAPDH, anti-RPS6 (ribosomal protein S6; Abcam), anti-PABP1 (poly-A-binding protein 1), anti-eIF4E (Sigma), anti-eIF4G, anti-argonaute2, anti-eIF2α (Cell Signaling) antibodies.
Renilla activity was measured using the renilla luciferase Assay System (Promega Co) in a Mithras (Berthold technologies) with 50 μl of substrate injection and 10 s of signal integration programme.
Samples were resolved on a SDS/PAGE (12% gel), dried and subjected to autoradiography for 12 h by the use of Kodak Biomax films (Fisher Scientific) and the signal was quantified by using a Molecular Dynamics PhosphoImager FLA 5100 (Fuji).
Rabbit reticulocyte lysate can be fractionated without loss of translational activity
We have adapted the method developed by Rau et al. , which consists of centrifuging the uRRL (Promega) to separate cytosolic components of the protein synthesis apparatus from the ribosomes. Both fractions can be rapidly frozen and stored at −80°C for several months (Figure 1A; Materials and Methods). In order to gain insight into the sublocalization of some components of the translational apparatus, we have performed Western blotting on the parental lysate (uRRL), the S100 supernatant (Su) and the ribosomal pellet [uRRL ribosomal fraction (Rurrl) Figure 1B]. Interestingly, eIF4G, PABP1 and eIF3e were exclusively found on the ribosomes whereas eIF2α, eIF4E and argonaute 2 were distributed in both fractions (Figure 1B). As expected, tubulin and GAPDH did not associate with the ribosomal pellet even after prolonged exposition of the membrane (Figure 1B).
The ribosomal and the cytoplasmic fractions from the uRRL were then re-assembled and used to translate the renilla construct driven by the β-globin 5′-UTR (Figures 1C and 1D). In order to refine the system, we have first determined the optimal amount of ribosomes (from uRRL) that can be mixed with the untreated lysate supernatant (Su) and found that to be one-tenth of the ribosome pellet (corresponding to 1 μg of ribosomes per 5μl of Su; Figure 1C). Then we compared the efficiency of the parental uRRL with the re-assembled lysate (Su+Rurrl; Figure 1D). It can be first observed that the supernatant fraction (Su) could not support translation and that the renilla construct was translated in the reconstituted lysate with an efficiency comparable to the efficiency of the parental lysate (Figure 1D; compare uRRL with Su+Rurrl). One microgram (0.1×) of ribosomal fraction was sufficient to restore translational efficiency of the parental uRRL. Therefore, a concentration of 1 μg of ribosomes was used in the remainder of the present study.
Biochemical features of the reconstituted lysate
We have next investigated how translation in the reconstituted lysate could be affected by different parameters such as ionic conditions, mRNA concentration and time of incubation. The luciferase gene construct driven by the glo 5′-UTR was translated in both the uRRL and the reconstituted lysate which was assembled with 1 μg of ribosomes. In a first set of experiments, the monovalent KCl concentration was adjusted from 0 to 150 mM final concentration added and translation was monitored for 30 min (Figure 2A). In both systems, we observed a peak of activity at 75 mM KCl added followed by a decrease for both lysates at the concentrations above. In a similar experimental setting, the divalent MgCl2 concentration was assayed and, likewise, we could see a similar behaviour for both lysates with an optimum at 0.75 mM MgCl2 added (Figure 2B) indicating that both in vitro systems are equally sensitive to changes in monovalent and divalent ion concentration. Next, we performed a dose curve from very low (0.2 nM) to very high (5400 nM) mRNA concentrations (Figure 2C) and this showed a very similar translational profile between the two lysates. Finally, the length of incubation was also investigated from 0 to 90 min from the time of addition of the mRNA in the translational assay (Figure 2D). In this case, we observed a plateau of activity in the parental lysate (uRRL) after 30 min whereas protein synthesis continued to rise in the reconstituted system (Ru+Su) for up to 60 min (see Figure 2D). Therefore, in order to allow a direct comparison for the remaining experiments, we have deliberately chosen to limit the incubation period to 30 min. mRNA integrity in both the uRRL and the Su+Ru was checked by qRT-PCR for a wide range of mRNA concentrations (from 0.27 to 2700 nM) and is plotted in Figure 2(E). This showed that no degradation has occurred over the 30 min incubation time and no difference in transcript stability could be observed between the two cell-free lysates.
Biochemical features of the hybrid lysate
Design of a hybrid system between the uRRL and ribosomes derived from cells
Having shown that the cytoplasmic components of the uRRL can be separated and reconstituted from the ribosomal pellet, we reasoned that we could take advantage of this flexibility to design a novel hybrid in vitro translation system, by assembling the supernatant from the reticulocyte lysate (Su) with the ribosomal fractions isolated from cultured cells. In order to do this, we followed the experimental procedure described in Figure 1(A) except that the ribosomes were pelleted from cultured cells. We used a spectrum of immortalized human cell lines from different origins such as HeLa (human adenocarcinoma), Jurkat (human lymphocyte from acute T-cell leukaemia), A549 (human lung carcinoma) and HuH7.3 (human hepatocarcinoma) but also undifferentiated mice stem cells (Figure 3A). Hybrid reconstituted systems were assembled by mixing different quantities of ribosomal fractions (R) isolated from these cells with the S100 supernatant from the uRRL (Su; see Materials and Methods). An in vitro transcribed globin–renilla reporter gene was translated for 30 min before analysis of renilla activity. Data showed that all ribosomal fractions derived from mammalian cell lines resulted in efficient protein synthesis to a level comparable to that observed with the parental lysate (Figure 3A; see HeLa, Jurkat, A549, HuH7.3 and stem cells). We next have investigated whether ribosomes that were directly isolated from organs could be used to assemble the hybrid system. For this, mice were killed by cervical dislocation and ribosomes were purified as described in ‘Materials and Methods’ from the brain, lungs, liver and heart. Translation of glo–renilla was performed as described above and results are presented in Figure 3(B). Production of luciferase could be observed in all cases but to a variable extent, with brain- and liver-derived ribosomes being more efficient than those obtained from lung and heart (Figure 3B). This shows that ribosomes can be extracted from any organs or tissues to be utilized in the hybrid system and that we can observe differences in translational efficiency depending on the tissues from which the ribosomes have been isolated.
Ribosomes isolated from animal cell or derived from organs can be used to assemble the hybrid system
The study was next extended to non-mammalian cells and we used ribosomes purified from avian (T2EC cell line), insect (RML12 cell line), fish (zebrafish caudal muscle), worms (full Caenorhabditis elegans) and plant (WGEs) cells. The experimental procedure used was the same as for mammalian cells (see ‘Materials and Methods’) and these ribosomal fractions were mixed with the supernatant from the uRRL. Data presented in Figure 3(C) show that only avian ribosomes were capable of supporting translation of the glo–renilla reporter construct to a level similar to that obtained in the uRRL. Nevertheless, some weak activity could be detected with insect and plant ribosomes to a lesser extent (Figure 3C, right panel) but no luciferase was produced in the lysate assembled with ribosomal fractions from fish and worms.
Functional features of the hybrid system
In order to establish functional characteristics of the hybrid in vitro translation system, we isolated ribosomes from a HeLa cell line as the latter has been widely used for the elaboration of cell-free lysates which could serve as a basis for comparison [22,25]. The cartoon in Figure 4(A) recapitulates the experimental procedure and Figure 4(B) shows a side by side comparison between the uRRL, the RRL, hybrid system using HeLa ribosomal fraction [Su+HeLa ribosomal fraction (Rh)] and a cell-free lysate made from HeLa. Translational efficiency of six different mRNAs was analysed by SDS/PAGE after [35S]-radiolabelling (Figure 4B), three of them (renilla, PAIP1 and eIF3g) are cellular mRNAs and the others (PB1, PA and NS1) are viral mRNAs from influenza virus. All of them were efficiently translated in the uRRL, RRL and hybrid system but not in the HeLa cell-free lysate where no labelled product could be detected (Figure 4B). To verify mRNA stability, we measured the level of endogenous ribosomal associated GAPDH mRNA and exogenous added renilla mRNA. Stability of these mRNAs in the hybrid system containing the HeLa ribosomal fraction was measured by qRT-PCR for different times of incubation (see ‘Materials and Methods’); no evidence for degradation over a 60 min period could be observed as illustrated in Figure 4(D).
Elaboration of a reconstituted in vitro translation system with ribosomes from Hela cells
Because translation in the HeLa lysate was too low to visualize radiolabelled proteins, we decided to monitor the enzymatic activity of luciferase as a readout in both the hybrid system and the HeLa cell-free lysate (Figure 4C). This confirms that, at all mRNA concentrations tested, protein synthesis in the hybrid system was considerably higher than in the HeLa cell free lysate (Figure 4C), with a 10-fold greater efficiency of translation for our hybrid system compared with the classical HeLa lysate, independently of the mRNA concentration. It should also be noted that translation in the hybrid system can be monitored even at extremely low mRNA concentration (less than 2.7 nM of mRNA).
The next step was to assess how capping and polyadenylation can affect translation of a reporter gene. This is an important point as the RRL cannot recapitulate the cap/poly(A) synergy that is observed in cells or even in cell-free lysates derived from cells . Therefore, we have used a reporter luciferase mRNA driven by the glo 5′-UTR which was synthesized in vitro in all possible combinations: capped and polyadenylated (annotated +/+), capped and non-polyadenylated (+/−), uncapped and polyadenylated (−/+) and uncapped and non polyadenylated (−/−) (see cartoon on top of Figure 5A). These transcripts were translated in the hybrid system assembled with a ribosomal fraction derived from HeLa cells or in the RRL as control. Results show that the hybrid system faithfully recapitulated the cap/poly(A) synergy observed both in living cells  and in competitive in vitro systems  (Figure 5A) and this was particularly striking with mRNAs that lacked the 5′-cap structure (glo −/+ and glo −/−) for which translational activity was almost abolished in the hybrid system. It is noteworthy that no degradation of the transcripts was observed over the incubation period, as previously demonstrated .
Translation of IRES-driven mRNAs in the hybrid system
Over the last two decades, in vitro translation assays have been widely used for studies on internal initiation which is the mechanism used by some mRNAs to recruit ribosomes via an IRES . However, due to the peculiar mechanism of IRES-dependent translation, some in vitro systems are unable to support internal initiation and this was notably shown for the WG lysate in which picornaviral mRNAs are not expressed . To assess translation of IRES-containing mRNAs, we have used a bicistronic construct coding for the firefly (first gene) and the renilla (second gene) in which the EMCV IRES was inserted in the intercistronic spacer. As seen in Figure 5(B), production of the two genes was equally efficient in both the RRL and the hybrid system (compare lanes 1 and 3). We next checked whether the hybrid system could recapitulate some selective conditions that are encountered in the course of picornaviral replication. One of those is created by the cleavage of the initiation factor eIF4G by the virally encoded L-protease from the picornal foot and mouth disease virus [37,38]. In infected cells, such a proteolytic event results in the collapse of cellular cap-dependent protein synthesis whereas translation from the viral IRES is stimulated [38–41]. In the RRL, the addition of the L-protease results in an attenuated effect of the viral enzyme in comparison with what happens in living cells [16,37]. Thus, we have added the in vitro translated L-protease to both the RRL and the hybrid system prior to translation of the firefly–EMCV–renilla bicistronic construct described above. Western blot analysis showed that eIF4G was similarly proteolysed in both assays (Figure 4B, upper panel) and protein production was quantified by [35S]-methionine incorporation in order to better compare the variations of expression of the two genes (Figure 4B, lower panel). This shows that the addition of L-protease had a much greater affect in the hybrid system by virtually abolishing cap-dependent translation (Figure 4B, compare lane 2 and lane 4).
One of the major differences between existing in vitro cell free lysates and the hybrid system resides in the possibility of assembling the latter with exogenous ribosomal fractions issued from any cell type. Thus, we reasoned that such a property could confer flexibility and adaptability on the hybrid system when it translates mRNAs that exhibit tropism for some specific cell types. For instance, it has been known for some time that some IRES-containing mRNAs are poorly expressed in the rabbit reticulocyte lysate due to the lack of IRES trans-acting factors (ITAFs) that can be found in the cellular environment . Therefore, as a proof of concept, we have used the PV IRES which was shown to be active in HeLa but not in BHK cells and which also shows reduced activity in the RRL system [17,18]. Luciferase mRNAs driven by the PV IRES or the glo 5′-UTR (Figure 5C, upper panel) were generated in vitro and electroporated into HeLa or BHK cells for 1 h before analysis of luciferase activity. Interestingly, whereas glo was expressed similarly in both cell types, this was not the case for the PV IRES containing mRNAs which exhibited a drastic reduction in activity in BHK cells, as reported previously  (Figure 5C, middle panel). Next, we have made hybrid systems assembled with the supernatant from the uRRL (Su) with ribosomes isolated from HeLa (Rh) or BHK [BHK ribosomal fraction (Rbhk)] cells (Figure 5C bottom panel). Translation of both globin and PV mRNAs was monitored and compared in the RRL, the hybrid system made with HeLa ribosomes (Su+Rh) and the hybrid system made with BHK ribosomes (Su+Rbhk). This mainly shows that translation of globin-driven mRNAs remained constant in the three in vitro systems whereas IRES-driven translation was enhanced in the presence of HeLa ribosomes (dark bars) but drastically impaired with BHK ribosomes (dark grey bars). Therefore, this reflects the situation observed upon electroporation of the cells (middle panel) and shows that addition of ribosomes derived from a given cell type can confer a cellular tropism for translation. Taken together, this makes the hybrid system a valuable tool to study expression of specialized genes whose translation is conditioned by a cellular tropism.
The rabbit reticulocyte lysate has been widely used over the last three decades to investigate molecular mechanisms of translation. However, it has often been criticized for the study of human genes or viral mRNAs that infect humans as it can lack factors that are required for their expression [17–20]. This was clearly demonstrated with seminal studies on internal initiation in which it was shown that efficient translation of the PV and rhinovirus IRESes in the RRL required the addition of an S10 extract from HeLa cells [17,21]. As an alternative to the RRL, mammalian cell-free lysates have been successfully developed from HeLa cells for the study of internal initiation ; however, these lysates often show reduced translational activity compared with the reticulocyte lysate . Although the reasons for low efficiency and sensitivity of cell-free lysates (other than the RRL) remain poorly understood, it has been suggested that DNA contamination, insufficient ribosome concentration or the presence of contaminants could be an explanation .
Therefore, the aim of the present study was to develop a novel cell-free system that could combine the efficiency and sensitivity of the RRL with the specificity of cellular extracts. For this, we have taken advantage of the flexibility of the uRRL to be fractionated and reconstituted to develop a novel adaptable CFPS (Figure 1). This is based on the reticulocyte lysate, depleted of its endogenous ribosomes that are replaced by exogenously added ribosomal fractions isolated from cultured cells, tissues or even directly from organs (Figure 3). This generates a novel in vitro translation system which retains the efficiency of the parental RRL and offers the adaptability of adding ribosomes isolated from any cell line (Figure 3). This demonstrates that ribosomes and ribosome-associated factors can be isolated and transferred without any loss-of-function, dismissing the hypothesis of ribosomes alteration. The hybrid system also recapitulates cap/poly(A) synergy, supports translation of IRES-driven mRNAs and exhibits a greater sensitivity to inhibitors of protein synthesis compared with the S7 RRL (Figure 5). It is also noteworthy that the hybrid system is much more efficient than a HeLa cell lysate (Figure 4) and offers much more flexibility. In fact, the hybrid system retains all major biochemical and functional characteristics of the rabbit reticulocyte lysate (efficiency, sensitivity, reliability, easiness to use etc.) and can be assembled with ribosomes derived from any type of cells.
Interestingly, we further showed that the cell type from which the ribosomes are derived greatly influences the translational characteristics of the system. This was illustrated by translating the PV IRES in the hybrid system assembled with HeLa (where PV translation is active) or BHK (in which PV translation is inactive; Figure 5).
It has been demonstrated that some ribosomes found in human pathologies contain modified rRNAs and ribosomal proteins  and these were suggested to play a role by altering the translational regulation pathway [43–45]. In that respect, the hybrid system could be a good tool to monitor, from a functional point of view, the impact of any cellular modifications that can affect the ribosome or any ribosome-associated factors.
Data showing weak or no activity with ribosomes isolated from insect, plant, fish and nematode suggest important differences in the nature and structure of ribosomes from different species (Figure 3). However, at this stage, we cannot completely rule out the fact that lack of activity of these ribosomes may be caused by differences in stability or integrity during the technical purification and this should be investigated further.
In addition, we also observed significant differences in efficiency between ribosomes derived from different organs from the same species (Figure 3). In this latter case, variations in activity could rather reflect essential differences in the composition of the ribosome-associated proteins. These ribosome associated proteins were recently defined as the ‘riboproteome’ by Pandolfi and colleagues  who characterized up to 1000 proteins among which are found constituents of the messenger ribonucleoprotein (mRNP) associated with core factors (initiation and elongation factors) and proteins (mainly ribosomal proteins) required for basic translation. They also showed that the amount and composition of the ribo-proteome greatly varies according to the cell type and the physiological status of the cell, which is in agreement with data presented in the present manuscript. Although it has been inferred for quite some time that the assembly of the mRNP is a determinant for the regulation of protein synthesis [47,48], this has always been demonstrated on particular cases of messengers or with specific constituents of the mRNP particle but never as a whole. Therefore, the utilization of the hybrid system could serve to directly address the role of these proteins in a functional in vitro assay.
Altogether, this adds to the growing evidence that the riboproteome is not homogenous within the cell and that it may be specialized for translation of a subset of mRNAs [42,49]. In that sense, the hybrid in vitro system could be the perfect tool to assess such a hypothesis and can be used both to investigate the role of individual ribosomal proteins through shRNA depletion and to identify and characterize the transcripts associated with non-canonical forms of ribosomes.
baby hamster kidney cells
cell free protein synthesis systems
eukaryote initiation factor
fetal calf serum
glyceraldehyde 3-phosphate dehydrogenase
internal ribosome entry site
influenza A non-structural protein 1
poly-A-binding protein 1
BHK ribosomal fraction
HeLa ribosomal fraction
nuclease treated rabbit reticulocyte lysate
uRRL ribosomal fraction
untreated rabbit reticulocyte lysate
wheat germ extracts
Baptiste Panthu designed the project, performed the experiments and assisted in writing the manuscript. Didier Decimo was involved in the designing of the experiments and provided technical support. Laurent Balvay was involved in the design of the experiments and assisted in the writing of the manuscript. Théophile Ohlmann designed the project and wrote the paper.
The authors wish to thank Dr L. Mazelin and Dr M. Delattre (LBMC, France) for the kind gift of mouse organs and C. elegans used in this study. Mouse stem cells and zebrafish were kindly donated by D. Aubert and L. Bernard (PRECI, France). RML12 cell line was provided by Dr M. Dreux (CIRI, France) and T2EC cell line given by Dr S. Giroud (CGphiMC, France). PHW2000-PB1, PA and NS were kindly donated by Dr M. Rosa Calatrava (VirPath, France).
This work was supported by the Agence Nationale de Recherche sur le Sida [grant number AO2014-14409]; and a PhD fellowship from the région Rhône Alpes.
The technique described in the present study is patented under number WO2014122231-A1.