Translational control of recombinant human acetylcholinesterase accumulation in plants
- 5.7k Downloads
Codon usage differences are known to regulate the levels of gene expression in a species-specific manner, with the primary factors often cited to be mRNA processing and accumulation. We have challenged this conclusion by expressing the human acetylcholinesterase coding sequence in transgenic plants in its native GC-rich sequence and compared to a matched sequence with (dicotyledonous) plant-optimized codon usage and a lower GC content.
We demonstrate a 5 to 10 fold increase in accumulation levels of the "synaptic" splice variant of human acetylcholinesterase in Nicotiana benthamiana plants expressing the optimized gene as compared to the native human sequence. Both transient expression assays and stable transformants demonstrated conspicuously increased accumulation levels. Importantly, we find that the increase is not a result of increased levels of acetylcholinesterase mRNA, but rather its facilitated translation, possibly due to the reduced energy required to unfold the sequence-optimized mRNA.
Our findings demonstrate that codon usage differences may regulate gene expression at different levels and anticipate translational control of acetylcholinesterase gene expression in its native mammalian host as well.
KeywordsCodon Usage AChE Activity Synonymous Codon Codon Usage Bias Rare Codon
synaptic isoform of AChE
codon adaptiveness index
cauliflower mosaic virus 35S promoter
sequence of the human gene encoding AChE-S
plant expression optimized gene
relative synonymous codon usage
ribulose bisphosphate carboxylase
tobacco etch virus
vegetative storage protein.
Different organisms preferentially use a different subset of the synonymous codons specifying a certain amino acid. In fact, codon usage is emerging as an important regulatory factor, and codon usage bias can be detected not only between organisms , but also in the same organism between differentially expressed genes , in different tissues  and developmental stages . A correlation was found in some cases, most notably among prokaryotes and in some metazoans (e.g. in dipterans and in nematodes), between the codon usage and the abundance of the cognate tRNAs, indicating that in these cases at least codon usage is probably governed by evolutionary selective forces [5, 6, 7, 8]. However, "translational selection" may not be the only adaptive process shaping codon usage, which can be biased due to interactions not (directly) involving codon-anticodon interactions such as splice-site recognition  and mRNA turnover . Particularly, while the available genome-scale analyses done for several plant species demonstrate that codon bias in plants is adaptive [11, 12], heterologous gene expression studies in plants led to a tentative conclusion that translation efficiency of the foreign genes is often considered secondary to mRNA processing and stability as a determinant for expressivity .
For many reasons, acetylcholinesterase, (AChE) is a useful model for a comparative study addressing the issue of codon bias and translation efficiency, as well for post-translational events [14, 15]. Vertebrate genes encoding the enzyme from several sources were expressed in a wide variety of systems such as bacteria , yeast , mammalian cell cultures , transgenic mice  and more recently in plants [20, 21]. The enzyme is known for its role in the termination of synaptic transmission by hydrolyzing acetylcholine, but during the past decade evidence has been gathered indicating its critical role in various processes such as development, stress responses, innate immunity, and bioscavenging . This complex array of functions is enabled through the equally complex molecular biology of the ACHE gene: the intricate control of its transcription, alternative splicing and translation and the equally intricate post-translational events governing its function such as subcellular targeting, glycosylation, proteolytic processing, membrane anchoring and protein-protein interactions . Interestingly, non-cholinergic functions of AChE are underscored by identification of AChE-like activity and genes potentially encoding the enzyme in several plant species [21, 22, 23].
Original reports from our laboratory indicated that plants can express this complex human enzyme. However, yields of plant-expressed AChE were modest because of limitations in the prototype expression construct and host plant [20, 21]. Here we demonstrate that AChE accumulation in plants is limited at the translation stage and that the enzyme levels can be dramatically increased as compared to our previously reported results, by conforming its codon usage and GC content to that of highly expressed plant genes. The optimized sequence results in increased translatability of the mRNA without affecting its accumulation or processing.
Design and cloning of the plant-expression optimized oACHE-Sgene
Optimized molecular features.
Potential methylation sitesb
5'-intron splice signalsd
Optimization of transgene enhances AChEaccumulation in plants
In optimizing the human ACHE-S gene sequence, we took a preemptive approach: modifying many molecular features that might have otherwise proven deleterious to the expression of the transgene or to the accumulation of its product. Thus we have modified key signals leading to transcriptional silencing (methylation), aberrant transcript processing (premature polyadenylation and unwarranted splicing) and translation attenuation (codon usage). It was therefore of interest to explore which of these changes contributed to the dramatic improvement in AChE accumulation observed with the optimized construct. To this end we removed the two 5'-splice sites, the potential poly-A site or all from the human sequence creating constructs pTM167, pTM165 and pTM164 respectively. Conversely, we reintroduced the same sites into the plant-expression optimized sequence, creating pTM195, pTM197 and pTM156 respectively. These new AChE-encoding plant expression constructs, alongside the pTM092 and pTM049 were introduced into N. benthamiana by Agrobacterium-mediated gene transfer (Fig. 3C–E).
Because of the integration position effect, transformants harboring the transgenes exhibited a range of recombinant gene expression levels, manifested in the distribution of AChE activities observed (Fig. 3D). Only kanamycin-resistant plants that had detectable (non-zero) AChE activity were considered for the following analyses. Accumulation of AChE activity significantly differed (P < 0.0001) between plants expressing variants of hACHES and plants expressing oACHES (and variants thereof). The highest AChE activity level obtained among the hACHES plants was 89 mU/mg protein, with a mean of 20 ± 3 mU/mg protein (n = 47, Fig. 3C). These results are somewhat higher but comparable to results previously described by us regarding the expression of the non-optimized sequence encoding the catalytic core of the enzyme in tomato plants . In sharp contrast, the highest specific activity level demonstrated for oACHES plants was 852 ± 28 mU/mg protein and the observed mean was 129 ± 22 mU/mg protein (n = 75, Fig. 3C).
The distribution of AChE accumulation data (Fig. 3E) is to a good approximation lognormal, and can be fitted to a logarithmic function, which allows one to predict the theoretical maximal level of AChE accumulation that may be obtained with each of the construct families. These values are in a good agreement with the actual observed maximal values, being about 108 and 953 mU/mg protein for the human and optimized constructs, respectively (Fig. 3E). It therefore appears unlikely that by screening additional transformants we could have identified hACHES lines that would accumulate significantly higher levels of AChE, than those we report. We conclude that expressing oACHES constructs leads to a 10-fold increase in AChE accumulation as compared to the hACHES constructs, and that the enhancement is due to the sequence optimization and not due to insufficient transformant pool size.
hACHE-S and oACHE-Stranscripts accumulate to similar levels in transgenic plants
Our analysis of the human ACHE-S gene shows it contains several plant RNA processing signals that theoretically could lead to the production of truncated, presumably inactive, AChE. However, removing these sequences in hACHE-S (pTM167, pTM165 and pTM164) or their reintroduction into the oACHES sequence (pTM195, pTM197 and pTM156) did not have a considerable effect on the enzyme accumulation levels (Fig. 3D). In contrast to the highly significant difference between the two construct families, no significant differences were found among plants expressing variants of the hACHES sequence (P > 0.12) or the one expressing variants of the oACHES sequence (P > 0.07). This implies that the three potentially deleterious sequences (the two 5'-intron splice sites and the polyadenylation sequence) do not play a major role in this case.
Transcript accumulation levels and integrity were further assayed by quantitative real-time PCR analysis using primers and probes to either the 5'-UTR (TEV leader sequence) or to the 3'-UTR (VSP terminator), common to both hACHE-S and oACHE-S, and were normalized to the co-determined levels of 18S rRNA. Normalized mRNA levels were correlated with normalized AChE activity (Fig. 4B, 4C). For each construct, there exists a statistically significant linear correlation between the accumulation levels of AChE activity and those of ACHE transcripts when assayed either with the TEV 5'-probe (hACHE-S: r2 = 0.73, P = 0.0005; oACHE-S: r2 = 0.60, P <0.0001) or the VSP 3'-probe (hACHE-S: r2 = 0.50, P = 0.0104; oACHE-S: r2 = 0.49, P <0.0001). As expected, this suggests that for either type of recombinant gene, the accumulation of the enzyme product depends on the levels of its cognate mRNA, which vary among the different plant lines for each construct due to the positional effect of transgene integration [29, 30], a phenomenon not unique to plants . However, for both constructs, the distribution of transcript accumulation data is statistically indistinguishable for either the TEV 5'-probe (t test, P = 0.46, Insert in Fig. 4B) or the VSP 3'-probe (t test, P = 0.22, Insert in Fig. 4C). Furthermore, the quantitative real time RT-PCR assays directed, respectively, at the common 5' and 3' UTRs correlate well with each other (r2 = 0.51, P <0.0001, data not shown). Thus, oACHE-S and hACHE-S transcripts accumulate to similar levels in the respective transgenic plants and it can be concluded that optimization of the ACHE-S sequence did not significantly increase transcript abundance, ruling out the latter as an explanation for the increased AChE activity in the oACHE-S plants. The explanation that neither transcription nor stability were affected seems more likely than the possibility that the optimization of the sequence resulted in simultaneous but opposite changes in transcription and stability (i.e. increased transcription with reduced transcript stability and vice versa).
Optimization affects translation of the ACHE transcript
Taken together, our results indicate that the higher levels of recombinant protein in the oACHE-S transgenic plants as compared to the hACHE-S plants are not due to transcriptional, post-transcriptional or to post-translational events. Rather, it appears that increased translation rates of the oACHE-S mRNA are responsible for the fact that AChE accumulates to higher levels in the oACHE-S plants.
Initiation is often considered to be the rate-limiting step for the translation process, and the regulation of mRNA sequences and secondary structures flanking the initiatory AUG are considered to play critical roles [34, 35, 36]. All of our constructs contained the same 5'-UTR (the TEV leader), but the 5'-proximal regions of the coding sequence were vastly different between the optimized and the human constructs in terms of their codon usage and GC content (Fig. 2). By in silico folding of the common 5'-UTR and the first 150 nucleotides of the two coding sequences (Mfold program ), it can be shown that in the six most-stable putative secondary-structures predicted for hACHE-S, the initiatory AUG is found within a relatively long and stable stretch of double helical RNA (ΔG° = 17.5 ± 0.0 kcal/mol, data not shown), while the equivalent region of oACHE-S is predicted to be at a base of a very short and considerably less-stable stem-loop structure (ΔG° = 6.3 ± 0.6 kcal/mol). It may therefore be speculated, based on these hypothetical structures, that the 5' proximal coding region plays an important role in controlling translation initiation.
The history of molecular biology and its applications during recent decades is inseparably linked to the ability to express proteins in heterologous systems – i.e. the expression of a "target" protein from one species in "host" cells of another species . From a biotechnology perspective, success in such efforts can be defined by "high-level" accumulation of the bona fide target in the host. This requires selecting the appropriate host species (and strain), defining its optimal growth conditions and/or optimizing genetic elements that control the expression and accumulation of the target. For example, expression in Escherichia coli of mammalian AChE, a glycoprotein, results in low yields of the unglycosylated, misfolded and destabilized protein , pointing to the need for a eukaryotic expression system. The importance of determining appropriate conditions is demonstrate by AChE of the electric ray, a cold-adapted fish, which, when expressed in mammalian cells, failed to accumulate if the cells were grown at 37°C, but could accumulate if the growth temperature was lowered . Similarly, human AChE produced in Pichia pastoris was poorly secreted into the medium as compared to AChE from the snake Bungarus fasciatus [17, 39], probably because of a transcription-attenuating GC rich region close to the 5'-end of the human coding sequence (Fig. 2) [31, 39].
Our initial attempt at expressing human AChE in transgenic plants was modestly succesful, based on the definition suggested above: the enzymatic activity was indistinguishable from that of authentic human AChE, but the level of activity was fairly low . An important factor responsible for the low observed activity in crude extracts was the presence of the anticholinesterase glycoalkaloid tomatine in the transgenic host – tomato plants . However, low expression/accumulation levels of foreign proteins in transgenic plants have unfortunately been the norm rather than the exception, despite an obvious, and rather natural, bias in the literature to focus on the latter (for a comprehensive case study see Diehn et al ). Low yields of the recombinant protein product can be the result of limitations at every stage of gene expression, from transcription, through transcript processing and half-life, translation, post-translational modifications and turnover. Several strategies have been developed to overcome such limitations, including expression in plastids [41, 42], use of virus-assisted expression systems [43, 44, 45], and optimizing the expression cassette to include strong promoters, appropriate 5'- and 3'-UTR, fusing targeting sequences, and conforming codon usage to that of plants [40, 46]. Plastid expression is precluded because AChE is a glycoprotein and viral systems able to support the expression of >70-kDa proteins have only just become available for testing . We therefore opted to focus on optimization of the gene sequence, which was synthesized de novo (Fig. 1, 3A), and was introduced into the low-alkaloid species N. benthamiana , lacking appreciable anticholinesterase activity (data not shown).
We present evidence for a substantial (>40 fold) increase in the accumulation of AChE activity in soluble protein extracts obtained from transgenic N. benthamiana plants expressing the full length S variant of human AChE as compared to our previously reported results concerning the expression of the AChE core domain in tomato plants (compare Fig. 3 and Fig 6 to Mor et al.  and Fletcher et al. ). In order to directly assess the effects of coding sequence optimization on the expression of the gene and the accumulation of its product, we compared the performance of the optimized gene construct (oACHE-S) to its native counterpart (hACHE-S) in N. benthamiana by transient expression assays (Fig. 3B) as well as in stable transformants (Fig. 3C). We demonstrate that optimization results in a 5 to 10 fold increase in the accumulation of the enzyme. Our results imply that a major contribution to the successful expression presented here is due to the optimization of the construct, although there were species- and isoform specific effects. For example, expression in tomato plants was considerably lower than in N. benthamiana and expression of the catalytic domain (constituting all but the 40-residue long C-terminal domain) was lower than that of AChE-S (unpublished observations). The latter may reflect the stabilizing effect of the carboxyl-terminal peptide of the S isoform . Our results further establish that expression of AChE in N. benthamiana appears to have been marginally, if at all, limited by transcriptional, post-transcriptional, or post-translational events (Fig. 3, 4, 5, 6). It stands to reason, therefore, that enhanced translation of oACHE-S transcripts as compared to those of hACHE-S is responsible for the better performance of the latter construct.
Without dismissing other explanations, most previous reports suggested that limitations in expression of transgenes in plants are mainly due to unfavorable mRNA metabolism, more specifically to reduced synthesis, premature polyadenylation, aberrant splicing and reduced stability of transcripts of foreign gene in plant cells (reviewed by Diehn et al. ). Indeed, the case is quite compelling for at least one well-studied group of transgenes encoding the insecticidal δ-endotoxins of Bacillus thuringiensis (B.t.). When introduced into plant nuclear genomes, B.t. genes expressed poorly, and the dismal expression levels correlated well with low abundance of the full-length transcripts. A systematic and thorough investigation into the molecular mechanisms of low B.t. mRNA levels (and the consequential low levels of B.t. toxins) by Pamela Green and co-workers, led to their conclusion on the importance of "instability of mRNA as a cause of limited expression of foreign genes" in general . B.t. genes, like many other eubacterial genes, are AT-rich and have multiple AU-rich destabilizing sequences and AU-rich plant polyadenylation signals in their mRNAs. All of these factors may contribute to rapid mRNA decay [13, 40]. The eubacterial codon-usage bias towards AU-rich codons was also concluded to contribute to mRNA decay [13, 40], although rare codons are clearly not sufficient to cause transcript degradation . We, like others, observe good correlation between the transgene transcript levels and protein accumulation levels. Nevertheless optimization did not increase mature mRNA abundance (Fig. 4) and we have no evidence for aberrantly shorter mRNA species in plants that express the non-optimized hAChE-S construct. Furthermore, eliminating three potentially deleterious RNA sequences from hACHE-S did not result in improved performance, while introducing them into oACHE-S did not lead to reduced expression. Thus, we conclude that mRNA metabolism is unlikely to be key element in controlling expression of the ACHE transgene in plants.
What could be, then, the factors that limit the expression of the human ACHE gene in plants? In many cases of heterologous protein expression, the problem that "proteins are often difficult to express outside their original context,"  can be overcome by changing codons to reflect frequencies observed in highly expressed proteins in a particular expression system. Codon usage bias is the primary factor in limiting heterologous gene expression in prokaryotes [50, 51], chloroplasts , and mammalian cells [53, 54]. Codon bias is associated with translational bias: highly expressed genes tend to utilize a limited set of codons and avoid the use of others, while codon preference positively correlates with their cognate tRNA content [1, 55]. Rare codons, especially when clustered in the 5'-coding region (as can be seen in hACHE-S, Fig. 2), can lead to ribosome stalling, which acts to reduce elongation rates , and in some cases (but evidently not the case here) may in turn destabilize the mRNA .
Codon bias is inherently linked to a skew in the sequence's GC content. Many of the genes whose expression was attempted in plants, including the B.t. genes, are bacterial and highly AU-rich. Because AU-rich sequences destabilize mRNA in plants , the rule of thumb was to elevate the GC content as part of the optimization process. At 65%, the GC content of human ACHE is particularly high, and in specific areas, such as the 5'-proximal region, it is 75% GC (Fig. 2). At the molecular level, the high GC content of the 5'-proximal region may act to stabilize secondary mRNA structures, which in the context of mammalian cells may attenuate the transcription machinery [31, 39]. While we do not observe differences in the overall steady state levels of the ACHE transcript accumulating in either oACHE-S or h-ACHE-S plants, secondary structures also affects translation initiation [56, 57, 58] and may also have a role in controlling the translation elongation rate . By creating chimeric constructs consisting of optimized and native human sequences we were able to show thatthe GC-rich 5' membrane proximal region comprising about 10% of the gene had a pronounced inhibitory effect on the expression level from the otherwise fully optimized gene (h-oACHE-S construct, Fig. 6). Interestingly, the reciprocal swap represented in o-hACHE-S didn't significantly improve the expression when compared to hACHES. This demonstrates that the optimization of the 5'-proximal region of the coding sequence, while critical for efficient translation, could not by itself account for the high level of AChE accumulation seen in oACHE-S expressing plants. Our results therefore suggest that in plants, both translation initiation and elongation are the most likely stages to be affected by sequence optimization of the ACHE gene. Our results also raise the interesting speculation that while in plants the high GC content of the gene may be a major obstacle for translation while in mammalian cells, normally growing at a much higher ambient temperature (typically 10°C to 20°C higher) the region provides for yet another potential checkpoint for the highly complex regulation of ACHE gene expression in mammals .
The work presented here leads to the conclusion that human AChE accumulation in plants is limited at the translation stage and that this limitation is specifically linked to its highly biased codon usage and GC content. Changing the sequence to conform it to those that typical of highly expressed genes in plants results in increased translatability of the mRNA without affecting its accumulation or processing. The contributions of codon-optimization and GC-content adjustment to expression enhancement are difficult to be separately assessed, but it is clear that both are important. While the overall GC content should be adjusted to resemble the general base composition of highly expressed genes of the expression host, one should also pay particular attention to more localized GC-content skews, revealed by the type of analysis shown here (Fig. 6). Similarly, there is evidence to reject "across the board" replacement of synonymous codons to the most preferred one for each amino acid and in fact rare codons at certain strategic points along the sequence play an important role in functional protein synthesis whereby pausing at such rare codons allows proper folding of the already synthesized protein domains [61, 62]. We are currently evaluating the relevance of this last point for the ACHE gene. Additional factors, such as subcellular targeting of the protein product and its subunit composition may affect the levels of its accumulation in the transgenic host. Indeed, we have preliminary evidence demonstrating that ER retention further increases the accumulation of some AChE variants (but not others) in stably transformed plants to about 1% of total soluble protein (Geyer, Fletcher and Mor, unpublished). Such a level of expression allowed us to efficiently purify the enzyme and we are currently testing its organophosphate scavenging potential in animal models.
Codon usage bias, GC content and mRNA folding analyses
Codon frequencies were obtained from the Codon Usage Database [63, 64]. Codon usage bias was calculated according to Sharp and Li . Briefly, the mRNA sequences encoding a group of highly expressed proteins from the model dicotyledonous plant Arabidopsis thaliana were selected to provide a reference for the codon usage bias. These included the small subunit 1B of RuBisCO, chlorophyll A/B binding protein 2, ribosomal protein L1 and L2, 40S ribosomal protein S2, S3 and S4, (accession nos. NM_123204, NM_102733, NM_202757, NM_201956 NM_115247, NM_115247 and NM_125228, respectively). The relative synonymous codon usage (RSCU = observed frequency/expected frequency assuming equal usage of synonymous codons for an amino acid), the relative codon adaptiveness (w, RSCU normalized to the most abundant synonymous codon for an amino acid) and the CAI values were calculated for the human ACHE gene sequence  according to Sharp and Li  (Fig. 2). Rare (or infrequently used) codons are arbitrarily defined here as having w values of less than 0.8. A moving CAI value (n = 50) was calculated for each codon. Similarly, the GC frequency was scored for each base as the average of a 100-base stretch centering on the particular base (Fig. 2). We have used the Mfold program  to predict mRNA secondary structures within the first 300 bases of the mRNA (centering on the initiatory AUG).
Construction of a plant-expression vector for the human ACHE-S gene (hACHE-S)
Oligonucleotides used in this study.
5' Sequence 3'
Construction of a plant-optimized ACHE-S gene (oACHE-S) and plant expression vector
The oACHE-S gene was assembled by the method of Stemmer et al. . Briefly, 78 partially overlapping oligodeoxyribonucleotides, collectively encoding both strands of oACHE-S were synthesized (Integrated DNA Technologies, Inc., Coralville, IA), ranging 3560 bases in length. Exact lengths of the oligonucleotides were determined so that the overlap T m would be 60 ± 2°C. Oligonucleotides representing three sections of the gene (bases 1–785, 786–1333, 1334–1848, Fig. 1, 3A) were separately mixed and subjected to PCR amplification. Next, 1/20th of the PCR-amplified mixtures were subjected to a second round of PCR using flanking primers to yield three fragments for the three sections. The fragments were gel-purified, cloned into pTOPOTA (Invitrogen), to yield pTM052, pTM053 and pTM057. An AscI-SacI fragment of pTM053 was cloned into the respective sites of pTM052 to yield pTM075. An MfeI-KpnI fragment of pTM057 was then cloned into pTM075 to yield pTM079, containing all three oACHE-S fragments. The plasmid was sequenced and a singlebase mutation was corrected (sequence of oACHE-S was deposited with GenBank, accession no. DQ140345, Fig. 3). An NcoI-KpnI fragment from pTM079 was used to replace the corresponding fragment in pTM034, yielding the intermediate vector pTM084, and the pGPTVkan based binary vector, pTM092 (Fig. 3).
Construction of variants of hACHE-S and oACHE-Sgenes
Alterations of specific sequences in the hACHE-S and oACHE-S genes were introduced by site directed mutagenesis as follows (for the oligonucleotides used, see Table 2). The pea poly-A signal in pTM048 was mutated (Fig. 3)  using primers oTM101 and oTM102 to yield pTM152. The two 5'-intron splice signals present in pTM048 were sequentially mutated by using oTM103 and oTM104 to disrupt one signal, yielding pTM153, and then using oTM105 with oTM106 to disrupt the other, yielding pTM161. Similarly, the two 5'-intron splice signals were mutated in pTM152 to yield pTM157, resulting in a hACHE-S variant lacking all three potentially deleterious sequences. We then replaced the NgoMI-KpnI fragment in pTM034 with the corresponding fragment in pTM152, pTM161, and pTM157, obtaining pTM160, pTM162, and pTM158, respectively. Following sequence confirmation, the expression cassette was cloned into the binary vector, pGPTV-kan, as described above to yield hACHE-S constructs, pTM163 (minus poly-A), pTM166 (minus 5'intron splice sites), and pTM159 (minus poly-A and 5'-intron splice sites).
The three potentially deleterious sequences were introduced back into the oACHE-S sequence in an analogous way. The mutagenesis oligonucleotides oTM095, oTM096 were used to recreate the poly-A signal in pTM079 to give pTM189. An NcoI-SacI fragment was used to replace the corresponding fragment in pTM034 to yield pTM194 and its sequence was confirmed. The two 5'-intron splice signals were introduced into the sequence of pTM053 and pTM057 with oligonucleotides oTM097, oTM098, oTM099 and oTM100 yielding pTM169 and pTM148 respectively. The plasmids were sequenced and the full-length oACHE-S coding sequence (with the splicing signals) was reassembled from an AscI-MfeI fragment of pTM169, an MfeI-SacI fragment of pTM148, and the SacI-AscI backbone of pTM084 by a triple-ligation cloning step, yielding pTM172. Reintroduction of all three potentially deleterious signals was similarly carried out yielding pTM151. Following sequence confirmation, the expression cassettes from pTM194, pTM172, and pTM151 were cloned into the binary vector pGPTV-kan, as described above to yield oACHE-S construct variants, pTM196 (+poly-A), pTM193 (+5'-intron splice sites), and pTM154 (+poly-A, +5'-intron splice sites).
Construction of mosaics of hACHE-S and oACHE-Sgenes
An AvrII restriction site was introduced into the hACHE-S plasmid, pTM048, and the oACHE-S plasmid, pTM084 (with the respective primer pairs, oTM165 and oTM166, and oTM167 and oTM168, see Table 2), yielding pTM241 and pTM235, respectively. Following sequence verification, a XhoI-SacI fragment from pTM235 was cloned into pTM084 to yield pTM238, while an NgoMI-KpnI fragment from pTM241 was used to replace the corresponding fragment in pTM034 to yield, pTM242 (Fig. 3A). The HindIII-AvrII fragment from pTM238 was used to replace the corresponding fragment from pTM238, yielding pTM252. Similarly, an AvrII-SacI fragment from pTM238 was used to replace the corresponding fragment from pTM242, to yield pTM253. The expression cassettes from pTM252 and pTM253, complete with the aforementioned control elements of pTM034 were then cloned into the binary vector, pGPTV-kan, as described above, to yield pTM256 (o-hACHE-S) and pTM257 (h-oACHE-S), respectively.
Stable and transient transformation of Nicotiana benthamiana
Binary vectors containing genes of interest were electroporated into Agrobacterium tumefaciens LBA4404 and were used in the subsequent transformation of N. benthamiana explants (>50 each) as described before . Kanamycin-resistant plants were regenerated as described . Agrobacterium-mediated transient expression assays were conducted essentially as described before [67, 68]. Briefly, A. tumefaciens cells harboring the appropriate expression vectors were grown for 24 hours, harvested by centrifugation and resuspended to culture density of 0.4 OD600 in 10 mM MES (pH 5.5) and 200 μM acetosyringone. Test constructs were co-expressed with the silencing suppressor protein, p19 . To this end, A. tumefaciens cells carrying each test plasmid were mixed (1:1) with A. tumefaciens cells carrying a CaMV 35S promoter-driven p19 expression vector (kind gift of Dr. Hugh Mason). After nicking the abaxial leaf epidermis with a needle, the bacterial suspension was infiltrated into the leaf through a blunt tip of a 1 ml syringe. Five days post-infection, the infiltrated-leaf tissue was harvested and homogenized as described below.
Biochemical and Molecular analyses
Genomic PCR was preformed on 0.1 μg total DNA isolated from kanamycin resistant plants (DNeasy kit, Qiagen) using either primers oTM087 and oTM090 or primers oTM126 and oTM047 to amplify hACHE-S or oACHE-S respectively (Table 1). RNA blot analysis on total leaf RNA (RNAqueous kit, Ambion) was done as described before  using a digoxigenin-labeled probes that were synthesized with primers oTM087 and oTM090 and pTM079 or pTM048 as templates.
For real-time RT-PCR analysis, total cellular RNA was prepared by homogenizing 100 μg of leaf tissue in a FastRNA-Pro-Green tube containing lysis matrix (QBiogene) in the presence of 800 μL Lysis/Binding solution (Ambion) using a FastPrep machine (Qbiogene). Lysates were then processed according to manufacturer's instructions for RNAqueous kit with the plant RNA Isolation Aid (Ambion). Residual genomic DNA was removed using the DNAfree system (Ambion). Yield and purity were determined spectrophotometerically and then RNA samples were reverse transcribed (RETROscript, Ambion) using random priming. To assess the quality of the cDNA, samples were screened by PCR with the primers oTM085 and oTM086 specific to the N. benthamiana actin gene (AY179605) and only those samples producing the expected 0.36 kb fragment were set aside for further processing (data not shown). Real-Time PCR was done as follows using an ABI Prism 7900HT sequence detection system (Applied Biosystems). Amplification of RNA samples (18 ng) was performed in triplicates in a 384-well plate with primers specific to the common 5'-UTR TEV leader and 3'UTR VSP terminator sequences (TEVf, TEVr, VSPf and VSPr, see Table 2) and custom-made TaqMan® FAM/MGB probes (TEVp and VSPp, Applied Biosystems). As an endogenous reference served 18S rRNA which was amplified using the TaqMan® eukaryotic 18S rRNA primers and FAM/MGB Probe (Applied Biosystem). Transcript levels were quantitated using two separate standard curves for the 18S rRNA reference and the ACHE transcript (using sample oACHE-S-24B) and normalized (ACHE value/18S value).
RT-PCR was also used to test for the presence of shorter, aberrantly spliced transcripts using cDNA samples (50 ng in a 25 μl reaction) and oTM089 and oTM049 as primers (35 cycles). In order to amplify even very rare templates, 1 μl samples of the above PCR reactions (total new reaction volume was 100 μl) were subjected to a second 35-cycle round of PCR, and the products were resolved by agraose gel electrophoresis.
To assess the ex-vivo stability of plant-expressed AChE, leaf discs (7 mM) from seedlings were floated, abaxial side up, in diH2O or in diH2O supplemented with 1 mM cyclohexamide [Sigma, 69]. At the indicated time points, leaf discs were sampled (n = 10), blotted dry, snap frozen in liquid N2 and stored at -80°C until further analysis.
Leaf samples were homogenized in ice-cold extraction buffer (1 M NaCl, 25 mM Tris, 0.1 mM EDTA, 10 μg/ml leupeptin (Sigma), and 0.5% Triton X-100, pH 7.4, 3 ml per 1 g tissue) using ceramic beads in a FastPrep machine. Ellman cholinesterase assays using acetylthiocholine (ATCh, Sigma) as substrate in the presence or absence of the selective AChE inhibitor 1,5-bis(allyldimethylammoniumphenyl)pentan-3 one dibromide (BW, 10 μM, Sigma) were done as described . BW-inhibitable ATCh hydrolysis rates were monitored at 405 nm for 30 min with a SpectraMax 340PC spectrophotometer, (Molecular Devices) and specific activity has been reported in mU/mg protein (AChE unit definition here is, 1 μmol ATCh hydrolyzed in 1 min at 25°C). Total soluble protein was determined as described before .
Student t test for two-group comparisons was performed with GraphPad Prism 4.0c for Macintosh (GraphPad Software, San Diego California USA). Multi-group comparisons were done using the non parametric exact permutation test with the PGD program  and P values were computed based on 10 repeat blocks of 1,000 permutations each.
This work was supported in part by a research contract from Defense Advance Research Project Agency (#N66001-01-C-8015) and by the NIH CounterACT Program http://www.ninds.nih.gov/funding/research/counterterrorism/index.htmunder the U-54-NSO58183-01 grant; a consortium grant awarded to USAMRICD and contracted to TSM under the research cooperative agreement number W81XWH-07-2-0023.
- 5.Ikemura T: Codon usage and tRNA content in unicellular and multicellular organisms. Mol Biol Evol. 1985, 2 (1): 13-34.Google Scholar
- 16.Fischer M, Ittah A, Gorecki M, Werber MM: Recombinant human acetylcholinesterase expressed in Escherichia coli: refolding, purification and characterization. Biotechnol Appl Biochem. 1995, 21 (Pt 3): 295-311.Google Scholar
- 18.Kronman C, Velan B, Gozes Y, Leitner M, Flashner Y, Lazar A, Marcus D, Sery T, Papier Y, Grosfeld H, Cohen S, Shafferman A: Production and secretion of high levels of recombinant human acetylcholinesterase in cultured cell lines: microheterogeneity of the catalytic subunit. Gene. 1992, 121 (2): 295-304. 10.1016/0378-1119(92)90134-B.CrossRefGoogle Scholar
- 29.Herrera-Estrella L, Van den Broeck G, Maenhaut R, Van Montagu M, Schell J, Timko M, Cashmore A: Light-inducible and chloroplast-associated expression of a chimaeric gene introduced into Nicotiana tabacum using a Ti plasmid vector. Nature. 1984, 310 (5973): 115-120. 10.1038/310115a0.CrossRefGoogle Scholar
- 31.Soreq H, Ben-Aziz R, Prody CA, Seidman S, Gnatt A, Neville L, Lieman-Hurwitz J, Lev-Lehman E, Ginzberg D, Lipidot-Lifson Y, Zakut H: Molecular cloning and construction of the coding region for human acetylcholinesterase reveals a G + C-rich attenuating structure. Proc Natl Acad Sci U S A. 1990, 87 (24): 9688-9692. 10.1073/pnas.87.24.9688.CrossRefGoogle Scholar
- 32.Bendtsen JD: SignalP. Center of Biological Sequence Analysis, Technical University of Dennemark. [http://www.cbs.dtu.dk/services/SignalP]
- 38.Gibney G, Taylor P: Biosynthesis of Torpedo acetylcholinesterase in mammalian cells. Functional expression and mutagenesis of the glycophospholipid-anchored form. J Biol Chem. 1990, 265 (21): 12576-12583.Google Scholar
- 42.Kumar S, Daniell H: Engineering the chloroplast genome for hyperexpression of human therapeutic proteins and vaccine antigens. Methods Mol Biol. 2004, 267: 365-383.Google Scholar
- 48.Sternfeld M, Ming G, Song H, Sela K, Timberg R, Poo M, Soreq H: Acetylcholinesterase enhances neurite growth and synapse development through alternative contributions of its hydrolytic capacity, core protein, and variable C termini. J Neurosci. 1998, 18 (4): 1240-1249.Google Scholar
- 53.Mossadegh N, Gissmann L, Muller M, Zentgraf H, Alonso A, Tomakidi P: Codon optimization of the human papillomavirus 11 (HPV 11) L1 gene leads to increased gene expression and formation of virus-like particles in mammalian epithelial cells. Virology. 2004, 326 (1): 57-66. 10.1016/j.virol.2004.04.050.CrossRefGoogle Scholar
- 56.Wu X, Jornvall H, Berndt KD, Oppermann U: Codon optimization reveals critical factors for high level expression of two rare codon genes in Escherichia coli: RNA stability and secondary structure but not tRNA abundance. Biochem Biophys Res Commun. 2004, 313 (1): 89-96. 10.1016/j.bbrc.2003.11.091.CrossRefGoogle Scholar
- 63.Nakamura Y: Codon Usage Database. [http://www.kazusa.or.jp/codon/]
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.