|Home | About | Journals | Submit | Contact Us | Français|
High-level expression of human trypsinogens as inclusion bodies in Escherichia coli requires deletion of the secretory signal sequence and placement of an initiator methionine at the N terminus. Trypsinogen preparations obtained this way contain a mixture of abnormal N termini, as a result of processing by cytoplasmic aminopeptidases. Here we describe an expression system that produces recombinant human cationic trypsinogen with a native, intact N terminus, using intein-mediated protein splicing and an aminopeptidase P (pepP) deficient Escherichia coli strain. As a first application of this system, the effect of the pancreatitis-associated mutation A16V on the autoactivation of human cationic trypsinogen was characterized. The use of the novel pepP knock-out Escherichia coli strain should be generally applicable to the expression of recombinant proteins, which undergo unwanted N-terminal trimming by aminopeptidase P.
The human pancreas produces the digestive pro-enzyme trypsinogen in three isoforms. On the basis of their relative isoelectric points and electrophoretic mobility, these are commonly referred to as cationic trypsinogen, anionic trypsinogen, and mesotrypsinogen. The isoenzymes are encoded by separate genes, the PRSS1 (protease, serine, 1), PRSS2 and PRSS3 genes (for a recent review see  and references therein). Cationic trypsinogen (PRSS1) and anionic trypsinogen (PRSS2) make up the bulk of secreted trypsinogens in the pancreatic juice, while mesotrypsinogen (PRSS3) accounts for 2–10 % [2–6]. Trypsinogens are synthesized as pre-pro-enzymes (pre-trypsinogens) with a signal peptide of 15 amino acids, followed by the 8 amino acid long pro-peptide, the trypsinogen activation peptide. The signal-peptide is removed upon entry into the endoplasmic reticulum lumen and the pro-enzymes are packaged into zymogen granules and eventually secreted into the pancreatic juice. Activation of trypsinogen to trypsin is achieved by proteolytic removal of the pro-peptide, which normally takes place in the duodenum catalyzed by enteropeptidase (enterokinase), a highly specialized serine protease in the brush-border membrane of enterocytes. Trypsin can also activate trypsinogen, a process termed autoactivation, which in the duodenum may have a physiological role in facilitating zymogen activation, whereas inappropriate autoactivation in the pancreas might cause pancreatitis. Mutations in the PRSS1 gene have been identified in patients with hereditary pancreatitis, familial or sporadic chronic pancreatitis [7–11]. To date, 22 PRSS1 gene variants have been described, which affect 18 different amino-acid positions in trypsinogen, and result in 20 different amino-acid substitutions. When classified according to frequency of occurrence, R122H (~70 %), N29I (~25 %) and A16V (~4 %) are the three most prevalent mutations [8–11].
In order to elucidate how mutations in cationic trypsinogen cause pancreatitis, biochemical analysis of the mutant proteins is essential. Pancreatic tissue or juice is not readily available in significant amounts from patients carrying PRSS1 mutations. In addition to limited availability, isolation of the heterozygously expressed mutant trypsinogen from the complex protein mixture poses another technical challenge. Therefore, heterologous recombinant expression in E. coli has been the method of choice to generate wild-type and mutant trypsinogen proteins for biochemical analysis. For reasons that are not apparent, cationic trypsinogen is not secreted to the periplasmic space of E. coli when expressed as a fusion with the alkaline-phosphatase signal-sequence, even though fusions with human anionic trypsinogen, mesotrypsinogen or rat anionic trypsinogen are secreted to low levels . Similarly, expression of full length native pre-trypsinogen in E. coli results in no detectable zymogen secretion (M. S-T., unpublished observations). To obtain high levels of human cationic trypsinogen expression in E. coli, the signal-sequence is replaced with an initiator methionine residue and trypsinogen is expressed in the cytoplasm, where it accumulates in inclusion bodies, which are isolated and re-natured in vitro. Re-folding procedures for denatured trypsinogen have been developed and perfected from the 1970s and applied in the expression-purification protocols of bovine and human isoforms [12–22]. In our laboratory, re-folded trypsinogen is purified to homogeneity by affinity chromatography using immobilized ecotin . Ecotin is a protease inhibitor with broad specificity, isolated from the periplasm of E. coli. Cationic trypsin preparations produced recombinantly or purified from pancreatic juice exhibit essentially identical catalytic activity, indicating that recombinant trypsinogen is a suitable experimental alternative to its native counterpart . However, here we report that subtle structural differences do exist between recombinant and native trypsinogen preparations, because cytoplasmic aminopeptidases process the N terminus of the recombinant protein in E. coli. This raises the possibility that the biochemical properties of pancreatitis-associated mutations characterized in the context of recombinant trypsinogen preparations may not accurately reflect the in vivo situation. Furthermore, the N-terminal heterogeneity prevents the analysis of the biochemical effects of the A16V mutation, which alters the very N-terminal amino-acid of mature trypsinogen. Therefore, development of an expression system which allows high-level production of secretory trypsinogens with authentic N termini would be valuable. Here we describe that utilization of self-splicing mini-intein fusions combined with deletion of the pepP gene circumvents the deleterious effects of aminopeptidases and allows the expression of human cationic trypsinogen with an authentic, native N-terminal sequence. As an application, wild-type and A16V mutant proteins were produced, purified, and properties of autoactivation were characterized. The use of the pepP knock-out E. coli strain should be generally applicable to the expression of other proteins, which undergo unwanted N-terminal trimming by aminopeptidase P.
Numbering of amino-acids of human cationic trypsinogen follows the genetic convention; and starts with the initiator methionine (Met1) of pre-trypsinogen. Because the secretory signal-peptide comprising the first 15 amino-acids is removed in the secretory pathway, the first amino-acid of mature trypsinogen is Ala16.
Construction of the pTrapT7-PRSS1 plasmid harboring the human cationic trypsinogen gene was described previously [12, 17]. Note that the nucleotide sequence of this gene construct differs at several positions from the PRSS1 cDNA sequence, but encodes the same protein (see ref. 12 for details). Intein fusion constructs were generated by overlap extension PCR. The ~460 nt DnaB mini-intein gene from the cyanobacterium Synechocystis sp. was amplified from the pTwin2 plasmid (New England Biolabs) using the sense “primer A” 5′-CGG GAG TCC ATG GCT ATC TCT GGC GAT AGT CTG ATC AGC-3′ (the Nco I site within the primer sequence is underlined) and the antisense “primer B” 5′-GAA AGG AGC GTT GTG TAC AAT GAT GTC ATT CGC-3. A ~230 nt portion of the recombinant human cationic trypsinogen gene (corresponding to amino-acids Ala16 – Gln86) was amplified from the pTrapT7-PRSS1 template with sense “primer C” 5′-ATT GTA CAC AAC GCT CCT TTC GAT GAT GAT GAC AAG-3′ and antisense “primer D” 5′-CTG CTC ATT GCC CTC AAG GAC-3′. Primer D anneals downstream of an EcoR I site within the cationic trypsinogen sequence (Fig 1 and online supplementary material). The underlined sequences in primers B and C are overlapping and thus allow the PCR products obtained in the first round of amplifications to hybridize. PCR products from the two reactions were gel purified, mixed and extended using primers A and D. The resulting PCR product was digested with Nco I and EcoR I and ligated into the pTrapT7-PRSS1 plasmid digested with the same enzymes (Fig 1 and online supplementary material). The sequence of the fusion construct has been deposited with GenBank under accession number DQ371396. Mutation N29I was introduced into the intein-PRSS1 fusion gene by PCR mutagenesis using the antisense mutagenic primer AAC AGA ATT CTC TTC GCA GAT GTA TCC TCC-3′ and “primer A”. Mutation A16V was created by 2-step overlap extension PCR mutagenesis. Mutagenic (“internal”) primers were A16Vsense 5′-GTA CAC AAC GTT CCT TTC GAT-3; and A16Vantisense 5′-ATC GAA AGG AAC GTT GTG TAC-3′. External sense and antisense primers were “primer A” and “primer D”. The PCR products were digested with Nco I and EcoR I and ligated into the pTrapT7-intein-PRSS1 plasmid digested with the same enzymes.
The pepP gene (GenBank accession number D00398) encoding aminopeptidase P was deleted from the E. coli chromosome using the method described by Datsenko and Wanner (2000) . Strains and plasmids used were kind gifts from Barry L. Wanner. Briefly, we PCR-amplified the kanamycin resistance gene using primers with 44–45 nucleotide extensions homologous to sequences flanking the pepP gene (Fig 2). The template was plasmid pKD4. PCR products were treated with Dpn I to eliminate methylated template DNA. Electrocompetent E. coli BW25113 cells harboring the helper plasmid pKD46 encoding the λ phage Red recombinase were prepared and PCR products containing the kanamycin resistance gene flanked by two FLP recognition target (FRT) sites were introduced by electroporation. Transformants were selected by kanamycin resistance. Red recombinase catalyzes homologous recombination between the PCR product and the bacterial chromosome, resulting in the exchange of the pepP gene with the kanamycin resistance gene. The helper plasmid was eliminated by growing at 37 °C and the new strain was named LG-1. Plasmid pCP20, harboring the FLP recombinase gene, was introduced into kanamycin resistant LG-1 cells by electroporation. Transformants were selected by growing at 30 °C in the presence of ampicillin. FLP recombinase excises the kan gene between the two FRT sites, leaving one FRT site. Plasmid pCP20 was then eliminated by growing at 43 °C, and the resulting aminopeptidase P deficient E. coli strain was designated LG-2. Finally, plasmid pGP1-2, harboring the T7 RNA polymerase gene under the control of a temperature-inducible λ promoter , was introduced into LG-2 cells by electroporation. The resulting strain was designated LG-3 and was used for the expression of the intein-trypsinogen fusions.
Intein fusion constructs harboring wild type and mutant trypsinogens were transformed into the aminopeptidase P deficient E. coli strain LG-3. Transformants were grown in 200 mL culture volumes at 30 °C. When culture density reached OD 0.5 at 600 nm (OD600), expression of trypsinogen was induced by shifting the incubation temperature to 42 °C for 30 minutes and adding isopropyl 1-thio β, D-galactopyranoside (IPTG) to a final concentration of 1 mM. The temperature shift induces the expression of T7 RNA polymerase encoded on plasmid pGP1-2 under the control of a temperature-sensitive λ repressor. IPTG was added because the pTrapT7 plasmid also contained a lac operator between the T7 promoter and the intein-trypsinogen fusion gene . After induction, cells were grown for an additional 5 hours at 30 °C. Cells were harvested, resuspended in 0.1 M Tris-HCl (pH 8.0), 5 mM K-EDTA at 0.8 mL per 10 mL original culture volume, and inclusion bodies were isolated by sonication (3 × 20 sec, Heat Systems Ultrasonics cell disruptor, Model W-200R with a microtip probe, continuous mode, power-setting 4) and centrifugation (13,200 rpm, 5 min, Eppendorf microcentrifuge, 4 °C). The pellet was washed twice with 1 mL 0.1 M Tris-HCl (pH 8.0), 5 mM K-EDTA and dissolved in 500 μL 4 M guanidine-HCl, 0.1 M Tris-HCl (pH 8.0), 2 mM K-EDTA, 30 mM dithiotreithol (final concentrations) and incubated at 37 °C for 30 minutes to reduce trypsinogens. Denatured trypsinogens were rapidly diluted in 50 mL refolding buffer (0.9 M guanidine-HCl, 0.1 M Tris-HCl (pH 8.0), 2 mM K-EDTA, 1 mM L-cysteine, 1 mM L-cystine), stirred under argon for 5 minutes and incubated overnight at 4 °C. The solution was diluted with 50 mL 0.4 M NaCl, centrifuged for 15 minutes at 15,000 rpm and loaded onto a 2 mL ecotin affinity column . The column was washed with 20 mM Tris-HCl (pH 8.0), 0.2 M NaCl and trypsinogens were eluted with 50 mM HCl. The ecotin-eluate contained about 90 % trypsinogen and 10 % uncleaved intein-trypsinogen fusion protein, which were separated in a final chromatographic step using a MonoS HR 5/5 strong cation-exchange column. Concentrations of the purified trypsinogen solutions were calculated from their ultraviolet absorbance at 280 nm, using a theoretical extinction coefficient of 36,160 M−1 cm−1. Trypsinogens eluted from the MonoS column were prone to spontaneous autoactivation, therefore, functional analysis was performed immediately after the chromatography.
MonoS-purified trypsinogens diluted to 2 μM final concentration were incubated at 37 °C in a final volume of 100 μL. For assays at pH 5.0, column fractions in 20 mM sodium acetate buffer (pH 5.0) were diluted with the same buffer. For assays at pH 8.0, column fractions were diluted with Tris-HCl buffer (pH 8.0) to a final buffer concentration of 0.1 M and 1 mM CaCl2 (final concentration) was also included. Reactions also contained ~0.1 M NaCl carried over from the MonoS chromatography and 2 mg/mL bovine serum albumin as inert protein to prevent binding of trypsinogens to plastic surfaces. To initiate the autoactivation reaction, 10 nM trypsin was added (final concentration). At the indicated times, 2 μL aliquots were removed from the reaction mixture and trypsin activity was determined using the synthetic chromogenic substrate N-CBZ-Gly-Pro-Arg-p-nitroanilide. Kinetics of the chromophore release was followed at 405 nm in 0.1 M Tris-HCl (pH 8.0), 1 mM CaCl2, at 22 °C.
The initial aim of our study was to test the effect of the pancreatitis-associated A16V mutation on human cationic trypsinogen. The mutation affects the first amino acid (Ala16) of the mature zymogen. First, trypsinogens cloned under the control of the T7 promoter in the pTrap-T7 plasmid were expressed in the E. coli Rosetta (DE3) strain (Novagen), which produces the T7 RNA polymerase in an IPTG-inducible fashion. The secretory signal sequence was deleted from the trypsinogen gene and an initiator methionine was inserted upstream of the native Ala16-Pro17-Phe18- N terminus [12, 17]. Unexpectedly, N-terminal protein sequencing of purified trypsinogens revealed an inconsistent N terminus with approximately 30% of trypsinogens having a Pro17-Phe18 sequence, whereas circa 70 % of trypsinogens contained the Met-Ala16-Pro17-Phe18- sequence (Table 1). Essentially no trypsinogen was recovered with an authentic, native-like N terminus.
Post-translational removal of the initiator methionine is a well-documented phenomenon in E. coli . Methionine aminopeptidase cleaves off the N-terminal methionine of cytoplasmic proteins, provided the next amino acid has a small, uncharged side chain. However, when recombinant proteins are expressed to high levels in E. coli, methionine aminopeptidase becomes saturated and only a fraction of the recombinant proteins is processed. This problem is alleviated by co-expression of methionine aminopeptidase [27, 28]. Clearly, in our experiments a large fraction (~70%, Table 1) of the over-expressed trypsinogen remained unprocessed, indicating that saturation of methionine aminopeptidase has occurred. Consistent with this interpretation, when a high-expressing mutant of cationic trypsinogen (D22A, see ref 29) was produced, more than 90% of recombinant trypsinogens retained the initiator methionine (not shown).
Processing of cationic trypsinogen by methionine aminopeptidase should result in the accumulation of a species with the native-like Ala16-Pro17-Phe18- N-terminal sequence. However, we did not find this protein in the purified trypsinogen fraction. Instead, approximately 30 % of trypsinogens contained the Pro17-Phe18- sequence at the N terminus, suggesting that recombinant cationic trypsinogen was further processed by a putative aminopeptidase P enzyme (Table 1). Generally, proline aminopeptidases excise N-terminal amino acids before a proline residue ; in this case Ala16 before Pro17. Indeed, a gene (pepP) coding for an aminopeptidase P enzyme was identified and cloned from Escherichia coli HB101 [31, 32], suggesting that the pepP gene product was responsible for N-terminal processing of recombinant cationic trypsinogen.
To study the biochemical effect(s) of the mutation A16V, it was imperative to use a trypsinogen preparation with native, intact N terminus. Therefore, we employed a two-step method to avoid cleavages by the two aminopeptidases mentioned above. First, we eliminated the need for an initiator methionine by fusing the N terminus of the trypsinogen gene with the C terminus of the DnaB mini-intein from Synechocystis sp (Fig 1 and online supplementary material). Inteins are self-splicing protein elements with a range of experimental applications in molecular biology [33–35]. Protein splicing is defined as the excision of an intervening sequence (the intein) from a protein precursor and the concomitant ligation of the flanking protein fragments (the exteins) to form a mature host protein (extein) and the free intein . Accordingly, inteins contain an N-terminal and a C-terminal splicing region which catalyze the respective splicing reactions via different mechanisms (for a more detailed description of the splicing mechanisms and references see the New England Biolabs website at http://www.neb.com/neb/inteins.html). In our fusion construct, the N-terminal splicing was disabled by replacement of the catalytic N-terminal Cys by an Ala residue. The C-terminal splicing region was directly fused “in-frame” with the Ala16-Pro17-Phe18- sequence of the trypsinogen N terminus, which now serves as the extein in this particular context (Fig 1 and online supplementary material). In vitro, the C-terminal cleavage reaction is usually induced by incubation of the fusion proteins at pH 7.0, 25 °C, however, cleavage can also spontaneously occur during expression in vivo. As expected, after C-terminal self-cleavage of the intein, a trypsinogen with the native N terminus Ala16-Pro17-Phe18- was released from the intein-trypsinogen fusion protein. However, this N terminus was still subject to cleavage by aminopeptidase P, as revealed by N terminal sequencing of intein fusion proteins expressed in Rosetta (DE3) cells (Table 1). In the next step, we set out to prevent cleavage by aminopeptidase P. We created an aminopeptidase P deficient E. coli strain by removing the gene coding for aminopeptidase P (pepP) [31, 32] with the method described by Datsenko and Wanner (2000) . Absence of the pepP gene was demonstrated by direct sequencing of the region, which revealed the characteristic ‘scar’ sequence, normally remaining after the procedure (Fig 2). Growth curves in LB medium for the knock-out strain as well as the parental strain BW25113 were essentially identical; indicating that pepP is not an essential gene under typical laboratory conditions. Importantly, all trypsinogens expressed as intein fusions in the aminopeptidase P deficient E. coli had authentic, native-like N termini (Table 1).
Intein fusion constructs with wild type and mutant trypsinogens were expressed in E. coli LG-3 as inclusion bodies (Fig 3A). SDS-PAGE analysis of solubilized inclusion bodies revealed that a large fraction of the fusion proteins have already been processed to mature trypsinogens in vivo. After in vitro re-folding, trypsinogens were purified by ecotin affinity chromatography . On SDS-PAGE gels, the majority of the preparation consisted of cleaved, intein-free trypsinogen with slower migrating bands corresponding to unprocessed fusion proteins (Fig. 3B). To remove the small fraction of unprocessed fusion proteins, ion exchange chromatography was employed. Samples were loaded onto a Mono S HR 5/5 column and proteins were eluted with a gradient of 0–0.5 M NaCl. Eluted trypsinogens were free from unprocessed forms (Fig. 3C).
The A16V mutation of cationic trypsinogen was identified by Witt et al.  as the 3rd most common genetic variant associated with chronic pancreatitis. Other groups have also found this mutation in their samples [10, 38–40], and to date more than 25 affected carriers have been documented worldwide. Previous biochemical analysis of the more frequent pancreatitis-associated R122H and N29I mutations indicated that the common phenotypic change is an increased propensity for autoactivation [12, 17, 18]. Therefore, it seemed logical to assume that the A16V mutation might exert its effect via a similar mechanism. On the other hand, the mutation is 7 amino-acids away from the Lys23-Ile24 cleavage site (see Fig 1) and it affects a region which is largely disordered in crystal structures [41, 42], suggesting it may have little, if any, effect on trypsinogen activation. The development of the unique expression system described in this communication allows these assumptions to be tested. In the experiments presented in Fig 4, we compared autoactivation of cationic trypsinogen harboring the A16V mutation to that of the wild-type zymogen. Mutation N29I, previously shown to enhance autoactivation [12, 17, 18], was included as a positive control. Autoactivation kinetics was determined under two conditions, at pH 5.0 and at pH 8.0 in the presence of 1 mM CaCl2. As observed previously, the N29I mutant exhibited moderately increased autoactivation under both conditions. Mutation A16V, on the other hand, had essentially no effect on autoactivation at pH 5.0 and caused a slight decrease at pH 8.0. These results indicate that enhanced autoactivation can be ruled out as a mechanism underlying the association of mutation A16V and chronic pancreatitis. Other mechanisms, such as an effect on the intracellular transport of pre-trypsinogen, as speculated by Witt et al.,  may be operational.
We have created an expression system to yield recombinant trypsinogen with a native, intact N terminus, using intein-mediated protein splicing and an aminopeptidase P deficient Escherichia coli. As a first application of this system, the effect of the pancreatitis-associated mutation A16V on the autoactivation of human cationic trypsinogen was characterized. The use of the pepP knock-out E. coli strain should be generally applicable to the expression of other proteins, which undergo unwanted N-terminal trimming by aminopeptidase P.
The authors thankfully acknowledge Barry L. Wanner (Department of Biological Sciences; Purdue University, West Lafayette, IN 47907) for sharing the strains and plasmids of the Red-recombinase based gene deletion system. Special thanks to Ron Kaback for his support. Protein sequencing was carried out by David McCourt (Midwest Analytical, Inc).
This work was supported by NIH grant DK058088 to M. S.-T.