Engineering of the Bacillus subtilis PhoD signal peptide to direct high-level, Tat-specific 1 export of a single-chain antibody fragment to the periplasm in Eschericha coli

26 Background : Numerous high-value proteins have been produced in E. coli, and a favoured 27 strategy is to export the protein of interest to the periplasm by means of an N-terminal signal 28 peptide. While the Sec pathway has been extensively used for this purpose, the Tat pathway 29 has potential because it transports fully-folded heterologous proteins. Most studies on the Tat 30 pathway have used the E. coli TorA signal peptide to direct export, because it is highly Tat- 31 specific, unlike many Tat signal peptides which can also function as Sec signal peptides. 32 However, the TorA signal peptide is prone to degradation in the cytoplasm, leading to 33 reduced export rates in some cases. Here, we have tested a range of alternative signal peptides 34 for their ability to direct Tat-dependent export of a single-chain antibody fragment (scFv). 35 Results : We show that the signal peptides of E. coli AmiC, MdoD and YcbK direct efficient 36 export of the scFv by both the Tat and Sec pathways, which may be a disadvantage when Tat- 37 specific export is required. The same applies to the Tat signal peptide of Bacillus subtilis 38 PhoD, which likewise directs efficient export by Sec. We engineered the PhoD signal peptide 39 by introduction of a Lys or Asn residue in the C-terminal domain of the signal peptide, and 40 we show that this substitution renders the signal peptide Tat-specific. These signal peptides, 41 designated PhoDk and PhoDn, direct efficient export of scFv in shake flask and fed-batch 42 fermentation studies, reaching export levels that are well above those obtained with the TorA 43 signal peptide. Culturing in ambr250 bioreactors was used to fine-tune the growth conditions, 44 and the net result was export of the scFv by the Tat pathway at levels of approximately 1g 45 protein/L culture. 46 Conclusions : The new PhoDn and PhoDk signal peptides have significant potential for the 47 export of heterologous proteins by the Tat system.


3
Background 51 52 Numerous proteins are transported across the bacterial plasma membrane by one of two 53 mainstream protein export pathways: the Sec and Tat pathways (reviewed in [1][2][3][4][5]). In E. coli, 54 proteins destined for export by the Sec pathway are synthesised with N-terminal signal 55 peptides that contain 3 domains: a positively-charged N-terminal (N) domain, hydrophobic 56 core (H) domain and a more polar C-terminal (C) domain ending with an Ala-Xaa-Ala 57 consensus domain that specifies cleavage by leader peptidase following export [4]. Substrates 58 are transported across the membrane through a SecYEG channel in an unfolded state, 59 primarily driven by the ATPase action of SecA [1]. 60 61 Tat substrates are likewise synthesised with N-terminal targeting peptides, but translocation 62 by the Tat pathway occurs by a completely different mechanism. Substrates are transported in 63 a fully-folded form by a membrane-bound TatABC translocon that is driven by the proton 64 motive force rather than ATP hydrolysis, and the available evidence (reviewed in [4,5]) 65 indicates that the translocation mechanism is totally different, although the actual process 66 remains poorly defined. Interestingly, there is mounting evidence that the system is not only 67 able to transport folded proteins, but is also inherently capable of preferentially selecting 68 correctly-folded proteins for translocation (reviewed in [6]). A wide range of incorrectly 69 folded proteins, including both natural Tat substrates and heterologous proteins, have been 70 shown to be quantitatively rejected by the Tat system [e.g. [7][8][9][10][11]. 71 72 The abilities of the Tat system have attracted increasing interest from a biotechnological 73 standpoint. The system is able to export a wide range of heterologous proteins if a Tat-74 specific signal peptide is attached; examples include green fluorescent protein [12], alkaline 75 phosphatase [7] and a series of biopharmaceuticals such as human growth hormone (hGH) 76 and antibody single chain variable fragments [13]. Until recently it was unclear whether the 77 Tat system could handle the large export rates required for industrial applications, but we 78 recently showed that hGH could be exported at very high rates (several grams of protein per 79 litre in fed-batch fermentation) in E. coli 'TatExpress' strains that over-express the tatABC 80 operon to cope with high-level export [14]. 81

82
The TorA signal peptide has been widely used for studies of the Tat pathway and for most of 83 the published attempts to export heterologous protein substrates. This is partly due to the high 84 export efficiency relative to many other Tat specific signal peptides, and partly because the 85 signal peptide acts in a Tat-specific manner; in contrast, a number of other Tat signals can 86 direct export of heterologous proteins via both the Tat and Sec pathways [15]. This reflects 87 the broadly similar 3-domain structures of the two types of signal peptide. However, in our 88 studies we have repeatedly observed that some constructs bearing the TorA signal peptide are 89 subjected to degradation in the cytoplasm [e.g. 13]. In many cases, this leads to the production 90 of what appears to be mature-size protein in the cytoplasm, and we have speculated that the 91 TorA signal peptide may be highly susceptible to proteolysis, leading to the appearance of 92 near-mature-size passenger protein in the cytoplasm. We therefore concluded that it should be 93 possible to improve export rates through the use of a more effective and robust signal peptide. 94 In this report we describe the design and testing of a new Tat signal peptide, based on the 95 PhoD signal peptide from Bacillus subtilis that is able to direct the export of an antibody 96 single chain variable fragment (scFv). The native PhoD signal peptide directs the export of an 97 6 of the TorA signal peptide readily occurs in this compartment; the full precursor form is 126 barely detectable. A mixture of precursor and the near-mature-size scFv is found in the 127 membrane fraction (M). Essentially no export is observed in Δtat cells, confirming that export 128 is Tat-specific. 129 130 Similar tests were carried out using AmiC-scFv, MdoD-scFv, YcbK-scFv and PhoD-scFv, 131 and the data show that these constructs are all efficiently exported in both wild type and Δtat 132 cells. This strongly indicates that export is taking place by both the Tat and Sec pathways. 133 Export is more efficient in wild type cells, but given that these export levels are due to a 134 combination of Tat-and Sec-dependent export, it is clear that Tat-dependent export is by no 135 means particularly dominant over Sec-dependent export. Several of these signal peptides have 136 been shown to mediate Sec-dependent export in previous studies (e.g. [15]) but these data 137 provide a clear picture that the level of Sec-dependent export can be substantial. 138

139
Engineering of modified PhoD signal peptides to confer Tat-specificity  140   141 We considered it important to generate an efficient Tat-specific signal as an alternative to the 142 TorA signal, and modified the PhoD signal peptide because it provided high levels of export 143 in our initial tests using scFv as a passenger protein (see Fig. 1). The sequence of the native 144 PhoD signal peptide is shown in Fig. 2, and in an effort to render the signal peptide Tat-145 specific, we replaced the Ala residue at the -6 position (arrowed) with Ile, Lys and Asn, 146 generating signal peptides designated PhoDi, PhoDk and PhoDn, respectively (see sequences 147 in Fig. 2). Leader peptidase, the enzyme responsible for the removal of signal peptides after 7 invariably short-chain [17], and the native cleavage site in B. subtilis ends with VNA as 150 shown in Fig. 2 (shown by a second arrow). 151

152
We reasoned that the introduction of a Lys at the -6 position in the PhoDk signal peptide 153 should reduce or block export by the Sec pathway because a number of studies have shown 154 that the presence of basic residues in the C-terminal domain of a signal peptide prevents 155 transport by Sec [18,19]. The introduction of Ile and Asn (PhoDi and PhoDn) was also carried 156 out, although this was judged less likely to confer complete Tat-specificity. Fig. 2A shows 157 initial export assays using the modified PhoD signal peptides in WT cells. After induction of 158 synthesis, cells were fractionated into spheroplast and periplasm samples (S, P) and the results 159 show that all three modified signal peptides direct efficient export of the scFv into the 160 periplasm (P). showing that these signal peptides are Tat-specific. In fact, PhoDn-scFv and PhoDk-scFv are 172 not detected in the cytoplasm or membrane fractions of the Δtat cells either, even in this 173 immunoblot which is highly exposed to detect even low signal levels, which is surprising given that the precursor form was clearly detectable in the cytoplasm of the wild type cells as 175 shown in Fig. 2A. This point was further tested in Fig. 2C, which shows similar export assays 176 carried out in Δtat cells after a range of induction times (60, 90 and 120 min post-induction). 177 The same result is obtained: PhoDi-scFv is efficiently exported (presumably by Sec) whereas 178 the other two constructs are not detected at any time and thus appear to be rapidly degraded in 179 the cytoplasm. We conclude that the PhoDn-scFv and PhoDk-scFv constructs are particularly 180 unstable in the cytoplasm of Δtat cells, for reasons that are completely unclear. The primary aim of this study was to generate a signal peptide that is capable of directing 185 high-level, Tat-specific export in industrial applications, and further expression and export 186 tests were carried out in fed-batch fermentation systems that are much more representative of 187 those used in industrial processes. We compared the export of scFv bearing the TorA and 188 We have purified the scFv from periplasmic fractions in order to quantify the levels of scFv 242 protein that can be generated from this type of fermentation process, but this particular scFv 243 does not bind well to Protein L (commonly used for purification of scFvs) or to affinity resins 244 that are used to bind His-tagged proteins. We believe that the C-terminal His tag is buried 245 within the native structure and inaccessible. Most of the protein is thus lost during 246 purification, which makes quantification difficult. However, the scFv has the same His tag as 247 that present on hGH which which was exported at levels of 2-5 g/L culture in a previous study 248 [14], and we have therefore used semi-quantitiative immunoblotting with the scFv and hGH 249 run on the same gels and blots. This indicates that the levels of scFv exported with the PhoDn 250 signal peptide reach approximately 1g protein per litre of culture, while the TorA-scFv is 251 exported at significantly lower levels. Recent studies have shown that the Tat pathway can export human growth hormone at levels 256 of several g/L in fed-batch fermentation cultures of E. coli 'TatExpress' strains that 257 simultaneously over-express the Tat machinery [14]. Clearly, this platform is capable of high-258 level protein secretion to the periplasm, but previous work has already pointed to potential 259 limitations in the current platform. In particular, the TorA signal peptide is very widely used 260 because it delivers high-level, Tat-specific secretion, yet appears to have the disadvantage that 261 it is prone to proteolytic degradation in the cytoplasm. This leads to a situation in which 262 export efficiency is heavily influenced by the relative rates of export vs degradation. We 263 commonly observe the presence of 'pseudo-mature' size protein in the cytoplasm when 264 expressing contructs bearing a TorA signal peptide, which suggests that the precursor proteins 265 are being cleaved near the signal peptide-mature protein junction [13]. 266

267
We therefore set out to identify an alternative signal peptide, and additionally sought to test 268 whether another industrially relevant protein (in this case an scFv) could be exported at high 269 levels by Tat. Studies on the other signal peptides used in this study provide an interesting 270 comparison. While the TorA signal peptide is extremely Tat-specific, the E. coli AmiC and 271 MdoD signal peptides direct efficient Sec-dependent transport in Δtat cells, as does the PhoD 272 signal peptide from B. subtilis. In these cases, export efficiency is approximately double in 273 wild type cells and our conclusion from these results is that the signal peptides are effective as both Sec or Tat signal peptides. A previous study has shown that a wide range of Tat signal 275 peptides can support Sec-dependent export using using an assay in which growth of malE -276 defective cells was dependent on export of signal peptide-maltose binding protein fusion 277 proteins [15]. Here, we have clearly shown that the above signal peptides can direct high rates 278 of Tat-independent (presumably Sec-dependent) transport of this scFv. 279

280
The wild type B. subtilis PhoD signal peptide directs very efficient export of the scFv but is 281 clearly directing transport by both the Tat and Sec pathways. However, a single amino acid 282 substitution renders the signal peptide Tat-specific, and the engineered PhoDn and PhoDk 283 signal peptides thus represent attractive alternatives to the well-known TorA signal peptide. In 284 this study we have set out to systematically test the capabilities of the PhoDn signal peptide, 285 using shake-flask, 'standard' fed-batch fermentation and the automated ambr250 fed-batch 286 systems. In each case, the export of the PhoDn-scFv was compared directly with that of a 287 TorA-scFv fusion protein and the data show that the PhoDn signal peptide delivers a 288 significantly enhanced level export of the scFv. By the end of the fermentation runs, the scFv 289 is a highly abundant periplasmic protein, demonstrating that the new engineered signal 290 peptide can be used for high-level export of this heterologous protein by the Tat pathway. The study sheds further light on the abilities of the Tat system to direct export of high-value 295 proteins to the E. coli periplasm. The availability of a new, engineered, Tat-specific signal 296 peptide means that yields of periplasmic scFv are much higher and this tool is a potentially 297 valuable complement to the engineered 'TatExpress' cells that are capable of higher rates of 298 protein transport due to their elevated levels of TatABC.