Mechanism of the secretion of the lanthipeptide nisin

Lanthipeptides are ribosomally synthesized and post-translationally modified peptides containing dehydrated amino acids and (methyl-)lanthionine rings. One of the best-studied example is nisin, which is synthesized as a precursor peptide comprising of an N-terminal leader peptide and a C-terminal core peptide. Amongst others, the leader peptide is crucial for enzyme recognition and acts as a secretion signal for the ABC transporter NisT which secrets nisin in a proposed channeling mechanism. Here, we present an in vivo secretion analysis of this process in the presence and absence of the maturation machinery composed of the dehydratase NisB and the cyclase NisC. The data clearly demonstrated that the function of NisC, but the mere presence of NisB modulated the apparent secretion rates. Additional in vitro studies of detergent-solubilized NisT revealed how the activity of this ABC transporter is again influenced by the enzymes of the maturation machinery, but not the substrate.


Introduction
In our study, we performed an in vivo and in vitro characterization of NisT to shed light on 111 the secretion mechanism of pre-NisA. We determined the kinetic parameter for the pre-NisA 112 secretion by analyzing the supernatant of pre-NisA secreting L. lactis strains via RP-HPLC. The 113 resulting apparent secretion rate (NisA•NisT -1 •min -1 ) of NisT was compared with the rate of 114 the NisBTC system and demonstrated a large enhancement in the presence of the 115 modification machinery. The in vitro characterization of NisT is the first study revealing 116 insights into the specific activity of a LanT lanthipeptide transporter and its modification 117 enzymes as well as its substrate. In conclusion, we demonstrate an enhancement of the 118 secretion through the maturation enzymes and a pivotal and bridging function of the 119 dehydratase NisB in the interaction of NisC and NisT. 120

122
In vivo secretion assay of pre-NisA 123 To obtain further insights into the mechanism of lanthipeptide secretion, the pre-NisA 124 secretion level of the L. lactis strain NZ9000 was investigated in the presence and absence of 125 the modification machinery. Here, the well-known nisin secretion and maturation system 126  3). The secretion of pre-NisA was reduced by a factor of 3.9 compared to strain NZ9000BTC. 172 The lowest amount of secreted peptide was determined in the supernatant of strain 173 NZ9000TC with a V max value of 38 ± 8 nmol. Here, a higher amount of uNisA was detected in 174 the cytoplasmic fraction, whereas no peptide was observed in strain NZ9000T. 175 In all strains, NisB, NisC and NisT were detected in their corresponding fraction 176   Figure 3D). The slope of the linear regression corresponds to V S app. of 192 NisA•NisT -1 •min -1 , which is 100.3 ± 35.2 for the strain NZ9000BTC ( Figure 2C) In order to determine the in vitro activity of a lanthipeptide ABC transporters, we purified 205 NisT the exporter of the class I lanthipeptide nisin as a deca-histidine tagged protein variant 206 (10HNisT). 10HNisT was homologously expressed in L. lactis NZ9000 and purified to high 207 purity (≥95% as estimated by Coomassie brilliant blue stained SDS-PAGE, Figure 3A   For the ATPase activity assay, the detergent was exchanged to CYMAL5 (Anatrace) 215 and the ATPase rate was expressed as specific ATPase rate (nmol•min -1 •mg -1 ). The kinetic 216 parameters of 10HNisT in detergent solution were determined and resulted in V max , K m and 217 k cat values for the transporter without its substrate (basal ATPase activity). The 218 concentration of 10HNisT was kept constant (1 µM), whereas the concentration of ATP was 219 varied from 0 to 5 mM and the reaction was stopped after 30 min. The basal ATPase rate of 220 10HTNisT had a V max value of 79.9 ± 2.9 nmol•min -1 •mg -1 , a K m value of 0.37 ± 0.04 mM 221 resulting in a k cat value of 143.1 ± 7.3 min -1 ( Figure 3C). As a control the H-loop mutant of 222 10HNisT (10HNisT H551A ; HA-mutant) was also purified following the same protocol and used 223 in the ATPase activity assay ( Figure 3B). The ATPase rate of the HA-mutant was reduced by a 224 factor of 17 (V max value 4.7 ± 2.5 nmol•min -1 •mg -1 ). The K m value increased by a factor 1.62 225 (0.60 ± 0.94 mM), whereas the k cat value was 10.2 ± 5.4 min -1 (14-fold lower than WT 226 10HNisT) ( Figure 3C). 227 228

In vitro ATPase activity with pre-NisA variants 229
To investigate the effect of substrate on the ATPase rate of 10HNisT, we added different pre-230 NisA variants. First, the pre-NisA peptides in different modification states (uNisA, dNisA and 231 mNisA, respectively) were purified (Supplementary Figure 4). Additionally, the leader 232 peptide of NisA (NisA LP ) was used in the ATPase assay to evaluate whether the isolated LP is 233 sufficient for recognition by NisT. 234 For the activity assay the ATP concentration was kept constant at 5 mM, while the 235 substrate concentration was varied from 0 to 40 µM. 10HNisT was pre-incubated with the 236 peptides prior to the activity assay. The basal activity of 10HNisT was set to 100% and the 237 ATPase rate with substrates was expressed as normalized ATPase rate. The ATPase rate of 238 10HNisT was slightly increased for all peptides. Some values like 20 µM uNisA showed a 239 stronger stimulation of the transporter (140 %, Figure 4B), while other (e.g. 20 µM mNisA) 240 showed a lower stimulating effect (110%, Figure 4D). However, a concentration dependent  NisC (see in vivo secretion assay) and therefore the ATPase rate of NisT might also be 248 influenced by these interaction partner. To investigate the effect of NisB and NisC on the 249 ATPase rate of 10HNisT the ATPase activity assay was repeated under the same conditions 250 (ATP concentration constant, various concentration of interaction partner). We observed 251 that the ATPase rate of NisT was independent of the various concentrations of NisB or NisC. 252 Thus, only fixed molar ratio of 10HNisT to the interaction partner was used (NisT:NisB/NisC 253 1:2; in the case of NisT:NisBC 1:2:2) ( Figure 5A). The basal ATPase rate of NisT was 62.5 ± 9.4 254 nmol•min -1 •mg -1 and was not changed within the experimental error in the presence of NisB 255 or NisC (54.7 ± 4.5 nmol•min -1 •mg -1 , 59.3 ± 4.9 nmol•min -1 •mg -1 ). If both proteins were 256 used in the assay (NisBC), the ATPase rate of 10HNisT was reduced by a factor of 1.3 (49.3 ± 257 4.4 nmol•min -1 •mg -1 ), but the difference was not significant ( Figure 5A). 258 Next, the ATPase rate of NisT with NisBC was investigated in presence of NisA LP and 259 mNisA, at concentrations ranging from 0 to 40 µM. The ATPase rate with the substrate 260 NisA LP was slightly increased but not in a concentration dependent manner ( Figure 5B  The same set up was used in the presence of the substrate mNisA CCCCA , which lacks 281 the last lanthionine-ring (ring E) and stabilizes the maturation complex of NisBC (Reiners, 282

Abts et al. 2017). This substrate showed no additional effect on the interaction of NisB with 283
NisT. However, the interaction of NisT and NisC was affected and the amount of co-eluted 284 NisC was strongly reduced ( Figure 6B). After addition of NisB, the interaction of NisB and 285 NisC with NisT was restored ( Figure 6B). Noteworthy, the addition of mNisA instead of the 286 ring-mutant shows an identical result on complex formation. Unfortunately, the analysis of 287 all elution fractions with an antibody against the LP gave no signals for the substrates 288 mNisA/mNisA CCCCA . This might be due to low concentrations of the peptides in the elution 289

fractions. 290
In summary, this is the first time that beside the interaction of NisT and NisC, an 291 interaction of NisT with NisB was shown. Even the co-elution of NisB and NisC with NisT as a 292 complex in the presence of the substrates mNisA and the ring-mutant mNisA CCCCA was 293

observed. Although no enhancement of the interaction between NisT and NisBC in presence 294
of substrates mNisA/ mNisA CCCCA was detected, NisB plays a pivotal for complex stability. 295

296
The mechanism of lanthipeptide modifications was subject of many studies, but still only 297 little is known about the secretion process of class I lanthipeptide ABC transporters (LanT).  Although, these studies clearly demonstrated the dependence of pre-NisA secretion on the 325 modification enzymes, it did not include a determination of the underlying kinetic 326 parameters. To determine these kinetic parameters, we quantified the amount of secreted 327 peptide via HPLC from different time points of various NZ9000 strains. 328 The first two kinetic parameters were V max and K 0.5 , which were obtained by an 329 allosteric sigmoidal analysis ( Figure 2). Generally, our results are consistent with the 330 aforementioned studies, in which strain NZ9000BTC had the highest V max and the lowest K 0.5 331 value reflecting a high secretion efficiency. The strains, which secreted uNisA (NZ9000T and 332 NZ9000TC), show lower V max and higher K 0.5 values. Interestingly, we observed in our 333 secretion assay some aberrations with respect to dNisA secretion. The expression of a 334 catalytic-inactive NisC (H331A mutant) in the NisBTC system (strain NZ9000BTC H331A ) did not 335 restore the secretion level of dNisA to the WT level. This is in contrast to Lubelski et al.,336 where a recovery of the pre-NisA secretion to WT level was observed (Lubelski, Khusainov et 337 al. 2009). The secretion of dNisA by NZ9000BT has a higher V max level and a lower K 0.5 value. 338 However, one has to consider the time scale of the secretion assays, which might explain 339 this difference. In our assay, the early kinetics of secretion (also see (van den Berg van 340 Saparoea, Bakkes et al. 2008)) might pronounce the differences between the strains more 341 clearly as an end-point determination after an overnight secretion. The precise 342 determination of secretion efficiency revealed a descending order of the pre-NisA secreting 343 strains (NZ9000BTC > NZ9000BT > NZ9000BTC H331A > NZ9000T > NZ9000TC). The secretion 344 efficiency clearly shows, that mNisA is secreted at high rates by strain NZ9000BTC and every 345 aberration of the secretion system reduced the rate at least by a factor of 2.2 (see strain 346 NZ9000BT). 347 The third kinetic parameter was the apparent secretion rate V S app. (NisA•NisT -1 •min -348 1 ), which we determined for the strains NZ9000BTC and NZ9000NisT, in which the secretion 349 efficiency was the highest for NisBTC system (Figure 2). These values are in general an 350 approximation to obtain a kinetic parameter, which can be related to other secretion 351 system, e.g. the SecA translocon. In comparison to the nisin modification and secretion 352 system (V S app : 191 ± 67 aa•s -1 ) the SecA translocon processes a secretion rate of 152-228 353 aa•s -1 per translocon (Robson, Gold et al. 2009 As ATP hydrolysis is clearly important for ABC transporter mediated substrate 360 translocation, we determined the in vitro activity of NisT in terms of ATPase rate without and 361 with substrate. Here, the basal ATPase activity had a value of 79.9 ± 2.9 nmol•min -1 •mg -1 362 with a K m value of 0.37 ± 0.04 mM, which is in the range of other ABC transporter 363 In 1996 a multimeric enzyme complex of NisBTC was proposed, but the isolation of 381 such a complex was not successful (Siegers, Heinzmann et al. 1996). Therefore, we choose to 382 study the specific interaction of NisT with NisB and NisC via a pull-down assay. Such a pull-383 down assay was performed with His-tagged pre-NisA, where NisB and NisC were co-eluted 384 from cytoplasmic fraction (Khusainov, Heils et al. 2011, Khusainov, Moll et al. 2013). In our 385 study, we expanded this set up and used purified NisT, NisB, NisC and pre-NisA ( Figure 6). 386 We observed a specific interaction of NisT with NisC, which is in line with previously 387

Cloning of nisBTC and nisBTC variants 504
The plasmid pIL-SVnisBTC was generated from pIL-SV and pIL3BTC . 505 The plasmid pIL3BTC was digested with the restriction enzymes NotI (NEB) and BstXI (NEB) 506 to receive a fragment BTC containing the genes nisB, nisT and nisC. Next 17,000xg for 20 min at 8°C. Supernatants were kept on ice before the RP-HPLC analysis. 538 Furthermore, 2 ml or 10 ml of the supernatant were precipitated by 1/10 volume (10%) TCA. 539 TCA samples were incubated at 8°C overnight. The TCA-precipitated peptide was centrifuged 540 at 17,000xg for 20 min at 8°C and consecutively washed three-times with ice-cold acetone. 541 The pellets were vacuum-dried and resuspended in 60 µl per OD 600 1 of 1x SDS-PAGE loading 542 dye containing 5 mM β-mercaptoethanol (β-ME). These resuspended TCA-pellets were 543 analyzed by Tricine-SDS-PAGE and Western blot. 544

545
In vivo secretion assay: Analysis of cell pellets 546 The resuspened cell pellets were thawed on ice and 1/3 (w/v) glass beads (0.3 mm diameter) 547 were added. Cells were disrupted on a vortex-shaker (Disrutor Genie, Scientific Industries). A 548 cycle of 2 min disruption and 1 min incubation on ice was repeated five times. A low spin 549 step at 17,000xg for 30 min at 8°C and subsequently a high spin step at 100,000xg for 120 550 min at 8°C was performed. The supernatant of the latter centrifugation step represents the 551 cytoplasmic fraction and the pellet corresponds to the membrane fraction. The SDS-PAGE 552 samples of cytoplasmic and membrane fractions were prepared by adding 4x SDS-PAGE 553 loading dye containing 5 mM β-ME and used for SDS-PAGE as well as Western blot analysis. The amount of secreted pre-NisA of the different L. lactis NZ9000 strains were plotted 570 against time and fitted using an allosteric sigmoidal fit (1). Note that y is the amount 571 secreted peptide (nmol), V max the maximal secreted amount, x is time (min), K 0.5 the time 572 point at which 50% of V max is present and h is the Hill slope indicating cooperatively. The 573 analysis was performed using Prism 7.0c (GraphPad). 574

575
(1) y = V max X ℎ K 0.5 ℎ +X ℎ 576 577 The apparent secretion rate (V s app ) was determined by plotting the amount of NisA and NisT 578 against time (min). The values were fitted using a linear regression (2). Note that y is the 579 amount of NisA molecules per NisT molecules (NisA•NisT -1 ), m is the slope V s app (NisA•NisT -580 1 •min -1 ), x is the time (min), and b the y -axis interception. The analysis was performed using 581 Prism 7.0c (GraphPad). 582

583
(2) y = mx + b 584 585 Expression and purification of NisT 586 L. lactis NZ9000 strain was transformed with pNZ-SV10HnisT and placed on SMGG17 agar 587 plates containing 5 µg/ml erythromycin. A GM17 (Erm) overnight culture was inoculated 588 with one colony and incubated at 30°C. A GM17 (Erm) main culture was inoculated to an 589 OD 600 of 0.1 with the overnight culture. After 3 h incubation, expression was induced by 590 adding 10 ng/ml nisin (powder from Sigma-Aldrich dissolved in 50 mM lactic acid) and 591 further grown for additional 3 h. Cells were harvested by centrifugation at 4000xg for 20 min 592 at 8°C and resuspened in R-buffer (50 mM Na-Phosphate buffer, pH 8, 100 mM KCl, 20% 593 glycerol) to an OD 600 of 200. To the resuspened cells 10 mg/ml lysozyme was added and 594 incubated at 30°C for 30 min. Prior to cell disruption, cells were incubated on ice for 15min. 595 The cell suspension was passed through a homogenizer (M-110P, Microfluidics System) at 596 1.5 kbar at least four times. The homogenized cell suspension was centrifuged at 12,000xg 597 for 30 min at 8°C. Subsequently, the supernatant was centrifuged at 100,000xg for 120 min 598 at 4°C to collect the membrane fraction. Membranes were resuspended with R-buffer 599 containing 10 mM imidazole and 0.5 mM AEBSF. The total membrane protein concentration 600 was measured by BCA assay (Thermo Fischer Scientific) and the concentration was adjusted 601 to 5-7.5 mg/ml. Membranes were solubilized with 1% (w/v) of the lipid-like detergents FC-16 602 In vitro ATPase activity assay 686 The ATPase activity of NisT was determined with the malachite green assay as described 687 previously with experimental alterations (Infed, Hanekop et al. 2011). In this assay the 688 release of inorganic orthophosphate after ATP hydrolysis was colorimetric quantified based 689 on a Na 2 HPO 4 standard curve. 690 All reactions were performed at 30°C in a total volume of 30 μl in activity assay buffer 691 containing 0.4% CYMAL5 and 10 mM MgCl 2 . 692 In each reaction ~2 μg of detergent-solubilized and purified NisT was used and the reaction 693 was started by adding ATP (0-5 mM). The background of the reaction was a sample without 694 For the reactions with substrates (pre-NisA variants) or interaction partners (NisB and NisC) 708 NisT was pre-incubated at 30°C for 10 min before ATP was added to start the reaction. All 709 reactions were performed at 30°C in a total volume of 30 μl in activity assay buffer 710 containing 0.4% CYMAL5, 400 mM glutamate and 10 mM MgCl 2 . In each reaction ~2 μg of 711 detergent-solubilized and purified NisT was used and the reaction was started by adding 5 712 mM ATP and stopped after 15 min following the procedure described above. In this reaction 713 the concentration of the different substrates (0-40 µM) and /or interaction partner was 714 varied and the ATPase activity was normalized to the specific ATPase activity of NisT without 715 substrate/interaction partner. In these cases, the background was subtracted prior to 716 normalization. 717 718

In vitro pull-down assay 719
The immobilization of 10HNisT to Ni-NTA magnetic beads (Quiagen) was performed as 720 described in the manufacture's manual. In brief, ~ 15 µg 10HNisT was incubated with Ni-NTA 721 magnetic beads for 30 min at 30°C. Excess of protein was removed by three washing steps 722 with activity assay buffer containing 0.4% CYMAL5 and 400 mM glutamate. Interaction 723 partners NisC and NisB were incubated in 1:1 molar ratio (but > 10x molar excess to NisT) 724 separately with or without mNisA/ mNisA CCCCA (20x molar excess to NisT) in activity assay 725 buffer containing 0.4% CYMAL5, 400 mM glutamate and 5mM MgATP for 15 min on ice. 726 Next, interaction partner were added to 10HNisT immobilized to Ni-NTA magnetic beads and 727 incubated for 1 h at 30°C. Positive control (only 10HNisT) and negative control (NisB, NisC) 728 samples were prepared by incubating the proteins with Ni-NTA magnetic beads separately. 729 After binding the Ni-NTA magnetic beads were washed six times with activity assay buffer. 730 Finally, 10HNisT was eluted by adding activity assay buffer containing 50 mM EDTA. The SDS-731 PAGE samples of pull-down assay fractions were prepared by adding 4x SDS-PAGE loading 732 dye containing 5 mM β-ME and used for Western blot analysis. 733 734

SDS-PAGE and immunoblotting for protein/ peptide analysis 735
In general the sodium dodecyl sulfate−polyacrylamide gel electrophoresis (SDS-PAGE) 736 experiments were performed using standard procedures (Laemmli 1970). In the SDS-PAGE 737 gels the acrylamide portion was 10% to have a separation range from 30 to 120 kDa for the 738 proteins NisC (~48 kDa), NisT (~69 kDa) and NisB (~117 kDa). 739 All peptides (e.g. pre-NisA variants) from secretion experiments or from cIEX purification 740 were analyzed by Tricine-SDS−PAGE (Schagger 2006). For Tricine-SDS-PAGE gels (12%) a 741 Mini-Protean system (Bio-Rad) was used. Tricine-SDS−PAGE and SDS−PAGE gels were stained 742 with colloidal coomassie (cc) (Dyballa and Metzger 2009). 743 All immunoblotting experiments were conducted following standard procedures. For the 744 quantification of NisT in the membrane fractions (in vivo secretion assay) various amounts of 745 a T NBD standard (stock solution 12.5 µg/ml) was added to create a calibration curve. The 746 band intensities on the Western blots were processed and determined by ImageJ (Schneider,