Nanopore sensing reveals a preferential pathway for the co-translocational unfolding of a conjugative relaxase–DNA complex

Abstract Bacterial conjugation is the main mechanism for the dissemination of antibiotic resistance genes. A single DNA strand of the conjugative plasmid is transferred across bacterial membranes covalently bound to a large multi-domain protein, named relaxase, which must be unfolded to traverse the secretion channel. Two tyrosine residues of the relaxase (Y18 and Y26 in relaxase TrwC) play an important role in the processing of conjugative DNA. We have used nanopore technology to uncover the unfolding states that take place during translocation of the relaxase–DNA complex. We observed that the relaxase unfolding pathway depends on the tyrosine residue involved in conjugative DNA binding. Transfer of the nucleoprotein complex is faster when DNA is bound to residue Y18. This is the first time in which a protein–DNA complex that is naturally translocated through bacterial membranes has been analyzed by nanopore sensing, opening new horizons to apply this technology to study protein secretion.


INTRODUCTION
Antibiotic resistance has become one of the most challenging problems in health care ( 1 , 2 ).The main mechanism for the spread of antibiotic resistance genes is bacterial conjugation ( 3 ).In this process, a DNA strand of the conjugati v e plasmid is transferred bound to a pilot protein, called relaxase.Relaxases are ssDNA endonucleases that recognize a specific sequence on the plasmid (the origin of transfer) and after a cleavage reaction on the nic site, remain covalently bound to the 5´-end of the DNA.The resultant nucleoprotein complex is transferred through a Type IV Secretion System (T4SS), a multiprotein complex that spans the inner and outer membranes of the donor bacteria (4)(5)(6).It is important to note that conjugati v e r elaxases ar e large multidomain proteins that, in some cases, like the relaxase of conjugati v e plasmid F, consist of > 1700 amino acid r esidues ( 7 ).Ther efor e, characterization of the mechanism of relaxase transport during conjugation is not only worthwhile for its biological relevance but also for the biophysical challenge of transferring these colossal pr oteins acr oss biological membranes.
R388 is one of the most well-studied and characterized conjugati v e plasmids.Its relaxase, TrwC, is composed of two domains.The N-terminal endonuclease domain (TrwC R , residues 1-293) catalyzes the cleavage and transfer of the conjugati v e ssDNA, whereas the C-terminal helicase domain (TrwC H , residues 296-966) is responsible for a 5´-3´DNA helicase activity ( 8 ).TrwC R has two catalytic tyrosines: Y18 and Y26.In vitr o , both ca talytic tyrosines are capable of bringing about the DNA cleavage reaction using oligonucleotide substrates containing the specific nic sequence, so the 5´-end of the cleaved DNA becomes covalently bound to the protein via Y18 or Y26 residues.Howe v er, in vivo , each tyrosine seems to play a distinct role.A translocation model was proposed ( 9 ), in which Y18 specificall y catal yzes the initial cleavage reaction and remains covalently bound to the DNA strand that is transferred to the recipient cell.Once in the recipient cell, the protein would perform a new str and-tr ansfer reaction to re-join both ends through Y26, leading to DNA r ecir cularization ( 9 , 10 ).Gi v en the large size of the relaxase and the internal diameter of a T4S channel --< 30 Å in diameter (11)(12)(13) --protein translocation must necessarily occur in an unfolded state.In fact, recent studies have shown that translocation through a conjugati v e T4SS r equir es the unfolding of the relaxase protein ( 14 ).Howe v er, whereas se v eral studies support the refolding of the transported protein in the recipient cell ( 10 , 15 ), the mechanism whereby the nucleoprotein complex is unfolded and secreted across bacterial membranes remains elusi v e.
The measurement of single-molecules translocating through membrane pores is the basis of nanopore technolo gy, w hich has experienced a major pro gress in the last few years as a platform for third generation DNA sequencing (Oxford Nanopore Technologies).In addition to DNA sequencing, ␣-haemolysin ( ␣-HL) nanopores are an excellent tool to study co-translocational protein unfolding at the single-molecule le v el.A heptameric ␣-HL protein forms a channel in a membrane, which allows the flow of ions in response to an imposed electrical field.A molecule translocating the nanopore produces a modulation in the current, which can be readily measured with sub-pA and sub-ms resolution.The ␣-HL nanopore has an upper entrance ∼24 Å wide and an internal constriction ∼14 Å wide ( 16 ).Interestingly, the electron microscop y structur e of the T4SS encoded by the R388 / pKM101 plasmids re v ealed an internal diameter of 10-20 Å ( 11 , 13 , 17 ), comparable to the internal diameter of the ␣-HL por e. Mor eover, the for ces applied in nanopore sensing are similar to those determined for the deli v ery of bacterial effectors (2.7-27 pN) ( 18 , 19 ), making ␣-HL nanopores a unique tool to study protein and DNA secretion.
Here, we have used nanopore technology to study the unfolding and transfer of a short ssDNA fragment of plasmid R388 covalently bound to its cogniti v e relaxase protein (TrwC R ) in an attempt to answer fundamental questions about the conjugati v e process, such as what are the for ces the r elaxase-DNA complex sustains as it unfolds and crosses the channel, or what is the differential behavior of the two specific r elaxase-DNA complex es.We have analyzed the co-translocational unfolding of TrwC R -DNA complexes with DNA either bound to Y18 (TrwC R (Y18)-DNA complex) or to Y26 (TrwC R (Y26)-DNA complex).Thus, w hen the ssDN A-protein comple x is dri v en to the pore in an electric field, the ionic current signal re v eals the threading of the DNA, the unfolding of the protein through distinct intermediate steps, and the translocation of the unfolded polypeptide.We have observed that both r elaxase-DNA complex es undergo co-translocational unfolding, showing a distinct cyclic pattern of se v en current le v els separated by se v en irre v ersib le steps.The duration and intensity of each le v el can be associated with the barriers that the protein encounters during its passage through the nanopor e. Inter estingly , the T rwC R (Y18)-DNA comple x trav ersed the pore 6 times faster than TrwC R (Y26)-DNA comple x.Ov erall, our r esults r e v eal a particularly favor able tr ansloca tion pa thway tha t depends on a tyrosine residue that is covalently bound to DNA and open up a ne w av enue in the study of protein secr etion by nanopor e sensing.

Protein expression and purification
TrwC R (wt) and variants TrwC R (Y18F) and TrwC R (Y26F) wer e expr essed in Esc heric hia coli C41(DE3) strain ( 20 ).Cells were lysed in buffer 100 mM Tris-HCl pH 7.5, 300 mM NaCl, 0.001% PMSF.Lysates were collected by centrifuga tion a t 40 000 rpm for 30 min a t 4 • C , diluted in buffer 100 mM Tris-HCl pH 7.5, 0.001% PMSF and applied to a P11 phosphocellulose column (W ha tman).Protein was eluted in a linear gradient of NaCl and TrwC Renriched fractions were pooled and applied to a HiTrap SP-HP column (GE Healthcare).After a subsequent elution in a linear gradient of NaCl, fractions were analyzed by SDS-PAGE.Protein concentration was estimated in a Nan-oDr op 2000c spectr ophotometer (Thermo Scientific) by UV absorbance at 280 nm using an extinction coefficient of 31.5 calculated in ProtParam tool from ExPASy bioinformatics r esour ce portal ( https://w e b.expasy.org/protparam/).Samples were mixed with 5% (w / v) glycerol and stored at -80 • C.

Cleav age r eactions and f ormation of T rwC R -DNA conjugates
Cleavage reactions were carried out by incubating TrwC R (wt) or TrwC R variants (20 M) with a 42-mer oligonucleotide (30 M) containing the R388 ni c site ( 12 + 30: 5'-TGC GTA TTG TCT / ATA GCC CAG ATT T AA GGA T AC CAA CCC GGC-3').The mixture was incuba ted a t 37 • C for 30 min in buffer 100 mM Tris pH 7.5, 1 mM MgCl 2 .After the cleavage r eaction, TrwC R r emains covalently bound to the 5´-end of the resultant 30-mer oligonucleotide.Reactions were stopped by adding 10 mM EDTA and then loaded onto a HiTrap Q HP column ( GE Healthcare ) to separate free TrwC R protein and DNA, from TrwC R -DNA conjugates.Fractions containing the complex TrwC R -DNA were detected by SDS-PAGE and quantified by Bradford protein assay ( 21 ).Samples were stored at -20 • C with 5% (w / v) glycerol.

␣-HL pores
Nanopore technology was used to analyze the cotranslocational unfolding of TrwC R -DNA complexes through ␣-HL nanopores.Wild-type monomers were expressed in an E. coli in vitro tr anscription / tr anslation system and oligomerized into heptameric pores by incubation with rabbit red blood cell membranes ( 22 ).Then, the heptameric pores were purified by SDS-PAGE, extracted from the gel (0.2 l, ∼1 ng / l) and added to the cis compartment of a bilayer apparatus as described in ( 19 ).

Electrical measurements and data analysis
A bilayer of 1.2-diphytanoylsn -glycero-3-phosphocholine (Avanti Polar Lipids, Alabaster, AL, USA) was made across an aperture with a diameter of 100 m in a Teflon film (Goodfellow), which separated two compartments of 0.6 ml each, cis and trans as previously described ( 23 ).Planar lipid bilayer r ecordings wer e carried out at 22 ± 1.5 • C. The buffer was 10 mM HEPES, 2 M KCl, pH 7.2.After the insertion of a single ␣-HL pore from the cis compartment, the solution was replaced with fresh buffer by manual pipetting in order to remove excess of ␣-HL.Ionic currents were measured using Ag / AgCl electrodes connected to a patchclamp amplifier (Axopa tch 200B , Axon Instruments).The signal was filtered at 5 kHz (low-pass Bessel filter) and data were collected at 20 kHz with a digitizer (Digidata 1440A, Axon Instruments).In GndHCl experiments, the denaturing agent was gradually added into the trans compartment.Once a functional ␣-HL nanopore was acti v ely translocating TrwC R (Y18)-DNA complexes at + 120 mV, different volumes (30, 60 or 90 l) of GndHCl (6 M) were gently mixed with the buffer of the trans compartment (0.6 ml).
Raw data were first analyzed with Clampex 10.5 software (Molecular Devices) to obtain the current and dwell time of the different le v els.Data organization was carried out with Microsoft Excel (Microsoft) and data analysis and plotting were performed with MATLAB R2018b (Math-Works).Dwell time distributions r epr esenting the probability density function (pdf) were fitted to a mono-or doublee xponential function, accor ding to the following formula: monoexponential pdf ( x ) = Ak exp( xk exp( x )), where A is the amplitude, k the rate, and x the natural logarithm of the dwell time.Rate constants ( k ) values, expressed as s -1 , are provided with 95% confidence interval in brackets ( n ≈ 300).Data were collected in independent experiments ( n > 3).The residual current of a le v el is defined as the ionic current of the le v el di vided by the ionic curr ent of the open por e. Error bars gi v en in the text and tables r epr esent the standard deviations (S.D.) between independent experiments ( n = 3).

Isolation of TrwC R -DNA conjugative complexes
The TrwC R domain catalyzes conjugati v e DNA cleavage, which occurs via a nucleophilic attack by the hydroxyl group of a tyrosine residue on the 5´-side of the DNA phospha te.This transesterifica tion r eaction r esults in a covalent linkage between protein and DNA ( 9 , 24 ) (Figure 1 ).
Her e, we generated TrwC R -ssDNA complex es by incubating TrwC R (20 M) with a 42-mer oligonucleotide (30 M) containing the nic sequence of plasmid R388.After the cleavage r eaction, TrwC R r emained covalently bound to the 5´-end of the resultant 30-mer oligonucleotide (Figur e 1 ).The r eaction r equir es Mg 2+ and is isoenergetic -the energy of the cleaved phosphodiester bond is stored in the form of a phosphotyrosine linkage-, which allows a subsequent ligation reaction ( 24 ).Therefore, TrwC is able to carry out cleavage and joining reactions and the resulting products r epr esent an equilibrium between cleaved and ligated DN A. Consequentl y , formation of T rwC R -DNA complexes is not an efficient process in vitro , which has so far pre v ented the analysis of such protein-DNA complexes ( 9 ).In this work, we succeeded in stabilizing the protein-DNA complexes by the addition of EDTA.Since DNA cleavage is a magnesium-dependent reaction ( 25 ), Mg-chelation after a 30 min reaction allowed us to enhance the formation of covalent protein-DN A adducts.TrwC R -DN A complexes were then isolated from free protein and DNA by anion exchange chromato gra phy (Figure 2 ).Fractions containing the protein-DNA complex were collected and analyzed by nanopore technology.
TrwC R has two catalytic tyrosines, Y18 and Y26, which are able to form covalent protein-DNA adducts.Substitution of tyrosine with phenylalanine abolishes their catalytic activity ( 9 ).Ther efor e, to obtain monodisperse protein-DNA complexes, T rwC R (Y26F) or T rwC R (Y18F) variants were used to get TrwC R -DNA complexes with the DNA either bound to residues Y18 or Y26, respecti v ely (TrwC R (Y18)-DNA or TrwC R (Y26)-DNA complexes).

T r anslocation of a TrwC R (Y18)-DN A comple x through an ␣-HL pore
According to the proposed model for bacterial conjugation, a TrwC (Y18)-DNA complex would mimic the nucleoprotein substrate that is produced after DNA processing in vivo .Figure 3 A shows a scaled representation of a TrwC R (Y18)-DNA complex and an ␣-HL nanopore used in this work (1zm5.pdband 7ahl.pdb,respecti v ely), highlighting the challenge of transferring a protein of 293 residues through the pore.
Membr ane tr ansloca tion of this conjuga ti v e comple x was examined as follows.First, a single ␣-HL was inserted in a lipid membrane that separated two compartments ( ciswhich was at ground-and trans ), filled with electrolyte solution (2 M KCl, 10 mM HEPES, pH 7.2).When an electrical potential of +120 mV was applied, we obtained a constant ionic current of ∼230 pA, which corresponds to the ionic current of the open pore ( I O ).In contrast, when we added a TrwC R (Y18)-DNA complex sample (1 M), we observed a multitude of events that partially blocked the por e. Mor e than 90% of the e v ents showed a pattern composed of se v en different ionic current le v els (Figure 3 B).The protein alone did not produce any significant signal and, ther efor e, we attribute the signal to the translocation of the complex through the nanopore by the electrophoretic capture of the ssDNA.
Data were collected and analyzed at fiv e different voltage values, from +100 mV to +140 mV.The capture frequency depended on the applied voltage, ranging from 12 e v ents / min at +100 mV to 55 e v ents / min at +140 mV.

Co-tr anslocational unf olding of a T rwC R (Y18)-DNA complex is a seven-step process
Translocation of a TrwC R (Y18)-DNA complex through an ␣-HL pore showed a pattern consisting of se v en ionic current le v els.The characteristic signal begins with a partial blockade of the pore, step O → 1, that leads to the first observed level, with a residual ionic current of I RES(%),120 mV = 50.5 ± 0.1% (mean ± standard deviation (S.D.), n = 60) (Figure 3 C).I RES (%) = ( I RES / I O ) × 100, where I RES is the current flowing during a blockade and I O the current through the unblocked pore.The distribution of dwell times for le v el 1 (step 1 → 2) was characterized by a r ate par ameter k 1 → 2 120 mV = 379 [315-443] s −1 ; 95% confidence interval (C.I.) in brackets; n = 300 (Table 1 ) (Supplementary Figure S1).This step was not observed in ∼11% of the e v ents, arguab ly because it proceeded to the next step faster than the time resolution.Since the estimated rate of this step was 380 s −1 and the time resolution was about 0.2 ms, at least ∼ 8% of the e v ents should be expected to fall below the detection limit in an exponential distribution, not being observed.The dwell time, the r esidual curr ent, and the lack of voltage effect on the dwell time, are similar to previous studies that assign this step to the initial contact of the oligonucleotide with the cis entrance of the ␣-HL pore, without threading yet the transmembrane region ( 26 ).
Next, the signal proceeded through a series of ionic current le v els (le v els 3, 4, 5, 6 and 7), which we attribute to the sequential unfolding and translocation of the protein.Le v el 3 is of lower conductance ( I RES (%), 120 mV = 9.5 ± 0.2) (Figure 3 C).It likely corresponds to the threading of Y18 and surrounding residues into the ␣-HL pore, trying to move to the trans compartment, and trapped by the still-folded TrwC protein in the cis side.It is important to note that the leader DNA strand is not bound to residue 1 in the protein, but to residue 18, which means tha t, a t pulling, residues 1-17 and residues 19-onwards will enter the nanopore all at once.In other w ords, tw o peptide segments must be threaded inside the pore at the same time, which must constitute a serious barrier to overcome.Indeed, a longer time is r equir ed to jump to the ne xt le v el (the rate value of step 3 → 4 is 12 [10][11][12][13][14][15] s −1 ; 95% C.I., n = 300) (Table 1 ).Interestingly, in addition to le v el 2, le v el 3 also showed a slight voltage dependence ( k 3 → 4 100 mV = 11 [10-13] s −1 versus k 3 → 4 140 mV = 16 [14-18] s −1 ; 95% C.I., n = 300, PCC r = 0.95) (Supplementary Table S1; Supplementary Figure S2).The fact that the net charge of the first protein seg- ment is negati v e, as shown in a window analysis of TrwC R sequence (Supplementary Figure S3), might help to thread and translocate this first peptide segment through the ␣-HL pore.
TrwC R is too large to enter the nanopore in a folded state and, ther efor e, subsequent steps (le v el 4 onwar ds) must corr espond to progr essi v e protein unfolding stages.None of them showed a significant voltage dependence (Supplementary Table S1).At level 7, after completion of unfolding, the remainder of the protein traverses the pore, which opens again (step 7 → O).Although it is not possible to assign each partial unfolding step to a particular protein segment, the se v en different steps described her e ar e observed in all the analyzed e v ents (Supplementary Figure S4) and, therefore, we can conclude that TrwC R unfolding always proceeds through the same route.

Pulling fr om tyr osine 26 leads to a less efficient cotr anslocational unf olding
Taking advantage of the catalytic activity of tyrosine 26, which also cleaves oligonucleotides containing the spe-cific nic sequence, monodisperse TrwC R (Y26)-DNA covalent adducts were obtained by using a TrwC R (Y18F) mutant.The protein, covalently attached to the resultant 30mer oligonucleotide, was pulled through the ␣-HL pore as described above (Figure 4 A) and data were analyzed at fiv e different voltage values, from +100 mV to +140 mV (Supplementary Table S2).Interestingly, all analyzed e v ents also showed se v en steps (Figure 4 B), as previously described for a TrwC R (Y18)-DNA complex.Steps 4 → 5 onwards proceeded at a similar rate (Table 1 , Supplementary Figures S1 and S5) and with similar I RES (%) values (Table 2 ).However, in steps 2 → 3 and 3 → 4, rates were significantly different from those observed for a TrwC R (Y18)-DNA comple x (Tab le 1 ).In particular, when DNA was bound to Y26, step 2 → 3 was 28 times faster (1051 [985-1116] s −1 versus 37 [35][36][37][38][39] s −1 , respecti v ely, at +120 mV) (95% C.I., n = 300).On the contrary, step 3 → 4 was 70 times slower in this construct.This step draws particular attention, since it is rate-limiting for the translocation process ( k 3 → 4,120 mV = 0.17 [0.15-0.2]s −1 ; 95% C.I., n = 300).As mentioned in the previous section, this step must correspond to the entrance of the N-  Values (s −1 ) were deri v ed from e xponential fits to dwell-time histo grams, in w hich the da ta from a t least thr ee independent experiments wer e compiled; 95% confidence interval is provided in brackets ( n ≈ 300).
Our results show that the co-translocational unfolding kinetics depend on the residue to which DNA is bound.T rwC R (Y18)-DNA and T rwC R (Y26)-DNA com-plexes proceed initially through two different routes, with distinct energy barriers to overcome, showing marked differences in their translocation kinetics.After le v el 3, both pr oceed thr ough the same r oute, which is not rate limiting the translocation process.The e v ent, as a whole, was 6 times slower in the case of the TrwC R (Y26)-DNA complex in comparison to a TrwC R (Y18)-DNA complex (median of 6.3 s versus 1 s, respectively, n = 300).The explanation for such an increase is the kinetic trap observed in le v el 3 in TrwC R (Y26)-DN A complexes, w hich re v eals an intermedia te tha t r equir es mor e than fiv e seconds to be unfolded.In vivo , TrwC is thought to be secreted bound to DNA through residue Y18 ( 9 ) and, as observed here, this combination leads to a faster, and thus more efficient secretion through the nanopore.

T rwC R (Y18)-DNA and T rwC R (Y26)-DN A comple x es can be distinguished according to their translocation kinetics
TrwC R wild type protein (TrwC R (wt)) was incubated with the same 42-mer oligonucleotide used for previous constructs, resulting in a mixture of TrwC R (Y18)-DNA and TrwC R (Y26)-DNA complexes ( 9 ).The sample was added to the cis compartment of an ␣-HL pore, and a transmembrane potential of +120 mV was applied.All the observed co-translocational unfolding e v ents showed se v en current le v els, but two dif ferent popula tions of molecules could be distinguished, showing different rate values in steps 2 → 3 and 3 → 4, which are the two steps shown to be different in the two types of protein-DNA complexes (Figure 5 ).Moreover, the dwell-time distribution in these two le v els was found to be bimodal, where each peak could be assigned to a particular construct (Figure 5 E, F, Table 1 ).The dwell-time distribution for le v els 1, 4, 5, 6 and 7 did not show significant differences in TrwC R (Y18)-DNA, TrwC R (Y26)-DNA or TrwC R (wt)-DNA constructs (Supplementary Figures S1, S5 and S6).I RES (%) values were also very similar for levels 1, 4, 5, 6 and 7, whereas the values at le v els 2 and 3 were a combination of the I RES (%) values obtained for each type of protein-DNA complex (Supplementary Figure S7), confirming the existence of two populations of molecules.In a previous analysis of TrwC R -DNA complexes based on suicide oligonucleotides, the authors concluded that an efficient Y26 DNA cleavage only occurred after a previous Y18 DNA break ( 9 ).This means that TrwC R protein molecules bound to DNA via residue Y26 would necessarily have another DNA molecule bound to Y18 (two 30mer oligonucleotides bound to a single protein molecule).Howe v er, in the work presented here, no such double complex es wer e observ ed, since all the analyzed e v ents showed the characteristic se v en-step pattern observ ed either for a TrwC R (Y18)-DNA complex or a TrwC R (Y26)-DNA com-plex.With this analysis we can conclude that DNA cleavage through Y18 is more efficient, since TrwC R (Y18)-DNA complex es wer e mor e abundant (Figur e 6 ).A k -means cluster analysis allowed us to assign each single e v ent of the TrwC R (wt)-DNA mixture to a particular complex, either TrwC R (Y18)-DNA or TrwC R (Y26)-DNA.The results showed that 78% of the molecules corresponded to a TrwC R (Y18)-DN A complex, w hereas onl y 22% had the DNA bound to residue Y26.Therefore, nanopore technology allowed us to go a step further and quantify the number of molecules of each complex, which reflects the potential of this technology to discriminate distributions within a population of molecules.

Effect of a denaturing agent on the kinetics of the translocation process
In bacterial conjugation, once TrwC protein is translocated to the recipient cell in an unfolded state, the protein must ra pidl y refold to be functionally acti v e again ( 10 , 27 ).We reasoned that, in T4SS, refolding of the relaxase protein in the recipient cell might act as a pulling force to complete translocation, as it occurs in T3SS ( 28 , 29 ).Ther efor e, if the unfolded translocated TrwC R substrate was not able to fold back into an acti v e conformation in the trans compartment of an ␣-HL nanopore system, an effect on the translocational pattern should be observed.
In order to carry out this experiment, once a functional ␣-HL nanopore was acti v ely translocating TrwC R (Y18)-DNA complexes, the denaturing agent guanidinium chloride (GdnHCl) was added gradually into the trans compartment, as described in Materials and Methods, preventing the correct refolding of TrwC at this compartment.We determined whether this had an effect on the char acteristic co-tr ansloca tional unfolding pa ttern of se v en steps.After the addition of 0.9 M GdnHCl, all the translocated TrwC R (Y18)-DNA complexes showed a different cotransloca tional unfolding pa ttern (Figure 7 ).W hereas levels 1 to 4 of the current trace remained unchanged (Supplementary Figure S8), all the e v ents lacked the characteristic ionic current le v els 5, 6 and 7. Instead, from le v el four onwards, differ ent curr ent b lockages with variab le dwell times and I RES (%) values, and without a reproducible pattern, were observed until the complex was finally translocated (Figure 7 ).In addition, the median duration of the translocation process was found to be ∼2 times longer in presence of GdnHCl (1.93 versus 0.99 s in absence of GdnHCl, n = 106).These rates allow us to conclude that The figur e shows thr ee differ ent e v ents of the co-translocational unfolding of a TrwC R (Y18)-DNA complex (panels A , B and C ), obtained at +120 mV in 2M KCl, after the addition of GdnHCl (0.9 M) at the trans compartment of the ␣-HL nanopore.Le v els 1 to 4 of the current trace were similar to those obtained in the absence of GdnHCl (the same color code for each le v el was applied).The e v ents lacked the characteristic ionic current le v el 5 and, instead, from le v el four onwar ds, differ ent curr ent blockages without a reproducible pattern were observed until the complex was finally translocated.
TrwC R (Y18)-DNA complex es ar e less efficiently translocated in the presence of the denaturing agent.It is important to note that the amount of GdnHCl added to the trans compartment (0.9 M) would not totally pre v ent TrwC R refolding since, at this concentration of denaturant, the protein is expected to be still partially folded.Although susceptibility to unfolding by GdnHCl varies fr om pr otein to protein, higher concentrations of GdnHCl (up to 6 M) are normally r equir ed for complete protein unfolding ( 30 ).

DISCUSSION
Bacteria can e volv e ra pidl y by acquiring new traits through horizontal gene transfer, such as virulence, metabolic properties, and most importantly, antimicrobial resistance.Bacterial conjugation r epr esents a major form of horizontal gene transfer and is one of the main mechanisms whereby bacteria become resistant to antibiotics ( 3 ).Plasmid DNA is transferred from a donor to a recipient cell while covalently bound to a protein (a relaxase).The transfer of this nucleoprotein complex across biological membranes is mediated by a T4SS.The internal diameter of this secretion channel is only 10-20 Å ( 11 , 13 , 17 ), which implies all proteins that transverse the channel must be unfolded for passage.According to this, it has been shown that the relaxase protein of plasmid R388, named TrwC, must be unfolded to be translocated through a conjugati v e T4SS ( 14) and refolded after its transport, since the protein is acti v e once in the recipient cell ( 10 ).Howe v er, little is known about the mechanism behind this process.In this work, we have used nanopore technology to analyze the co-translocational unfolding of the relaxase domain of the conjugati v e protein T rwC (T rwC R ) covalently bound to a 30-mer oligonucleotide.This protein-DNA complex is the result of the cleavage by TrwC R of a longer oligonucleotide containing the R388 sequence at the origin of transfer.Therefore, this complex mimics the nucleoprotein substrate that is transferred in bacterial conjugation.It is important to note that TrwC R domain is sufficient to carry out cleavage and transesterification reactions and, in fact, some mobilizable plasmids code for a functional relaxase protein that only consists of this domain (all relaxases belonging to class MOB C ) ( 31 ).
TrwC R has two catalytic tyrosines, Y18 and Y26.In vivo , each tyrosine seems to play distinct roles.It was proposed that DNA is transferr ed to the r ecipient cell covalently bound to Y18 residue ( 9 ).Once in the recipient cell, Y26 would perform a new str and-tr ansfer reaction to rejoin both ends, leading to DNA r ecir cularization.In order to test this model, we analyzed the co-translocational unfolding of TrwC R -DNA complexes with DNA either bound to Y18 (TrwC R (Y18)-DNA complex) or to Y26 (TrwC R (Y26)-DNA complex.W hen transloca ted through ␣-HL nanopores, both types of constructs showed se v en co-translocational unfolding steps, consisting of subsequent transloca tion intermedia tes a t various stages of unfolding.The number of steps observed and, ther efor e, the complexity of the process seems to be linked to the size and the nature of the protein fold.Thus, the artificially DNA-tagged thioredoxin protein (ThrX), with 107 residues, showed only four steps ( 19 ).It is important to emphasize that the relaxase-DNA covalent complex analyzed in this work is a biological substrate transferred in bacterial conjugation and, in contrast to other proteins analyzed by nanopores so far ( 23 , 32 ), does not r equir e any artificial tag to traverse the por e. Mor eover, TrwC R has 293 r esidues, almost thr ee times as many as the ThrX protein.To our knowledge, this is the largest protein translocated through an ␣-HL nanopore.In addition, it does not r equir e additional proteins, such as the bacterial unfoldase ClpX at the trans compartment, used in other studies to facilitate the transport by capturing the protein at the exit of the nanopore ( 33 ).
Analyzing the co-translocational unfolding pattern, the duration and intensity of each step can be associated with a particular structural arrangement.Ther efor e, an inspection of the thr ee-dimensional structur e of TrwC (Figure 8 , Supplementary Figure S9) allowed us to describe, for each le v el, the obstacles that the protein is likely to overcome as it crosses the nanopore.Le v el 1 must correspond to the initial contact of the oligonucleotide with the cis entrance of the ␣-HL ( 26 ).Thus, we attribute le v el 2 to a state where TrwC R is placed at the top of the pore, with the oligonucleotide covalently linked through Y18 threading into the ␣-HL pore, as previously described for the ThrX protein ( 19 ), where similar r esidual curr ents for the thr eaded oligonucleotide state were observ ed.Accor dingly, the results show that this le v el (step 2 → 3) exhibited a slight increase in the rates at higher voltage values, due to the electronegati v e charge of the DNA.Le v el 3, of lower conductance, likely corresponds to TrwC R entrance into the ␣-HL pore.Since the leader DNA is not bound to residue 1 in the protein, but to either residue 18 in a TrwC R (Y18)-DNA complex or residue 26 in a TrwC R (Y26)-DNA complex, the barrier to overcome at this stage must be different and, in fact, this is what we observed.In a TrwC R (Y26)-DNA complex, level 3 was 70fold slower than in a TrwC R (Y18)-DNA complex.The duration of this step draws particular attention, since it is the longest of the entire transloca tion process, meaning tha t the protein needs more than 4 s to overcome this barrier (step 3 → 4), in comparison to the 75 ms r equir ed w hen DN A is bound through residue Y18.This kinetic trap is clearly ratelimiting for the translocation process of TrwC R (Y26)-DNA complex.
TrwC R structur e pr esents two domains.The first domain comprises residues 1-168 and is a two-layer ␣/ ␤ sandwich domain, also known as 'palm' domain, consisting of an antiparallel fiv e-stranded ␤-sheet with two helices flanking one face of the sheet ( 24 ).When TrwC is pulled from Y18, residues 1-17 and residues 19-onwards will enter the nanopore at the same time.Residues 1-17 can be released by dismantling first a short helix (residues 11-17) and then unzipping the ␤1-strand (r esidues 1-9), wher eas amino acids on the other side of the DNA attachment site (residues 19 -33) probably do not offer so much resistance, since they are part of an unstructured region, poorly resolved in the crystal structure (Figure 8 A).By contrast, the number of residues that enter the nanopore from both sides of the attachment site increase in a TrwC R (Y26)-DNA complex.Ther e ar e 25 r esidues from the N-terminus plus r esidues 27onwards on the other side (Figur e 8 d).Mor eover, r esidues 33-37 are part of the ␤2-strand that has to be unzipped from the other edge of the central ␤-sheet, which would offer higher mechanical resistance to unfolding in comparison with a TrwC R (Y18)-DNA complex.Such a change in the structure near the point where the force is applied (either residue Y18 or Y26) would explain the long unfolding time observed in level 3 for the TrwC R (Y26)-DNA complex.
Le v el 4 presents the lowest conductance of all the steps in the translocation process but, comparati v ely, it resolv es v ery ra pidl y.It might correspond to the co-translocational unfolding of the two ␣-helices that connect ␤2 and ␤3 strands (residues 38-61) and the next destructured loop (residues 62-79), which would diffuse easily through the pore before encountering the central ␤-sheet.On the contrary, step (5 → 6) presents the lowest rate (Table 1 ) and it likely corresponds to the unfolding of such a central core ( ␤3-␤7 strands, residues 80-168) (Figure 8 , Supplementary Figure S9).Ther efor e, in TrwC R (Y18)-DNA complex es, le v el 5 is the le v el that r equir es mor e time in the translocation process.Le v els 6 and 7 would correspond to the unfolding of the second domain: the helical C-terminal domain known as 'fingers' ( 24 ).At le v el 7, after completion of unfolding, the remaining segment of the protein traverses the pore, which is then unblocked (step 7 → O).The se v en different  43) and its topological diagram.When TrwC R is pulled from residue Y18, two peptide segments must be threaded inside the nanopore: 17 amino acids that stretch from the N-terminus to Y18 (in green), and an equivalent segment from residue Y18 onwards (in red) (panels A -C ).When TrwC R is pulled from residue Y26, the number of residues from each segment that enter the nanopore is higher.There will be 25 residues from the N-terminus (in green) plus residues 27-onwards on the other side (in red).In this latter case, the second segment includes the ␤2-strand and, ther efor e, this first unfolding e v ent involv es the disruption of the antiparallel fiv e-stranded ␤-sheet (panels D -F ).Arrows correspond to ␤-strands while helices are represented as cylinders.le v els described here are observed in all the analyzed e v ents (from le v el 4 onwar ds, rates and I RES (%) values were similar in both constructs) and, ther efor e, allow us to conclude that TrwC R unf olding alwa ys pr oceeds thr ough the same route.Moreover, we can also conclude that a TrwC R (Y18)-DNA complex is more ef ficiently transloca ted through the pore (each e v ent occurs in less than 1 s, in comparison with the 6 s on average r equir ed for TrwC R (Y26)-DNA translocation).The success in transferring the genetic material depends, to a great extent, on how ra pidl y the process occurs, since maintaining close contacts between donor and recipient cells is a limiting factor in some environments.Thus, this result would support the idea that DNA is translocated to the recipient cell covalently bound to residue Y18.The tr anslocation r ate observed in our experiments is comparable to that of Type III secretion systems (T3SS), used by bacteria to deli v er virulence proteins into the cytosol of host cells also in an unfolded state ( 29 ).In Salmonella , for instance, the rate of protein secretion has been measured at 7 to 60 proteins per cell per second ( 34 ).Bacterial conjugation, howe v er, demands more time since the process requires the transfer of a complete plasmid ( 35 ).
Within this work, we also demonstrate the potential of nanopore technology to distinguish and classify different protein-DNA complexes in a bulk solution, since each molecule shows a particular and consistent pa ttern.W hen TrwC R (wt) is incubated with the same oligonucleotide used for the previous constructs, a mixture of TrwC R (Y18)-DN A and TrwC R (Y26)-DN A complexes is obtained.After a ppl ying this sample to an ␣-HL nanopore, the same se v en current le v els were observ ed, but two different populations of molecules could be distinguished, showing different I RES (%) values in le v els 2 and 3.These differences were similar to those observed for TrwC R (Y18)-DNA and TrwC R (Y26)-DNA complex es.Mor eover, the dwell-time distribution in each of these two le v els was found to be bimodal, where each peak could be assigned to a particular construct.Ther efor e, each single e v ent of the TrwC R (wt)-DNA mixture could be assigned to a particular complex.A k -means cluster analysis allowed us to conclude that 78% of the molecules corresponded to a TrwC R (Y18)-DNA complex, which supports the idea that Y18 is more effecti v e than Y26 in the first cleavage reaction.
Other relevant questions, such as how the system is energized or the complex translocated into the recipient cell, remain unclear.In T3SS, an hexameric ATPase at the base of the channel seems to be involved in the unfolding process ( 36 ).It was suggested that this ATPase could also provide energy for pushing the substrate through the secretion channel, driving protein export.T4SS present hexameric ATPases at the base of the secretion channel that might also provide energy for the initial protein-DNA pumping (37)(38)(39).Ne v ertheless, as it occurs in T3SSs, there is a large distance between the ATPases and the host membrane -at least 30 nm ( 40 )-and this primary energy supply would be inefficient for the transfer of the protein-DNA complex ( 41 ).In that sense, in T3SS, it has been reported that folding of the secreted substrate upon exit from the secretion channel provides energy for pulling protein effectors through the apparatus ( 28 , 29 ).Ther efor e, another important aspect to be considered in powering the transport of the nucleopro-tein complex in bacterial conjugation, is the contribution of the relaxase unf olding-ref olding process.TrwC must be refolded in the recipient cell in order to recircularize the ssDNA plasmid str and.Thus, a r apid refolding of TrwC might also play an important role as a pulling force to complete the translocation process.In our nanopore system, we reasoned that if TrwC R refolding was playing such a role, we should observe a dif ferent co-transloca tional unfolding pattern in the presence of a denaturing agent.After the addition of 0.9 M GdnHCl to the trans compartment, levels 1 to 4 of the current trace remained unchanged but, in subsequent steps, different current blockages without a r eproducible pattern wer e observed until the complex was finally translocated.Such an unpredictable pattern might reflect TrwC R movement attempting to advance along the pore.On aver age, the tr anslocation process was faster in the absence of GdnHCl (median duration of the translocation process of 1.93 s versus 0.99 s).Ther efor e, our r esults ar e compatible with the idea that protein refolding after secretion helps to pull the complex through the pore, which is in agreement with biophysical studies that show that upon dena tura tion of the relaxase by temperature or chaotropes the protein is able to recover the nati v e state without the assistance of external agents ( 15 ).
In summary, this is the first time that the transfer across membranes of a conjugati v e comple x has been studied at a single molecule le v el, uncov ering the different partly unfolded sta tes tha t take place during transloca tion.Other important questions, such as the energy source required for translocation in the cellular process, remain unsolved.It is tempting to specula te tha t intrinsic structural properties in TrwC that allow rapid refolding have been selected by evolutionary pr essur e, favouring a particular unfolding and transloca tion pa thway.

Figure 1 .
Figure 1.Formation of TrwC R -DNA covalent complex es.( A ) TrwC R structur e consists of a two-layer ␣/ ␤ open sandwich core, also known as 'palm' domain (pdb: 1zm5) (44).( B ) 42-mer oligonucleotide with the R388 plasmid sequence that contains the nic site for TrwC.( C ) After the cleavage reaction, TrwC remains covalently bound to the 5´end of the resultant 30-mer sequence.( D ) Schematic r epr esentation of the nucleophilic attack by the hydroxyl group of a catalytic tyrosine residue on the 5´-side of the DNA phosphate.This transesterification reaction results in a covalent linkage between the tyrosine residue and DNA.Adapted from ( 9 ).

Figure 2 .
Figure 2. Purification and characterization of TrwC R -DNA conjugati v e comple xes.In or der to obtain TrwC R (Y18)-DNA comple xes, TrwC R (Y26F) protein (20 M) was incubated with a 42-mer oligonucleotide (30 M) containing the R388 ni c site, and sample was loaded onto a HiTrap Q HP column ( GE Healthcar e ) to separa te free TrwC R pr otein and DNA.( A ) Elution pr ofile.( B ) SDS-polyacrylamide gel of the obtained fractions: (M) SDS-PAGE standards low range; (R) complex formation reaction; peaks have been numbered in order of elution (1-4).Fractions from peak 3 were harvested and used in nanopore studies.Panels ( C ) and ( D ) show the same procedure to obtain TrwC R (Y26)-DNA complexes except that, in this case, TrwC R (Y18F) protein was used to form the protein-DNA complex.

Figure 3 .
Figure 3. Co-translocational protein unfolding of a TrwC R (Y18)-DNA complex is a seven-step process.( A ) Size-scale r epr esentation of a TrwC R (Y18)-DNA complex threading into the cis entrance of the ␣-HL por e. ( B ) Repr esentati v e current trace of the co-translocational unfolding of a TrwC R (Y18)-DNA complex, obtained at +120 mV in 2 M KCl. ( C ) Event histograms of the residual current le v els, e xpressed as I RES (%) at +120 mV, observed during the co-translocational unfolding of the TrwC R (Y18)-DNA complex.I RES (%) = ( I RES / I O ) × 100, where I RES is the current flowing during a blockade and I O is the current through the unblocked pore.Each le v el is represented with a different color code.

Figure 4 .
Figure 4. Pulling from tyrosine 26 leads to a less efficient co-translocational unfolding.( A ) Size-scale r epr esentation of a TrwC R (Y26)-DNA complex threading into the cis entrance of the ␣-HL pore.( B ) Representati v e current trace of the co-translocational unfolding of a TrwC R (Y26)-DNA complex, obtained at +120 mV in 2M KCl. ( C ) Event histograms of the residual current le v els, e xpressed as I RES (%) at +120 mV, observed during the co-translocational unfolding of the TrwC R (Y26)-DNA complex.Color codes for current le v els are similar to those represented in Figure 3 .

Figure 5 .
Figure 5. Lifetimes of le v els 2 and 3 of TrwC R -DNA constructs during translocation.Event histograms of the dwell times in le v el 2 (step 2 → 3) ( A ) and le v el 3 (step 3 → 4) ( B ) for a TrwC R (Y18)-DNA complex at +120 mV.The fit is to a single exponential function and yields the rate constants k 2 → 3 = 37 s −1 and k 3 → 4 = 12 s −1 at +120 mV.Panels ( C ) and ( D ) are the e v ent histograms of the dwell times for a TrwC R (Y26)-DNA complex in le v el 2 (step 2 → 3) (C) and le v el 3 (step 3 → 4) (D), respecti v ely.Data were also fit to a single exponential function, which allowed the determination of the rate constants k 2 → 3 = 1051 s −1 and k 3 → 4 = 0.17 s −1 at +120 mV.Panels ( E ) and ( F ) are the e v ent histograms of the dwell times for a TrwC R (wt)-DNA complex in le v el 2 (step 2 → 3) (E) and le v el 3 (step 3 → 4) (F), respecti v ely.In this case, the fit is to a doub le e xponential function and yields the rate constants k 2a → 3 = 32 s −1 and k 2b → 3 = 966 s −1 , and k 3a → 4 = 10 s −1 and k 3b → 4 = 0.19 s −1 at +120 mV.

Figure 6 .
Figure 6.Scatter plot of the dwell time values associated to le v els 2 and 3 of TrwC R -DNA constructs.Points r epr esent dwell times values in le v els 2 (step 2 → 3) and 3 (step 3 → 4) for TrwC R (Y18)-DNA ( ), TrwC R (Y26)-DNA ( ), and TrwC R (wt)-DNA ( ) constructs, collected at +120 mV.Two dif ferentia ted popula tions ar e observed, corr esponding to TrwC R (Y18)-DNA and TrwC R (Y26)-DNA samples, respecti v el y.TrwC R (wt)-DN A samples show a combination of the values observed for these two populations, with a majority of e v ents associated to a TrwC R (Y18)-DNA popula tion sample.Da ta from a t least three dif ferent experiments were used.

Figure 7 .
Figure 7. Current trace of the co-translocational unfolding of a TrwC R (Y18)-DNA complex in the presence of guanidinium chloride.The figur e shows thr ee differ ent e v ents of the co-translocational unfolding of a TrwC R (Y18)-DNA complex (panels A , B and C ), obtained at +120 mV in 2M KCl, after the addition of GdnHCl (0.9 M) at the trans compartment of the ␣-HL nanopore.Le v els 1 to 4 of the current trace were similar to those obtained in the absence of GdnHCl (the same color code for each le v el was applied).The e v ents lacked the characteristic ionic current le v el 5 and, instead, from le v el four onwar ds, differ ent curr ent blockages without a reproducible pattern were observed until the complex was finally translocated.

Figure 8 .
Figure 8. Structural barriers to overcome in TrwC R unfolding.The figur e r epr esents the 3D structure of TrwC R (pdb: 1zm5) (43) and its topological diagram.When TrwC R is pulled from residue Y18, two peptide segments must be threaded inside the nanopore: 17 amino acids that stretch from the N-terminus to Y18 (in green), and an equivalent segment from residue Y18 onwards (in red) (panels A -C ).When TrwC R is pulled from residue Y26, the number of residues from each segment that enter the nanopore is higher.There will be 25 residues from the N-terminus (in green) plus residues 27-onwards on the other side (in red).In this latter case, the second segment includes the ␤2-strand and, ther efor e, this first unfolding e v ent involv es the disruption of the antiparallel fiv e-stranded ␤-sheet (panels D -F ).Arrows correspond to ␤-strands while helices are represented as cylinders.