Mu Transposition in the Absence of the Target-capture Protein MuB Reveals New Roles of MuB in Target Immunity and Target Selection, and Redraws the Boundaries of the Insular Ter Region of E. coli

The target capture protein MuB is responsible for the high efficiency of phage Mu transposition within the E. coli genome. However, some targets are off-limits, such as regions immediately outside the Mu ends (cis-immunity) as well as the entire ∼37 kb genome of Mu (Mu genome immunity). Paradoxically, MuB is responsible for cis-immunity and is also implicated in Mu genome immunity, but via different mechanisms. In this study, we tracked Mu transposition from six different starting locations on the E. coli genome, in the presence and absence of MuB. The data reveal that Mu’s ability to sample the entire genome during a single hop in a clonal population is independent of MuB, and that MuB is responsible for cis-immunity, plays a lesser role in Mu genome immunity, and facilitates insertions into transcriptionally active regions. Unexpectedly, transposition patterns in the absence of MuB have helped extend the boundaries of the insular Ter segment of the E. coli genome.

Phage Mu uses transposition to amplify its genome ~100-fold during its lytic cycle in E. coli, 27 making it the most efficient transposable element (TE) described to date [1][2][3] (Fig. 1A). Mu 28 transposes by a nick-join pathway, where assembly on Mu ends of a six-subunit MuA transposase 29 complex (transpososome) is followed by introduction of nicks at both ends; the liberated 3′-OH 30 groups at each end then directly attack phosphodiester bonds spaced 5 bp apart in the target 31 DNA, covalently joining Mu ends to the target [4]. The resulting branched Mu-target joint is 32 resolved by replication, duplicating the Mu genome after every transposition [5]. At the end of the 33 lytic cycle, Mu copies are excised for packaging by a headful mechanism that cuts and packages 34 host DNA on either side of Mu [1,6]. The latter finding has been exploited to examine target site 35 preference in vivo by sequencing the flanking host DNA packaged in Mu virions [7,8]. 36 The B protein of Mu (MuB), a non-specific DNA-binding protein and AAA+ ATPase, is essential 37 for the efficient capture and delivery of the target to the transpososome via MuB-MuA interaction; 38 MuB also plays critical roles at all stages of transposition by allosterically activating MuA (see [3, 39 9]). MuB forms ATP-dependent helical filaments, with or without DNA [10][11][12]. Because of a 40 mismatch between the helical parameters of the MuB filament and that of the bound DNA, it has 41 been proposed that the DNA at the boundary of the MuB filament deforms, creating a DNA bend 42 favored by MuA as a target [11,13,14]. While most TEs display some degree of target selectivity 43 [15], Mu is perhaps the most indiscriminate, with a fairly degenerate 5bp target recognition 44 consensus [7,8,16,17]. Even though MuB facilitates target selection, recognition of the 45 pentameric target site is a property of MuA, and is independent of MuB. 46 Several bacterial TEs, including members of the Tn3 family, Tn7, and bacteriophage Mu, display 47 transposition immunity [15,[18][19][20][21][22]. These elements avoid insertion into plasmid DNA molecules 48 that already contain a copy of the transposon (a phenomenon called cis-immunity), and it is 49 thought that this form of self-recognition must also provide protection against self-integration (TE 50 polymerization of MuB on the genome, forming a barrier against self-integration, was proposed 75 [27]. 76 The present study investigates the role of MuB in the three diverse functions discussed above -77 target capture, cis-immunity, and Mu genome immunity in vivo. Through comparison of insertion 78 patterns of wild-type (WT) Mu and Mu∆B prophages placed at six different locations around the 79 E. coli genome, we show that cis-immunity depends on MuB, while Mu genome immunity is only 80 slightly breached in its absence. The data also reveal a previously unappreciated role for MuB in 81 facilitating Mu insertion into transcriptionally active regions, as well as several interesting and 82 hitherto unknown aspects of Mu target choice in vivo. An unanticipated outcome of this study is 83 the finding that the Ter segment of the E. coli genome, which is more isolated from the rest of the 84 genome, is larger than previously estimated. We recently exploited the DNA-DNA contact mechanism of phage Mu transposition to directly 90 measure in vivo interactions between genomic loci in E. coli [29]. Thirty-five independent Mu 91 prophages located throughout the genome were allowed to go through one round of transposition. 92 The data showed that in a clonal population, Mu is able to access the entirety of the genome with 93 roughly equal probability regardless of its starting genome location, suggesting widespread 94 contacts between all regions of the chromosome. The data led us to conclude that the 95 chromosome is well-mixed and shows a 'small world' behavior, where any particular locus is 96 roughly equally likely to be in contact with any other locus. The exception was the Ter region, 97 reported by Mu as being less well-mixed than the rest of the genome. 98 While MuB is essential for target capture in vitro [4], transposition is still detectable in vivo in the 99 absence of MuB at an efficiency nearly two orders of magnitude lower than WT [30]. To examine 100 how MuB influences the target selection in vivo, we monitored insertion patterns of a subset (six) 101 of the Mu prophages used in the original study [29] (Fig. 2A), after a single round of transposition, 102 in the presence and absence of MuB (WT vs ∆MuB) (Fig. 2B). For analysis, the genome was 103 partitioned into 200 equally sized bins (each bin ~23.2 kb) ( Fig. 2A). To generate sufficient 104 insertion resolution, transpositions were analyzed using a target enrichment protocol [29] and 105 deep sequencing of 10 million reads or more. Due to lower transposition frequencies of ∆MuB 106 prophages, these were sampled ~50% more with a 15 million read depth. The data plotted in 107 Figure 2B show similar insertion profiles for both WT and ∆MuB throughout the genome after 108 normalizing to the read depth for both prophages. Thus, like WT, the ∆MuB prophages transpose 109 to every bin of the genome in a clonal population, allowing us to conclude that the ability of Mu to 110 sample the entirety of the genome in one transposition event is independent of MuB. 111 The color-coded map of the E. coli genome shown in Figure 2A depicts the length and boundaries 112 of chromosomal regions deduced by prior methodologies to be either unreactive or partially 113 reactive with the other regions [31]. With the exception of Ter, the Mu methodology failed to detect 114 all such boundaries [29]. The Ter region has unique properties shaped by the activity of MatP [32] 115 and the condensin MukBEF [33,34], and has been shown by several methodologies to be more 116 isolated from the rest of the chromosome [29,35,36]. Comparison of WT vs ∆MuB insertion 117 patterns supported this conclusion while revealing more details. For example, the ∆MuB prophage 118 located in Ter (Ter-Mu) had >40% of its total insertions occur within the Ter region, which only 119 comprises ~20% of the genome (light red profile). While Ter-∆MuB prophage sampled the DNA 120 around its starting location more efficiently than it did the rest of the genome, the ∆MuB prophages 121 at the five other locations showed a converse pattern in that they could not access Ter as easily 122 (dark red profile). The latter prophages had <15% of their total insertions within Ter. Comparison 123 of both the outgoing and incoming ∆MuB profiles all lined up precisely, giving us a clearer view of 124 the boundaries flanking Ter. According to Valens et al. [31], the Ter region extends from 125 nucleotide position 1128 kb (26' on the genetic map) to 2038 kb (47'). According to the 126 transposition patterns of ∆MuB prophages, the Ter region extends from nucleotide position 911 127 kb (21') to 2200 kb (47'), expanding the left boundary by more than 217 kb (Fig. 2B). We note that 128 ΔMuB prophages did not reveal other boundaries (as demarcated by the colored segments in Fig prophages in sampling Ter, it follows that MuB must weaken the Ter boundary conditions. The 142 property of MuB to nucleate as helical filaments on DNA [11], may be responsible for displacing 143 the boundary-guards. These results imply that the Ter segment is even less well-mixed than 144 determined in the study utilizing WT Mu [29]. 145 146

MuB facilitates transposition into highly transcribed regions 147
Two prior microarray data sets have shown a negative correlation between transcription and Mu 148 transposition [37,38], although one these studies found several exceptions to this rule, and 149 suggested that some other cellular feature controls these insertion events [38]. We examined this 150 issue using our higher-resolution data set. Fig. 3  downstream. However, the transposition difficulty was exacerbated in ΔMuB prophages, which 160 showed an increase in an exclusion zone starting near the TSS for transcriptionally active genes 161 and to a lesser extent for the comparatively less transcriptionally active lacZY. Interestingly, two 162 different WT Mu insertion patterns were observed within the lac operon, whose lacZ and lacY 163 genes are repressed by the activity of the lacI repressor, which is expected to be transcribed [42]. 164 The number of Mu insertions in lacI were roughly half those in lacZY, with a strong suppression 165 of insertions around the TSS and +1 nt region of lacI for WT. This observation is in agreement 166 with the previous findings of a negative correlation between transcription and transposition. 167 Of six potential promoters in the flhDC operon that control flagellar gene transcription in 168 Salmonella, only two (P1 and P5) were seen to be functional [39]. These two sites are each 200-169 300 bps upstream of the +1 nt [39]. On the other hand, the specific transcriptional start site for 170 dnaJ is 2 kb away, as dnaJ is always co-transcribed with dnaK, with a small 370 nt RNA candidate 171 tpke11 between the two genes [40,43]. WT prophages show a near uniform sampling across 172 flhD, with reduced insertion around the TSS, while ΔMuB prophages show in addition a secondary 173 exclusion zone upstream from the +1 that encompasses both P1 and P5 promoter regions. Even 174 though TSS is absent in dnaJ, WT Mu shows an insertion exclusion zone around +1 of this gene. 175 ΔMuB prophages show an exclusion zone upstream of dnaJ not seen in WT, around the position 176 of tpke11, while revealing an unusually permissive region upstream of dnaK. The latter permissive 177 region in both WT and ΔMuB corresponds to the 377 bp intergenic region between yaaI and the 178 dnaKJ operon promoter. While this set insertion patterns overall is consistent with the negative 179 correlation between transcription and transposition, particularly around the TSS and +1 for WT, 180 the insertion patterns in dnaJ reveal that the +1 region presents a transposition barrier 181 independent of the promoter region, and is likely reflective of the translation activity of the mRNA 182 near this genomic site given that transcription and translation are coupled in bacteria. 183 To examine Mu insertion patterns in genes that are transcribed but not translated, we looked at 184 both ribosomal RNA operons and tRNA genes. E. coli has 7 ribosomal RNA operons that are 185 highly transcribed [44]. We observed a large variation of insertion profiles in these regions (Fig.  186   S1). For example, the insertion frequency of WT Mu is highest in rrnA, uniform across the entire 187 operon, and independent of MuB. rrnE and rrnH receive more insertions in the 23S compared to 188 the 16S region, and are responsive to MuB. rrnG shows a large increase in sampling only at the 189 5' end of the 16S region (note that rrnG is on the negative strand). There seems to be an equal with the newer data, in that Mu activity is highest within rrnA, and lowest near the promoter region 199 of rrnE (Fig. S1). Regardless of the rrn operon, there seems to be a small window between the 200 16S and 23S subunits in each operon that is marked by an increase in insertion frequency. This 201 window contains non-coding sequence as well various tRNA sequences. The latter are highly 202 undersampled by Mu insertions even when they occur elsewhere in the chromosome as 203 discussed below. 204 Mu insertion patterns into 86 tRNA genes scattered throughout the E. coli genome [47], is shown 205 in Figure S2. Mu shows an interesting selectivity for inserting into 30 of these genes, avoiding the 206 region that would ultimately be the mature tRNA sequence, as exemplified by the large hole or 207 gap with no insertions seen in the bottom half of the WT Mu panel. Note that Mu is more actively 208 inserting into the genomic regions associated with the 5' leader and 3' tailing sequences of pre-209 tRNA. This would suggest that there is some genomic feature (fold, DNA-binding protein) that is 210 ultimately protecting this region of DNA from Mu insertion. ΔMuB prophages incidentally were 211 less likely to insert into the entire pre-tRNA sequence, suggesting that the transcriptome 212 machinery provided a much higher barrier of access to the ΔMuB prophages over the WT 213 prophages. Using genome-wide transcription propensity data [48], we were able to compare the 214 levels of transcription for each of the tRNA sequences along with the likelihood that Mu (WT and 215 ΔMuB) would transpose within them. Although the transcriptional information was quantitatively 216 sparse amongst most of the tRNA genes, the accessibility of insertion into 36 tRNAs that are the 217 lowest transcriptionally active genes, and exclusion of insertion into the highest transcriptionally 218 active ileY and selC (marked with red asterisks), is unmistakable. In these two regions, there are 219 no insertions in the entire pre-tRNA CDS in both WT and ΔMuB. 220 We conclude that the level of availability of a target for Mu insertion is highly correlated with its In vivo, a preference for 5'-CGG as the central triplet was derived from cloning 100 Mu-host 248 junctions from packaged phage particles [8]. To re-examine target preference using our current 249 data set, we pooled first-hop insertion data totaling over 120 million targeted Mu reads for both 250 the WT and ΔMuB constructs. We observed that in the genome, sequences with the triple-'G' 251 consensus and their reverse compliment were 3-4 times more abundant than the 5'-CYSRG-3' 252 sequences, explaining the preference for 5'-CGG in the earlier study (Fig. S3A). Sequencing data 253 suggest that there is a 7-fold preference for the 8 possible 5'-CYSRG-3' consensus sequences 254 over the other 1016 remaining pentamer sequences (Fig. S3B). 255 256

MuB is responsible for cis-immunity 257
The cis-immunity phenomenon has been studied in vitro exclusively by the Mizuuchi group, from condition, we let the experiment run for 2 hours, which allowed WT to complete its lytic cycle (in 267 ~50 min) and ΔMuB prophages to accumulate 5 to 10 copies of Mu on average per cell as 268 predicted by genome abundance, assuming an even distribution of Mu copy number among the 269 population. All six prophage strains were used for EST experiments, and one WT plus all six 270

ΔMuB prophages for LST experiments. 271
During EST, WT Mu (bottom row of all plots) does not transpose within 1.5 kb outside each of the 272 starting Mu positions, consistent with the cis-immunity phenomenon (Fig. 4A). That the absence 273 of transposition in this region is not due to an intrinsic resistance to insertion within this DNA, is 274 seen from the pooled profiles of the other prophages for the same region (WT pool). Figure 4B  individual ends (Fig. S4). LST samples for WT OPL-Mu show that cis-immunity remains intact 279 through multiple rounds of transposition (Fig. 4C). We observed that the insertion patterns outside 280 Mu ends have a sigmoidal nature in both samples, suggesting that cis-immunity is not entirely 281 dependent on linear genomic distance. The previously described ratchet-model suggests that 282 intrinsically clustered MuA would hydrolyze proximal MuB-ATP during dynamic loop formation 283 due to Brownian motion [25]. We propose that the sigmoidal pattern of cis-immunity arises from 284 three-dimensional features of the genome, and that dynamic loop formation is the necessary 285 factor in creating the conditions that generate such a pattern. 286 In the absence of MuB, we expected to see a maximal insertion frequency (consistent with that 287 for bulk DNA) around 150 bp outside Mu ends, which is the in vitro persistence length of DNA 288 [52]. Instead, EST ΔMuB prophages exhibited a more gradual increase in insertion frequency, 289 starting between 500 to 600 bp and reaching bulk transposition efficiency around 7 kb from the 290 ends ( Fig. 4B and Fig. S4). This pattern was different in the LST samples, where cis-immunity 291 remained intact for the WT OPL-Mu (Fig. 4C left), but was completely abrogated in ∆MuB OPL-292 Mu alone (compare Fig. 4C middle with left), and well as in all six ∆MuB prophages combined 293 (Fig. 4C right). For the ∆MuB prophages, insertions started at 98 bp (at a distance smaller than 294 the in vitro persistence length; [53]) and reached bulk efficiency between 2 and 6 kb. Why are the 295 ∆MuB insertion patterns so different between the EST and LST samples? We think the lower 296 transposition efficiency of ∆MuB prophages did not provide sufficient opportunity to sample 297 nearby space during EST, whereas the increased ∆MuB copy number in the LST samples 298 provided a greater opportunity to saturate the cis region. We conclude that MuB is indeed 299 responsible for cis-immunity in vivo. 300 What is the importance of cis-immunity in the life of Mu? Avoiding insertion into regions flanking 301 Mu ends would avoid destroying flanking Mu copies when packaging begins, since the DNA 302 packaging machinery resects on average 100 bp of host DNA flanking the left end and 1.5 kb of 303 DNA flanking the right end. Negating this concern, however, is the finding that Mu samples the E. 304 coli genome extensively in a distance-independent manner (Fig. 2) [29]. A more likely possibility 305 is that cis-immunity is an evolutionary remnant of MuB-and MuA-like functions in an ancestral 306 transposon, where additional partner proteins directed transposition to specific sites. For example, 307 Tn3 and Tn7 exhibit target immunity much further than Mu [22,54,55]. Tn7 has two proteins 308 TnsB and TnsC that are thought to play roles similar to MuB and MuA respectively. Tn7 has two 309 partner proteins, TnsD and TnsE, that promote different target choices. Han and Mizuuchi [25] 310 discuss how the Mu cis-immunity system may have evolved from a Tn7-type target site search. 311 Mu apparently discarded these partners during an evolutionary trajectory more suited to its viral 312 lifestyle, acquiring features that unfettered its ability to choose. 313 314

MuB is only partially responsible for Mu genome immunity 315
The cis-immunity phenomenon depends on MuB removal from DNA adjacent to and outside Mu 316 ends. By contrast inside Mu, the MuB was observed to bind strongly during the lytic cycle, 317 implicating a role for bound MuB in Mu genome immunity [27]. In the EST insertion data shown 318 in Fig. 4A, there were no observable self-insertions (SI) in either WT or ΔMuB (the latter have 319 1.5x the depth of sequence reads compared to WT). SI was also not detected in the EST data for 320 35 WT prophages reported earlier [29]. To determine if this immunity is still intact at the end of 321 the lytic cycle, we examined LST counts in the two prophage populations (Fig. 5). The WT OPL-322 Mu was still immune to SI (not shown), but the ΔMuB prophages, which have higher copy numbers 323 in LST, now showed evidence of self-insertion. However, out of 90 Million Mu targeted reads from 324 deep sequencing, 85 instances of SI were observed, spread across all 6 starting ΔMuB 325 prophages. We conclude that, unlike cis-immunity which is completely abrograted in the absence 326 of MuB (Fig. 4C), genome immunity is only slightly violated. Therefore, the bulk of genome 327 immunity is determined by factors other than MuB. 328 Mu ends (L and R) define a boundary separating two modes of MuB binding and immunity [27]. 329 We had proposed that Mu genome immunity arises from a special structure that Mu adopts, aided 330 by both specific Mu sequences and by general cellular NAPs. In the center of the genome is the 331 strong gyrase-binding site (SGS), which is essential for Mu replication in vivo and is believed to 332 function by influencing efficient synapsis of the Mu ends [56][57][58]. The SGS is thought to act by 333 localizing the 37 kb Mu prophage DNA into a single loop of plectonemically supercoiled DNA upon 334 binding of DNA gyrase to the site. We had proposed that an SGS-generated Mu loop, sealed off 335 at the Mu ends by either the transpososome or NAPs, serves as a scaffold for nucleating MuB 336 filaments in the Mu interior, providing a barrier to Mu integration. Evidence for a separate, stable 337 prophage Mu domain, bounded by the proximal location of Mu L and R ends, was indeed obtained 338 [28]. Formation/maintenance of the Mu domain was dependent on SGS, the Mu L end, MuB 339 protein, and the E. coli NAPs IHF, Fis and HU. Of these components, SGS is essential for Mu 340 transposition in vivo [59,60], hence its contribution to Mu genome immunity cannot be assessed. 341 To examine the contribution of the NAPs, we analyzed our published data where we had 342 monitored Mu transposition in all NAP mutants of E. coli (these were collected during EST) [29]. 343 We observed no instances of Mu self-transposition in any of the NAP deletions examined. 344 345

Summary 346
MuB is critical for Mu's ability to efficiently capture targets for transposition. We show in this study 347 that besides enabling efficient targeting, MuB also makes refractory targets more facile, likely by 348 displacing bound proteins. By weakening/altering boundary features that demarcates the Ter 349 region, MuB allows Mu to access Ter more readily. Transposition patterns in the absence of MuB 350 have allowed us to more accurately measure the Ter boundaries, revealing that this region is 351 larger than previously estimated. Perhaps in a similar manner, MuB also provides access to 352 targets engaged in transcription/translation. We have mapped the range of cis-immunity more 353 accurately, and show that it persists well into the lytic cycle for WT prophages, but is abolished in 354 ΔMuB strains. We show that Mu genome immunity also persists through the lytic cycle for WT 355 prophages, and is only rarely infringed upon in ΔMuB prophages, showing conclusively the 356 distinction between these two forms of immunity. There is clearly more to be learned about what 357 enables genome immunity. prophages. Lysogen genomic DNA was purified using a commercially available gDNA purification 372 kit (Wizard, Promega). gDNA samples were stored at -20 ºC in a 10 mM Tris pH 8.0, 1 mM EDTA 373 buffer until ready for target enrichment. 374

Target Enrichment 375
Primer y-link1 has a hand mixed random 6 nucleotide barcode to identify PCR duplicates in 376 sequencing. Y-link adapters were annealed by mixing equivalent amounts of primers y-link1 and 377 y-link2 at room temperature and heating to 95 ºC then cooled down to 4 ºC using a temperature 378 ramp of 1 ºC per second. Genomic DNA was digested with the frequent cutter HinPI (NEB) and 379 then ligated with the y-link adapter using a quick ligase kit (NEB). The ligation product was purified 380 using magnetic beads (Axygen). Mu insertion targets were enriched, by PCR amplification of the 381 ligation product using y-link_primer and Mu_L31, an initial melting temp of 95 ºC for 1 min and 8 382 cycles of 95 ºC for 20 s, 68 ºC for 20 s, 72 ºC for 1 min. A final extension of 72 ºC was added for 383 5 minutes. The PCR product was purified using magnetic beads (Axygen) and frozen at -20 ºC 384 until ready for sequencing. 385

Genomic Sequencing 386
Target enriched samples were submitted to the Genomic Sequencing and Analysis Facility 387 (GSAF) at UT Austin for sequencing. Libraries were prepped by GSAF using the facility's low-388 cost high throughput method. Sequencing was done on an Illumina NextSeq 500 platform using 389 2X150 paired ends targeting 10 to 15 million reads. All sequencing data discussed in this work is 390 available at https://www.ncbi.nlm.nih.gov/sra/PRJNA597349. 391

Identifying Mu Insertion Locations. 392
Mu transposition targets were identified using lab software entitled Mu Analysis of Positions from 393 Sequencing (MAPS) as described earlier [29]. MAPS has been modified since initial publication 394 to provide nucleotide precision for target enriched samples and provide self-insertion information. Availability of data and material: All strains generated in this study are available without restriction. 401 The sequencing data presented in this paper can be accessed on the SRA database under the 402 project number PRJNA597349. Software used to analyze the sequencing data can be accessed 403 by github (DOI 10.5281/zenodo.3762807). single-strand nicks, joining these to MuB-captured target DNA. MuB binds DNA non-specifically, polymerizing in short filaments, and increases the catalytic efficiency of target capture. B. Both cis-immunity and Mu genome immunity operate by two distinct mechanisms to prevent Mu insertion. Cis-immunity is characterized by the lack of insertions outside Mu ends, typically within 5 kb. Mu genome immunity is characterized by absence of insertion anywhere within the ~37 kb Mu genome.  Table 1 for their exact locations). These locations were chosen because they are spread throughout the chromosome, and therefore ideally suited for sampling features across the genome. oriC in the Ori region is the site where bi-directional replication begins (green arrow), terminating at the dif site, exactly opposite to oriC within the Ter region (cyan arrow). OPL, Ori proximal left; OPR, Ori proximal right; TPL, Ter proximal left; TPR, Ter proximal right. The boundaries of the various colored regions are taken from [1]. B. The genome was partitioned into 200 equally sized bins (A), and the normalized number of unique insertions into each bin for each prophage was computed, as displayed by the color bar. The highest number of unique insertions for any non-starting bin was ~8000 insertions corresponding to just under 1.0. Each starting bin position can be identified by its high number of counts (deep blue bins); the large number of reads associated with the starting position information were retained to aid in identifying this initial position shown on the map in B. The multi-color strip on top of each panel corresponds to chromosomal the regions shown in A. The Ter region (cyan) as explored by the Ter-∆MuB prophage is 217 kbp larger than earlier estimates [1]. This is recognizable as a square block of lighter red insertions in the Ter-∆MuB prophage, which lines up with identical blocks of darker red insertions in the other five ∆MuB prophages. Figure 3: MuB is responsible for capturing target sites near highly transcribed/translated genes. Twenty-three highly transcribed genes, plus the lac operon, flhD and dnaK-dnaJ, were selected for comparison between WT and ∆MuB insertion patterns. For WT transposition, the EST counts (early stage transposition; 15 min induction of transposition) were pooled together from all six prophage locations with an average of 5 million reads per prophage. ΔMuB experiments pooled all six prophages with an average of 20 million reads per prophage. Each gene is oriented to where the +1 nt of coding sequence of the gene starts at the tick mark labeled +1, and downstream nucleotides follow to the right. Upstream nucleotides are marked by negatively labeled tick marks. The expected transcription start site labeled <TSS> is 125 nt away from the +1 site.