Note: this is a search engine friendly version of my lab notebook, please see the pdf version of this document for a more human friendly, printer friendly version.

Chapter 5
Experimental Operon Determination

THIS CHAPTER/PROJECT IS DEAD
(or at least not-active)
Please see the last paragraph of this chapter for conclusions that might still be interesting regarding what I found in this chapter. It seems that a huge chunk of the bacterial genome is transcribed - even regions that you wouldn't expect. However, I'm leaving this project aside to focus on the PET sequencing project that will hopefully provide similar/better information for less work, less money, and in a more generalized framework. If I pick this project up again, I definitely need to try doing qPCR to these regions.
Tue Feb 14 17:25:50 EST 2006
I'm not interested in operon predictions, known operons, or complex operons. However, as I explained in my Feb 2, 2006 labmeeting talk, it is important to know all of the operons in a genome if you really want to verify predicted regulatory interactions using ChIP. You have to know the most likely binding site of the TF (which is in front of the first gene in the operon that the TF is predicted to regulate) to design a primer set that quantifies the enrichment of your TF for your target. It was for this reason that I initially became interested in developing a high-throughput method of operon determination.
The method I planned to use is simple and the low-throughput version is not-at-all novel. I can't find too many recent papers describing it, but I'd guess it's the standard way operons were found in the past. The strategy is shown visually in Figure . The idea is this: You have 3 genes you think might be in an operon together. To test this, design 5 sets of primers and make sure the primers work and are specific by testing them on genomic DNA. Using cDNA produced from an mRNA lysate of a particular condition or a mix of several conditions, run a PCR reaction for each of the primer pairs. Amplification of the gene itself (Fig orange arrows) is a necessary precursor to determine operons (Fig purple arrows). Successful amplification of of the genes lets you know if the gene is present in the cDNA. If it isn't, you can learn nothing about this particular genes operon from this particular cDNA sample. If two consecutive genes are present (two adjacent orange arrow pairs are amplified) and you can amplify from one gene to the next (the purple arrows), then the genes must be transcribed as one unit and are in an operon together. By tiling this procedure along the genome you need 4 primers per gene * 4000 genes = 16000 primers.
Please see the pdf version for figures
Figure 5.1:
16000 primers is 8000 PCR reactions and 8000 gels. However, by multiplexing the PCR reactions (5-10 per tube) and running microfluidic gels using something like the Agilent 5100 ALP, it should be possible to cut this down to around 1000-1500 PCR reactions. With a mediocre PCR machine and a robot plus the 5100 ALP (or LabChip 90), all the operons for one species could be checked in 24-48 hours. Specifically for the ALP, the cost is $130,000, plus $400 for reagents for 3000 samples, plus $900 for the microfluidics chips to run 6000 samples. Assuming the multiplexing could bring the number of reactions to 1000, the cost per operon validation is $288 - less than a microarray. This means it wouldn't be too hard to run a few studies in different conditions to test for complex operons. There is also an initial fee for the primers needed to do all the PCR reactions: 16000 primers would be $20-40K. The machine can handle about 340,000 samples a year and has a 10% downtime (sale rep says it is a new machine that needs to be repaired a fair amount), still that's 170 potential full operon checks a year.
A smart algorithm could pick primers that would potentially work in multiple related species. Also, the size ranges of the PCR amplification would need to be chosen so that they are different enough that when each reaction is run on an e-gel all 5-10 bands (for the amplified PCR reactions) are easily distinguishable. The e-gel chip works for 25-1000bp and at least in their literature easily distinguishes stuff more than 7bp difference in length. In practice we'd probably try for at least 50bp between the fragment lengths. To increase the chance of having more genes present and therefore more operons determined, it might be necessary to pool mRNA from many different conditions. This would destroy any complex operon information.
The more I think about this method the more I like it. One thing that it could address better than any previous study is: what fraction of an organism's genes are present in the cell at any given time. Microarrays kinda give you this information, but while my experience with Affy chips leads me to believe the are fairly precise for E. coli, I have no idea how sensitive they are. PCR is certainly more sensitive, and comparing just the gene amplifications (orange arrows) to the Affymetrix call values (Present, Absence, Marginal) would be interesting by itself. In addition, once you have all these primers you can do some really nice spike-in studies to really test the range of microarray quantitation and how cross-hybridization affects accuracy (see my prospectus notes for details).
This method will not work if a complex operon is transcribed in multiple forms in the same condition. For this method to work it is VERY IMPORTANT that the mRNA prep be completely clear of genomic DNA. Otherwise it'll be a real pain-in-the-ass to set some theshold to chop things off that you think are amplified just because of the contamination. I think it would be wise to use the new Qiagen preps that are designed to reduce genomic DNA PLUS a DNAse digestion just to be safe. To check for this, it would probably be useful to run a few PCR reactions on the cleaned-up mRNA to make sure nothing amplifies.

5.1  Initial Tests

I don't think the PI is too fond of this idea. My guess is he thinks it is expensive for what you get, especially since he's not the Broad Institute with deep pockets. However, some of the network inference predictions to unknown targets have a known target right next too them. The simplest explanation for the interaction is that the two genes are in an operon together, which is why the inference algorithm says they are both regulated by the same TF. To validate the new target, you don't need to do a complicated ChIP experiment, all you need is to do a two-gene version of what is shown in Figure 5.1.
I ordered primers to test this for genes flgK and flgL. flgK is a known target of the flagella regulators, flgL is not but it is right next to flgK. Now that I'm typing this, I realize I should order primers for a few other genes just to make sure this works for known genes.

5.1.1  flgK, flgL operon primer tests

Mon Mar 20 12:32:21 EST 2006
I'm going to test the primers by simply doing a PCR of genomic DNA. I'm also going to try doing a completely unintelligent multiplex by dumping all the primers into one reaction assuming no nasty interactions are going to happen. In practice I don't plan on multiplexing the same operon in the same tube, since there might be steric problems with amplifying some many regions that are very close to each other.
I'm running a 25ml PCR with the following reaction:
PCR Reaction composition
H2O 8.5 ml
Qiagen Master Mix 10 ml
Forward and reverse primer1.5 ml
final primer concentration150 pM
template DNA 100 ng
Thermal cycler conditions
Initial denaturation5 min 95° C
2-Step cycling
Denature:30 sec 95° C
Anneal/Extend:45 sec 60° C
Number of Cycles:28
Final Extention: 7 min 72° C

The predicted lengths of the amplified regions of each primer pair are:
ID forward primer reverse primer length (bp)
A flgK F flgK R 161
B flgL F flgL R 220
C flgK-L F flgK-L R 307
Eight combinations will be tested: blank, A, B, C, AB, AC, BC, ABC.
Unfortunately, after the PCR finished the ABC reaction tube was empty. I think these PCR tube strips I have are too thin and it must have had a little hole in the tube and the stuff inside leaked out. There was just a tiny amount. The other tubes on the strip looked fine.
Please see the pdf version for figures
Figure 5.2: 80 ml, 2% agarose gel with 0.5 ul of 1% ethidium bromide run for 40 min at 100 volts. 10 ml of Fisher BioReagents exACTGene Low Range DNA Ladder, with bands:weights(ng) of 2000:105, 1500:87, 1000:68, 750:59, 500:94, 300:27, 150:34, 50:25 was used. Product sizes for the 3 amplifications should be: A=flgK=161bp, B=flgL=220bp, C=307bp. 10ml of the PCR product was run on the gel.
Brief Conclusions:   The results certainly were not too bad for a first try. All the single-plex reactions worked and were the correct length and easily differentiable. The multiplex (as explained above this wasn't an ideal multiplex situation) reactions didn't fair as well. But one of the three tested worked. This is only a 2-plex reaction. Unfortunately the 3-plex reaction evaporated. I think another attempt and at the same time a run on cDNA should be attempted. This would be much better if I had a few operons to test so I didn't have to multiplex reactions that fight for the same template DNA initially. I'm not sure what the effect of increasing the genomic DNA and/or descreasing the primer concentration would have. It'd be more appropriate to get two known operons to add in the mix with this unknown one before delving into improving the protocol.

5.1.2  PCR tips

here are some PCR tips I pulled off of this excellent website.

5.2  Moving to a more realistic scale

Tue May 2 16:08:15 EDT 2006
I have to work on a large scale (i.e. > 24 PCR reactions) otherwise optimizing the multiplex PCR to follow size, location, and primer-dimer constraints is very difficult. I'm going to focus on small contiguous chunks of genomes. The first chunk of 19 genes is in Figure . Working in little chunks is helpful because it makes the primer sets somewhat self-contained (i.e. like a programming function is makes the project seem smaller, stresses the big picture, and localizes errors). Each chunk will occupy a 96-well plate. I haven't yet decided whether the multiplex set will come from combining chunks (to minimize steric constraints) or within a chunk. Making cDNA with random hexamers removes all but the most extreme steric problems (i.e. two amplifications that overlap), but those are trivial to prevent. The only time there might be a problem is with running the positive control on genomic DNA. Last, it is relatively easy to find syntenic regions of  19 genes across multiple organisms allowing the generation of primer sets that work on multiple organisms. Longer stretches run into problems that will have to be dealt with on a per species basis.
Description of Genome Chunk1   Tue May 2 16:08:15 EDT 2006 Genome Chunk 1 contains 19 genes: lpxL yceA yceI yceJ yceO solA yceP dinI pyrC yceB grxB mdtH rimJ yceH mviM mviN flgN flgM flgA
In Figure , known operons are designated by a gray rectangle background.
Pros: Cons:
Please see the pdf version for figures
Figure 5.3: Genes whose operon will be tested in the initial pilot study

5.2.1  Getting started

Sun Jun 4 17:27:43 EDT 2006
It was difficult to design multiplex primers using only one chunk in a 96-well plate. So I added a second chunk consisting of the following 11 genes: ydgD, mdtI, mdtJ, ydg, pntB, pntA, ydgH, ydgI, folM, ydgC, and rstA. In total the first plate of operon primers contains 30 genes and 28 gene-spans (the oligos amplifying from the end of one gene to the beginning of the next).
The plates were filled with each columning containing primers that are entirely compatible with each other (based on computational estimates). So each plate can potentially be plexed up to 8 reactions. The rules in designing the primers were:
  1. primers must work across W3110 (lab strain), EDL933 (O157:H7 pathogen), MG1655 (most common lab strain), and Sfl2457T (shigella very common 3rd world pathogen)
  2. primer size: 23+/-3
  3. melting temp 61+/-3
  4. GC content 50%+/-10
  5. remove first and last 20 base pairs (in case the ends of the genes were incorrectly annotated or unevenly transcribed in the cDNA reaction; making sure the primers work across four species also helps be certain that all the gene boundaries are good.)
  6. no self complementation or primer dimer formation across the pairs
  7. size between 75 and 650 bp
The rules for multiplexing were:
  1. 8-plex to fill up an entire column of a 96-well plate
  2. lengths of products must be at least 10% different so they can be differentiated on an agarose gel
  3. no primer dimers are allowed between any of the primers in the plexed reaction
The multiplex rules were satified by a greedy search for a set of primers that fit the rules. The search often does not converge to an agreeable solution (i.e. all primer sets are 8-plex), but the search is restarted until an agreeable solution is found (typically 5-30 tries). The software used for this was: primer3 for designing the pairs, ntdpal (basic alignment software that comes with primer3) for checking multiplex primer-dimer formation, and a couple perl scripts to glue the pipeline together.

5.2.2  Preping genomic DNA

Mon Jun 5 17:58:52 EDT 2006
I attempted to prep genomic DNA for use in the PCR experiments with very little yield. Up-to the lysis step worked very well. They lysate was clear after 1 hr. Previously the solution had been very viscous even after the prep was finished. I tried to clean it up a little by doing an initial ethanol precipitation followed by incubation with RNAse cocktail to the remove RNA yielding cleaner genomic, but the solution was so goopy that I had a hell of a time getting rid of it without out getting rid of my DNA.
Two preps were done yielding:
Sample DNA (ng/ul) 260/280 260/230
1 2.2
2 36.0
Brief Conclusions:   The yield was pathetic and handling the DNA in the goopy lysate was a real pain-in-the-ass. Tomorrow I'm going to try again but starting with 2 ml of culture (the original protocol) and going straight to a phenol:chloroform using gel-phase-lock, followed by addition of RNAse Cocktail, followed by a second phenol:chloroform extraction with a gel-phase-lock tube. Maybe if I want to increase the yield with the larger volume, I need to also increase the amount of TE I'm working with. I can double the concentrations and still fit it in a gel-phase-lock 2 ml tube. I just would have to use isopropanol precipitation. I plan on using isopropanol tomorrow anyways (that's what the original protocol suggests).
Brief Conclusions:   Tue Jun 6 17:01:03 EDT 2006
Just an update. I think a large part of the problem is that it takes forever for the genomic DNA to go back into solution. I rechecked today and there is more DNA in the tube not a huge amount but significantly more than before. Some of the genomic DNA can be seen in Figure 3.8 and again in Figure . The DNA was much too long for the type and voltage of the gel.
The yields are now:
Sample DNA (ng/ul) 260/280 260/230
1 28.2
2 40.0
I think the way to go might be to shear it by sonication after the lysis is complete. That would more represent the cDNA that I'm actually going to be using anyways.

5.2.3  Preping genomic DNA: 2nd try

Tue Jun 6 17:04:17 EDT 2006
Tried 3 samples. One without RNA digestion (sample 1). Two with RNA digestion in between the two phenol:chloroform extractions (samples 2 and 3). The third sample used double of everything (4 ml of culture, etc)
The yields as of Wed Jun 7 16:51:09 EDT 2006 (about 18 hours for resuspension) are:
Sample DNA (ng/ul) 260/280 260/230 total yield (ug)
1 327.3 167.7
2 82.6 41.3
3 117.4 58.7
All three samples were run on a 1% agarose gel (see Figure ).
Please see the pdf version for figures
Figure 5.4: Genomic DNA on a 1% agarose gel. The genomic DNA is too long and doesn't run properly. Samples 1 and 2 on the right are from the previous attempt at making genomic DNA. Sample 1 on the left did not have an RNAse digestion step (as can be seen by the large amount of RNA)
Brief Conclusions:   I'm pretty confident the genomic prep is working. I think working with genomic DNA sucks because it is goopy, slimy, sticky, hard to pipette, and hard to resuspend. HOWEVER, it looks like the prep is working. For an easier time, I would shear the DNA right after the lysis step. The RNA digestion also seems to be important.

5.2.4  Preping genomic DNA: with shearing

June 12, 2006
I ran the genomic DNA prep protocol but this time I sheared the lysate by sonication after the lysing step. The hope is that the shearing will make the prep easier and the resulting DNA should be more similar to cDNA created by random hexamers (i.e. shorter fragments not continous long pieces). Samples 1-3 were sheared at 20% power for 30 seconds, which resulted in a foamy mess. Samples 4-6 were sheared at 10% for 30 seconds, which worked much better (no foaming).
The yields as of Jun 13 are (samples in 300 ml total volume):
Sample DNA (ng/ul) 260/280 260/230 total yield (ug)
1 186.7 56
2 284.5 85.4
3 246.1 73.8
4 237.2 71.2
5 246.2 73.9
6 152.9 45.9
All 6 samples were run on an agarose gel (see Figure ).
Please see the pdf version for figures
Figure 5.5: Sheared genomic DNA on a 1% agarose gel.
Brief Conclusions:   The genomic DNA was easier to handle and sheared into a broad range. Shearing size was better when using 10% power and avoiding the foaming problem. Samples 4-6 will be used as the input DNA to further operon PCR reactions.

5.2.5  Preparing the oligos

The oligos were shipped at 200 uM in two plates (one forward one reverse). I'm going to dilute and mix the forward and reverse primers into a new plate. The mixing is:
This gives a final concentration of 1 uM for each primer. For a 5 ml PCR reaction and 150 nM of each primer, I just need 0.75 ml of the mix.

5.2.6  Realistic Scale: first experiments ® single-plex

All 58 primers are going to be tested in a single-plex reaction using genomic DNA. I'm going to use sample 2 of the genomic DNA prep on June 6th.

Single-plex first attempt

PCR run on Jun 7, 2008.
The plate is:
Single-plex oligo plate
- 1 2 3 4 5 6 7 8 9 10 11 12
A ydgH_106 mdtH_122 solA_77 yceI_191 yceO_75 yceJ_yceO_522 ydgD_mdtI_263 flgM_flgA_261 - - - -
B pntB_pntA_220 yceO_solA_329 mviN_166 flgN_268 mdtJ_ydgG_623_622 mdtI_174 yceJ_101 ydgI_folM_348 - - - -
C rimJ_yceH_165 flgM_98 yceH_112 mdtI_mdtJ_216 yceA_yceI_167 ydgH_ydgI_371 mviM_mviN_444 - - - - -
D dinI_94 pntA_461 grxB_201 grxB_mdtH_322 lpxL_yceA_466 yceA_424 flgN_flgM_178 - - - - -
E yceI_yceJ_121 yceP_dinI_509_512 ydgG_372 pyrC_137 solA_yceP_393 dinI_pyrC_316 mdtH_rimJ_337 - - - - -
F flgA_184 pyrC_yceB_363 ydgD_454 lpxL_420 mviN_flgN_341 yceH_mviM_235 folM_389 - - - - -
G yceB_grxB_266 mdtJ_170 ydgI_272 mviM_89 rimJ_299 ydgG_pntB_271 yceP_79 - - - - -
H ydgC_79 folM_ydgC_195 ydgC_rstA_235 pntB_111 rstA_230 yceB_197 pntA_ydgH_638 - - - - -
PCR Reaction composition
H2O 1.5 ml
NEB PCR Master Mix 5 ml
Forward and reverse primer(1.5 ml ) 150 nM
template DNA (2 ml ) 150 ng
Thermal cycler conditions
Initial denaturation30 sec 95° C
3-Step cycling
Denature:30 sec 95° C
Anneal:30 sec 54° C
Extend:120 sec 65° C
Number of Cycles:30
Final Extention: 5 min 70° C
Samples were run on a 2% gel Jun 8, 2006 (see Figure ).
Please see the pdf version for figures
Figure 5.6: 300 ml 2% agarose gel run at 4.8 V/cm in TAE. 1.6 ml EtBr. The lengths here don't correspond to the expected lengths. Something is weird here. Missing wells (less than column 8) are due to evaporation in the wells on the edge of the PCR plate.
Brief Conclusions:   Nothing (not even easy PCR reactions) work right the first time. The lengths of the products in Figure 5.6 don't correspond to the expected product sizes of the tested genes. I'm not sure why. Beyond this problem, the 96-well PCR plate also had some serious evaporation issues on the edges (see Figure ). Overall I think nothing can be concluded from this (except don't use the Costar 96-well plates next time).
Please see the pdf version for figures
Figure 5.7: The Costar 96-well PCR stuff just doesn't seem to work well on the edges of the plate. I think I need to move to BioRad (the maker of the PCR machine).

Single-plex second attempt

Jun 8, 2006
I switched to 8-well BD Falcon PCR-strips figuring that for sure this would fix the evaporation problems. WRONG!!! Just for the record, BD Falcon PCR strips SUCK!!! How hard is it to make a freaking plastic tube that stays sealed when a heated-lid is pushing down on it? Apparently it is difficult for the folks at BD Falcon. Same evaporation issues I had with the Costar plate occured with the stupid strips (no picture because in my anger I grabbed all the tubes and forcefully threw them away w/o thinking about a picture). I've never had this problem with a PCR strip before.
I lowered the amount of genomic DNA (because I'm running low) and the extension time (to speed things up).
PCR Reaction composition
H2O 1.5 ml
NEB PCR Master Mix 5 ml
Forward and reverse primer(2 ml ) 150 nM
template DNA (1 ml ) 75 ng
Thermal cycler conditions
Initial denaturation30 sec 95° C
3-Step cycling
Denature:30 sec 95° C
Anneal:30 sec 54° C
Extend:60 sec 65° C
Number of Cycles:30
Final Extention: 5 min 70° C




Single-plex third attempt

Jun 9, 2006
I switched to MP (Molecular BioProducts) PCR strips. And repeated the PCR for all the primers.
PCR Reaction composition
H2O 1.5 ml
NEB PCR Master Mix 5 ml
Forward and reverse primer(2 ml ) 150 nM
template DNA (1 ml ) 75 ng
Thermal cycler conditions
Initial denaturation30 sec 95° C
3-Step cycling
Denature:30 sec 95° C
Anneal:30 sec 54° C
Extend:60 sec 65° C
Number of Cycles:30
Final Extention: 5 min 70° C
Please see the pdf version for figures
Figure 5.8: 300 ml, 2% agarose gel with 1.6 ml EtBr. The primers all seem to work as designed. Unfortunately the ladder looks like hell.
Brief Conclusions:   Finally, all the primers amplified the correct size with no failed primer (Figure 5.8). The ladder in the gel is too blurry for the sizing I want to be able to do with the Versadoc software. I think I need to reduce the amount of agarose I use and run at a higher voltage (and maybe try TBE or SB buffer). It's time to multiplex!!!!!

5.2.7  Multiplexing chunks1 and 2

Jun 11, 2006
Of course, with my luck (or perhaps inattention to detail is a better word) I had a bug in my software for choosing multiplex sets. It was supposed to only allow product lengths 10% different from each other in the same well (so they could be differentiated on the gel). In general this was the case EXCEPT, the bug allowed an exception when the product lengths were almost exactly the same length (i.e. around 0% different). This is the worst possible bug since it drastically reduces the options for plexing this thing up.
Still the primers work and I've learned a lot so far and I can still learn from making the best of this situation. Looking by eye I came up with a way to cut the eight rows down to three (one 4-plex and two 2-plexes). The rows are:
a-plex = rows A, B, D, G
b-plex = rows C, H
g-plex = rows E, F
PCR Reaction composition
H2O 1.5 ml
NEB PCR Master Mix 5 ml
Forward and reverse primer(1.5 ml ) 100 nM
template DNA (2 ml ) 150 ng
Thermal cycler conditions
Initial denaturation30 sec 95° C
3-Step cycling
Denature:30 sec 95° C
Anneal:30 sec 54° C
Extend:60 sec 65° C
Number of Cycles:30
Final Extention: 5 min 70° C
Please see the pdf version for figures
Figure 5.9: 300 ml, 2% agarose gel with 1.6 ml EtBr. Using 1.0 mm comb at 6 V/cm. Here A is a-plex, B is b-plex and C is g-plex
I ran the remaining 6 ml (including dye) of sample to try and get a nicer gel by running it hotter. I ran them out for a long time to try and resolve the lanes with similar sized fragments (see Figure 5.9). I also used the versadoc software to detect the ladder and use it as a standard to estimate the length of each of the bands (see Figure ).
Please see the pdf version for figures
Figure 5.10: 250 ml, 2% agarose gel with 1.6 ml EtBr. Using 1.0 mm comb at 8 V/cm. Here A is a-plex, B is b-plex and C is g-plex
Please see the pdf version for figures
Figure 5.11: Multiplex results annotated by the versadoc.
Brief Conclusions:   The versadoc software isn't dead-on with the bp estimates (could be my fault if gel is slightly angled or could just be the imprecision in agarose gels). However, it is close enough (+/- 40bp) and the ratios of lengths within a well are good enough to computational know if there are bands of a particular known length or not. I think ALL of the multiplex reactions worked, which is great and all I could wish for. The only problem is that the mixes I picked by hand weren't the best. Columns 1 and 13 (A1 and A7) appear to have only three bands, but I think they are two bands of damn near the same size (A1: 94bp, 106bp; A7: 79bp and 101bp). If I had a clearer gel, I think the A7 should be resolvable (and if you look real close and use your imagination a little you can see them both already). I think running hotter with less agarose and less sample (Figure 5.10) helped out a bit (think slow gel = Figure 5.9. Certainly the move to the 1 mm comb cleans things up. It's only the smaller bands that are smearing. I think 7 ml total volume (including the dye), a 200 ml gel (or 180 ml but that's a pain to measure) will help more. Also need to try TBE and SB buffers and raising the temperature even more. Thankfully the PCR bands are much clearer than the ladder so all I have to do is clean up the thing that's normally the cleanest thing on the gel. Last, if I'm lowering the volume the Fisher dye isn't concentrated enough. I'll have to make the dye from DNA by mixing it with buffer. Also, I might try one of the more expensive low-bp agarose gels. But they are almost 2x the cost (making a gel can be almost $8)

Better looking, faster running 96-well multiplex gels

Sat Jun 17, 2006 A 25 ml multiplex PCR was run to allow different gels to be tried on the same DNA.
The set-up of the multiplex plate is:
2-plex oligo plate
- 1 2 3 4 5 6 7 8 9 10 11 12
A ydgH_106 mdtH_122 solA_77 yceI_191 yceO_75 yceJ_yceO_522 ydgD_mdtI_263 flgM_flgA_261 - - - -
pntB_pntA_220 yceO_solA_329 mviN_166 flgN_268 mdtJ_ydgG_623_622 mdtI_174 yceJ_101 ydgI_folM_348 - - - -
B rimJ_yceH_165 flgM_98 yceH_112 mdtI_mdtJ_216 yceA_yceI_167 ydgH_ydgI_371 mviM_mviN_444 - - - - -
ydgC_79 folM_ydgC_195 ydgC_rstA_235 pntB_111 rstA_230 yceB_197 pntA_ydgH_638 - - - - -
C dinI_94 pntA_461 grxB_201 grxB_mdtH_322 lpxL_yceA_466 yceA_424 flgN_flgM_178 - - - - -
yceB_grxB_266 mdtJ_170 ydgI_272 mviM_89 rimJ_299 ydgG_pntB_271 yceP_79 - - - - -
D yceI_yceJ_121 yceP_dinI_509_512 ydgG_372 pyrC_137 solA_yceP_393 dinI_pyrC_316 mdtH_rimJ_337 - - - - -
flgA_184 pyrC_yceB_363 ydgD_454 lpxL_420 mviN_flgN_341 yceH_mviM_235 folM_389 - - - - -
To make this plate from the single-plex plate you need:
New Row Old Rows
A A, B
B C, H
C D, G
D E, F
Mon Jun 19, 2006
The multiplex is working well, but the bands are all fuzzy. The 1 mm comb improved things, but now I'm trying new buffers that are more recommended for shorter DNA and can run at higher voltages without melting. In addition, the volume is being reduced to make a thinner gel which is clearer. The first gel was run on TBE. The second was run on SB buffer (this is a recently published buffer see ). Both gels were 180 ml 2% agarose gels run for 40 min at 250 V (10V/cm) with 1.6 ml EtBr.
Please see the pdf version for figures
Figure 5.12: Multiplex results on a TBE gel. 2% agarose, 1.6 ml EtBr. 10 V/cm (250 V total). 40 min. 6 ml PCR sample.
Please see the pdf version for figures
Figure 5.13: Multiplex results on a SB gel. 2% agarose, 1.6 ml EtBr. 10 V/cm (250 V total). 40 min. 6 ml PCR sample.
Brief Conclusions:   The good: both gels ran in 40 minutes, didn't melt, and were much clearer than the TAE gel (particularly the ladder) (see Figures 5.12 and 5.13. I changed a lot of variables (ran at a higher voltage, different buffer, thinner gel), so it is hard to know which variables increased the clarity of the gel and how much. However, the TBE gel was run near the maximum voltage for that buffer type while SB should be able to run much hotter (though the power supply we have will only go 50 V higher). The SB gel was also clearer than the TBE gel and SB is cheaper than TBE so that's probably my buffer of choice for now. The software based estimates of DNA length are closer with these clearer gels. The bad: The plex in well C1 is barely visible. I might need to increase the amount of sample I put on the gel.

5.3  Moving on to cDNA from a few conditions

Everything seems to be working now on genomic DNA. I can prep sheared genomic with no problems, I can get 2-plex reactions to work on the  60 primer sets I have for the two gene chunks, I have gels that run fast, have good resolution, and use a minimal amount of agarose. Now its time to really see if this thing works! Can this multiplex PCR method work on cDNA to differentiate the operons for these 20-30 genes being tested here???? Finally, a time for results that will add something to knowledge about the cell (even though it is still only a small set of genes).

5.3.1  Initial chosen conditions

Four conditions are being chosen around the two things I know will cause differential expression. 1) +/- norfloxacin will influence the DNA damage operon 2) +/- amino acids will induce the lrp regulated operon.
The four conditions to be tested are:
  1. LB
  2. LB with 75 ng/ml norfloxacin
  3. minimal Davis media with glucose (0.5%)
  4. minimal Davis media with glucose (0.5%) casamino acids (0.2%)
I'd prefer minimal media throughout, but I don't want to run to many experiments to start off with and using rich-undefined media like LB provides an easy way to really change the expression of a lot of genes relative to the minimal with glucose so hopefully, I'll really see a lot of genes in their on and off over the course of the 4 experiments. Minimal with casamino acids provides nitrogen and amino acids. LB provides more amino acids, peptides, and vitamins (from the yeast extract). There is also a pH difference between the buffered Davis media and the unbuffered LB media.

5.3.2  Growing cells and RNAprotect

5 ml of cells were grown overnite in LB. The next day 2 ml was washed in Davis media. 250 ml was placed into 25 ml of media (1/100 dilution) of one of the four media from section (5.3.1) (in a 125 ml baffled flask).
Below are the OD measurements (minus background) of the cells. Emphasized text is when the sample was taken.
Post-incubation time (hr:min) Sample1 OD Sample2 OD Sample3 OD Sample 4 OD
2:00 0.29 0.254 0.027 0.083
2:21 0.412 0.42
2:44 0.031 0.168
3:35 0.892 0.505 0.036 0.386
3:45 0.451
15:31 1.734 1.142 1.266 1.646
For samples 1, 2, 4, 2 ml of RNAprotect was added to a 15 ml centrifuge tube and mixed with 1 ml of culture. For sample 3, I mixed 400 ml of sample with 800 ml of RNAprotect. For all samples, I followed the RNAprotect instructions and the samples were placed in the freezer where they are supposely good for 2 weeks (i.e. I can wait two weeks before doing an RNAeasy prep). As the table above shows, sample 3 with glucose only was growing very slow. Next time I should use a 1/50 or a 1/25 dilution for this condition or start earlier. As it was, it was getting late and I decided to go with a late stationary sample the next morning. Hopefully, the lrp controlled amino acid genes will still be expressed.
Brief Conclusions:   I tracked the LB samples longer because I was afriad the norflox wasn't doing anything. With more time, it was clear it was doing something. As time went on the LB + norflox ended up being passed in OD by the LB only culture. Hopefully, the use of late stationary phase minimal + glucose won't mess anything up.

5.3.3  RNApreps and cDNA

Sun Jul 23 18:38:09 EDT 2006
Been busy with the network inference paper resubmission. But the RNAprotect says it's good for 2 weeks at -20° C. It's only been a week. I used the lysozyme + proteinase K protocol in the RNAprotect manual for the lysis step and then followed the RNAeasy protocol for the rest. I used 1 mg/ml lysozyme in TE and 10 ml of 20 mg/ml Proteinase K per ml of lysis solution. Yields after the RNA easy kit were a little low:
Sample RNA (ng/ul) 260/280 260/230
1 453.9
2 490.8
3 316.7
4 507.9
I want the RNA to be very clean (i.e. free of DNA) so I used the DNA-free kit from Ambion on the RNA samples after the RNApreps were finished (I used 35 min at 37° C incubation instead of 30 min). Yields dropped quite a bit more after using the DNA-free kit, either because you always lose stuff when you switch tubes or because there was a fair amount of DNA probably both. The yields are still plenty high to get a good amount of cDNA:
Sample RNA (ng/ul) 260/280 260/230 Amount (ml ) for 500 ng
1 322.6 1.55
2 138.0 3.62
3 107.1 4.67
4 161.2 3.10
500 ng of RNA was used in a Superscript II cDNA reaction according to the Invitrogen instructions. 100 ng of random hexamers and 10 mM dNTP were used. The RNA was placed in -80C and the cDNA was placed in -20C until the PCR reactions are done (probably tomorrow).
Brief Conclusions:   Everything looks good as far as the preps go. I still need to run a little on a gel and have a look tomorrow.
Brief Update Mon Jul 24 13:45:15 EDT 2006:   I noticed (a little late unfortunately), that Ambion also sells a newer product called TURBO DNA-free that is supposed to eliminate much more genomic DNA. If the DNA-free kit I tried isn't sufficient I'll should try that one.
Wed Jul 26, 2006
Finally got around to running those RNA samples on a gel. Ran 5 ml of each (so the total amount of RNA was different for each (Figure ).
Please see the pdf version for figures
Figure 5.14: 1% agarose, 0.5 ml EtBr. (100 V total). 60 min. 5 ml RNA sample
Brief Conclusions:   Things are looking a little sparse in Sample 3 (Figure 5.14. That is the stationary phase culture, so perhaps it's not too surprising. Even the rRNA bands are week though.

5.3.4  2-plex PCR on cDNA sample 3 (norflox)

July 24, 2006
I'm going to use sample 2 (LB + norflox) in a first test of the multiplex with cDNA. Genomic DNA will be used as a positive control. Sample 2 RNA only will be used as a negative control (to make sure the genomic contamination left in my prep is sub-detection by PCR and agarose gel-electrophoresis).
I'm using the 2-plex oligoplate (section 5.1 page pageref) for the + control samples and sample 2. For - control samples, I'm using just rows C and D from the single-plex plate (section 5.1 page pageref).
The positive control params (using genomic 4 from page pageref) were:
PCR Reaction composition
H2O 3.7 ml
NEB PCR Master Mix 5 ml
Forward and reverse primer(1 ml ) 75 nM
template DNA (0.3 ml ) 75 ng
Thermal cycler conditions
Initial denaturation30 sec 95° C
3-Step cycling
Denature:30 sec 95° C
Anneal:30 sec 54° C
Extend:60 sec 65° C
Number of Cycles:30
Final Extention: 5 min 70° C
For Sample 2, I added 4 ml of cDNA to the entire master mix (this is 4 ml of the 20 ml total cDNA reaction; into a master mix for 68 samples). For the RNA - control, I added 1 ml of RNA to the entire master mix (for 17 samples), which is still more RNA than is present in the cDNA mixture.
Please see the pdf version for figures
Figure 5.15: 2% agarose, 1.6 ml EtBr. SB buffer (250 V total). 45 min. 7 ml RNA sample; top two panes are + control on genomic DNA
Brief Conclusions:   It looks like there might be a little evaporation of the primers in my oligo plate that I keep in the fridge. I much prefer to store my working stock in the fridge, but perhaps in the future I should be more attentive to use it up faster. The gels, though not the most beautiful, do portray what I'd expect. Almost all the + control wells worked (those that didn't had evaporation issues) (Figure 5.15). Then the PCR reactions on the cDNA are more sparse, showing that not all genes (and operons!) are active (or connected). I will save all my gels to a file in the future, but as I'm going to be running more and more I'll probably not put them all in this document otherwise it'll take too much time and the real analysis in the near future is not going to be done looking at these gels (otherwise I'll never be able to scale up) but rather in the tab-delimited text file I export from the BioRad Versadoc software that gives me the estimated length of each band and the intensity.
Brief Update Sat Jul 29 20:55:11 EDT 2006:   The aluminum cover is definitely NOT sufficient to prevent evaporation; well A1 was completely dry and the other wells are much lower than they should be given the amount of primer I've used so far. I bought some polypropelyene mats (like how the primers from IDT come) that are supposed to work much better.




5.3.5  2-plex PCR on cDNA all samples

July 25, 2006
I ran 20 ml of all primers on on all 4 samples. 8 reactions were run on each of the RNA samples.
July 26, 2006
I ran an two SB gel with 7 ml of each sample . The gel was prestained with EtBr.
The raw data is: test
Please see the pdf version for figures
Figure 5.16: 2% agarose, 1.6 ml EtBr. SB buffer (250 V total). 45 min. 7 ml RNA sample; top two panes are + control on genomic DNA
Please see the pdf version for figures
Figure 5.17: 2% agarose, 1.6 ml EtBr. SB buffer (250 V total). 45 min. 7 ml RNA sample; top two panes are + control on genomic DNA
Brief Conclusions:   talk about results and dif between sybr and etbr, that I need to repeat fixing the dried out plate problem and with replication. 4 new conditions

5.4  cDNA from 4 conditions with replication

5.4.1  Growth and RNAProtect

Fri Aug 4, 2006
Due to problems with consistency across the SYBR gold and EtBr gels and what looked possible genomic DNA contamination in my RNA preps (only detectable with SYBR gold), I'm making the following modifications: (1) I'm running 3 replicates of each condition so I can better judge consistancy (2) I'm using the DNA-free TURBO kit which is supposed to be much much better than the previous DNA-free (not turbo) kit that I used to get rid of genomic DNA in my RNA prep. Regarding point (2), I also used the lengthier protocol where you add 1 ml of DNAase, incubate 30 min, add another 1 ml of DNAse, incubate another 30 min. This is 2x the DNAse I used before.
I modified the conditions slightly to hopefully give me a little more diversity in expression that the previous four:
  1. LB broth with 100 ng/ml norfloxacin (this is up 25 ng/ml from before); log phase
  2. Davis minimal media with 0.5% glucose (as in previous experiment BUT in this experiment I sampled them in log phase)
  3. EC broth; log phase
  4. EC broth and 100 ng/ml norfloxacin; stationary phase (they had kinda a weird stationary phase at a very low OD; I think 100 ng/ml norfloxacin was pretty harsh so they could only reach around OD 0.2 (see table below and Figure )
An overnite culture was washed in Davis minimal media and used as a starting culture into the 25 ml media volume. Cells were grown in 250 ml baffled flasks (I wanted to use 125 ml baffled flasks but all of them were dirty). For the minimal media cultures I used a 1/25 dilution from the washed overnite. For the other conditions, I used 1/100.
Growth with time is shown below. Samples taken are shown in italics.
E. coli growth for 4 conditions 3 replicates
Time min 1 2 3 4 5 6 7 8 9 10 11 12
13:50 70 0.063 0.056 0.07 0.08 0.076 0.08 - - - - - -
14:20 100 - - - - - - 0.095 0.064 0.099 0.109 0.099 0.071
14:50 130 0.191 0.188 0.226 0.09 0.07 0.084 - - - - - -
15:15 155 - - - - - - 0.212 0.157 0.235 0.177 0.186 0.202
15:40 180 0.385 - 0.402 0.092 0.078 0.085 - - - - - -
15:51 191 0.418 0.403 0.467 - - - - - - - - -
16:00 200 - - - - - - 0.412 0.323 0.457 0.234 0.26 0.237
16:15 215 - - - - - - 0.533 0.381 0.523 - - -
16:50 250 - - - 0.117 0.114 0.114 - - - 0.234 0.239 0.312
17:30 290 0.405 0.31 0.325 - - - - - - 0.204 0.222 0.377
17:45 305 - - - - - - - - - 0.195 0.191 0.398
17:50 310 - - - - - - 0.81 0.868 0.858 - - -
18:00 320 - - - 0.193 0.193 0.187 - - - - - -
18:40 360 - - - 0.239 0.271 0.265 - - - - - -
19:10 390 0.3 0.213 0.256 - - - 1.028 1.03 1.02 0.163 0.159 0.621
19:25 405 - - - 0.399 0.4 0.386 - - - - - -
19:30 410 - - - - - - 1.081 1.072 1.027 0.154 0.145 0.637
19:50 430 - - - 0.42 0.473 0.461 - - - - - 0.75
21:51 (next day) > 800 - - - 1.598 1.524 1.335 - - - - - 1.85
For the EC and LB+nor conditions I added 1 ml of culture to 2 ml RNAprotect. For the Davis condition, I used 1.5 ml of culture and 3 ml of RNAprotect. For the EC+nor sample (which wouldn't reach a high OD), I used 2ml of sample in 4 ml of RNAprotect
Please see the pdf version for figures
Figure 5.18: Growth curve for the 12 samples to be used for operon determination
Brief Conclusions:   So far so good it seems; I'm a little worried that the EC+nor cells are too sickly. I will know more after looking at the RNA on a gel. Also, the EC norflox sample 3 (see Figure 5.18) grew much better than the other two. I only added 2 ml of my stock norfloxacin to reach the 100 ng /ml . I should probably have diluted it a bit. I'm afriad that the difference in growth is likely do to my pipetting error, since it was very hard to get the little pipette into the bottom of my 250 ml flask. THe growth curves aren't the best, since I didn't sample often or long enough, but that really wasn't the point. They're good enough to get the idea that norfloxacin really slows them down at 100 ng/ml and even more so when there are bile salts in the media (for the EC broth).

5.4.2  RNA preps, DNAse digestion and cDNA

Sat Aug 5 21:07:10 EDT 2006
Samples were randomized before RNApreps as:
Condition Original ID Randomized ID
LB and 100 ng/ml nor 1 12
LB and 100 ng/ml nor 2 2
LB and 100 ng/ml nor 3 6
Davis and 0.5% glucose 4 5
Davis and 0.5% glucose 5 11
Davis and 0.5% glucose 6 8
EC 7 4
EC 8 1
EC 9 10
EC and 100 ng/ml nor10 3
EC and 100 ng/ml nor11 7
EC and 100 ng/ml nor12 9
This randomization should prevent biases in miniprep, PCR, or gel order from being systematic (as opposed to just random noise). I'll use the randomized order from here on.
For the RNAeasy preps I used 1 mg/ml lyzozyme in TE with 10 ml of Proteinase K (20 mg/ml) for each 100 ml of lysozyme. I had a little problem with the pellets not dissolving easily (they did eventually). Maybe I let the lyzozyme stay with the proteinase K too long before I started (apprx 10 min) and it chopped up the lyzozyme?
I prepped all 12. Then I used Ambion's DNA-free TURBO kit (got a free sample one supposed to work much better than the original DNA-free kit). I used 1 ml TURBO DNAse incubated 30 min. I added additional 1 ml TURBO DNAse and incubated an additional 30 min.
Yields for the 12 samples, post-DNAase digestion are:
RNA prep yields post DNAse digestion
Sample ID ng/ml A260 260/280 260/230 Constant ul in 1 ug
1 98.2 2.455 40 10.2
2 112.67 2.817 40 8.9
3 280.49 7.012 40 3.6
4 675.74 16.894 40 1.5
5 362.49 9.062 40 2.8
6 451.65 11.291 40 2.2
7 222.91 5.573 40 4.5
8 370.16 9.254 40 2.7
9 532.62 13.315 40 1.9
10 424.56 10.614 40 2.4
11 287.9 7.198 40 3.5
12 373.7 9.343 40 2.7
Add gel of the RNA
I made cDNA with 1 ug of RNA for each sample (final RNA conc in the cDNA is appx 50 ng/ml ), this is 2x the amount I used last time, so hopefully this will improve yield a little on the PCRs. I also made an RNA + water negative control sample (the cDNA minus mix) made 30 ml of 50 ng/ul solution. I will used this sample RNA sample for running a negative control PCR.

5.4.3  384-well qPCR

Mon Aug 7 11:55:35 EDT 2006
I'm going to use the new 384-well block on our Icycler PCR machine. I'm using a total volume of 14 ml . Since there are 4 x 8 samples, I can run 12 samples per plate like so:
Layout for 384-well PCR of operon samples
- 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
A 1 1 1 1 3 3 3 3 5 5 5 5 7 7 7 7 9 9 9 9 11 11 11 11
B 2 2 2 2 4 4 4 4 6 6 6 6 8 8 8 8 10 10 10 10 12 12 12 12
C 1 1 1 1 3 3 3 3 5 5 5 5 7 7 7 7 9 9 9 9 11 11 11 11
D 2 2 2 2 4 4 4 4 6 6 6 6 8 8 8 8 10 10 10 10 12 12 12 12
E 1 1 1 1 3 3 3 3 5 5 5 5 7 7 7 7 9 9 9 9 11 11 11 11
F 2 2 2 2 4 4 4 4 6 6 6 6 8 8 8 8 10 10 10 10 12 12 12 12
G 1 1 1 1 3 3 3 3 5 5 5 5 7 7 7 7 9 9 9 9 11 11 11 11
H 2 2 2 2 4 4 4 4 6 6 6 6 8 8 8 8 10 10 10 10 12 12 12 12
I 1 1 1 1 3 3 3 3 5 5 5 5 7 7 7 7 9 9 9 9 11 11 11 11
J 2 2 2 2 4 4 4 4 6 6 6 6 8 8 8 8 10 10 10 10 12 12 12 12
K 1 1 1 1 3 3 3 3 5 5 5 5 7 7 7 7 9 9 9 9 11 11 11 11
L 2 2 2 2 4 4 4 4 6 6 6 6 8 8 8 8 10 10 10 10 12 12 12 12
M 1 1 1 1 3 3 3 3 5 5 5 5 7 7 7 7 9 9 9 9 11 11 11 11
N 2 2 2 2 4 4 4 4 6 6 6 6 8 8 8 8 10 10 10 10 12 12 12 12
O 1 1 1 1 3 3 3 3 5 5 5 5 7 7 7 7 9 9 9 9 11 11 11 11
P 2 2 2 2 4 4 4 4 6 6 6 6 8 8 8 8 10 10 10 10 12 12 12 12
Brief Conclusions:   Thu Aug 24 19:55:43 EDT 2006
I haven't been keeping detailed notes of my results lately. In summary, I've gotten the PCR volume down to 8ml with no evaporation problems (I prepare 4 extra reactions (i.e. for each 32 gene set: one condition; I prepare enough for 36 genes) for pipetting error so I don't run out of master mix. For these reactions I use 0.8 ml primer (100 nM), 4 ml master mix, 3.2 ml H2O . For each 36 sample set, I use 2.3 ml of cDNA - total.
I ran 10 of the conditions for all 32 genes, each with a negative control RNA only sample (see the gelResults in binary format). There are some problems with genomic contamination, see the binary format data and Figure .
Things I learned: (1) you can run a SB buffered owl large gel for 40 min at 300 V (12V/cm) for nice separation and relatively fast running time. SB buffer can be used at least 4 times with no noticable effect in gel quality. If SB buffer sits in the rig for more than a couple days it gets a cob-web, white moth-ball kinda substance in it (I always switch buffers when this happens). (2) I seem to be able to amplify most of the genes and a significant part of the operon-spanning fragments (3) EtBr is easier to work with than SYBR the SYBR gold stained stuff shows too much and and it is hard to tell the difference between PCR crap and real signal; any time there really was a real signal I could also see it in the EtBr gel (see Figure ).
Problems I encountered: (1) the PCR machine must not be consistent across the wells. The left half of the gel is always has faint amplification (and therefore faint bands) relative to the right. If I load the samples in the gel in the opposite order the gradient goes the other way, so it seems the problem is not related to the gel itself. (put gel figure and reversed gel figure). I called Bio-Rad and they sent a replacement heat block. I tried again today with the new block and the problem doesn't seem to have been fixed. My only hope is to try the block in the other PCR machine. (2) There was some amplification of the RNA-only control. So the TURBO DNA-free kit was not sufficient to remove all the genomic contamination. (3) I've read a few papers on transcript read through, and it seems like it is pretty frequent for intrinsic terminator sequences (though the papers are based on only a few examples, the almost always find read through). I'm afriad this makes it almost impossible with gel-based methods to determine if the operon spanning region is amplified because the gels are being transcribed as a single unit or because the first gene is reading through its terminator sequence sometimes. The problem might be overcome by qPCR, but then scaling up is impossible as qPCR master mix would be prohibitively expensive. I plan to try and do paired-end tag (PET) sequencing, hopefully using a highly parallel sequencing method. Now I'm probably going to move this project towards a small study on these 60 genes and intergenic regions using qPCR perhaps with absolute quantitation. One interesting thing would be to use qPCR, microarrays, and sequencing on the same 4 conditions and see how the quantities match up. But for now I'm going to try qPCR. The cloning and preparation of cDNA for sequencing will be reported in its own Chapter. By sequencing PETs you determine both the start and end of a transcript, which if I can get long cDNA (close to full length hopefully) will provide a real nice way to determine operons.
Interesting things: I seem to be able to amplify very often (see binary expression file) a region spanning the 5' regions of two genes that are on opposite strands, here there shouldn't be a transcript read through problem, so it could be any interesting result. Perhaps and interesting direction for this work would be to only design primers for genes on opposite strands that are 5' to 5', and see how often this occurs. I'm going to try and make a few RNA samples that have more negligible amounts of cDNA and test these regions again to make sure this is a real thing I'm finding and not some genomic contamination kinda trick. If it's real, I'll probably buy a few primers for some other reginos and try this out in a separate chapter of my lab notebook. Could do a Northern blot to try to determine how long the RNA fragments are that are in these weird regions.
Please see the pdf version for figures
Figure 5.19: 2% agarose, 1.6 ml EtBr. SB buffer (250 V total). 45 min. 7 ml RNA sample; top two panes are + control on genomic DNA

5.5  gDNA free RNA

test