From OpenWetWare

Jump to: navigation, search

iGEM 2008

Home        People        Project        Protocols        Notes        Ideas        Changes       


Engineered Gut Microbiota

Josh K. Michener 20:49, 4 June 2008 (UTC): Okay, my grand (partly) unified (and only partly feasible) vision: We have the cells metabolizing lactose and feeding that carbon into central metabolism. The excess flux is diverted into folate biosynthesis (rather than lactic/acetic acid production). The folate is periodically liberated from cells by inducing prophage expression - the prophage lyses the cell and as a side effect releases the vitamins. I still can't work Doug's project into it, though.

  • Cbeisel 04:33, 16 June 2008 (UTC):How do we envision our engineered gut bacteria will be used? One idea is distributing it in yogurt and baby formula. Thinking about it from a marketing standpoint may help develop the final story.


  • Vitamin choice: General Reference
    • Folate: 400-600ug RDA. Naturally produced by E. coli. Six enzymatic steps. Decent bioavailability (not enough for your whole RDA, but a measurable contribution). I haven't found any indication of the rate limiting step.[1, 2, 3, 4, 5, 6]. Folate deficiency in pregnant women can cause birth defects.
      • Victoria Hsiao 10:13, 11 June 2008 (UTC)In Zhu et al [6], they cited a paper that suggested folK was the rate limiting enzyme (pyrophosphokinase).
      • Victoria Hsiao 10:13, 11 June 2008 (UTC)What are the differences between the pathway in e coli and that in L.lactus or B.subtilis? I'm having a hard time finding any papers that have the folate synthesis pathway in e coli.. I found this site that has the pathway mapped out, [1], but would the order that the enzymes appear in correspond to the order that their corresponding genes need to be in in the gene cluster?
      • Victoria Hsiao 10:13, 11 June 2008 (UTC) In Sybesma et al [7] they suggest that overexpression of folKE and folC is best for the intracellular accumulation of folate in L.lactis. folKE(2-amino-4-hydroxy-6-hydroxymethyldihydropteridine pyrophosphokinase & GTP cyclohydrolase I), folC (polyglutamylfolate synthetase). They also noted that overexpression of GTP cyclohydrase I can be effective for increasing flux since it is the first enzyme in the pathway.
        • Josh K. Michener 19:52, 11 June 2008 (UTC): Look at Bermingham et al. [3] Figure 1 for the E. coli folate biosynthesis pathway. Keep in mind that it introduces PABA and glutamate without explicitly listing their synthetic pathways. I think the subtilis folK is the coli HPPK. If I read it right, mtrE is the same as folE, which in coli is GTPCH1.
      • Victoria Hsiao 09:59, 15 June 2008 (UTC) Wegkamp et al used the folate gene cluster from L.lactis to turn a folate consuming bacteria into a folate producing one! [8]
        • Victoria Hsiao 09:59, 15 June 2008 (UTC) I've found the corresponding genes for E. coli, but maybe we should also use the L.lactis gene cluster since it would be less regulated?
      • Victoria Hsiao 09:59, 15 June 2008 (UTC) The same group also characterized the pabA,pabB, and pabC genes from L.lactis and determined and that are 3 are important and necessary in the production of folate. [9]
    • Beta carotene: Need 3-6mg per day[2]. Unclear how much the Keasling lab made, since it's all reported as relative production[10].
      • Cbeisel 13:08, 2 June 2008 (EDT):We currently have some of the plasmids used in the paper, so it shouldn't be difficult to initially produce Beta carotene and begin optimizing production.
      • Victoria Hsiao 17:08, 4 June 2008 (UTC)Just to make sure I understood the paper correctly, they were able to vary the production rate of the beta-carotene by changing the length of the hairpin 5' of the crtI?
      • Josh K. Michener 21:58, 4 June 2008 (UTC): They were changing the relative stability of the mRNAs, which changes the relative expression of the genes in the operon.
    • Biotin: 30 ug RDA. The Liao lab got 100ug/L in an engineered pathway[11]
      • Victoria Hsiao 21:47, 3 June 2008 (CDT)From Wikipedia: "Deficiency is extremely rare as intestinal bacteria generally produce an excess of the body's daily requirement. For that reason statutory agencies in many countries (e.g. the Australian Department of Health and Aging) do not prescribe a recommended daily intake." [3] Would that make biotin production not as relevent a concern?
      • *Cbeisel 21:53, 3 June 2008 (CDT):Based on the wikipedia reference, biotin overproduction would not be an especially worthwhile pursuit. It sounds like you can focus on other vitamins.
    • Vitamin D: You probably get enough from sunlight[4]
    • Vitamin E: 10mg RDA, which is high. I haven't found any bacterial pathway reported.
    • Vitamin K - natural model[12]
  • How do we get the vitamins out - secretion, or lysis?
  • Lysis
    • --Victoria Hsiao 05:14, 2 June 2008 (EDT)In this article they used both the expression of lysis gene E and expression of staphyococcal nuclease in E coli to lyse the bacteria. [13] The gene E expression was thermally induced (they kept the bacteria at 24C to repress E, and then raised the temperature to 42 to induce expression). Since 37C(body temp) is closer to 42 then 24.. temperature probably isn't the best switch. For the 2nd type of lysis they used staphylococcal nuclease A (SNUC), which they induced chemically with IPTG (isopropyl-β-d-thiogalactopyranoside)
      • Cbeisel 13:58, 2 June 2008 (EDT):From the paper [13], it appears that expression of lysis gene E w/ or w/o SNUC results in the formation of bacterial ghosts characterized by a relatively intact cell membrane and a cavernous intracellular space. Looking online, it appears bacterial ghosts were being developed as vaccine delivery vehicles [14] -- interesting stuff. For us, the intact membrane could be problematic if Beta carotene localizes here, but maybe not. Any lysogenic factors that can effectively break down the cell membrane?
    • Victoria Hsiao 05:14, 2 June 2008 (EDT)Ideally I guess we would want the E coli to lyse once the vitamins in the cytoplasm have reached a certain concentration.
      • Victoria Hsiao 05:14, 2 June 2008 (EDT)Is lysis gene E naturally present in E coli?
        • *Cbeisel 14:05, 2 June 2008 (EDT):It appears lysis gene E comes from the phage ΦX174, so it shouldn't be naturally expressed.
      • Victoria Hsiao 05:14, 2 June 2008 (EDT)How easy is it to implement a concentration switch?
      • Victoria Hsiao 05:27, 2 June 2008 (EDT) Actually, if we just make it so that the e coli are unable to stop producing the vitamin, is there some way we can make the cell burst on its own due to osmotic pressure? We would have to deal with the petidoglycan layer between the inner and outer membranes. In this study, [5] they pretreated their cells w/ lysozyme to break down the peptidoglycan.
        • Victoria Hsiao 05:41, 2 June 2008 (EDT)Can we make something like: High vitamin concentration induces lysozyme expression, which leads to lysis via osmotic pressure.
        • Josh K. Michener 13:57, 2 June 2008 (EDT): Look at this [15, 16].
          • Victoria Hsiao 16:30, 4 June 2008 (UTC)If the lysis is triggered automatically by low glucose concentration, would high vitamin concentrations eventually lead to low glucose concentrations by crowding out the glucose, or would that take way too long? Is pBlueLysis+ (from Yun et al) something that we'd be able to get?
          • Victoria Hsiao 16:36, 4 June 2008 (UTC)If the lysis is triggered by IPTG, we'd have to make the e coli produce IPTG as well...
        • Josh K. Michener 14:45, 2 June 2008 (EDT): Or, could we trigger prophage expression and use the phage to liberate the vitamins?
        • Cbeisel 14:48, 2 June 2008 (EDT): Designing a concentration switch may be really difficult. An alternative is to identify allosteric activators or repressors used in the regulatory scheme for natural vitamin production. If one recognizes our vitamin, we could use it to trigger cell lysis. If we can't find any, how else could you ensure (or hedge your bets) that cell lysis occurs after saturating vitamin levels are produced?
          • Victoria Hsiao 16:36, 4 June 2008 (UTC)Could we just make it a timing thing, and not have a switch at all? i.e the lysis gene and the vitamin production gene are both constantly being expressed, but at a rate such that the cell does not actually lyse until a time when we're pretty sure the bacteria has already produced a significant amount of vitamin. So then we'd just have to figure out the rate of vitamin production, and then engineer the rate of lysis gene expression to match that time. Could we use the hairpin approach used in this paper [10] to match the relative production rates?
        • Josh K. Michener 15:26, 2 June 2008 (EDT): Looks like there's a biotin sensor[17]. Still looking for a folate one.
  • Josh K. Michener 17:17, 9 June 2008 (UTC): As a fall back plan - do we necessarily need to sense when to trigger lysis? At steady state, the vitamin concentration is only going to vary by a factor of two (just after division to just before division). So we could randomly trigger lysis and get pretty good release.
    • Victoria Hsiao 10:47, 11 June 2008 (UTC) In this paper, Templin et al [6] deleted the gene ldcA which encodes a cytoplasmic L,D-carboxypeptidase. The mutant with the ldcA deletion was unable to make the pentapeptide precursor needed to recycle the murein(peptidoglycan) in the cell wall) and instead steadily accumulated tetrapeptide in the cytoplasm. As a result of the weakened cell wall, this lead to spontaneous autolysis during the stationary growth phase. This sounds promising, as long as most of the folate production occurred prior to reaching the stationary growth phase. Although i don't know how we'd ensure that enough bacteria lived on to maintain a stable population. Also, ampD is the gene that controls the murein recycling, so maybe we could also do something that that; when they deleted both ldcA and ampD, the e coli did not produce the mutant tetrapeptide.
      • Josh K. Michener 20:01, 11 June 2008 (UTC): Question - do we think that the cells would reach stationary phage in the gut? Or, more correctly, would they have sufficient stress to cause autolysis?
      • Josh K. Michener 20:01, 11 June 2008 (UTC): It looks like we have several options for lysis - prophage, the T4 holin, the T7 lysozyme, the ldcA deletion. I'd probably try several options.
        • Josh K. Michener 23:02, 12 June 2008 (UTC): This paper [7] looks at expressing the holin by itself, and sees nearly complete lysis within 30 minutes.
      • *Cbeisel 00:28, 13 June 2008 (UTC):Here's a paper that shows 30-60 min lysis when various holin variants are expressed from a plasmid [18][8]. At first glance, it appears we can make point mutations or truncations that increase the lysis time...assuming that's of any use to us.

Lactose intolerance

  • Secreted or intracellular?[19, 20]
    • Robert 22:18, 22 May 2008 (EDT): For article 7, does B. longum contain the beta-gal gene? Or are they testing another way to reduce lactose by using B. longum? I googled the strain and I couldn't find much about it, but that it does ferment sugars into lactic acid, so I am guessing it has the gene, or one of similar kind.
      • Josh K. Michener 23:40, 22 May 2008 (EDT): From the paper: "We hypothesized that bifidobacteria could exert a positive effect on lactose digestion because of their substantial &gal activity (6, 16)."
      • Robert 23:56, 22 May 2008 (EDT): Thank You! I see it now.
  • What's the proximate cause of the lactose intolerance phenotype?
    • Usually, lactose is split into glucose and galactose by lactase present in the villi of the small intestine. If lactase is not present, the lactose continues into the colon, where it is metabolized by gut flora, resulting in in vivo fermentation. Wikipedia
      • Josh K. Michener 12:25, 19 May 2008 (EDT): Lactose causes two problems: first, metabolism into H2 (and possibly further into methane) or it changes the osmolarity of the gut and causes diarrhea [9]. So there are a number of directions we can attack this from - consume H2, do a better job of making lactase, digesting the glucose and galactose first, importing and sequestering the lactose, etc. The natural pathway seems to be sugar->lactate/acetate->H2->methane. So if we can get a good aerobic culture going, that might be enough.
  • Our lactase-producing bacteria would have to somehow outcompete the lactose-metabolizing bacteria in the colon.
    • It would at least have to get to the lactose first, so that the other bacteria don't get a chance to ferment it.
    • Or it could produce so much extracellular lactase that any lactose entering the colon would immediately be broken down.
    • Would we want our bacteria to live in the small intestine?
    • Robert 14:27, 29 May 2008 (EDT): Do we know the strain, or strains involved in metabolizing lactose? It would be interesting to see what would happen if we block or stop metabolizing lactose and block off the strains that convert this chain of events from lactose -> release of H2 -> methane, etc. In any event, I would think this is much harder. As mentioned above, if we have our strain live in the small intestine, they would hopefully come in contact with the lactose before the other strains, so if we can break it down at the small intestine, this would hopefully resolve our problem?
      • Josh K. Michener 14:42, 29 May 2008 (EDT): One of the big advantages of working with the Nissle 1917 strain is that we don't have to engineer it to live in the gut. The downside is that we probably don't want to try tinkering with that. Bacterial counts in the small intestine are very low - most things live in the large intestine. So we probably have to compete directly with the other gut bacteria rather than pre-empting them by going to the small intestine.
    • *Cbeisel 14:55, 2 June 2008 (EDT):It may be possible to reengineer the regulation of lactose transport such that our cells will be primed to quickly uptake lactose before the other gut microbes. We'd still have to figure out what our bacteria would do with the lactose (ie. hopefully not ferment it), although expedited uptake may be a start. The ref from Doug's 'healthbot' presentation may be a good start [21].
      • Robert 15:21, 2 June 2008 (EDT): I don't know if it's my computer, but I still can't access it. Is anyone else having this problem? If so, can you e-mail me the article? *Cbeisel 15:35, 2 June 2008 (EDT):Link fixed. Apparently capitalization matters....
      • Robert 17:27, 3 June 2008 (CDT):I was reading on p. 946 in the section Regulation of Carbon Metabolism in Gram-Negative Enteric Bacteria, and they share an idea that might help our strain. They showed ptsHI mutants, more specifically, mutations in the crr gene, which encodes for EIIA(GLUC) allowed the strain to grow on non-PTS carbon sources, ours being lactose. It also made it not able to grow in the PTS carbon sources, such as glucose. Later, they discussed the phosphorylation and dephosphorylation of EIIA(GLUC) and P~EIIA(GLUC). Unphosphorylated EIIA(GLUC) can bind and inhibit our lacY protein, along with other non-PTS carbon sources. Is it possible to block the binding of EIIA(GLUC), or keep EIIA(GLUC) ALWAYS phosphorylated?
      • *Cbeisel 18:24, 3 June 2008 (CDT):You have the right idea. EIIA(GLUC) will bind lacY, thereby inhibiting transport when glucose is present. Mutating or deleting EIIA would allow uninhibited transport of lactose, although the cells can't uptake glucose. Another option is mutating lacY to prevent the inhibitory binding interaction. It appears a few groups have identified point mutations in lacY that abolish EIIA inhibition without affecting lactose transport. This paper is a good start [22] and may be a simple solution to this aspect of lactose intolerance.
    • Robert 18:42, 4 June 2008 (UTC): I was looking for enzymes, to convert H2 or methane, with the sites Doug shared w/ the team, and the only enzyme I found was the one he pointed out in his e-mail, methane hydroxylase. I am wondering if this approach using enzymes is possible? Also, for the reaction with methane hydroxylase, methanol is produced. Would methanol cause any other problems? I tried google and pubmed, but I wound up empty.
      • Josh K. Michener 20:45, 4 June 2008 (UTC): If I understand correctly, you're proposing to use a methane monooxygenase (MMO), correct? That's basically impossible to express in E. coli - it's a multi-component system from an archaea. We might be able to use the hydrogen, but methane is probably impossible.
        • Robert 20:49, 4 June 2008 (UTC): Oh ok. I really wasn't sure about how we would use enzymes that convert methane and H2 as one of the possible ways to target lactose intolerance. Should I disregard that goal?
          • Josh K. Michener 20:50, 4 June 2008 (UTC): I'd see if you can find a way to use the hydrogen.
      • Robert 00:56, 15 June 2008 (UTC):For glucose uptake in the large intestine (in a canine), one paper showed that when adrenaline is present, there is a 150% increase in glucose uptake. The main idea is glucose uptake takes place in the large intestine. The numbers they found were 28.28 ± 20mg/min without adrenaline present.Glucose uptake in dogs
        • Josh K. Michener 01:05, 15 June 2008 (UTC): From some poking around, it looks like milk is about 5% lactose (50mg/mL)[10]. So an 8oz glass of milk has ~12g of lactose. At the dog's rate (keep in mind humans would be faster), that would take 6.7 hours to absorb. Can we find a ratio of the surface area of a dog's large intestine to a human?
      • Robert 01:55, 15 June 2008 (UTC):Here is an approximate of the human intestine. It's about 1.5m long, and a diameter of about 2.5 inches. After calculation, SA is appx. 3000 sq cm. Although in this paper, Rat Microvilli showing that the microvilli of a rat's intestine enlarges the SA by as large as a factor of 20, so 6 sq meters is a rough estimate for a human. For dogs, an appx length of a regular sized 40 lb dog is 16 inches. If we assume both the dog's and human's large colon are proportional, it's about 1/4 the size.
        • Josh K. Michener 04:09, 15 June 2008 (UTC): Okay, that gives us ~100 minutes to absorb the lactose from a glass of milk. So our timescales are similar (though it would be nice if absorption were a little faster).
      • Robert 02:10, 15 June 2008 (UTC): One question I will try to research is how much lactose would cause producing methane gas, and stomach pain. A few lactose intolant people I know are not able to consume milk, but they are able to consume yogurt with no problem. If we can reduce the amount of lactose being metabolized and converted into H2 and Methane, to an amount that doesn't produce enough methane to have an effect, would this be reasonable?
      • *Cbeisel 22:32, 15 June 2008 (UTC):Keep in mind that we intend to release B-gal into the gut to quickly break down lactose into glucose and galactose. Can you find anything on what an excess of glucose in the large intestine does? As long as B-gal breaks lactose down fast enough, gut adsorption may be significant enough to prevent major side effects. As a side note, feeding our gut microbes can be a good thing [11] as long as it happens on a somewhat regular basis.
        • Robert 05:00, 16 June 2008 (UTC):I was not able to find anything about excess glucose in the large intestine, but I found a lot of pages about diabetes. I'm going to look at what causes diabetes and see if the excess glucose will have any effect (I don't know much about diabetes but I'll present what I find at our meeting tomorrow).
  • Adaptation [23]

Prophage targeting other bacteria

  • lamB is sufficient for lambda infection.
  • Looking at other receptors
  • SPO2 is a lysogenic subtilis phage
  • How do we get the phage inside the cell? Take a resistant strain (coli w/o lamB for lambda, coli for a subtilis phage), add the necessary receptor on a plasmid. Infect, select for lysogens. Grow under nonselective conditions, then counterselect for loss of the plasmid. Voila. Tetracycline can be counterselected[24].
  • Works in cows[25]
  • We need to select a suitable phage for the project. For convenience, here is a list of phages we can order from atcc: [12]
  • SPO1, Phie, and SP8 all belong to the same family [13].

They have linear, double stranded DNA however, it appears that they have thymine replaced with hydroxymethyluracil. It is not well characterized how well this works with cloning...

  • SPP1 looks promising even though it is not a propahge. It possesses double stranded linear DNA. Furthermore, this paper [14] describes creating a fusion virus of SPP1/lamda. They can infect E. Coli, and express both lamda and SPP1 genes within E. Coli. This shows some promise and offers new ideas. Perhaps we can construct a fusion phage?
  • SPP1 complete genome can be found here:[15] and restriction mapping can be found here: [16].
  • We need to transform large plasmids, transformation efficiency generally goes down, however, electroporation conditions can be used to compensate. [17]

Population Variation

  • Slipped-strand mispairing (SSM)[26] can produce a heritable variation in the expression from a promoter. Roughly one in 1000 divisions produces a change in expression. Couple this expression to a selectable/counterselectable marker. Under any given condition (selection, say), the population thrives, but with a small group of the opposite phenotype (non-expressing). Switch conditions (to counterselecting), and the population can use these revertants to recover. The switching is stochastic by nature and can be directly compared to both natural [27] and synthetic [28] systems that utilize stochastic switching to adapt to variable and fluctuating environments.
  • Also use FimE (below)?
  • Allen Lin 16:04, 29 May 2008 (EDT): To make sure that I'm understanding this, the promoter goes between the two fim recombination sites, and we can place one gene downstream of the switch, and other gene upstream of the switch, but on the opposite strand. The cell would then have two states. The cell would then switch from one state to the other when FimE is active. So unless we add FimB, a cell that has switched to the second state can't switch back, but it doesn't need to, because there will be some number of cells in the first state left. Is this correct?
    • Allen Lin 16:14, 29 May 2008 (EDT): To continue Doug's idea, then main initial state could be the production of vitamins or something, and the second state would be the production of oxidase. This way, so cells with FimE turned on will die, but not all cells are in that state. Would this be a way to compete against other native bacteria? Keasling's paper said that they were able to eliminate sporadic expression of FimE by using a weaker RBS.
    • Josh K. Michener 16:54, 29 May 2008 (EDT): Yeah, that's the general idea. Ideally we'd like to have multiple states for fimE (so you switch from the undifferentiated vitamin/lactase state to either the phage or the ROS burst, say). Also, for what we're doing, fimE might not be critical - the ROS burst or the phage will both kill the cell, so we don't necessarily need to turn off the vitamin/lactase production. The SSM mechanism is required, though.

ROS bursts

  • H2O2 [29, 30]
    • Turn on xanthine oxidase[31, 32] (or galactose oxidase[33]), turn off catalase
      • Doug Tischer 03:23, 17 May 2008 (EDT)I like using xanthine oxidase because it can be suddenly "turned on" (really, changed from xanthine dehydrogenase to xanthine oxidase) through proteolytic cleavage. However, this is only true of mammalian xanthine oxidases. It is further complicated by the fact that xanthine oxidase can be revirsibly turned on/off by oxidation/reduction of some disulfide bonds. I'd be worried that because it is a mammalian protein with some disulfide bridges, that we wouldn't get good expression in bacteria. I tried to see if anyone has expressed it in E. coli, but haven't had any luck. There are bacterial versions of xanthine oxidase, but these can't be turned on/off like their mammalian cousins. We could try to make the bacterial xanthine oxidase turn on and off by preventing their natural dimerization by adding some interfering peptide sequence. This would be linked to the original protein by a linker domain that has a protease site. Once the protease is expressed, xanthine oxidase would be trimmed, it would dimerize, and we would get a sudden burst of H2O2. So this is where I'm stuck, since I don't have enough experience. Is it worth it to try and express the mammalian xanthine oxidase in E. coli and hope it can be done relatively easily or should I start looking into strategies for turning bacterial xanthine oxidase on and off?
        • References? Yamaguchi et al[31] showed that they could express the human XO in E. coli, but they only looked for activity in vitro. Not clear that they'd have activity in vivo.
        • Doug Tischer 05:46, 18 May 2008 (EDT)Looks like it is very tricky to express mammalian xanthine oxidase in bacteria, mostly because of the molybdenum center. The paper above reported that only 4% of their expressed xanthine oxidase was active. I'm now thinking it might be best to take an enzyme that produces H2O2 that is naturally found in bacteria and is a homodimer (like glucose oxidase, I think). Then, express one of the monomers as a fusion protein on the cytoplasmic C-terminal tail of a integral membrane protein, linked by a linker region with a protease site. This setup would prevent dimerization until the monomers are released into the cytoplasm when the protease is expressed
        • Josh K. Michener 14:59, 18 May 2008 (EDT): Remember, try to make this an incremental progression, not a moon shot. Also, keep in mind that the simplest solution is the most likely to work. So the first step would be expression of an oxidase and measurable production of peroxide in vivo. I'd probably take a couple different oxidases (try the galactose oxidase[33], for instance, in addition to a glucose oxidase - that one's upstairs in the Arnold Lab) and just see how much peroxide you can make. Then see how well you can regulate it just with transcriptional regulation - try it in BL21(DE3)pLysS, for instance. You could probably do that in parallel with the protein engineering for your protease cascade, but don't rely on any given thing working.
    • Trigger by conjugation? [34, 35]


Error fetching PMID 14532068:
Error fetching PMID 16534780:
Error fetching PMID 16554821:
Error fetching PMID 18362885:
Error fetching PMID 14617664:
Error fetching PMID 18048927:
Error fetching PMID 7916011:
Error fetching PMID 6259126:
Error fetching PMID 11676567:
Error fetching PMID 16040799:
Error fetching PMID 12663927:
Error fetching PMID 15708311:
Error fetching PMID 15767015:
Error fetching PMID 6374667:
Error fetching PMID 2561954:
Error fetching PMID 11022933:
Error fetching PMID 17301077:
Error fetching PMID 2656662:
Error fetching PMID 1366680:
Error fetching PMID 11707617:
Error fetching PMID 7775420:
Error fetching PMID 8157597:
Error fetching PMID 8792277:
Error fetching PMID 11157352:
Error fetching PMID 8527227:
Error fetching PMID 6127606:
Error fetching PMID 18396082:
Error fetching PMID 8613033:
Error fetching PMID 16081276:
Error fetching PMID 12111724:
Error fetching PMID 17698004:
Error fetching PMID 10858213:
Error fetching PMID 17631628:
Error fetching PMID 15113564:
Error fetching PMID 12788700:
Error fetching PMID 16885287:
Error fetching PMID 11386882:
Error fetching PMID 18051328:
Error fetching PMID 17158705:
Error fetching PMID 16269750:
Error fetching PMID 10194817:
Error fetching PMID 2667763:
Error fetching PMID 8830713:
Error fetching PMID 8694025:
Error fetching PMID 11255004:
Error fetching PMID 6090713:
Error fetching PMID 4564201:
Error fetching PMID 6091132:
Error fetching PMID 18029799:
Error fetching PMID 8144476:
Error fetching PMID 8889178:
Error fetching PMID 2985547:
Error fetching PMID 1534329:
Error fetching PMID 9618442:
Error fetching PMID 15128580:
Error fetching PMID 17308179:
  1. Error fetching PMID 16081276: [asrar]
  2. Error fetching PMID 8613033: [camilo]
  3. Error fetching PMID 12111724: [bermingham]
  4. Error fetching PMID 17698004: [gabelli]
  5. Error fetching PMID 12788700: [sybesma2]
  6. Error fetching PMID 16269750: [zhu]
  7. Error fetching PMID 15113564: [sybesma1]
  8. Error fetching PMID 15128580: [wegkamp1]
  9. Error fetching PMID 17308179: [wegkamp2]
  10. Error fetching PMID 11676567: [smolke2001]
  11. Error fetching PMID 18396082: [bernstein]
  12. Error fetching PMID 8527227: [suttie]
  13. Error fetching PMID 14532068: [haidinger]
  14. Error fetching PMID 10194817: [Eko1999]
  15. Error fetching PMID 11386882: [morita]
  16. Error fetching PMID 18051328: [yun]
  17. Error fetching PMID 2667763: [cronan]
  18. Error fetching PMID 11255004: [ramanculov2001]
  19. Error fetching PMID 8792277: [jiang]
  20. Error fetching PMID 11157352: [devrese]
  21. Error fetching PMID 17158705: [Deutscher2006]
  22. Error fetching PMID 8830713: [Hoischen1996]
  23. Error fetching PMID 8694025: [hertzler]
  24. Error fetching PMID 6259126: [bochner]
  25. Error fetching PMID 16885287: [sheng]
  26. Error fetching PMID 14617664: [phase]
  27. Error fetching PMID 16554821: [fluct1]
  28. Error fetching PMID 18362885: [fluct2]
  29. Error fetching PMID 2561954: [tiina]
  30. Error fetching PMID 7775420: [gonzalez1]
  31. Error fetching PMID 17301077: [yamaguchi]
  32. Error fetching PMID 11022933: [stevens]
  33. Error fetching PMID 11707617: [sun]
  34. Error fetching PMID 2656662: [mazodier]
  35. Error fetching PMID 1366680: [berg]
  36. Error fetching PMID 16534780: [fim]
  37. Error fetching PMID 18048927: [holden]
  38. Error fetching PMID 7916011: [gally]
  39. Error fetching PMID 16040799: [rao]
  40. Error fetching PMID 12663927: [paulsen]
  41. Error fetching PMID 15708311: [westendorf]
  42. Error fetching PMID 15767015: [schultz]
  43. Error fetching PMID 6374667: [benson]
  44. Error fetching PMID 8157597: [gonzalez2]
  45. Error fetching PMID 6127606: [bentley]
  46. Error fetching PMID 10858213: [pericone]
  47. Error fetching PMID 17631628: [regev]
  48. Error fetching PMID 6090713: [raden]
  49. Error fetching PMID 4564201: [gemski]
  50. Error fetching PMID 6091132: [devries]
  51. Error fetching PMID 18029799: [bielke]
  52. Error fetching PMID 8144476: [deng]
  53. Error fetching PMID 8889178: [baxa]
  54. Error fetching PMID 2985547: [elledge]
  55. Error fetching PMID 1534329: [lindsey]
  56. Error fetching PMID 9618442: [vanbelkum]
All Medline abstracts: PubMed HubMed
Personal tools