Research Paper Presentation

1. Research Paper Presentation • for • An Introduction to Computational Molecular Biology • CS502, Spring 2014 • University of Illinois at Chicago • Md Abu Naser Bikas • Dept. of CS, PhD Program • mbikas2@uic.edu • UIN: 651644268 20th March, 2014

2. Mechanism-Independent method for predicting response to multidrug combinations in bacteria • Authors • Kevin Wood, Harvard University, Cambridge • Satoshi Nishidaa , Harvard University, Cambridge • Eduardo D. Sontag , Harvard University, Cambridge • Philippe Cluzela, Rutgers University, Piscataway, NJ • Published in • Proceedings of the National Academy of Sciences (PNAS) • July 24, 2012

3. Outline • Basic Idea of the Paper • Specific Aim of Research • Experimental Design • Experimental Result • Research Outcome • Limitations • Conclusion

4. Understanding the Title Mechanism-independent method for Predicting Response to Multidrug Combinations in Bacteria

5. What is this Paper About? Drugs are commonly used in combinations larger than two for treating bacterial infection. Impossible to infer directly from the effects of individual drugs the net effect of a multidrug combination. Mechanism-independent method for predicting the microbial growth response to combinations of more than two drugs. Drug pairs are sufficient to infer the effects of larger drug combinations.

6. What is this Paper About? (cont.) Microbial responses to drug combinations can be predicted using a simple formula Rational design of candidate therapies using combinations of more than two drugs. Multidrug response in bacteria obeys statistical or chemical ?

7. Experimental Elements Bacteria used to perform experiments: Escherichia coli (Gram-negative) Staphylococcus aureus (Gram-positive) Drug combinations used Protein synthesis inhibitors macrolides, aminoglycosides, tetracyclines, lincosamides, and chloramphenicol DNA synthesis inhibitors fluoroquinolones and quinolones Folic acid synthesis inhibitors sulfonamides and diaminopyrimidines Cell wall synthesis inhibitors polypeptide antibiotics, preservatives, analgesics

8. Bacteria E-Coli Gram-negative rod-shaped bacterium Commonly found in the lower intestine of human and animal Can cause serious food poisoning in humans Most of this bacterium is the most human friendly and can benefit their hosts by producing vitamin K

9. Bacteria Staphylococcus aureus Gram-positive , also known as ‘Golden staph’ About half of us carry this organism in our skin and nasal passages If you have an infected cut or sore, it can contain large numbers of Staph Keep any cuts or sores well covered if you are handling foods Raw milk can also be a source of this bacteria Likes to grow in salty and sweet foods(custard, hams, frankfurters, salads etc.) Takes only a very short time to make someone sick

10. Drugs Protein synthesis inhibitors DNA synthesis inhibitors Folic acid synthesis inhibitors Cell wall synthesis inhibitors

11. Specific Aim of Research Combinations of three or more drugs have been studied in both clinical and laboratory settings as potential treatments for severe microbial infections. Drug interactions have typically been studied using descriptive, rather than predictive Two drugs when used in combination on microbial growth effect of one drug is reduced in the presence of another => antagonistic effect of one drug is enhanced in the presence of another => synergistic

12. Specific Aim of Research (cont.) Because of these antagonistic and synergistic druginteractions the effects of drug combinations cannot be predicted based on the effects of the drugs alone. Little is known about how more than two drugs combine to yield higher-order effects on bacterial growth Is it possible to understand and to predict the effects of these larger drug combinations without relying on specific mechanistic details?

13. Specific Aim of Research (cont.) Consider a Classic three-drug combination Example: Protein synthesis inhibitor chloramphenicol 1.5 μg/mL + DNA synthesis inhibitor ofloxacin 40 ng/mL + folic acid synthesis inhibitor trimethoprim 0.3 μg/mL. The growth rate of E. coli treated with each drug alone is about 0.58, 0.47, and 0.39

14. Specific Aim of Research (cont.) Chloramphenicol + ofloxacin => growth rate of 0.53 significantly higher than expected from a naive multiplication of the single drug rates consistent with antagonism between DNA synthesis inhibitors and Protein synthesis inhibitors ofloxacin + trimethoprim => growth rate < 0.01 lower than expected from single drug growth rates consistent with synergy between two ofloxacin + trimethoprim => growth rate of 0.16 slightly smaller than multiplication from single drug growth rates consistent with antagonism between two The effects of all three pairs of drugs differ significantly from that predicted by multiplication of single drug effects.

15. Specific Aim of Research (cont.) Based on the drug pair response seems like a very little hope that such an assumption of independence will be useful when all three drugs are combined Surprisingly, the growth rate in the presence of all three drugs => product of single drug growth rates suggesting the drugs act independently Why have the previously strong interactions between drug pairs been eliminated when the three drugs are combined? The authors answer this question using a quantitative framework to provide insight into how the cell integrates during drug combinations

16. Experimental Setup Bacterial Strains E-Coli Staphylococcus aureus Drugs All drug solutions prepared from solid stocks Stored in the dark at −20 ° All drugs were thawed and diluted in sterilized media Media LB broth for experiments on E. coli Soy Broth for experiments on S. aureus

17. Experimental Design Lets: g1, g2 , g3 , …….. , gN=> growth rates of cells D1, D2, D3, ……..., DN=> N drugs Need to Estimate single drug gi and two-drug gij growth rates Based on this these growth rates multidrug growth rates will be predicted using a formula (Entropy Maximization)

18. Experimental Design (cont.) The authors model the effect of each drug, Di, Using an associated stochastic variable, Xi where Measured growth rate = mean (i.e., expectation) of Xi gi =< Xi> Similarly, in the presence of two drugs, i and j, gij =< XiXj> In general, the normalized growth in presence of a combination of N drugs, g1...N, = <X1. . .XN>

19. Experimental Design (cont.) An absence of correlation between variables Xi and Xj indicates that the drugs do not interact, and therefore gij is equal to the product of the independent growth rate gi and gj In the absence of interactions between the drugs, this statistical model is equivalent to the well-known Bliss independence model [1]in pharmacology. 1. Bliss CI, “The calculation of microbial assays”, Bacteriol Rev, 1956, 20:243–258

20. Experimental Design (cont.) Drug Interactions as a Mechanism-Independent Statistical Problem To characterize the interactions between drugs (i.e., synergies and antagonisms) the authors introduces a probability density P(x) P(x) = P(x1, x2, . . ., xN) that describes the joint distribution of these random variables can be estimated using experimental data. Specifically, estimated the probability density P(x) using only the growth rate data in response to single drugs and drug pairs Lets call this estimate Ppair(x), because it depends only on the interactions between drug pairs and the effects of the drugs alone

21. Experimental Design (cont.) To estimate Ppair(x) from experiments, entropy maximization [2] [3] is used Well-established statistical technique that guarantees that Ppair(x) contains only the information from our one-drug and two-drug data sets Maximum entropy distribution Used: Z = Normalization Constant hi= Resilience coefficients characterize the single drug response Determined from the measurements of single drug effects at each dosage Jij= Drug–drug coupling coefficients that characterize the response to pairs of drugs Determined from the measurements of pairwise drug effects at each dosage 2. Cover TM, Thomas JA (2006) Elements of Information Theory, XXIII (Wiley-Interscience, Hoboken, NJ). 3. Jaynes ET (1957) Information theory and statistical mechanics. Phys Rev 106:620–630.

22. Experimental Design (cont.) Growth in response to multiple drugs can be predicted from the growth in response to those drugs singly and in pairs using maximum entropy.

23. Experimental Design (cont.) Three- and Four-Drug Interactions Arise from Accumulation of Pair wise Interactions For calculating the expected growth response to a larger combination of drugs take the average of the all estimated distributions Ppair(x) This prediction would match experimental results only if the net effects of the drug combination were to arise entirely from the accumulation of pairwise interactions but not from higher drug interactions.

24. Experimental Results Compare with the Prediction To test this framework, the authors calculated expected growth response to various combinations of N drugs. ( n = 3 or n = 4) Directly measured bacterial growth in the presence of these drug combinations Need to compare this results with Ppair(x)

25. Experimental Result (cont.) Their predictions are identical with this result. No hidden correlations Only correlations from pairwise and single drug effects Demonstrates that the pairwise interactions are indeed sufficient to accurately predict the growth response to the combination of the multiple drugs Experimented With: 120 unique drug dosage combinations comprised of 93 unique 3-drug combinations

26. Experimental Design (cont.) Distribution Ppair(x) calculated from entropy maximization can also be described by simple algebraic expressions: The response to three drugs (gijk) gijk = gigjk + gjgik + gkgij − 2gigjgk; The response to four drugs (gijkl) gijkl = gijgkl +gikgjl + gilgjk −2gigjgkgl Fully consistent with maximum entropy predictions Can be derived from the famous Isserlis theorem [4] The simple expressions provide a way to predict the effect of a drug combination on growth without using the sophisticated maximum entropy framework 4. Isserlis L, “On a formula for the product-moment coefficient of any order of a normal frequency distribution in any number of variables”, Biometrika, 1918, 12:134–139

27. Experimental Result (cont.) • Predictions highlight ways that pairwise interactions accumulate to yield higher-order interactions.

28. Research Outcome The net effect of drug combinations is dominated by the accumulation of pairwise drug interactions Provide a powerful strategy for the rational design of candidate therapies using combinations of three or more drugs in case of mechanistic descriptions are not available. As their findings do not depend on any specific details about the cellular system, it may be applicable to other bacteria and even to eukaryotes. The chemical complexity underlying the cellular response to drug combinations often does NOT exceed that of drug pairs. These findings therefore raise the possibility that the multidrug response in bacteria obeys statistical, rather than chemical, laws for combinations larger than two.

29. Some Practical Limitations The author measured the distribution Ppair(x) for a particular bacterial strain which cannot in general, be used to predict the multidrug response in a different strain. Possible to design an ad hoc example in which any pairwise model is likely to fail If one drug were an enzyme that required two substrates, then the combination of the enzyme with both substrates might yield a completely novel three-body interaction that could not be predicted from the pairwise effects

30. Conclusion Pairwise interactions are indeed sufficient to accurately predict the growth response to the combination of the multiple drugs There may exist other pairwise models that could also be used to estimate the effects of larger drug combinations. The success of their finding is that now at least one such pairwise model exists that provides excellent predictive power in multidrug combinations.

31. Open Research Questions To validate the prediction of multidrug combinations in viral infections To check how well their prediction method works in Human body treated with multidrug against cancer To test the multidrug prediction method with RNA synthesis inhibitor

32. Thank You