newcastle igem presentation 2008

BugBuster:BugBuster:Computational design of a bacterial Computational design of a bacterial

biosensorbiosensor

2008 Newcastle University iGEM team

M. Aylward, R. Chalder, N. Nielsen-Dzumhur, M. Taschuk , J. Thompson & M. Wappett

2

BackgroundBackground

• Bacterial infection is a major cause of disease and death, particularly in developing countries

• Resistant strains are becoming a major problem

• Quick, cheap and accurate diagnostics are invaluable

• We want to engineer a diagnostic tool to identify these infections, that can be used in situations where laboratory access, refrigeration and expensive chemicals are not available

3

Sensing BacteriaSensing Bacteria

• Gram positive bacteria secrete ‘fingerprints’ of signal peptides, unique to the species or even the strain

• They also sense these peptides, to facilitate cell-cell communication within the strain

• We could potentially use the sensors for these peptides to design a bacterium which ‘works out’ what Gram positive bacteria are present in its environment

• Fluorescent proteins can provide a discriminatory output

4

Choosing a ChassisChoosing a Chassis

• Quorum sensing is well characterized in Bacillus subtilis

• Bacillus subtilis sporulates– Spores are extremely resilient

– Can be rehydrated as required

• Bacillis subtilis 168 is a well-characterized laboratory strain

– Genetically amenable

– Competency can be induced

• Considerable expertise based in Newcastle in Cell and Molecular Biosciences

5

The ChallengeThe Challenge

• There are potentially many peptides to sense

• Not just presence or absence, but also relative levels of input

• Only limited outputs possible

• Want the choice of output to reflect the presence of pathogenic bacteria

• This is a classical example of a multiplexing problem

• A standard technique from computing science for addressing these kinds of problems is Artificial Neural Networks

The challenge: To implement an ANN in our bacterium, using genetic regulatory cascades to mimic the “neurons”.

6

7

Meeting the ChallengeMeeting the Challenge

• Designing this kind of system by hand is not tractable– Too many interactions

– Too many parameters to tune

– Not enough time to ‘try it out’ in biology

• Computational approaches are required– Evolutionary computing explores a large range of designs with many

different interactions

– Computational modelling of these designs evaluates the parameter space

– Thousands of different designs with many parameterisations can be simulated before making even one engineered bacterium

• Computational solutions can then be implemented in vivo

• Quantification of these biological constructs can feed back into the computational design process

8

short-dis.aviWorkbench

M2SConverter

EvolutionaryAlgorithm

PartsRepository

ConstraintsRepository

Sequence

Feedback

Synthesize

Clone

Analyze

Implementation

9

Modelling with CellMLModelling with CellML

• Parts, and interactions between parts, have associated CellML models

• CellML is modular. Each component:– Captures the dynamic behaviour – Describes how it influences the behaviour of the parts it is attached

to– Supports building complex, multi-component systems from small,

modular descriptions – ‘bottom up’ modelling

• The Evolutionary Algorithm assembles models of the complete system from these part and interaction models – Simulations predict the behaviour– Comparison to our specification to evaluate ‘fitness’

10

The peptide receiver device: designThe peptide receiver device: design

• The wet-lab and the in silico parts of the project were proceeding in parallel

• We decided to build a peptide receiver device to test if our B. subtilis 168 was capable of sensing and responding to the subtilin quorum peptide (a lantibiotic) produced by B. subtilis ATCC6633

• This was modelled bottom up using CellML

11

The peptide receiver device: The peptide receiver device: implementationimplementation

• We designed a device by assembling multiple virtual parts

• The resulting DNA sequence (2.2k) was synthesized by GenScript Corporation

12

pUC57-ncl08

4908bp

Synthesis and cloningSynthesis and cloning

7899bp

pUC57

2708bp

ncl108

2200bp

10099bp

8399bp

pGFP-rrnB

Newcastle device in pUC57

Bacillus integration vector

T4 DNA ligaseTransform into E. coli

Ncl108

BBa_K104001

pGFP-rrnBIntegration Vector

13

Genomic IntegrationGenomic Integration

14

Characterizing BBa_K104001Characterizing BBa_K104001

• Grow ATCC6633, and extract supernatant containing subtilin

• Culture BBa_K104001-transformed 168 in subtilin supernatant at concentrations of:– 0%

– 1%

– 10%

• Image under microscope

• Quantify using Flow Cytometry

15

Characterization of the Characterization of the peptide receiver peptide receiver devicedevice

16

Cell Sorting ResultsCell Sorting Results

Subtilin Fluorescence0% 7.701% 14.7710% 21.95

17

ConclusionsConclusions

We have:1. Demonstrated a bottom-up modelling approach for composing

systems from small functional modules, based upon CellML2. Designed and implemented a software system for the computational

design of complex regulatory networks3. Successfully integrated a two-component quorum sensing system into

Bacillus subtilis, demonstrating that our sensor approach is feasible– Designed, modelled and submitted a working, standard BioBrick

(BBa_K104001) for sensing the quorum communication peptide subtilin, that works as predicted

4. Sent information and developed a B. subtilis website to help the Cambridge University team

5. Taken the Cambridge 2007 BBa_I746107 AIP-inducible promoter P2 and GFP reporter, cloned it into an integration vector and successfully integrated it into the chromosome of 168, ready for further characterization

18

Future work…if we had more timeFuture work…if we had more time

• Characterize BBa_K104001 in more detail

• Characterize other relevant two-component quorum sensors, to expand the detection range and sensitivity

• Implement and characterize the computationally-generated networks in vivo

• Modify or replace the existing spaRK promoter to be constitutive, rather than linked to sporulation (SigA, not SigH)

• Explore a wider range of output reporters

• Produce the bacterium for use in the field

19

AcknowledgementsAcknowledgements

• Our instructors:

– Dr. Jen Hallinan, School of Computing Science– Dr. Matt Pocock, School of Computing Science– Prof. Anil Wipat, School of Computing Science

• Our advisors:

– Jan-Willem Veening, Institute for Cell and Molecular Bioscience

– Leendert Hamoen, Institute for Cell and Molecular Bioscience

– Colin Harwood, Institute for Cell and Molecular Bioscience

– James Lawson, Auckland Bioengineering Institute– Michael T. Cooling, Auckland Bioengineering

Institute– Glen Kemp, NEPAF– Achim Treuman, NEPAF

Our sponsors: