May 27th 2010 BIOSENSORS 2010: 20th Anniversary World Congress on Biosensors Pierre R. Marcoux (1,2), Mathieu Dupoy (1,2), Raphael Mathey (1,3), Armelle Novelli Rousseau (1,3), Pierre Joly (1,2), Florence Rivera (1,2), Sophie Le Vot (1,2), Jean-Pierre Moy (1,2), Frédéric Mallard (1,3) 1 bioMérieux – CEA joint team, 17 rue des Martyrs, 38054 Grenoble cedex9, France. 2 Commissariat à l'Energie Atomique (CEA), LETI, MINATEC, Grenoble, France. 3 bioMérieux, Grenoble, France Today, rapid detection and identification of bacteria in microbiological diagnosis is a major issue. Methods relying on the observation of few or single bacteria will allow much faster delivery of clinical information. We report here a method for rapid detection of single bacteria and real-time monitoring of one of their metabolic activities. We have used reverse emulsions (water in perfluorinated oil) as a way of encapsulating Escherichia coli into microreactors using MMFD (dire par quelle technologie). Every aqueous homodisperse droplet (200 pL) is either empty or includes a single bacterium, provided that the concentration of bacteria in aqueous phase is low enough. Moreover, this phase contains a fluorogenic substrate (MUG), so that we can monitor the glucuronidase activity through fluorescence microscopy. Such a confinement provides us with two major advantages: we can monitor the enzymatic activities of single bacteria and the concentration of fluorescent bacterial metabolites is strongly increased. We have also discovered a perfluorinated formulation that is non toxic for bacteria. Working on a 2.105 cfu/mL sample and using phenotypic characteristic of the bacteria (glucuronidase activity) , we demonstrated it is possible to detect bacteria in less than 2 hours. We also showed that metabolic detection enables rapid and reliable enumeration in less than 10 hours. Our innovative enumeration technique has been validated by reference methods (growth on agar dish plates CPS3). Furthermore, thanks to confinement of single bacteria, we studied the heterogeneity of a clonal population of bacteria: over ~200 single bacteria on a 24h-period, we monitored glucuronidase enzymatic activity and growth for every confined cell.

1. Random micro-confinement of bacteria into picolitre emulsion droplets for rapid detection and enumeration by enzymatic activity determination. bioMérieux – CEA joint team, Grenoble (France). Pierre R. Marcoux, Mathieu Dupoy, Pierre L. Joly, Florence Rivera, Sophie Le Vot, Jean-Pierre Moy. Armelle Novelli-Rousseau, Raphaël Mathey, Frédéric Mallard. BIOSENSORS 2010, Parallel Session 5D: Enzyme-based biosensors.

5. Intro: rapid detection and enumeration of bacteria excitation of the reference fluorophore (fluorescein) excitation of the fluorophore produced by bacteria (4-MU) reverse emulsion (water in oil) aqueous sample with bacteria 480 nm 360 nm The ratio yields the number of bacteria / mL.

7. Experimental: a three-step process encapsulation the emulsification process must be stable (homodisperse droplets). storage reading interpretation emulsification incubation fluo. measurement 1. Impede compositional ripening and coalescence (efficiency of confinement). 2. Avoid any movement of droplets during incubation. Measure fluorescence ( fluorescein, 4-MU, DsRed ) as a function of time in a maximum number of stored droplets. reference fluorophore 4-MU

10. 2 h 5 h 9 h 11 h 17 h 6 h 10 h 14 h

11. Results: enumeration based on enzymatic activity E. coli at 37°C in droplets, fluorescence kinetics of 4-MU (t=0 is the time when bacteria are encapsulated into droplets): 155 positive droplets were counted among all the observed droplets Enumeration result: positive droplets/explored volume = 155 cfu/0,811 µL = 1.9 × 10 5 cfu/mL 2 nd method: based on Poisson’s law, the number of empty droplets (negative drops) is compared with the number of filled droplets (positive drops), and we assume that all the filled drops include a single bacterium at t=0 . 155 filled drops for a total amount of 3902 observed drops  filling ratio = 155/3902 = 4 % We deduce cV = 41 × 10 -3 . If we assume that all the drops have the same volume V = 200 pL, then c = 2.0 × 10 5 cfu/mL .

12. Results: Two kinds of control regarding enumeration 1. DsRed labelling: plasmid coding for a fluorescent protein DsRed, 214 filled drops in the explored volume (811 nL)  2.6 × 10 5 cfu/mL Poisson’s law: filling ratio = 5.5 %, it yields cV = 0.058 and c = 2.8 × 10 5 cfu/mL 2. Streaking on agar plates with the liquid sample of bacteria (dilution 1/100, then 0.1 mL are spread on a Petri dish) : 188 cfu per plate  1,9.10 5 cfu/mL LB (lysogeny broth) chromID CPS3 Streaking on CPS3 medium is a standard method for the enumeration of E. coli.

13. Results: Enzymatic enumeration vs. DsRed labelling In the explored volume V = 0.811 µL: DsRed labelling yields 214 cfu but only 155 of them provided a detectable signal of 4-MU fluorescence after 22 h of incubation at 37°C  only 72% of encapsulated bacteria are detected in 22h à 37°C (this ratio is coherent with the enumeration results we got from the nebulisation device). DsRed + reference fluorophore (fluorescein) 4-MU

14. 4-MU fluorescence (normalised with respect to fluorescein) Only droplets 1 and 3 show a detectable enzymatic activity. Results: Enzymatic enumeration vs. DsRed labelling 1: filled 2: filled 4: filled 3: filled 5: filled 6: empty

15. Results: Growth vs. enzymatic activity Fast growth and high enzymatic activity. Slow growth and no detectable enzymatic activity. Fast growth, but without any detectable enzymatic activity. Slow growth, but high enzymatic activity. empty drops

16. Conclusions 74% More than a fast detection and enumeration method, we have a tool for the study of single cells (metabolism, growth, etc). Our enumeration results are in good agreement with the standard agar plate method. Detection: in less than 2 h. Enumeration: A plateau is reached after 10 h.