Long Dis­tance Chats

by Eam­mon Ri­ley and Alex Meeske

Writ­ing this blog en­try with a fel­low grad­u­ate stu­dent nearly 3000 miles away taught us both a valu­able les­son in com­mu­ni­ca­tion: it is sim­ply chal­leng­ing to com­mu­ni­cate ef­fec­tively across great dis­tances. Not so with mi­crobes. Re­cently sci­en­tists have dis­cov­ered that bac­te­ria can over­come the prob­lems of long-range com­mu­ni­ca­tion. With the right ge­netic rewiring, they can be co-opted into cheap, use­ful tech­nol­ogy.

Biosen­sors and Os­cil­lat­ing Biopix­els: Can You Hear Me Now?

The abil­ity of bac­te­ria to com­mu­ni­cate has been har­nessed to de­velop biosen­sors –  de­vices that use bi­o­log­i­cal mech­a­nisms to de­tect the pres­ence of mol­e­cules. Biosen­sors can poten­tially be used to de­tect pol­lu­tants in wa­ter, con­t­a­m­i­nants in food, or to di­ag­nose med­ical con­di­tions. The mech­a­nisms that mi­crobes uti­lize to sense and re­spond to chem­i­cal infor­ma­tion can be rewired to trans­mit a sig­nal. The fu­ture of biosen­sors is promis­ing. In addi­tion to their porta­bil­ity and low cost, biosen­sors of­fer fur­ther ad­van­tages: they pro­vide unique, tun­able, and highly sen­si­tive out­puts that can be reg­is­tered in real-time. No need for time-con­sum­ing and la­bor-in­ten­sive lab­o­ra­tory analy­ses.

Fig­ure 1. "Can you hear me now?" Source

Bac­te­r­ial biosen­sors of­ten rely on ge­netic cir­cuits that are pro­grammed to gen­er­ate an os­cil­lat­ing feed­back in re­sponse to a sig­nal. One ob­sta­cle to con­struct­ing re­silient cir­cuits is the in­ter­cel­lu­lar vari­abil­ity aris­ing from a noisy en­vi­ron­ment that im­pedes large-scale syn­chro­niza­tion of in­di­vid­ual cells. Fur­ther­more, biosen­sors must do more than sim­ply sense their tar­get; they must also re­li­ably con­vey that in­for­ma­tion to hu­mans. But such tech­ni­cal hur­dles can be over­come by uti­liz­ing fre­quency-based cir­cuits, which are more ro­bust than their am­pli­tude-based coun­ter­parts. It turns out that am­pli­tude-based cir­cuits are acutely sen­si­tive to noise-gen­er­at­ing en­vi­ron­men­tal fac­tors, whereas fre­quency-based cir­cuits are eas­ily dig­i­tized, read­ily up­dated, less sen­si­tive to am­bi­ent con­di­tions (con­sider the sound qual­ity of FM ra­dio over AM ra­dio), and pro­vide a dy­namic read­out. Con­se­quently, fre­quency-based cir­cuits have come to per­me­ate our every­day life, com­pris­ing the ba­sis of AC power, GPS, and lasers. As such, they are an at­trac­tive fea­ture for the fu­ture de­vel­op­ment of biosen­sors.

Fig­ure 2. The ba­sic mi­croflu­idic de­vice used in this study. Me­dia con­tain­ing vari­able con­cen­tra­tions of a pol­lu­tant to be stud­ied, in this case ar­sen­ite, is fed through the cell port, flow­ing past the biopix­els ar­ray into the cell and waste port. Dur­ing load­ing, pres­sure is in­creased at the cell port and de­creased at the waste port to re­verse the flow and al­low cells to pass by the trap­ping re­gions. Source

Quo­rum Sens­ing: A Mol­e­c­u­lar Mi­cro­phone

A ma­jor im­ped­i­ment to the de­vel­op­ment of ef­fec­tive biosen­sors is that the en­doge­nous mech­a­nisms of bac­te­r­ial com­mu­ni­ca­tion are only ef­fec­tive on a lo­cal scale. Many bac­te­ria are ca­pa­ble of sens­ing chem­i­cal sig­nals gen­er­ated by their neigh­bors and elic­it­ing a be­hav­ioral re­sponse to these signals—a phe­nom­e­non known as quo­rum sens­ing. The dis­cov­ery of quo­rum sens­ing demon­strated that bac­te­ria are well versed in the art of cou­pling gene ex­pres­sion within a small pop­u­la­tion of cells. How­ever, har­ness­ing the power of quo­rum sens­ing to pro­duce strong, long-range in­ter­ac­tions ca­pa­ble of in­stan­ta­neously syn­chro­niz­ing gene ex­pres­sion presents a chal­lenge. The biggest ob­sta­cle is that the sig­nals that evoke a quo­rum sens­ing re­sponse must dif­fuse through a liq­uid medium, a slow process on a macro­scopic scale.

Gas Phase Com­mu­ni­ca­tion: A Mol­e­c­u­lar Tele­phone

Whereas quo­rum sens­ing is slow to sig­nal over long length scales, gas phase-me­di­ated com­mu­ni­ca­tion is rapid, al­beit tran­sient and weak. Re­searchers have ex­ploited the ad­van­tages of gas phase sig­nal­ing, namely speed and range, to rapidly syn­chro­nize the be­hav­ior of thou­sands of bac­te­r­ial colonies. Sig­nal­ing in the gas phase typ­i­cally comes at the ex­pense of sig­nal sus­tain­abil­ity. The Hasty lab at UCSD, how­ever, has rewired a quo­rum sens­ing-based ge­netic cir­cuit to be re­in­forced by gas phase sig­nal­ing, thereby pro­duc­ing co­or­di­nated, os­cil­lat­ing bac­te­r­ial feed­back that is quickly trans­mit­ted.

Syn­er­gis­tic Syn­chro­niza­tion: Now You're Speak­ing My Lan­guage!

To over­come the slow speed and range lim­i­ta­tions of quo­rum sens­ing on the mil­lime­ter scale, the Hasty lab de­vel­oped a gas per­me­able mi­croflu­idic ar­ray com­posed of thou­sands of dis­tinct bac­te­r­ial colonies, coined "biopix­els." Cells in the biopix­els carry a ge­netic cir­cuit that gen­er­ates an os­cil­lat­ing sig­nal in re­sponse to the quo­rum sens­ing in­ducer acyl ho­moser­ine lac­tone (AHL). The re­searchers de­signed ge­netic cir­cuit to be strongly re­in­forced by, of all things, the sim­ple volatile mol­e­cule H₂O₂ (what could be sim­pler than hy­dro­gen per­ox­ide?). In this ge­netic cir­cuit, quo­rum sens­ing syn­chro­nizes within a pixel, while the va­por species mod­u­lates long-range cou­pling be­tween pix­els. Here, the weak but global in­ter­ac­tions achieved by gaseous dif­fu­sion co­or­di­nate the strong, lo­cal in­ter­ac­tions me­di­ated by quo­rum sens­ing. Whereas each sys­tem does not pro­duce co­her­ent os­cil­la­tions in­di­vid­u­ally, nest­ing these two meth­ods of com­mu­ni­ca­tion re­sults in what the au­thors call "syn­er­gis­tic syn­chro­niza­tion". Us­ing these meth­ods, the Hasty lab was able to syn­chro­nize over 50 mil­lion cells arranged in a 24mm x 12mm ar­ray of over 12,000 biopixels—a dis­tance of greater than 5,000 cell lengths—with high tem­po­ral pre­ci­sion.

Fig­ure 3. An ar­ray of syn­chro­nously os­cil­lat­ing biopix­els built by the Hasty Lab. Source

So, why are more biopix­els bet­ter than one? Larger num­bers of biopix­els yield sig­nals strong enough to be de­tected with­out the need for mag­ni­fi­ca­tion and ex­pen­sive op­ti­cal equip­ment, al­low­ing the biosen­sor to re­main portable and in­ex­pen­sive. Fur­ther­more, in­creas­ing the num­ber of biopix­els al­lows for the gen­er­a­tion of a rich di­ver­sity of out­puts. Us­ing mul­ti­ple biopix­els of vary­ing size and sep­a­ra­tion, the Hasty lab hopes to make cir­cuits con­sist­ing of switches and logic gates that are far too com­plex to be achieved by a sin­gle os­cil­lat­ing biopixel.

Fig­ure 4. Ge­netic con­fig­u­ra­tions of os­cil­la­tory cir­cuits. Source

How does the ge­netic cir­cuit work? (Warn­ing: Start think­ing here like an elec­tronic en­gi­neer.) The cir­cuit is com­prised of three in­ter­con­nected mod­ules that drive these ro­bust and syn­chro­nous os­cil­la­tions on the macro­scopic scale. A quo­rum sens­ing mod­ule forms the ba­sis of the os­cil­la­tor and me­di­ates syn­chro­niza­tion at the in­di­vid­ual biopixel level. Here, both luxI, which en­codes an acyl-ho­moser­ine lac­tone (AHL) syn­thase, and aiiA, which en­codes an AHL-de­grad­ing lac­tonase, are ex­pressed from a spe­cial pro­moter un­der the dual con­trol of LuxR and H₂O₂. LuxR is a tran­scrip­tional ac­ti­va­tor that dri­ves ex­pres­sion of these two genes upon bind­ing the AHL au­toin­ducer, a chem­i­cal sig­nal that ini­ti­ates the quo­rum sens­ing re­sponse. There­fore, syn­the­sis of AHL re­sults in self-re­in­forc­ing os­cil­la­tory be­hav­ior be­cause the au­toin­ducer si­mul­ta­ne­ously stim­u­lates both its own pro­duc­tion and degra­da­tion.

The ef­fect of the AHL au­toin­ducer is re­stricted lo­cally by high flow rates in the chan­nels con­nect­ing the pix­els. Thus, an ad­di­tional cou­pling mod­ule is nec­es­sary for in­ter­pixel syn­chro­niza­tion. This is achieved by wiring the gene en­cod­ing NADH de­hy­dro­ge­nase (ndh), which gen­er­ates H₂O₂ va­por, to the same H₂O₂-and LuxR-re­spon­sive pro­moter dri­ving the genes in the os­cil­la­tor. The re­sult­ing ef­fect is that os­cil­lat­ing lev­els of the slowly dif­fus­ing AHL au­toin­ducer yield syn­chro­nized os­cil­la­tions in H₂O₂ lev­els that al­low rapid trans­mis­sion of the sig­nal over large dis­tances. The H₂O₂ rapidly dif­fuses through the ar­ray, syn­chro­niz­ing the os­cil­lat­ing re­sponses among the biopix­els. Fi­nally, ex­press­ing sfGFP from an ad­di­tional H₂O₂-AND LuxR-re­spon­sive pro­moter gen­er­ates a pho­to­met­ric read-out of the syn­chro­nized os­cil­la­tions. But don't take our word for it; take a look at the movies!

Proof of Prin­ci­ple: Ap­pli­ca­tions

Fig­ure 5. Heat map and tra­jec­to­ries de­pict­ing time-lapse out­put of 500 in­di­vid­ual biopix­els un­der­go­ing rapid syn­chronization. Sam­pling time is 2 min. Source

We know what you're think­ing. This cir­cuit is all well and good, but does it ac­tu­ally work as a biosen­sor? The au­thors' next step was to ap­ply this tool to­wards de­vel­op­ing a biosen­sor for the de­tec­tion of a com­mon wa­ter pol­lu­tant: ar­sen­ite. One of the ad­van­tages of the cir­cuit is its mod­u­lar­ity: the in­di­vid­ual com­po­nents can be in­ter­changed and ad­di­tional com­po­nents lay­ered on top with­out al­ter­ing the ba­sic struc­ture of the cir­cuit. Sim­ply in­clud­ing a copy of luxI en­cod­ing an AHL syn­thase un­der the con­trol of an ar­sen­ite-re­spon­sive pro­moter ef­fec­tively cou­ples ar­sen­ite de­tec­tion with the quo­rum sens­ing re­sponse. In the ab­sence of ar­sen­ite, the ArsR re­pres­sor binds DNA to block tran­scrip­tion of the ad­di­tional quo­rum sens­ing mod­ule, al­low­ing only base­line os­cil­la­tions. The pres­ence of ar­sen­ite, how­ever, causes ArsR to re­lease from the DNA and ac­ti­vate tran­scrip­tion of this mod­ule, which causes an in­crease in the os­cil­la­tory pe­riod that is pro­por­tional to the con­cen­tra­tion of ar­sen­ite. In this con­fig­u­ra­tion, pe­riod mod­u­la­tion can be used to de­ter­mine, with high pre­ci­sion, the lev­els of tox­ins in a sam­ple.

Fig­ure 6. Up­per: With a pe­riod mod­u­la­tion out­put, ad­dition of 0.8 μM ar­sen­ite re­sults in an in­crease in the os­cil­la­tory pe­riod from 69 min to 79 min. Lower: With a thresh­older out­put, ad­di­tion of 0.25 μM ar­sen­ite trig­gers a shift from rest to os­cil­la­tory be­hav­ior within 20 min. Source

A vari­a­tion on this cir­cuit in­volves plac­ing the gene en­cod­ing the LuxR tran­scrip­tional ac­ti­va­tor un­der the con­trol of an ar­sen­ite-re­spon­sive pro­moter as the only copy of luxR in the genome. Ex­po­sure to ar­sen­ite is there­fore re­quired for LuxR ex­pres­sion and the ac­ti­va­tion of the rest of the cir­cuit. Os­cil­la­tory be­hav­ior is only ob­served once the level of toxin reaches a crit­i­cal thresh­old. Thus, the cir­cuit can be wired in such a way that it is only ac­ti­vated upon ex­po­sure to a given con­cen­tra­tion of toxin.

In ad­di­tion to ge­netic al­ter­ations to the cir­cuitry, phys­i­cal ma­nip­u­la­tions to the mi­croflu­idic de­vice can also be made to gen­er­ate unique out­puts. Al­ter­ing the dis­tance be­tween biopix­els re­sults in anti-phase syn­chro­niza­tion, where neigh­bor­ing biopix­els ex­hibit op­po­site be­hav­ior in ac­ti­va­tion and de­ac­ti­va­tion of the cir­cuit. Al­ter­na­tively, con­struct­ing a de­vice in which there is a mixed pop­u­la­tion of biopixel sizes re­sults in the phe­nom­e­non of 2:1 res­o­nance, where the fre­quency of the larger biopix­els is twice that of the smaller traps. All told, there is a rich di­ver­sity of out­puts that can be gen­er­ated with only mi­nor ad­just­ments to the un­der­ly­ing cir­cuit.

In­ex­pen­sive and re­spon­sive biosen­sors have tremen­dous util­ity for de­vel­op­ing coun­tries and the mil­i­tary. A tremen­dous ad­van­tage of biosen­sors in test­ing wa­ter qual­ity is that there is no need for bulky, hi-tech, la­bor-in­ten­sive lab­o­ra­tory equip­ment. Fur­ther re­search and de­vel­op­ment may well re­veal many more ap­pli­ca­tions for this kind of tech­nol­ogy. But we will still need the bac­te­ria. Un­til hu­mans mas­ter gas phase com­mu­ni­ca­tion, how­ever, we are stuck with tele­phones.

 

Ref­er­ence

Prindle A, Samayoa P, Razinkov I, Danino T, Tsim­ring LS, Hasty J (2011). A sens­ing ar­ray of rad­i­cally cou­pled ge­netic 'biopix­els'. Na­ture, 481 (7379), 39–44 PMID: 22178928

 

Eam­mon Ri­ley is a grad­u­ate stu­dent in the Di­vi­sion of Bi­o­log­i­cal Sci­ences at the Uni­ver­sity of Cal­i­for­nia, San Diego. Alex Meeske is a grad­u­ate stu­dent in the Bi­o­log­i­cal and Bio­med­ical Sci­ences Pro­gram at Har­vard Uni­ver­sity.