Phototaxis in Volvox e 22, 2010 U vol. 107 U no. 25 U 11147–11650 In This Issue PNAS Jörn Dunkel Proceedings of the National Academy of Sciences of the United States of America o the light www.pnas.org 20.416 otosynthesis, algae such as Volvox carteri swim from sunlight. To execute this motion, known hese microorganism colonies must coordinate housands of flagellated cells despite the organcentral nervous system. Using analytical and ods, Knut Drescher et al. (pp. 11171–11176) at V. carteri spins about its swimming direction hat likely coevolved with the organism’s flagelmaximize photoreactivity. To characterize the g of the organisms, the authors measured the produced by the flagella and modeled the modynamic equations. Using the model, the aua theoretical optimal spinning frequency and ng experimentally by observing how well the media with increased viscosities that inhibited Multicellular colony Volvox carteri. ability to spin. According to the authors, the exonstrated that with a decreased rotation rate the algae were unable to execute phototaxis as accurately esting that in V. carteri, flagellar beating and spinning are linked adaptations. By better understanding anisms coordinate multicellular processes, the findings may provide insight into key evolutionary steps led to higher organisms with central nervous systems. — T.J. Tuesday, January 14, 14 dunkel@mit.edu perimentally. spot at different moments in time (6).The Many species do this for by Details of the Mathematical Model. mathematical model beating direction (17). Instead of quantifying average photo• U, the translational swimming speed, which fixes thethe amplitude phototaxis of Volvox on measured parameters swimming on helical paths relies alongonly which their eyespot acts asanda of v0 . Forby therecording simulations webeating used U ¼ 390 μm∕s,of theeach mean of response the frequency flagellum of is able continuously to give detailedsearching predictionsspace of the for swimming light antenna bright characterisspots (3). the populations we investigated experimentally. and the ability toward the light. It based on a knowlHigher tics eukaryotes haveto aturn nervous system tois integrate visual • ωr , the rotation rate without a light stimulus, which fixes the edge of the fluid velocity at the edge of the flagellar layer of information different sources and orchestrate coordinated Author contributions: K.D., the R.E.G., and I.T. designed K.D., R.E.G., and I.T. amplitude of w0 . For simulations we usedresearch; ωr ¼ 2.3 rad∕s, Volvoxfrom and how this fluid velocity changes when parts of the surperformed research; K.D. analyzed data; and K.D., R.E.G., and I.T. wrote the paper. as shown in Fig. 7 of the main text. responses (7,are 8).exposed to a light stimulus. face •TheThe θ dependence of theofsurface authors declare no conflict interest.velocity. For the simulations Multicellular organisms of that intermediate such asTuval The coupled equations make thecomplexity, model 1are given in the Knut Drescher, Raymond E.upGoldstein , and Idan we approximated v ðθÞ by a superposition of two associated 0 Submission. main alga text. To determine the relatives time evolution of theevolved system of the colonial Volvox and its (9), have a This article is a PNAS Direct 1 1 ðcos θÞ, as shown by Legendre functions, −P1 ðcosRoad, θÞ þ 0.25P 2 coupled equations, we solved the coupled partial differential Department of phototaxis Applied Mathematics and Theoretical Physics, University of Cambridge, Wilberforce Cambridge CB3 0WA, United Kingdom means of high-fidelity without a central nervous sysTo whom correspondence should be addressed. E-mail: R.E.Goldstein@damtp.cam.ac.uk. the dashed magenta line in Fig. 5 of the main text. Using a equations for pðθ;ϕ;tÞ and hðθ;ϕ;tÞ numerically with a built-in soltem and, in many cases, even in the absence of intercellular This articlesin supporting information online at www.pnas.org/lookup/suppl/ simple dependence forfor v0 ðθÞ gives qualitatively similar reby Harry L. Swinney, University of Texas, Austin, TX,t and May contains 6,θ 2010 (received review January 28, 2010) verEdited in Mathematica (Wolfram Research) between times and approved communication through cytoplasmic connections (10). Volvox doi:10.1073/pnas.1000901107/-/DCSupplemental. sults. We assume that w has the same θ dependence as v0 . 0 t þ δt. Due to the integral in the equation for Ω, we used an Euler • βðθÞ, responsivity of theof fluid flow to light For Chlamydomona ^ single Alongtothe evolutionary path for from cells multicellular or- the carteri consists thousands ofstimulation. biflagellated method then solve the equation IðtÞ at every timetostep. We the full model, we used a close approximation to atthethe βðθÞ ensured convergence of thenervous results by choosing small enough like somatic cells sparsely distributed of a passi ganisms with a central system areaspecies of intermediate www.pnas.org/cgi/doi/10.1073/pnas.1000901107 PNAS ∣ June 22, 2010 ∣ vol. 107 ∣ no. 25 ∣ surface 11171–11176 shown in the inset in Fig. 5A of the main text. For the reduced step size δt. spherical matrix, small of germ ce complexity that move in ways suggesting high-level coordination, model, we used βðθÞextracellular ¼ 0.3, the mean of theand βðθÞaused fornumber the In addition to finding the angle of the Volvox axis with the light inside the sphere (Fig. 1A). During development the flagella or yet have none. Instead, organisms of this type possess many autonfull model. direction, the model can also be used to determine the organism such that rotates about respectively. its swimming direction, th omous cells endowed with programs that haveStone evolved • τr and τa ,ent the response andVolvox adaptation time scales, swimming velocity U, via another result from andto achieve For the simulations, we used the its values measured for a light trait that gave Volvox name (11). Coordination of the somat concerted to 2010 environmental Here experiment June 22, U vol. 107 U no.stimuli. 25 U 11147–11650 Samuel (18) responses −2 s−1 , as displayed in intensity of 16 μmol PAR photons m cells resembles orchestrating a rowboat with thousands of ind and theory are used to develop a quantitative understanding of Z Fig. 2B of the mainrowers text. but without a coxswain (9). Nature’s solution is 1 pendent how cells of such organisms coordinate to achieve phototaxis, UðtÞ ¼ uðθ;ϕ;tÞdS; [S1] • τbh , the bottom-heaviness time scale, is defined by considering response program at the single-cell level that produces an acc by using the colonial 4πR alga2 Volvox carteri as a model. It is shown a flagellaless Volvox that is tilted at an angle ζ from the vertical. rate steering mechanism, an emergent that the surface somatic cells act as Proceedings individuals but are orchestrated axisofof Volvox would relax back to the vertical atproperty a rate at the coloni of the National Academy ofThe Sciences thethis United States of America www.pnas.org which allows trajectories of the organism to be reconstructed. level.bhYet remains to beweunderstood what by their relative position in the spherical extracellular matrix ζ_and ¼ − sinðζÞ∕τ . Foritthe simulations, used τbh ¼ 14 s, asform the respon A solution of the photoresponse pðθ;ϕ;tÞ is plotted in Fig. 6 of program measured in ref. 19.must take to coordinate the cells and to yield hig their photoresponse functiondefined to achieve cothe maincommon text, using the “reduced model” in thecolony-level main fidelity phototaxis in the presence of the steering constraints Moving to the lightthat range ordination. Analysis models the minimal to the text. A decomposition ofofthis photoresponse intofrom spherical harIn order to compare the results from this mathematical model m monics Y l ðθ;ϕÞ is given in Fig.that, S7.algae The photoresponse p coma viscous environment. biologically faithful shows because flagellar beating disTo optimize photosynthesis, such as the Volvox carteri swim with the measurements of the phototactic ability as a function of puted “fullormodel” during a phototactic is light, shown More than a century ago, Holmes (12) proposed that the s toward away from sunlight. To execute thisturn motion, knownthe colony playsbyanthe adaptive down-regulation in response to viscosity, we implemented a viscosity dependence in the model. in Fig. S8, neglecting bottom-heaviness. phototaxis, microorganismdirection colonies must facing a source of light down-regulate their flagell needs to as spin aroundthese its swimming andcoordinate that the response For this we matic definedcells u ¼ ½u& The initial conditions of the model were a horizontal light diw ηw ∕η and τbh ¼ ½τ bh &w η∕ηw , where η is the beating of thousands of flagellated cells despite the organactivity, a hypothesis latervalues confirmed kineticsan andupward-pointing natural spinningposterior-anterior frequency of the colony appearthe toviscosity be and the subscript w denotes in water.by several investigato rection, axis, and ism’s lack of a central nervous system. Using analytical and (13–16). Although this control principle will initially turn the co mutuallyempirical tuned methods, to give the maximum photoresponse. These Knut Drescher et al. (pp. 11171–11176) ony towards the light, the colony might adapt (14, 15) to the lig further predict phototactic ability decreases dra-H, Hegemann 1.models Kirk DL, Kirk MM (1983) Protein patterns during asexual life cycle of 11. Harz P (1991) Rhodopsin-regulated calcium currents in Chlamydomonas. demonstrate that Vsynthetic .that carterithe spins about itsthe swimming direction Volvox carteri. Dev Biol 96:493–506. Nature 351:489–491. a frequency likelydoes coevolved theat organism’s flagel-frequency, before good alignment with the light direction has been reache maticallyatwhen the that colony notwith spin its natural 2. Solari CA, Ganguly S, Kessler JO, Michod RE, Goldstein RE (2006) Multicellularity and 12. Schaller K, Uhl R (1997) A microspectrophotometric study of the shielding properties lar kinetics to maximize photoreactivity. To characterize the Surprisingly, this observation has not been synthesized into a pr atheresult confirmed by of phototaxis assaystransport. in which colony functional interdependence motility and molecular Proc Natl Acad rotation of eyespot and cell body in Chlamydomonas. Biophys J 73:1573–1578. flagellar beating of the organisms, the authors measured the Sci USA 103:1353–1358. 13. Gerisch G (1959) Cellularquantitative differentiation in Pleodorina and thewith organisation dictive, modelcalifornica consistent the principles of flu was slowed by increasing the fluid viscosity. produced by the flagella modeled 3. Sakaguchi H,fluid Iwasavelocities K (1979) Two photophobic responses inand Volvox carteri. the PlantmoCell of colonial Volvocales (translated from German). Arch Protistenkunde 104:292–358. dynamics, nor are there data on Volvox phototaxis that can b tion with hydrodynamic equations. Using the model, the auPhysiol 20:909–916. 14. Hoops HJ (1993) Flagellar cellular and organismal polarity in Volvox carteri. J Cell Sci compared with such a theory. Here we use a combination of e 4.adaptation Schletz K (1976) Phototaxis in∣ Volvox—Pigments involved in the∣ perception of and light 104:105–117. thors identified aflagella theoretical optimal spinning frequency ∣ evolution ∣ fluid dynamics multicellularity 15. Coggin SJ, Kochert G (1986) Flagellar development and regeneration in Volvox carteri direction. Z tested Pflanzenphysiol 77:189–211. the finding experimentally by observing how well the periment and theory to show that adaptation and colony rotatio (Chlorophyta). J Phycol 22:370–381. 5. Halldal P (1958) Action spectra of phototaxis and related problems in Volvocales, Ulva algae swam in media with increased viscosities that inhibited playMutants key the phototaxis mechanism Multicellular colonyroles Volvoxin carteri. 16. Huskey RJ (1979) affecting vegetative cell orientation in Volvox carteri.of DevV. carteri. By qua gametes and Dinophyceae. Physiol Plant 11:118–153. he most primitive “eyes” evolved longthe before brains the organism’s to spectral spin. According authors, ex- and even Biol 72:236–243. 6. Mast SO (1917) The relation ability between color and to stimulation in thethe lower tifying the flagellaras photoresponse of V. carteri in detail, we sho before forms nervous system organization ap-unable demonstrated thatof with a decreased rotation rate the algae to execute accurately 17. were Herraez-Dominguez JV, Gilphototaxis Garcia de Leon F, Diez-Sales O, Herraez-Dominguez M organism. J periments Expthe Zool simplest 22:471–528. that it actsByasbetter a band pass filter allows adaptation to differe asOA, before, suggesting in VTwo . organisms carteri, flagellar beating linked adaptations. (2005) Rheological characterization of understanding two viscosity grades ofthat methylcellulose: An ap7.peared Sineshchekov Jung K-H, Spudich (2002) rhodopsins mediate phototaxis tosenseare on Earth (1, 2). JLthat Many are ableandtospinning and to the modeling of the behaviour. Polym Sci 284:86–91. low- and high-intensity in Chlamydomonas reinhardtii. Proc Natl USAfindings proach light environments, minimizes the influence of fast light fluctu how simplelight organisms coordinate multicellular processes, may provide insight into keythixotropic evolutionary steps Colloid respond to light stimuli, an ability essential toAcad theScithe optimization 18. Stone HA, Samuel ADT (1996) Propulsion of microorganisms by surface distortions. 99:8689–8694. that eventually led to higher organisms with central nervous systems. — T.J. and maximizes the response to stimuli at frequencies th 77:4102–4104. 8.of MatPIV is an open source PIVthe software toolbox written Matlab. Downloads and and photosynthesis, avoidance of for photodamage, the Phys useRev Letttions, 19. Drescher K, correspond et al. (2009) Dancing Hydrodynamic bound swimming These measur details are at http://www.math.uio.no/∼jks/matpiv/. to Volvox: the rotation rate of states the of organism. of light as a regulatory signal. One of the more striking responses a probiotic form of Escherichia coli, Phys Rev probiotic may Water in early lunar algae. Lett 102:168101. 9. Yoshimura Modified K, Kamiya R (2001) The sensitivity of Chlamydomonas photoreceptor is ments suggest that the response kinetics and colony rotation ha called Nissle, to express CAI-1, andCA, Kessler isoptimized phototaxis, inagainst which motile photosynthetic microorganisms ad20. Solari JO, Michod RE (2006) A hydrodynamics approach to the evolution of for the frequency of cell body rotation. Plant Cell Physiol 42:665–672. protect cholera magmas evolved be and mutually tuned and inoptimized for phototax tested the bacteria as ainprophylactic multicellularity: Flagellar to motility germ-somadunkel@math.mit.edu differentiation Volvocalean 10.just Huth their K (1970)swimming Movement and path orientation of Volvox aureus (translated from with respect toEhrbg. incident light a finely Whereas low-density Vibrio cholerae Recent studies have argued that hygreen algae. Am Nat 167:537–554. German). Z Pflanzenphysiol 62:436–450. against V. cholerae virulence in an Furthermore, we develop a mathematical theory that predic bacterial accumulations the hu- relies on sensory inputs from tuned manner (3, 4). Thisinsteering droxyl ions in lunar minerals indicate infant mouse model. The researchTuesday, January 14, 14 the phototactic fidelity of Volvox as the rotation rate and oth Fidelity of adaptive phototaxis 1 In This Issue T PNAS Knut Drescher Princeton Idan Tuval Mediterranean Institute for Advanced Studies Ray Goldstein Cambridge dunkel@math.mit.edu Tuesday, January 14, 14 Why is Volvox interesting ? • germ-soma differentiation ‘technique’ In Thisreproduction Issue • interesting asexual • metachronal waves • locomotion • phototaxis June 22, 2010 U vol. 107 U no. 25 U 11147–11650 PNAS Proceedings of the National Academy of Sciences of the United States of America www.pnas.org Moving to the light To optimize photosynthesis, algae such as Volvox carteri swim toward or away from sunlight. To execute this motion, known as phototaxis, these microorganism colonies must coordinate the beating of thousands of flagellated cells despite the organism’s lack of a central nervous system. Using analytical and empirical methods, Knut Drescher et al. (pp. 11171–11176) demonstrate that V. carteri spins about its swimming direction at a frequency that likely coevolved with the organism’s flagellar kinetics to maximize photoreactivity. To characterize the flagellar beating of the organisms, the authors measured the fluid velocities produced by the flagella and modeled the motion with hydrodynamic equations. Using the model, the authors identified a theoretical optimal spinning frequency and tested the finding experimentally by observing how well the algae swam in media with increased viscosities that inhibited Multicellular colony Volvox carteri. the organism’s ability to spin. According to the authors, the experiments demonstrated that with a decreased rotation rate the algae were unable to execute phototaxis as accurately as before, suggesting that in V. carteri, flagellar beating and spinning are linked adaptations. By better understanding how simple organisms coordinate multicellular processes, the findings may provide insight into key evolutionary steps that eventually led to higher organisms with central nervous systems. — T.J. dunkel@math.mit.edu Tuesday, January 14, 14 Evolution of multicellularity m in Applied Mathematics, and ¶BIO5 Institute, University of Arizona, Providence, RI 02912 EudorinaJanuary 22, 2006) Volvox oved AprilChlamydomonas 18, 2006 (received for review reinhardtii elegans carteri nes lar a an ent lly ges nd By uid in ary Gonum Pleodorina Volvox on Fig. 1. Volvocine green algae arranged pectorale californica according to typical aureus colony radius R. The lineage ranges from the single-cell Chlamydomonas reinhardtii (A), to us, Short et al, PNAS 2013 Gonium pectorale (B), Eudorina elegans (C), to the somaundifferentiated ng dunkel@math.mit.edu differentiated Pleodorina californica (D), to the germ–soma differentiated V. Tuesday, January 14, 14 er- Volvox carteri somatic cell 200 ㎛ cilia daughter colony from germ cell http://www.youtube.com/watch?v=fqEHbJbuMYA dunkel@math.mit.edu Tuesday, January 14, 14 Volvox carteri somatic cell cilia 200 ㎛ daughter colony from germ cell ... and they can dance ! Drescher et al (2010) PRL Tuesday, January 14, 14 dunkel@math.mit.edu Volvox carteri somatic cell cilia 200 ㎛ daughter colony Drescher et al (2010) PRL Tuesday, January 14, 14 dunkel@math.mit.edu Volvox carteri 200 ㎛ 10 ㎛ Chlamydomonas reinhardtii dunkel@math.mit.edu Tuesday, January 14, 14 Chlamydomonas alga 10 ㎛ ~ 50 beats / sec Goldstein et al (2011) PRL Tuesday, January 14, 14 10 ㎛ speed ~100 μm/s dunkel@math.mit.edu Chlamydomonas Merchant et al (2007) Science Tuesday, January 14, 14 dunkel@math.mit.edu Model organism for studying meta-chronal waves Brumley et al (2012) PRL dunkel@math.mit.edu Tuesday, January 14, 14 ertial (dynamic) pressure ⇤U 2 and viscous shearing str µU/L can be characterized by the Reynolds number4 Swimming at low Reynolds number R ⌅ U L⇤/µ = U L/⇥. Example: Swimming in water with ⇥ = 10 6 m2 /s • fish/human: L ⌅ 1 m, U ⌅ 1 m/s ⇧ R ⌅ 106 . R • bacteria: L ⌅ 1 µm, U ⌅ 10 µm/s ⇧ R ⌅ 10 U L⇥/ ⇥ 1 Geoffrey Ingram Taylor 5 James Lighthill If the Reynolds number is very small, R ⇥ 1, t NSE (8) can be approximated by the Stokes equation 0 = µ ⌥2 u 0 = ⌥ · u. ⌥p + f , (10 (10 These equations+must still be endowed time-dependent BCs with appropri initial and boundary conditions, such as ,e.g.,6 Edward Purcell u(t, x) = 0, as Shapere & Wilczek PRL p(t, x) = p⇥(1987) , Tuesday, January 14, 14 |x| ⇤⌃ . (1 Superposition of singularities 2x stokeslet = symmetric dipole stokeslet rotlet -F F r̂ · F p(r) = + p0 2 4⇥r (8⇥µ) 1 vi (r) = [ ij + r̂i r̂j ]Fj r flow ~ r 1 F r 2 ‘pusher’ Tuesday, January 14, 14 r 2 Volvox carteri swimming speed ~ 100 ㎛/sec PIV 100 ㎛ ⇤✓⌧⇡⌧ ⌧ ⇡⇠ ⌅⌥ ↵⇡⌧# ⇥⇡◆ ⌧ ⇣ ⌃ ⇠ ⌫ ⇡⇣⌘⌧ ⇣ ⌧ $ ⇣⇢⌥ ⇤⇣⇧⌘⌅ ⇣ ⌧ ⇣ ◆ ✓ ⇠"⇡$⌥ ⇤✓⌅ ✏⇡⇣ ⇥⇤ ⇥⇤ ⇥⇥⇥⇥⇥ ⇥⇥⌥ ⇤⇣⌅ ⇣ ◆ ⇤⌘⌅ ⌫⇣" ⌧⇢ ⇤✓⌅⌥ ⌅⇤ ⇠ ◆⇠✓⇣ ⇣"⇠ $⌥ 5=@;43 0B @==; B4;>4@0BC@4 E8B7 B74 :0A4@ >@=D838<6 B74 =<:G :867B A=C@24↵ *4 5=2CA43 =< 0 >:0<4 ✓ ⇥; 8<A834 B74 270;14@ B= ;8<8;8H4 AC@5024 4⇥42BA 0<3 @42=@343 ;=D84A 0B ✏✓ 5>A ⌃ 0AB20; &⇠⇣ $7=B@=<⌥↵ ⌧027 ;=D84 E0A 0<0:GA43 E8B7 AB0<30@3 0:6=@8B7;A B= B@029 1=B7 24::A 0<3 B@024@A↵ =@ 4027 24:: AE8;;8<6 0:=<6 B74 5=20: >:0<4 5=@ ;=@4 B70< A ⌃⇧ 1=3G :4<6B7A⌥ E4 2=::42B43 B74 8<AB0<B0<4=CA D4:=28BG =5 0:: B@024@A <=@;0:8H43 1G B74 AE8;;4@⇧A A>443 C> B= 0 38AB0<24 =5 ⌘ ↵ '74 @4AC:B⌦ 8<6 ⇣⌅⇣ ⇤ ⌃ D4:=28BG D42B=@A E4@4 18<<43 8<B= 0 ✏⌅✓ ⇥; A?C0@4 6@83 ⌃A7=E< 8< 86↵ ⌘ 14:=E⌥ 0<3 B74 ;40< =5 B74 E4::⌦@4A=:D43 0CAA80< 38AB@81CB8=< 8< 4027 18< E0A B094< 0A 0 :=20: ;40AC@4 =5 B74 ⌅=E ⇤4:3↵ < 1=B7 4F>4@8;4<BA ⇥ E8:: 8<3820B4 B74 AE8;;4@⇧A A>443 E78:4 ⌅⌃⇤⌥ 0<3 ⇧⌃⇤⌥ ⌫ ⌅⌃⇤⌥ ⇥ 0@4 B74 D4:=28BG ⇤4:3 8< B74 :01=@0B=@G 0<3 2=;=D8<6 5@0;4A @4A>42B8D4:G↵ ⇠ BG>820: 4F>4@8;4<B0: ⌅=E ⇤4:3 0@=C<3 ⇤ ⇣ ⌘ 8A A7=E< 8< 86↵ ⌃0⌥↵ *4 ⇤B B74A4 ⇤4:3A B= 0 AC>4@>=⌦ A8B8=< =5 0 C<85=@; 10296@=C<3 D4:=28BG ⌃⇥ ⌥ 0 &B=94A:4B ⌃&B⌥ 0 AB@4AA:4B ⌃AB@⌥ 0<3 0 A=C@24 3=C1:4B ⌃A3⌥ ⇡$ #⇠⇢⇢⇠ ⇥ ⇠ !◆ ⇣ ◆ ⇣⇢⌃ ⇠" ⌧ ⌧ ⌫ ⇥ ⌫ ⇣⇢ ✓⌧⇡⌧ ⌘⇣ ⇧ ⌅ ⇧✏ ⌃AB@08< ('⌧+ ⌥ E0A 6@=E< 0F4<820::G 8< ;438C; -✏ . =< 0< =@18B0: A7094@ 1=B7 8< 0 38C@<0: 7 270;14@ E8B7 ◆ 7 8< 0@B8⇤280: 2==: 30G:867B ⌃⇧ CF⌥ 0B ✏⇥ ⇢ 0<3 7 8< B74 30@9 0B ✏◆⇥ ⇢↵ '74 :0@64 4<24 8< =@60<8A; A8H4 14BE44< ⇤ ⇣ ⌘ 0<3 ⇥⌥ ⌅⌦✓ Drescher al (2010) PRL ↵⌅ @4?C8@43 BE= 38AB8<2Bet;4B7=3A B= ;40AC@4 B74 B74G 2@40B4↵ ⇠ ⇢⇢ 20;4@0 ⌃$894 ⇠::843 )8A8=< <=:=684A⌥ ;=C<B43 =< 0 2=<B8<C=CA:G⌦5=2CA01:4 ;8⌦ Tuesday, January 14, 14 ⇧⌅ ⇧ ⌃⇤⌥ ⌫ / ⌦ ⌃ ⌥ ⌃ ⇥ ⌃ /⇤/⇤⌥ ⇥ ⌃ / ⇣⌃ ⌥ ⌥ ⇥ ⇥ /⇤ ⌃ ⌥ ⌥⇤ ⇤ ⇤ ⇣ ⌅ /⇤/⇤ ⇥ ⌃ / / 8A B74 C>E0@3 D4@B820: C<8B E74@4 8A B74 C<8B B4<A=@ ⌃ D42B=@ /⇤ ⌫ ⇤⌥ 0<3 ⇤ 8A ;40AC@43 5@=; B74 24<B4@ =5 B74 =@60<8A; ⌃ ⇥ ⇧ ⇥ ⌥↵ '74 =@84<B0B8=< =5 0:: ;C:B8⌦ >=:4A 8A ⇤F43 B= 14 0:=<6 B74 D4@B820: 0<3 E4 0@4 :45B E8B7 A8F >0@0;4B4@A ⌃⌦ ⇧ ⇧ dunkel@math.mit.edu ⌥ ⌃ ⇧ ⌥⇤ ⇧ ⇥ ⇧ ⇥ ⌥↵ '74 ⇤BA 34A2@814 @4;0@901:G E4:: B74 4F>4@8;4<B0: ⌅=E 0:;=AB How does Volvox achieve phototaxis ? Approach: • light response of individual cells • effects of size & spinning frequency • mathematical modeling • check predictions of model dunkel@math.mit.edu Tuesday, January 14, 14 Fig. S1. Experimental setup Spectra of growth and stimulus light sources. Fig. S1. Spectra of growth a as tiv th fu th SI aw τr (F Fig. S2. (A) Schematic diagram of the sample chamber. (B) Photograph of a micr the focal plane and pointing toward the fiber. (Scale bar: 200 μm.) re (2 Fig. 1. Geometry of V. carteri and experimental setup. (A) The beating flaIn gella,(B) twoPhotograph per somatic cell (Inset), create a fluid flow from the anterior to the e chamber. of a micropipette holding a V. carteri colony and the ofo posterior, withμm.) a slight azimuthal component that rotates Volvox about its ber. (Scale bar: 200 tim posterior-anterior axis at angular frequency ωr . (Scale bar: 100 μm.) (B) 2 L Studies of the flagellar photoresponse utilize light sent down an optical fiber. to dunkel@math.mit.edu Ca Tuesday, January 14, 14 Spectra of light sources Fig. S1. Spectra of growth and stimulus light sources. Fig. S1. Spectra of grow Fig. S2. (A) Schematic diagram of the sample chamber. (B) Photograph of a the focal plane and pointing toward the fiber. (Scale bar: 200 μm.) f the sample chamber. (B) Photograph of a micropipette holding a V. carteri colony and th ard the (Scaleofbar: 200and μm.) Fig.fiber. S1. Spectra growth stimulus light sources. bright-field 𝝀>620, 100 fps dunkel@math.mit.edu Tuesday, January 14, 14 Photo-response at different intensities 0.25Hz amplitude of the photoresponse for top-hat stimuli of frequency 0.25 Hz, at different stimulus light intensities. dunkel@math.mit.edu Tuesday, January 14, 14 photoresponse variable that is large 18–20), fluid inertia isofnegligible andFig. the flagella-induced flowdecrease light-induced in flagellar activity and vanishes when is a direct measure the flagellar activity. 2A shows a typical there is no such change in flagellar activity.decrease The empirically detime trace of of the photoresponse, measured in terms of shows the light-induced in flagellar act a direct measure the flagellar activity. Fig. 2A a typical illumination of the termined constant β > 0 quantifies the amplitude of the decrease flagella-generated flow speed uðtÞ, normalized by the flow speed there such in τflagellar acti For a model of pðtÞ is thatno captures the change two time scales in uðtÞ∕u me trace under of time-independent the photoresponse, in#30° terms of 0 .the , and averaged over a illumination u0 measured dependence of the and τr , we require a second variable hðtÞ, which we define as a from the anterior pole. We found that a step up in light intensity termined constant β > 0 quantifies the a agella-generated flow speed uðtÞ, normalized by the flow speedrepresentation of the hidden internal biochemistry dimensionless elicits a decrease in flagellar activity on a response time scale τr , responsible for adaptation (24, 25). system of coupled For a model ofequapðtÞ that captu in uðtÞ∕u stimulation. For th followedadaptive by a recovery tophotoresponse. baseline activity a time scale averaged over #30° 0. A nder illumination u0 ,onand sticstime-independent of the (A)τa The local flagellations that is consistent with the measured uðtÞ∕u0 is and τr , we require a second variable hð om the anterior pole.measured We found that a step in light intensity peed uðtÞ (Blue), with PIVupjust above the flagella photoresponse τr p_ ¼ ðs − hÞHðs − hÞthe − p; [1] dimensionless representation of the hidd icits a decrease in flagellar activity on a response time scale τr , τ h_ ¼ s − h; [2] a nollowed light intensity, serves as a measure of flagellar activity. The responsible for adaptation (24, transforms of25). p Aan by a recovery to baseline activity on a time scale τa where the light stimulus sðtÞ that is a dimensionless measurewith of the the measur tions is consistent ed in the dark is u0 ¼ 81 μm∕s for this dataset. Two time Adaptive photo-response neglecting the Hea photoreceptor input that incorporates the eyespot directionality. The Heaviside step function Hðs − hÞ is used to ensure that a step _because ¼ ðs it− hÞHðs −wit h associated down in light stimulus cannot increase u above uτ0r, p keeps p ≥ 0. In these equations, the values p& ¼ 0 and h& ¼ s1tivity upon a s _ a suffiare stable rand global attractors in the sense that,τafter s −the h; ability of ah ¼ ciently long time under constant light stimulus s1 , the pair (p, h) relaxes to (p& , h& ). However, if s increases from s1 for t <further below. for t > 0 the the solution 0 to s2 for t ≥ 0, then where light isstimulus sðtÞtheisresponses a dimen : a short response time τr and a longer adaptation time oretical curve (Red) is from Eq. 4. (B) The times τ (Squares) ry smoothly with the stimulus light intensity, measured in or bars are standard deviations. Rð SI Text), allow that incorporates th [3] away from theu The step function Hðs − hÞ is ðs2 Heaviside − s1 Þ −t∕τ pðtÞ ¼ − e−t∕τ Þ: [4]τ r is always a f 1 − τr ∕τ down ina ðelight stimulus cannot (Fig. increase 2B), con nas.org/cgi/doi/10.1073/pnas.1000901107 p ≥is a0.sharp Intransient theseincrease equations, the va th Volvox, there in Although When τr ≪ τa , as forkeeps † pðtÞ [and decrease in uðtÞ], peaking and at a time t ∼ τr lnðτ be are stable global attractors in the a ∕τ r Þ,rents have followed by a slow relaxation back to zero, as in the measured(22), their con ciently time under constant ligh shownsetup. inlong Fig.(A) 2A.The Fig. 1. Geometry of V.flagellar carteriphotoresponse and experimental beating flaIn Volvox,ifa sst & , h&periodic The rotation of Volvox about its axis andto the (p resulting ). However, (p, h)flow relaxes gella, two per somatic cell (Inset), create a fluid from the anterior to the illumination of the photoreceptors suggest an investigation of theof 1 ms (22) is posterior, with a slightdependence azimuthal of component rotates about its t > 0 the solu t ≥Volvox 0, then for 0 to that s2 for the photoresponse on the frequency of sinusoidal time for Ca2þ a Þ; hðtÞphotoreceptor ¼ s1 e−t∕τa þ s2 ð1 − e−t∕τinput a r 1𝜇m tracers L ∕D ∼ 0.2 s ( , suggest ¼ s e to τ þ s2 ð1 − 10µm from hðtÞ cilium Ca at the ba every somatic cell, we measured the fluid motion produced by theðs − eses 23). A s Þ(22, −t∕τ u(t) = average -30° ... +30° flagellar beating by using particle image velocimetry (PIV). pðtÞ This ¼ ðe of th like that posterior-anterior at angularForfrequency bar: 100 dependence μm.) (B) of 2 the above ω model this frequency r . (Scale Fig. 2. Characteristics of the adaptive photoresponse. (A) The local flagella-axisstimulation. generated fluid speed uðtÞ (Blue), measured with PIV just above the flagella photoresponse is R light ¼ j~p∕sent s~j, where p~ an andoptical s~ are the Fourier Studies of the flagellar the photoresponse utilize down fiber. −t∕τ during a step up in light intensity, serves as a measure of flagellar activity. The transforms of p and s, respectively. R is well-approximated by a r 1 baseline flow speed in the dark is u0 ¼ 81 μm∕s for this dataset. Two time 2þ neglecting the Heaviside function in Eq. 1 (see SI Text) to give scales are evident: a short response time τr and a longer adaptation time τa . The fitted theoretical curve (Red) is from Eq. 4. (B) The times τr (Squares) ωτ 2 1 [5] Rðωs Þ ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffisffiffiffiffiaffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi : and τa (Circles) vary smoothly with the stimulus light intensity, measured in 2 2 2 2 ð1 þ ωs τr Þð1 þ ωs τa Þ terms of PAR. Error bars are standard deviations. 11172 ∣ Tuesday, January 14, 14 a τr ∕τa is only ∼ approach implicitly averages over several neighboring flagella,1 − latter and, by measuring the fluid velocity just above the flagellarDrescher tips, et al.unknown. www.pnas.org/cgi/doi/10.1073/pnas.1000901107 we obtain a natural input for the hydrodynamic τa , as of forphotoVolvox, there a sh When τr ≪models The is measu dunkel@math.mit.edu tactic turning described further pðtÞ below. Because of theinlow speed at ju [and decrease uðtÞ],fluid peaking Reynolds number associated with followed flows generated by V. carteri scribed by a slow relaxation backbytouðz The rotation of illumination of the dependence of the stimulation. For th the photoresponse transforms of p an neglecting therevers Hea the responses to step up and step down stimuli are eynolds number associated with flows generated by V. carteri scribed by uðtÞ∕u0 ¼ 1 − βpðtÞ, where pðtÞ is a photoresponse variable that is large when th 8–20), fluid inertia is negligible and the flagella-induced flow light-induced decrease in flagellar activity and a direct measure of the flagellar activity. Fig. 2A shows a typical there is no such change in flagellar activity. The me trace of the photoresponse, measured in terms of the associated adaptation; there wasβ >no changethein flage termined constant 0 quantifies amplitude agella-generated flow speed uðtÞ, normalized bywith the flow speed stics of the adaptive photoresponse. (A)over The localinflagellaFor a model ofThis pðtÞ that captures theun tw , and averaged #30° nder time-independent illumination u tivity0 upon a step down in uðtÞ∕u light0 .intensity. response andflagella τr , we require a second variable hðtÞ, which eed (Blue), measured above the om theuðtÞ anterior pole. We found thatwith a step PIV up injust light intensity the ability of V . carteri to turn toward the light, as ex dimensionless representation of the hidden intern icits a decrease in flagellar activity a response scale τr , activity. The light intensity, serves as a on measure oftime flagellar responsible adaptation (24, A system o further Atscale very light for intensities and25).long stimu llowed by a recovery to baseline activitybelow. on a time τa high ed in the dark is u0 ¼ 81 μm∕s for this dataset. Two time tions that is consistent with the measured uðtÞ∕u Adaptive photo-response : a short response time τr and a longer adaptation time τr p_ ¼ ðs − hÞHðs − hÞ − p; SI Text), allowing Volvox to avoid photodamage by sw retical curve (Red) is from Eq. 4. (B) The times τr (Squares) _ ¼ s − h; h τ a away from light. Irrespective in ry smoothly with the stimulus lightthe intensity, measured in of the stimulus lightRð a second, several s τr is always a fraction ofwhere the light whereas stimulus sðtÞ τisa ais dimensionless or bars are standard deviations. input that incorporates the eyespo (Fig. 2B), consistent withphotoreceptor early observations (14, 15). The Heaviside step function Hðs − hÞ is used to e Although the kinetics down and inbiochemistry of photorecep light stimulus cannot increase u above & keeps p ≥ 0. In these equations, the values p rents have been studied in Chlamydomonas (2, 21) and¼ nas.org/cgi/doi/10.1073/pnas.1000901107 are stable and global attractors in the sense th (22), their connection tociently the flagellar photoresponse is u long time under constant light stimul & , h& ). However, if s increases 2þ (A) The beating fla(p, h) relaxes to (p In Volvox, a step stimulus elicits a Ca current whose tim 0 to s2 for t ≥ 0, then for t > 0 the solution is m the anterior to the of 1 ms (22) isFig.too short to ofaccount for the measured τr . 2. Characteristics the adaptive photoresponse. (A) The local flagella tes Volvox about its −t∕τ þ s ð1 − e−t∕τ Þ; 2þgenerated fluid speed uðtÞ (Blue), measured hðtÞ ¼ s e with PIV just above the flagella 1 2 time for Ca to diffuse the length of the flagellum L e bar: 100 μm.) (B) 2+ during a step up in light intensity, serves as−5 a measure of flagellar activity. The 2 2 s (for ∼speed 15 μm, D∼ cm which is L ∕D ∼ 0.2(?) ðs2for −∕s), s1 Þ dataset. 𝜏r :fiber. Ca -diffusion baselineL flow in the dark is u10 ¼ 81 μm∕s this Two time −t∕τ −t∕τ own an optical pðtÞ ¼ ðe −e Þ: a longer adaptation time scales are evident: a short response time τ and 1 −triggers τr ∕τa , suggesting that the photocurrent an in to τ r 𝜏a : unknown τ . The fitted theoretical curve (Red) is from Eq. 4. (B) The times τ (Squares and τ of (Circles) smoothly with stimulus light intensity, measured inh thevary flagella, previous Ca2þ at the base τa , the as for Volvox,with there is a sharp tran When τr ≪ consistent terms of PAR. Error bars are standard deviations. produced by the pðtÞ [and decrease in uðtÞ], at a time eses (22, 23). Although the dependence of τapeaking on light inte dunkel@math.mit.edu followed by a slow relaxation back to zero, as i þ metry (PIV). This current in photoresponse Volvox, theshown decay constant like that of the H ∣ www.pnas.org/cgi/doi/10.1073/pnas.1000901107 11172 flagellar in Fig. 2A. a 0 a a r r a r a Tuesday, January 14, 14 The rotation of eses (22, 23). Although the dependen light-induced decrease in flagellar activity and 2A shows a typical elocimetry (PIV). This illumination of the current in Volvox like that of the H there is no such change in flagellar activity. The in terms of the dependence of the neighboring flagella, latter is only ms (22); bioche termined constant β >∼75 0 quantifies thethe amplitude by the flow speed stimulation. For th bove the flagellar tips, unknown. . For a model of pðtÞ that captures the tw in uðtÞ∕u eraged over #30° the photoresponse models of photoand τ , weThe require a secondadaptive variable hðtÞ, which measured response pamic in light intensity transforms of p an dimensionless representation of the hidden intern Because of τthe fluid speed just above the colony su , low onse time scale neglecting the Hea Thea Heaviside step function Hðs −1 hÞ is−βpðtÞ, used to ensur responsible for adaptation (24, 25). A system o and longer adaptation time : a short response time τ r generated by V. carteri _ ¼ − wher scribed by uðtÞ∕u τ p ¼ ðs hÞHðs − hÞ − p; n a time scale τa 0 down in light stimulus cannot increase u above u0 retical curve (Red) is fromtions Eq. 4. that (B) The times τr (Squares) is consistent with the measured uðtÞ∕u _ − h; τ h ¼ s that photoresponse variable is flagella-induced flow & larg ¼ 0 keeps p ≥ 0. In these equations, the values p Rð ry smoothly with the stimulus light intensity, measured in where the ðs light stimulus sðtÞ− is a dimensionless light-induced decrease in flagellar a Fig. a typical are stable and global attractors in the sense that, a or bars2A areshows standard deviations. _ τ p ¼ − hÞHðs hÞ − p; r photoreceptor input that incorporates the eyespo long istime under constant light there noThe such change in flagellar Heaviside step function Hðs − stimulus hÞ is used toa e ured in terms of theciently &, _ &in down light stimulus cannot increase u above h ). However, if s increases fro (p, h)termined relaxes to (p h ¼ s − h; τ a constant β > 0 quantifies the ized by the flow speed0 to s for t ≥ 0, then keeps p ≥ 0. In these equations, the values p ¼ nas.org/cgi/doi/10.1073/pnas.1000901107 for t > 0 the solution is 2 are stable and global attractors in the sense th alongmodel of constant pðtÞ that cap in uðtÞ∕u0 . For nd averaged over #30° ciently time under light stimul where the light stimulus sðtÞ is, ha). dimensionless −t∕τ −t∕τ However, if s Þ; increases (p, h) relaxes to (p and τ , we require a second variable hðtÞ ¼ s e þ s ð1 − e 2 t > 0 the ep up in light intensity t ≥ 0, then for is s for1incorporates photoreceptorr input0 tothat thesolution eyespo dimensionless representation of the hi response time scaleThe τr , Heaviside step function Hðs − hÞ is used to ðs2 − shðtÞ e þ s −t∕τ ð1 − e Þ;en 1 Þ ¼ s−t∕τ pðtÞfor ¼ cannot ðe − e u 25). Þ: A responsible adaptation (24, down in light stimulus increase above ty on a time scale τa 1 − τr ∕τa ðs − s Þ pðtÞ ¼with the ðe − e & Þ: tions that is consistent meas − τ ∕τvalues p ¼ keeps p ≥ 0. In these equations, 1the that consistent the0 ¼measured uðtÞ∕u eynolds number associated with flows tions generated by V.iscarteri 1 − βpðtÞ, where pðtÞ 0is is a scribed with by uðtÞ∕u photoresponse þ variable that is large when th 8–20), fluid inertia is negligible and the flagella-induced flow τr p_ ¼ ðs −decrease hÞHðsin−flagellar hÞ − p;activity and light-induced a direct measure of the flagellar activity. Fig. 2A shows a typical there is no such change in flagellar activity. The me trace of the photoresponse, measured in terms of the _ termined constant β > 0 quantifies the amplitude agella-generated flow speed uðtÞ, normalized by the flow speed h ¼ s − h; τ auðtÞ∕u . For a model of pðtÞ that captures the tw stics of the adaptive photoresponse. (A)over The localinflagella#30° 0 nder time-independent illumination u0 , and averaged 0 andflagella τr , we require a second variable hðtÞ, which eed (Blue), measured above the om theuðtÞ anterior pole. We found thatwith a step PIV up injust light intensity photo-response dimensionless representation of the hidden intern lightτr , activity. stimulusThe sðtÞ is a dimensionless me r ofthe icits a decrease in flagellar activity awhere response time scale light intensity, serves as a on measure flagellar variable responsible for adaptation (24, 25). A system o llowed by a recovery to baseline activity on a time scale τ a photoreceptor input that incorporates themeasured eyespot d ed in the dark is u0r ¼ 81 μm∕s for this dataset. Two time tions that is consistent with the uðtÞ∕u Photo-response model r a & a & & a 2 −t∕τa a 1 2 −t∕τa 1 r Tuesday, January 14, 14 2 r −t∕τa −t∕τr a τa ,global as for Volvox, is athe sharp areWhen stableτr ≪ and attractors in tha as for Volvox, there sense is atransien sharp tran When τr ≪ τa , there †− _ τ p ¼ ðs − hÞHðs pðtÞ [and decrease in uðtÞ], peaking at a time ∼ pðtÞ [and decrease in uðtÞ], peaking at a time t r dunkel@math.mit.edu ciently long time followed underby constant light stimul a slow relaxation back to zero, as i followed by a slow relaxation back to zero, as in th & & flagellar photoresponse shown in Fig. 2A. (p, h) relaxes to (p , h ). However, if s increases Heuristic response model BIOPHYSICS AND COMPUTATIONAL BIOLOGY outlined above remains. However, having only a small photoresponsive region complicates the heuristic picture: If the eyespots could only direct an all-or-nothing response as they move from the shaded to the illuminated side of the sphere, the best possible phototactic orientation is drawn in Fig. 4C. Such a mechanism Fig. 4. Heuristic analysis of the phototactic fidelity. A–C illustrate simplified phototaxis models. Photoresponsive regions are colored green, the region that actually displays a photoresponse is in shades of red, and shaded regions are gray. (A) If τa ¼ ∞, ωr ¼ 0, and the responsive region is as drawn, the posterior-anterior axis k will achieve perfect antialignment withdunkel@math.mit.edu the light diTuesday, January 14, 14 rection I. The time scale for turning τ t ∼ 3.3 s can be estimated by assuming Let’s try to be more quantitative ... dunkel@math.mit.edu Tuesday, January 14, 14 For Volvox, which generally near the anterior pole (Fig. 5), yet the mathematical model with predictive power. posterior-anterior axis is th outlined above remains. However, havin In general, phototactic orientation is due to an asymmetry of photoresponse kinetics, as sh sponsive region complicates the heuristi the flagellar behavior between the illuminated and shaded sides could essential only direct for an all-or-nothing high-fidelityresp ph of the organism. The mechanism that achieves this asymmetry the is shaded to the illuminated side ofmay the of adaptation. Spinning phototactic orientation is drawn in Fig. species-dependent, but it is instructive to consider a hierarchy of Frequency dependence of photo-response of unsymmetrical colony de ingredients. First, consider a nonspinning spherical organism that Fig. 3. Photoresponse frequency dependence and colony rotation. (A) The normalized flagellar photoresponse for different frequencies of sinusoidal stimulation, with minimal and maximal light intensities of 1 and 20 μmol PAR photons m−2 s−1 (Blue Circles). The theoretical response function (Eq. 5, Red Line) shows quantitative agreement, using τr and τa from Fig. 2B for 16 μmol PAR photons m−2 s−1 . (B) The rotation frequency ωr of V. carteri as a function of colony radius R. The highly phototactic organisms for which photoresponses were measured fall within the range of R indicated by the Tuesday, January 14, 14 organisms with a restricted of eyespots, such as Chlamy also required for detecting In Volvox colonies, the near the anterior pole (Fig outlined above remains. Ho sponsive region complicates could only direct an all-or-n the shaded to the illuminate phototactic orientation is d Fig. 4. Heuristic analysis of the phototactic fid phototaxis models. Photoresponsive regions ar that actually displays a photoresponse is in shad are gray. (A) If τa ¼ ∞, ωr ¼ 0, and the respon posterior-anterior axis k will achieve perfect an rection I. The time scale for turning τt ∼ 3.3 s ca that the fluid velocity on the illuminated side is value and using Eq. 8 without bottom-heaviness photoresponse may decay before the optimal o After the initial transient in A has decayed, the flagellar down-regulation) is in the region that an illustration, the configuration drawn in this p the organism would turn away from the light, i ientation is reached the steering is stopped at a dunkel@math.mit.edu with I. A remedy against this orientational limita best attainable orientation towards the light is al input for the hydrodynamic models offollowed photoThe measured adaptive response ofgenerally the to flag For Volvox, which by a slow relaxation back near the anterior pole (Fig. 5), yet −t∕τa þ s ð1 − e−t∕τa Þ; the mathematical model Because with predictive hðtÞ ¼ s e escribed further below. of thepower. low fluid speed just above the (Fig. 2 posterior-anterior axis is th 1 colony 2 surface outlined above remains. However, havin In general, phototactic orientation is due toscribed anphotoresponse asymmetry of ¼ photoresponse shown inthe Fig. r associated with flows generated by V.flagellar carteri 1− βpðtÞ, where pðtÞ isas ash by uðtÞ∕u kinetics, sponsive region complicates heuristi 0 ðs − s Þ thenegligible flagellar and behavior between the illuminated and shaded sides 2direct1 an all-or-nothing −t∕τa −when r resp could only photoresponse variable that is ertia is the flagella-induced flow essential for high-fidelity ph pðtÞ ¼ ðelarge e−t∕τ Þ:the The rotation of Volvox about its axis of the organism. TheFig. mechanism achieveslight-induced this asymmetry is shaded −toflagellar τthe the side of the r ∕τilluminated a decrease activity and e of the flagellar activity. 2A showsthat a typical of1in adaptation. Spinning may orientation is drawnsugg in Fig. species-dependent, but it is ofsuch theofphototactic photoreceptors therea ishierarchy no change flagellar activity. Thede of in unsymmetrical colony he photoresponse, measured in instructive terms illumination of to theconsider τa , that as for Volvox, there isthe a sharp transieo When τr ≪ ingredients. First, consider a nonspinning spherical organism termined constant β > 0 quantifies amplitude organisms with a restricted d flow speed uðtÞ, normalized by the flowdependence speed † of the photoresponse on th pðtÞ [and decrease in uðtÞ], peaking at a time t pðtÞ that captures the tw in uðtÞ∕u0 . For a modelofofeyespots, such as Chlamy endent illumination u0 , and averaged over #30° followed byrequire a slow relaxation back to hðtÞ, zero,which as in stimulation. For the above model this and τ , we a second variable onse. (A) The local flagellaalso required for detecting r pole. We found that a step up in light intensity flagellar photoresponse shown in Fig. 2A. dimensionless representation of the colonies, hidden interna In Volvox , in flagellar activity on a response time scale τ r ~ PIV just above the flagella ~ the photoresponse is R ¼ j p ∕ s j, The rotation of Volvox about its axis andwhere thethe resu responsible for adaptation (24, 25). A system of near the anterior pole (Fig covery to baseline activity on a time scale τa illumination of the photoreceptors suggest an inves ure of flagellar activity. The tions that theabove measured uðtÞ∕u transforms ofispofconsistent and s,withrespectively. R outlined remains. Ho dependence the photoresponse on the frequency sponsive region complicatesd for dataset. Two time stimulation. For theτrabove this stics ofthis the adaptive photoresponse. (A) The localneglecting flagellathe Heaviside in E p_ ¼ ðsmodel −function hÞHðs − frequency hÞ − p; could direct pan all-or-n eed uðtÞ (Blue), adaptation measured with PIVtime just above the flagella ~ and the photoresponse is R ¼ jp~only ∕s~j, where s~ ar a longer _the illuminate light intensity, serves as a measure of flagellar activity. The ¼ sshaded − h; to the transforms of p andτa h s, respectively. R is well-app .d(B) The τrμm∕s (Squares) in the darktimes is u0 ¼ 81 for this dataset. Two time phototactic is d τSI neglecting the Heaviside function orientation in Eq. 1 ω (see s a a short response time τr and a longer adaptation time ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffimffi Þ ¼ Rðω where the light stimulus sðtÞp is ffiaffiffiffidimensionless ght intensity, measured in s retical curve (Red) is from Eq. 4. (B) The times τr (Squares) 2eyespot 2 Þð1 ωofsþ τthe Fig. 4.that Heuristic analysis phototactic fid a the photoreceptor input incorporates ð1 ω τ ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi p Þ ¼ : ar Rðω s r y smoothly with the stimulus light intensity, measured in s phototaxis models. Photoresponsive regions 2 2 2 2 The Heaviside step is used ð1Hðs þ aω− þ ωiss τto Þen s τhÞ r Þð1 thatfunction actually displays photoresponse ina shad r bars are standard deviations. down in light stimulus cannot increase above are gray. (A) If τa ¼ ∞, ωr ¼ 0, anduthe respon posterior-anterior axis k will perfect ¼ keeps p ≥ 0. In these equations, theachieve values p& an rection I. The time scale for turning τt ∼ 3.3 s ca as.org/cgi/doi/10.1073/pnas.1000901107 are stable and global inthethe senseside that that the attractors fluid velocity on illuminated is 01107 value and using Eq. 8 withoutlight bottom-heaviness ciently long time under constant stimulu photoresponse may decay before the optimal o Fig. 3. Photoresponse frequency dependence and colony rotation. (A) The & , h& ). However, if s increases (p, h) relaxes to (p After the initial transient in A has decayed, the normalized flagellar photoresponse for different frequencies of sinusoidal t ≥ 0, flagellar then for t > 0 the issolution is that to1 sand down-regulation) in the region stimulation, with minimal and maximal light intensities0 of 20 μmol 2 for Frequency dependence of photo-response PAR photons m−2 s−1 (Blue Circles). The theoretical response function (Eq. 5, Red Line) shows quantitative agreement, using τr and τa from Fig. 2B for 16 μmol PAR photons m−2 s−1 . (B) The rotation frequency ωr of V. carteri as a function of colony radius R. The highly phototactic organisms for which photoresponses were measured fall within the range of R indicated by the Tuesday, January 14, 14 an illustration, the configuration drawn in this p the organism would turn away from−t∕τ the light, i −t∕τ a a Þ; hðtÞ ¼ iss1reached e þ steering s2 ð1 −isestopped ientation the at a dunkel@math.mit.edu with I. A remedy against this orientational limita best attainable ðs2 orientation − s1 Þ towards the light is pðtÞ ¼ ðe−t∕τa − e−t∕τr Þ: outlined above remains. However, hav sponsive region complicates the heuris could only direct an all-or-nothing res the shaded to the illuminated side of th phototactic orientation is drawn in Fi Spinning frequency vs size Fig. 4. Heuristic analysis of the phototactic f phototaxis models. Photoresponsive regions Fig. 4. Heuristic a that actually displays a photoresponse is in sha models are gray. (A) If τa phototaxis ¼ ∞, ωr ¼ 0, and the resp posterior-anteriorthat axis k will achieve perfect actually displaa rection I. The time scale for turning τt ∼ 3.3 s areongray. (A) If side τa that the fluid velocity the illuminated value and using Eq. 8 without bottom-heavine posterior-anterior photoresponse may decay before the optima Fig. 3. Photoresponse frequency dependence and colony rotation. (A) The rection I. has The timet After the initial transient in A decayed, normalized flagellar photoresponse for different frequencies of sinusoidal flagellar down-regulation) is influid the region th stimulation, with minimal and maximal light intensities of 1 and 20 μmol that the veloc −2 −1 an illustration, the configuration drawn in thi PAR photons m s (Blue Circles). The theoretical response function (Eq. 5, value and using Eq the organism would turn away from the light Red Line) shows quantitative agreement, using τr and τa from Fig. 2B for ientation is reached the steering is stopped photoresponse maa 16 μmol PAR photons m−2 s−1 . dependence (B) The rotation frequency ωr of V. carteri Fig. 3. Photoresponse frequency and colony rotation. (A) The with I. A remedy against this orientational lim as a function of colony radius R. The highly phototactic organisms for which After the initial tra normalized flagellar photoresponse for different frequencies of sinusoidal best attainable orientation towards the light photoresponses were measured fall within the range of R indicated by the down-reg stimulation, with maximal intensities ofapproxi1 and 20 μmol in a flagellar is localized small anterior region, and purple minimal box, and the and distribution of R canlight be transformed into an response as they move from the s distribution function (PDF) of ωr (Inset), by using the noisy an illustration, the s−1 (Blue Circles). The theoretical response functionor-nothing (Eq. 5, PAR photons mate m−2 probability (D) Measurements of the eyespot (Orange) pl curve of ωr ðRÞ. The purple box in A marks the range of ωr in this PDF (green organism woul Red Line) shows quantitative from 2B for (E) Volvoxthe is bottom-heavy, because th line indicates the mean), agreement, showing that the using response τtime scalesτ a and colonyFig.SI Text). r and the geometric center of the as frequency optimized to maximize the photoresponse. ientation is colony reache 16 μmol PARrotation photons m−2are s−1mutually . (B) The rotation frequency ωr of V.from carteri dunkel@math.mit.edu with I. A remedy ag as a function of colony radius R. The highly phototactic organisms for which Drescher et al. PNAS ∣ June 22, 2010 best attainable orie Tuesday, January 14, 14photoresponses were measured fall within the range of R indicated by the the flagellar behavior between the illuminated and shaded sides of the organism. The mechanism that achieves this asymmetry is species-dependent, but it is instructive to consider a hierarchy of ingredients. First, consider a nonspinning spherical organism that dunkel@math.mit.edu Tuesday, January 14, 14 Fig. 3. Photoresponse frequency dependence and colony rotation. (A) The normalized flagellar photoresponse for different frequencies of sinusoidal stimulation, with minimal and maximal light intensities of 1 and 20 μmol PAR photons m−2 s−1 (Blue Circles). The theoretical response function (Eq. 5, Red Line) shows quantitative agreement, using τr and τa from Fig. 2B for 16 μmol PAR photons m−2 s−1 . (B) The rotation frequency ωr of V. carteri as a function of colony radius R. The highly phototactic organisms for which photoresponses were measured fall within the range of R indicated by the purple box, and the distribution of R can be transformed into an approximate probability distribution function (PDF) of ωr (Inset), by using the noisy curve of ωr ðRÞ. The purple box in A marks the range of ωr in this PDF (green line indicates the mean), showing that the response time scales and colony Fig. 3. Photoresponse frequency dependence and colony rotation. (A) The normalized flagellar photoresponse for different frequencies of sinusoidal stimulation, with minimal and maximal light intensities of 1 and 20 μmol PAR photons m−2 s−1 (Blue Circles). The theoretical response function (Eq. 5, Red Line) shows quantitative agreement, using τr and τa from Fig. 2B for 16 μmol PAR photons m−2 s−1 . (B) The rotation frequency ωr of V. carteri as a function of colony radius R. The highly phototactic organisms for which photoresponses were measured fall within the range of R indicated by the essential f of adaptat of unsym organisms of eyespo also requi In Volv near the outlined a sponsive r could onl the shade phototacti Fig. 4. He phototaxis that actuall are gray. (A posterior-a rection I. Th that the flui value and u photorespo After the in flagellar do an illustrati the organis ientation is with I. A re best attaina Optimal response ! How about spatial structure ? dunkel@math.mit.edu Tuesday, January 14, 14 ses 23). Although the dependence of τsurface, intensity is a on light nds(22, of individual flagella on the colony we þ current in Volvox, the decay constant of the of ke that of the H nuum approximation in which there is a temporally ad atter is only ∼75 ms (22); the biochemical origin of τ remains an a varying surface If θ on and ϕ are the polar and of the thousands velocity. of individual flagella the colony surface, we az of the thousands of individual flagella on the colony surface, we nknown. adopt a continuum approximation in which there is a temporally ma adopt a continuum u approximation in which there is a temporally gles on a sphere, respectively, the surface velocity spatially varyingadaptive surface velocity.response If θ and ϕ are the polar and loc and spatially varying surface velocity. If θ and ϕ are the polar and The and measured of the flagella-generated ^ ^ azimuthal the surface velocity u the ex mposed intoangles u ¼on va θsphere, þ wrespectively, ϕ. We interpret u as azimuthal anglesveon a sphere, respectively, the surface velocity u ^ ^ ^ We interpret u as the vemay be decomposed into u the ¼ vθ þcolony wϕ. We interpret u as the ve-be decomposed In uid speed just above surface (Fig. 2A) can may into ube ¼ vθ^ deþ wϕ. edge of the flagellar layerlayer (32); forpractical practical locity at the edge of the flagellar (32); for reasons ra locity atreasons the edge of the flagellar layer (32); for practical reasons pðtÞ is a dimensionless cribedexperimental by uðtÞ∕u pr measurements of uβpðtÞ, are madewhere just above that layer. 0 ¼1− experimental measurements of u are made just above that layer. measurements of u are made just above that layer. ch In the absence of a light stimulus, u ¼ u0isandlarge we assumewhen that In the the absence of a light stimulus, u ¼ u0 and we assume that the hotoresponse variable that there is a large sh is constant on the colony surface because of the ratio on¼ theu colony surface because of thev0 ðθÞ∕w ratio v0 ðθÞ∕w 0 ðθÞ 0 ðθÞ is constant e of a light stimulus, u and we assume that the 0 cells (9). activity re Fig. 4.vanishes Heuristic analysis of the cells phototactic fidelity.step A–C illust orientational order of somatic (9). Following precise orientational order of Followingprecise step ght-induced decrease insomatic flagellar and when ma changes inthe light intensity, measurements offlagella vðθ;ϕ;tÞ on at fixed ϕ gre models. Photoresponsive regions are colored changes in light intensity, measurements of vðθ;ϕ;tÞ atbecause fixed ϕ phototaxis ðθÞ is constant on the colony surface of the of thousands of individual the colony s 0 all here isshow nothat such change in flagellar activity. The empirically deshow that in each region, the surface velocity displays a photoactually displays a photoresponse is in shades of red, and s in each region, the surface velocity displays a photo- that adopt a continuum approximation in which there is a response of the form shown in Fig. 2A but that the overall tational order of somatic cells (9). Following step response of the form shown in Fig. 2A but that the overall are gray. (A) If τ ¼ ∞, ω ¼ 0, and the responsive region is ermined constant β > 0 quantifies the amplitude of the decrease and spatially varying surface velocity. If θ and magnitude varies with θ axis (Fig.k 5A). We thusperfect model antialignment uðθ;ϕ;tÞ ϕbyare the magnitude varies with θ (Fig. 5A). We thus model uðθ;ϕ;tÞ by posterior-anterior will achieve wi ght intensity, measurements of vðθ;ϕ;tÞ at fixed ϕ allowing the quantities β, p, and h to depend on position: azimuthal angles on a sphere, respectively, the surface allowing the a quantities β, p, h to that dependcaptures on position: therection model ofandpðtÞ two I.time scales n uðtÞ∕u The time scale forτturning ∼ 3.3^ s can be estimate 0 . For au ¼ vθ^τ þ Th may be decomposed into w ϕ. We interpret u that the fluid velocity on the illuminated side is reduced to 0.7 each region, the surface velocity displays a photo− βðθÞpðθ;ϕ;tÞ&: [6] ¼ u0 ðθÞ½1 nd τr , we require a ¼second variable hðtÞ, which we define as aflagellar locityand atuðθ;ϕ;tÞ the edge the layer (32); [6] value uðθ;ϕ;tÞ u0 ðθÞ½1 − βðθÞpðθ;ϕ;tÞ&: using Eq. 8 of without bottom-heaviness. (B)for If τ practi < τ th ,a the form frequency shown in Fig. but that the overall experimental measurements are just above photoresponse decay before the optimal orientation has Photoresponse dependence and 2A colony rotation. (A) The un imensionless representation of the hidden internal biochemistry The measured βðθÞ is may shown in the inset of in u Fig. 5A.made The measured βðθÞ is shown in the inset in Fig. 5A. In the of light stimulus, uwe ¼ make uthe assum ab After theabsence initial transient in A has decayed, largest alized flagellar photoresponse for different frequencies of sinusoidaluðθ;ϕ;tÞ To define stimulus s ona the colony surface, usewe of photo 0 and aries with θ (Fig. 5A). We thus model by To define the stimulus s on the colony surface, we make use of esponsible for adaptation (24, 25). A system of coupled equa^ defined ^ colony try the angle ψðθ;ϕ; IÞ ¼ n^ · I, where n^surface isturned the bec ðθÞ∕w is constant on− the ratio v0down-regulation) flagellar iscos inψthe region that just int ation, with minimal and ^maximal light intensities of 1 and 20 μmol 0 ðθÞthrough ^ the angle ψðθ;ϕ;IÞ defined through cos ψ ¼ −n^ · I, where n^ unit is thenormal po to the surface. Whenorder ψ ¼ 0 of (π),somatic theinlight is directly an illustration, the configuration drawn this cells panel surprising s (Blue Circles). The theoretical response function (Eq. 5, hotons m quantities β, p, and h to depend on position: precise orientational (9). Foll is ons that is consistent with the measured uðtÞ∕u unit normal to the surface. When ψ ¼ 0 (π), the light is directly 0 a given surface patch. The light-shadow asymmeto above (behind) Front-back asymmetry a r t a t −2 −1 the organism turn away from the light, indicating that ne) shows quantitative agreement, using τr and τa from Fig. 2B for changes in would light intensity, measurements of vðθ;ϕ;tÞ above (behind) a given surface patch. The light-shadow asymmewi try in s can therefore be modeled by a factor Hðcos ψÞ. Superimientation is reached the steering is stopped at acomponent suboptimal o mol PAR photons m−2 s−1 . (B) The rotation frequency ωr of V. carteri Fig. 5. Anterior-posterior asymmetry. (A) The anterior-posterior show that in each region, the surface velocity display try in s can therefore be modeled by a factor Hðcos ψÞ. Superimde posed on this factor may be another functional dependence on ψ I. A remedy against thisshown orientational limitation would of the with fluid flow, measured 10 μm above the beating flagella, following athatbet unction ofposed colony R. Themay highly phototactic organisms for which r be another response of the form inforward Fig. 2A but dim onradius this factor functional dependence on ψ to account for the eyespot sensitivity in the direction, 0 step up in illumination at time t ¼ 0 s. The dashed line indicates the approxbest attainable orientation towards the light is drawn, if the p responsestowere measured falleyespot within sensitivity the range in of the R indicated by the account for the forward Fig. direction, with experiments on Chlamydomonas (28) supporting a depenmagnitude varies with θ (Fig. 5A). We thus model 5.imation Anterior-posterior asymmetry. (A) The anterior-posterior model. (Inset) βðθÞ is blue (with pcomponen norv 0 ðθÞ used in the isfto localized in The a numerical small anterior region, and the eyespots d Fig.and 5. Anterior-posterior asymmetry. (A)be Thetransformed anterior-posterior component e box, theexperiments distribution of R can into an approxidence ðψÞ ¼ cos ψ. class of models we consider for the with on Chlamydomonas (28) supporting a depenmalized to unity), and thequantities mean β isabove red.β,(B)the The probability of flagella to pos allowing the p, and h to depend on of the fluid flow, measured 10 μm beating flagella, following a (PDF) of the fluid flow, measured 10 μm above the beating flagella, following onent or-nothing as they move from the shaded to illu probability distribution function of ω by using theanoisy dimensionless is response therefore r (Inset), respond to light scorrelates with the size of dashed the somatic cell eyespots.the Thethe dence f ðψÞ ¼ cos ψ. The class of models we consider for the step up in illumination at time t ¼ 0 s. The line indicates approx W step up in illumination at time t ¼ 0 s. The dashed line indicates the approxwing a (D) Measurements offlow the occurs eyespot (Orange) placement yield κ light-induced decrease in fluid beyond the region of flagellar The purple box in A marks the range of ω in this PDF (green of ωimation r ðRÞ. dimensionless r s is therefore to v 0 ðθÞ used in the numerical model. (Inset) βðθÞ is blue (with p norusedofinthe the numerical model. (Inset)−βðθÞ is blue (with p no imation to v 0 ðθÞ ve βðθÞpðθ;ϕ;tÞ&: uðθ;ϕ;tÞ ¼ udynamics. ^ bottom-heavy, pprox0 ðθÞ½1 response because(E) nonlocality of fluid ψÞ: [7] sðθ;ϕ; IÞ ¼ f ðψÞHðcos SI Text). Volvox is because the center of mass dicates the mean), showing that the response time scales and colony malized to unity), and the mean β is red. (B) The probability of flagella malized to to unity), and the mean β is red. (B) The probability of flagella t p nor^ from the geometric center of the colony as indicated. ψÞ: [7] sðθ;ϕ; IÞ ¼ ðψÞHðcos respond to are lightmutually correlates with the size offmaximize the somatic the cell eyespots. The on frequency optimized to photoresponse. ella to above specification ofsize the ofdynamics the Withto the respond light correlates with the cellsurface Th The measured βðθÞthe is shown in somatic theofinset ineyespots. Fig.it 5A. light-induced decrease in fluid flow occurs beyond the region of flagellar may be sufficient for Volvox in natural environments, because s. The velocity, the angular of the colony is (31) light-induced decrease in velocity fluid flow occurs beyond the region of flagella dunkel@math.mit.edu response because the nonlocality of fluid dynamics. the ofabove specification of the dynamics of the surface With To define the stimulus s on the colony surface, would robustly navigate Volvox closer to the light, even though the we m agellar Zfluid dynamics. response because of the nonlocality of wh velocity, the angular velocity of the colony is (31) 1 3 ^ ^ erTuesday, et al. January PNAS ∣ June 22, 2010 ∣ vol. 107 ∣ no ^ organism does not swim directly toward the light. The orientathe angle ψðθ;ϕ; IÞ defined through cos ψ ¼ − n · I, wh 14, 14 g^ × k^ − n^ × uðθ;ϕ;tÞdS; [8] ΩðtÞ ¼ an τ p_ ¼ ðs − hÞHðs − hÞ − p; [1] − βðθÞpðθ;ϕ;tÞ&: [6] uðθ;ϕ;tÞ ¼ u ðθÞ½1 [2] τ h_ ¼ s − h; d βðθÞ is shown in the inset in Fig. 5A. he stimulus s on the colony we make use of of the here the light stimulus sðtÞ issurface, a dimensionless measure ^ definedinput ^the ;ϕ;IÞ through ψ ¼ −n^ · I, where n^ isdirectionality. the hotoreceptor thatcos incorporates eyespot o the surface.step When ψ ¼ Hðs 0 (π), the light to is ensure directlythat a step he Heaviside function − hÞ is used r owards the light is drawn, if the photorespon Eye-spot measurements erior region, and the eyespots display an a move from anterior polethe shaded to the illuminated sid The amplitude of the photoresponse for top-hat stimuli of frequency 0.25 Hz, at different stimulus light int espot (Orange) placement yield κ ¼ 57° $ 7° (se 𝜽=50° 𝜽=0 heavy, because the center of mass (Pink) is offs of the colony as indicated. ∣ Fig. S4. The amplitude of the photoresponse for top-hat stimuli of frequency 0.25 Hz, at different stimulus light intensities. 20𝜇m June 22, 2010 ∣ vol. 107 ∣ no. 25 ∣ 111 teri somatic cells at the anterior pole have their orange eyespots facing away from the fluid-mechanical anterio Fig. S5. (A) The V. carteri somatic cells at the anterior pole have their orange eyespots facing away from the fluid-mechanical anterior pole. (B) The somatic cellsθand at polarthe angleanterior. θ ¼ 50° from the anterior.bars: (Scale bars: μm.) (C) Illustration of the eyespot in the somatic cells and the relation olar angle ¼eyespots 50° from (Scale 2020μm.) (C) Illustration of placement the eyespot placement in theto somatic c the posterior-anterior axis k. In contrast to this schematic drawing, V. carteri colonies consist of thousands of somatic cells, as shown in Fig. 1A of the main text axis k. Inand contrast toin this as measured ref. 20.schematic drawing, V. carteri colonies consist of thousands of somatic cells, as shown in F f. 20. dunkel@math.mit.edu Tuesday, January 14, 14 Basic ingredients of a‘full’ model • self-propulsion • bottom-heaviness • photo-response kinetics • photo-response spatial variation dunkel@math.mit.edu Tuesday, January 14, 14 onstant β a>light 0 quantifies the^inamplitude of the decrease posed toangle stimulus. magnitude varies with θ (Fig. 5A). We thus mode e instead continuously changes with the angle at ugh theSciences measured βðθÞ is ^and shown the inset in Fig. 5A. th the atResearch he Council (K.D.), the Engineering and Biological Sciences trythe inksrotational can therefore be of modeled by a•factor Hðcos ψÞ.posteriorSuperimdrag the sphere (20). The second term is where g and are the directions of gravity and the sence of a light stimulus, u ¼ u and we assume that the and the ht. Together with an appropriate e The θ dependence of the 0 ^ sposed For aappropriate model of pðtÞ that captures thefactor two time scales τSciences pled equations that make upthe the model are given in the responsible for phototactic steering, w ^ allowing the quantities β, p, and h to depend on p a and where g and k are directions of gravity the posterioran o define the stimulus on the colony surface, we make use of rientads program of the Biotechnology and Biological Research Council, on this may be another functional dependence on ψ espots receive light. Together with an appropriate responsible for phototactic steering, where the integral is taken we approximated v ðθÞ by anterior axis, respectively. The first term in Eq. 8 arises from athis directionality leads to the uðθ;ϕ;t 0 Þ∕w ðθÞ is constant on the colony surface because of the require a second variable hðtÞ, which we define as a To determine the time evolution of the system of responsi ^ ^ 0anterior over of the sphere of radiu ty leads theaxis, respectively. The first insurface Eq. arises from ere angle ψðθ;ϕ; IÞ defined cos ψ¼ − n^ term · I,the where n^ in is8Legendre the theto Human Frontier Science Program (I.T.), the US Department of Energy, tothrough account for the eyespot sensitivity the forward direction, 1ð e if, as over the surface of the sphere of radius R. In a reference frame functions, −P bottom-heaviness and represents aFollowing balance between thea depentorque 1 as ement (Fig. 4D), this directionality leads to the mmetry between illuminated and uations, we solved the coupled partial differential ess representation of the hidden internal biochemistry rientational order of somatic cells (9). step with experiments on Chlamydomonas (28) supporting ve and the Schlumberger Chair Fund (R.E.G.). illuminated and bottom-heaviness and represents a balance between the torque ðθÞ½1 − βðθÞpðθ;ϕ;tÞ&: uðθ;ϕ;tÞ ¼ u where the Volvox at the origin with a normal to the surface. When ψ ¼ is0 at (π), the light is directly over the 0theisdashed of the where the Volvox the origin with a fixed orientation, the light magenta line ior-posterior component e that acts when the posterior-anterior axis is not parallel to gravity pðθ;ϕ;tÞ and hðθ;ϕ;tÞ numerically with a built-in solThe measured βðθÞ is efor for adaptation (24, 25). A system of coupled equaentation toward the light has been dence f ðψÞ ¼ cos ψ. The class of models we consider for the ^ ^ he light has been a response asymmetry between illuminated and that acts when the posterior-anterior axis is not parallel to gravity ^ ^ n light intensity, measurements of vðθ;ϕ;tÞ at fixed ϕ e (behind) a given surface patch. The light-shadow asymmedirection evolves as d I∕dt ¼ −Ω × I. direction evolves as d I∕dt ¼ −Ω × I. ngle at simple sin θ dependence fo ing flagella, following a where th hematica (Wolfram Research) between times t and at is s consistent with the measured uðtÞ∕u and the rotational drag of the sphere (20). The second term is dimensionless s is therefore To define the stimul 0signalling 2þ and the rotational drag of the sphere (20). The second term is Biol yes The measured βðθÞ is shown in the inset in Fig. 5A 23. Tamm S (1994) Ca channels and in cilia and flagella. Trends Cell n s can therefore be modeled by a factor Hðcos ψÞ. Superimne indicates the approxThe above coupled equations can be solved numerically (see The above coupled equations can b topriate in each region, the surface velocity displays a photosults. We assume that w s until perfect orientation toward the light has been 0 ^ to the integral in the equation for Ω, we used an Euler e direction the angle ψðθ;ϕ; IÞaxis defi responsible for phototactic steering, where the integral is taken ents can be op)dβðθÞ is blueenvironments (with p nornatural can be op4:305–310. responsible for phototactic steering, where the integral is taken ^ To define the stimulus s on the colony surface, on this factor may be another functional dependence on ψ SI Text), e.g., to determine the angle αðtÞ of the organism • to βðθÞ, the responsivity ofwe th _ ¼ equation τ p ðs − hÞHðs − hÞ − p; [1] SI Text), e.g., determine the angle ^ ψÞ: [7] sðθ;ϕ; IÞ ¼ f ðψÞHðcos of the form shown in Fig. 2A but that the overall r to the then solve the for IðtÞ at every time step. We ecreated probability of flagella to Jülicher 89. unit normal to the sur over the surface of the sphere of radius R. In a reference frame by the 24. Friedrich BM, F (2007) Chemotaxis of sperm cells. Proc Natl Acad Sci USA ^ ^ The a over the surface of the sphere of radius R. In a reference frame 0), which may be created by the with the light direction. It is interesting to consider two special ^ ccount for the eyespot sensitivity in the forward direction, the angle ψðθ;ϕ; IÞ defined through cos ψ ¼ − n · I, w the full model, we used with the light direction. It is interestin matic cell eyespots. The nvergence of the results by choosing a small enough ed and _ Rev edororientation varies with θh; (Fig. 5A). We thus model uðθ;ϕ;tÞ by 104:13256–13261. above (behind) a given h ¼ s − [2] τ wind-driven the above specification of the dynamics of the surface With where the Volvox is at the origin with a fixed orientation, the light cnisms, in natural environments can be opa where the Volvox is at the origin with a fixed orientation, the light cases of the model class outlined above. In the biologically faithshown in the inset in Fig.lig5 convection, or wind-driven experiments on Chlamydomonas (28) supporting a depenunit normal to the surface. When ψ ¼ 0 (π), the d. the region of flagellar SI Text), cases of the model class outlined abov as been 25. Spiro PA, Parkinson JS, ^Othmer (1997) A on model of excitation and adaptation in ^HG ^position: velocity, the angular velocity of the colony is (31) nphototaxis the quantities β, p, and h to depend even ^ try in s can therefore b direction evolves as d I∕dt ¼ −Ω × I. mics. ful “full model,” we use the measured βðθÞ and the realistic eyedirection evolves as d I∕dt ¼ −Ω × I. model, we used βðθÞ ¼ 0.3 e f ðψÞ ¼ cos ψ. The class of models we consider for the bient vorticity (30), which may be created by the above (behind) a given surface patch. The light-shad hat can counteract phototaxis even on to finding the angle of the Volvox axis with the light Z ful “full model,” we use the measured bacterial chemotaxis. Proc Natl Acad Sci USA 94:7263–7268. with the light stimulus sðtÞ is a dimensionless measure of the e to a property posed on this factor m The above coupled equations can be solved numerically (see spot directionality f ðψÞ ¼ cos ψ. In the “reduced model,” we 1 3 full model. The above coupled equations can be solved numerically (see ensionless s isalso therefore try in s can therefore be modeled by a factor Hðcos he model can be used to(1983) determine the organism ^of−directionality experiments is due to a property 26. Walsh P, Legendre L Photosynthesis natural phytoplankton under high ^ ^ g × k n × uðθ;ϕ;tÞdS; [8]In ΩðtÞ ¼ her nearby organisms, convection, or wind-driven spot f ðψÞ ¼ cos ψ. t be opptor input that incorporates the eyespot directionality. pChlamydomonas 3 to account for the eye cases of consider only a light-shadow response asymmetry—i.e., f ðψÞ ¼ en • τ and τ , the response and SI Text), e.g., to determine the angle αðtÞ of the organism axis SI Text), e.g., to determine the angle αðtÞ of the organism axis τ 8πR ðθÞ½1 − βðθÞpðθ;ϕ;tÞ&: [6] uðθ;ϕ;tÞ ¼ u0fluctuations r a functional ronments, because it bhStone velocity U, via another result from and frequency light simulating those induced by be sea another surface waves. Limnol depe posed on this factor may cellular ancestor Chlamydomonas side step function Hðs − hÞ is used to ensure that a step consider only a light-shadow respons om their center e by the AlluC 1—and use the mean of the measured βðθÞ—i.e., βðθÞ ¼ 0.3. with experiments on .ight, A mechanism that can counteract phototaxis even For the simulations, we ^ with the light direction. It is interesting to consider two special even though the with the light direction. It is interesting to consider two special ψÞ: [7] sðθ;ϕ; IÞ ¼ f ðψÞHðcos ) Oceanogr 28:688–697. ful “full ^ to account for the eyespot sensitivity in the forwa ^ where g and k are the directions of gravity and the posteriorof mass is offset from their center asymmetry. (A) The anterior-posterior component , because it ght stimulus cannot increase u above u ure srior-posterior due to clustern light. use the mean of the measured 0of1—and -driven other features are shared between the models. intensity of 16 μmol PAR e The orientadence f ðψÞ ¼ cos ψ. T cases of the model class outlined above. In the biologically faith27. Jennings HS (1901) On the significance the spiral swimming of organisms. Am Nat cases of the model class outlined above. In the biologically faitholled laboratory experiments is due to a property & & anterior axis, respectively. The first term in Eq. 8 arises from with experiments on Chlamydomonas (28) support sured βðθÞ is shown in the inset in Fig. 5A. spot dir low, measured 10 μmisZ above the beating flagella, following a ¼ 0 and h ¼ s 0. In these equations, the values p eads to a torque ottom-heaviness due to cluster1 n A phototactic turn of a hypothetical non-bottom-heavy Volvox Fig. 2B of the main text. n be overcome if, as other features are shared between th h06. the above specification of the dynamics of the surface is even dimensionless seyeis ther 35:369–378. 1 model,” ful “full model,” we use the measured βðθÞ and the realistic eyevs terior-posterior component ful “full we use the measured βðθÞ and the realistic bottom-heaviness and represents a balance between the torque dence freduced ðψÞ ¼ cos ψ. The class of models we con mination at time t ¼unicellular 0 on s.inThe dashed line indicates the approxand global attractors the sense that, after a suffihares with its ancestor Chlamydomonas UðtÞ ¼ uðθ;ϕ;tÞdS; [S1] he vertical on a ne the stimulus s the colony surface, we make use of ), the strength of the • τ , the bottom-heaviness simulated by the model is shown in Fig. 6, indicating an or (Fig. 4E) and leads to a torque y consider 28. Schaller K, David R, Uhl R (1997) How Chlamydomonas keeps track of the light once it 2 bh city, the angular velocity of the colony is (31) A phototactic turn of a hypothetica roperty spot directionality f ðψÞ ¼ cos ψ. In the “reduced model,” we that acts when the posterior-anterior axis is not parallel to gravity ating flagella, following a ia- used in the 4πR spot directionality f ðψÞ ¼ cos ψ. In the “reduced model,” we ðθÞ numerical model. (Inset) βðθÞ is blue (with p nordimensionless s is therefore , the pair g time under constant light stimulus s es with the angle at ^ has ^ where totaxis inIÞ Volvox 1 n a^ center flagellaless Volvox that is intricate link between organism rotation, adaptation, and reached the right phototactic orientation. Biophys J 73:1562–1572. as ^ Z ψðθ;ϕ; defined through cos ψ ¼ − · I, n is the sðθ Their center of mass is offset from their gae: direction with the vertical on a and the rotational drag of the sphere (20). The second term is d line indicates the approxconsider only a light-shadow response asymmetry—i.e., f ðψÞ ¼ simulated by the reduced model is sho omonas 1—and u &appropriate & ). consider only a light-shadow response asymmetry—i.e., f ðψÞ ¼ nity), and the mean β is red. (B) The probability of flagella to 1 3 with an , h However, if s increases from s for t < xes to (p self-propulsion, 29. Schaller K,the Uhl Rresponsible (1997) In A microspectrophotometric study of the shielding properties The axis of this Volvox wou 1 steering, steering. reality, however, Volvox is bottom-heavy, which is parfor phototactic where the integral is taken ws trajectories of organism to be reconstructed. coer ^ set) βðθÞ is blue (with p northful theory of phototaxis in Volvox All 1—and use the mean of the measured βðθÞ—i.e., βðθÞ ¼ 0.3. mal to the surface. When ψ ¼ (π), the light is directly ^0use ^0×the g × k − n uðθ;ϕ;tÞdS; [8] ΩðtÞ ¼ intricate link between organism r center ght correlates with the size of the somatic cell eyespots. The ^ For Volvox, this bottom-heaviness is due to clusterAll 1—and the mean of measured βðθÞ—i.e., βðθÞ ¼ 0.3. onality leads to the _ 3 other fe ψÞ: sðθ;ϕ; IÞ ¼ f ðψÞHðcos t ≥ 0, then for t > the solution is of eyespot and cell body in Chlamydomonas . Biophys Jdirection 73:1573–1578. ζR.¼ − sinðζÞ∕τ .frame For the ticularly important when the light is horizontal. In this the above speci With over the surface of the sphere of radius In a reference . photoresponse τ 8πR bh on of the photoresponse pðθ;ϕ;tÞ is plotted in Fig. 6 of e probability of flagella to rbh other features are shared between the ast four features: self-propulsion, decrease in fluidand flow occurs beyond the region ofmodels. flagellar clustersteering. In reality, however, Volvox is b een illuminated hind) a given surface patch. The light-shadow asymmeother features are shared between the models. 30. Durham WM, Kessler JO, Stocker R (2009) Disruption of vertical motility by shear where the Volvox is at the origin with a fixed orientation, the light ells in the posterior (Fig. 4E) and leads to a torque measured in ref. 19. case, we previously observed (33) that the organisms reach a final velocity, the angular v A pho ext, using the “reduced model” defined in the main somatic cell eyespots. The e ause of the nonlocality of fluid dynamics. A phototactic turn of a hypothetical non-bottom-heavy Volvox ^¼ −t∕τ −t∕τ ard theand light has been ^ layers. ^non-bottom-heavy nse kinetics, and photoresponse the above specification of thetorque dynamics of With a þ a Þ; torque ticularly important when the light dire ^ triggers formation of thin phytoplankton Science 323:1067–1070. re g k are the directions of gravity and the posteriordirection evolves as d I∕dt ¼ −Ω × I. hðtÞ s e s ð1 − e [3] A phototactic turn of a hypothetical Volvox n therefore be modeled by a factor Hðcos ψÞ. Superimset by the balance of the bottom-heaviness and the angle α 1 2 omposition of this photoresponse into spherical harnd the region ofby flagellar f gn the swimming direction with the vertical on a a simulate simulated the reduced model is shown in Fig. 6, indicating an In order to compare the1 re 31. Stone HA, Samuel ADT (1996) Propulsion of microorganisms by surface distortions. eus velocity, the angular velocity of the colony is (31) The above coupled equations can be solved numerically (see al on a rior axis, respectively. The first term in Eq. 8 arises from case, we previously observed (33) that eynolds number simulated by the reduced model is shown in Fig. 6, indicating an phototactic torque. We therefore define the “phototactic abilðθ;ϕÞ is given inopFig. S7. The photoresponse pdependence comthis intricate factor may be another functional on ψ amics. ΩðtÞaxis ¼of th onments can be ox link between organism rotation, adaptation, and ðs − s Þ Phys Rev Lett 77:4102–4104. SI Text), e.g., to determine the angle αðtÞ of the organism with the measurements Z ∼ 14 s (20). A faithful theory of phototaxis in Volvox 2 1 intricate −t∕τ −t∕τ om-heaviness and represents a balance between the torque ed and angular Volvox set by the balance of the bottom angle α a between r Þ: turn ity” A ¼ ðswimming speed toward the lightÞ∕ðswimming speedÞ. τ intricate link organism rotation, adaptation, and he “full model” during a phototactic is shown ficient for Volvox in natural environments, because it pðtÞ ¼ ðe − e [4] f bh y be created by the nt for the eyespot sensitivity in the forward direction, 1 3 om n, 32. Blake1 In JR (1971) A spherical envelope approach to ciliary propulsion. J Fluid Mech with the light direction. It is interesting to consider two special steering. reality, however, Volvox is bottom-heavy, which is par− τReynolds ^ − η dunkel@math.mit.edu viscosity, implemented a he fluid velocity r ∕τ aIn acts when the posterior-anterior axis is not parallel to gravity Both models predict that as isthe viscosity is3weincreased, while ^ ^ g × k n × uðθ;ϕ;tÞdS ΩðtÞ ¼ axis. In the low number ulsion, neglecting bottom-heaviness. re include at least four features: self-propulsion, phototactic torque. We therefore def steering. reality, however, Volvox bottom-heavy, which is parstly navigate Volvox closer to the light, even though the steering. tion, or wind-driven 46:199–208. ^ cases of thethe model class outlined above. In the biologically faithe riments on Chlamydomonas (28) supporting a depenticularly important when light direction is horizontal. In this ^ where g and k are½u&thη τ 8πR vironments, because it Forτ this we defined u¼ totactic steering fixed, the phototactic keeping the internal parametersbhτ and Hydrodynamic model Tuesday, January 14, 14 ‘Simple’ squirmer model ny behavior during a phototurn. A–E show the colony axis k (Red Arrow) tipping toward the light direction I (Aqua Arrow). Colors re Fig. 6. Colony behavior during a phototurn. A–E show the colony axis k (Red Arrow) tipping toward the light direction I (Aqua Arrow). Colors represent the ðtÞ of amplitude the down-regulation of flagellar beating in ain simplified model of phototactic showsofthe location of colonies in A–E pðtÞ of the down-regulation of flagellar beating a simplified model of phototactic steering. steering. F shows the Flocation colonies in A–E along the swimming trajectory. ajectory. of the fidelity of phototaxis in Volvox and that a quantitative understanding can be obtained if a realistic of the fidelity of phototaxis in eyespot Volvoxdirectionand that a qu dunkel@math.mit.edu ality and anterior-posterior response asymmetry are included. eyespot understanding can be obtained if a realistic These models further illustrate that if all somatic cells were SICAL and therefore a reduced phototactic torque. The sharp transition Fig. 7 occurs when the phototactic ore a inreduced phototactic torque.torque Thebecomes sharp comparable transition to the other torques in thetorque system. The simulations neglected curs when the phototactic becomes comparable torques Tuesday, Januarydue 14, 14to ambient fluid motion and included only the ‘Full’ squirmer model m Fig. S7. The photoresponse p may be decomposed into the spherical harmonics Y m l ðθ;ϕÞ via the equation pðθ;ϕ;tÞ ¼ ∑l;m alm ðtÞY l ðθ;ϕÞ. The decomposition was done for the photoresponse shown in Fig. 6 of the main text–i.e., using the reduced model. For this model, the dominant modes are the constant Y 00 , the 0 Y $1 1 modes that give a ϕ dependence similar to the light-shadow asymmetry, and Y 1 , which gives an anterior-posterior asymmetry that becomes important in this model when the organism has turned significantly toward the light. B–G display the spherical harmonics on a sphere. Fig. S8. The behavior of the photoresponse pðθ;ϕ;tÞ during a phototactic turn, using the full model defined in the main text, neglecting bottom-heaviness. A– E show the colony axis (Red Arrow) tipping toward the direction of light (Aqua Arrow) over time. The color scheme illustrates the magnitude of p. F shows the location of colonies in A–E along the swimming trajectory. dunkel@math.mit.edu Tuesday, January 14, 14 Squirmer model movie provided by K. Drescher Tuesday, January 14, 14 dunkel@math.mit.edu the flagellar behavior between the illuminated and shaded sides of the organism. The mechanism that achieves this asymmetry is species-dependent, but it is instructive to consider a hierarchy of ingredients. First, consider a nonspinning spherical organism that dunkel@math.mit.edu Tuesday, January 14, 14 Fig. 3. Photoresponse frequency dependence and colony rotation. (A) The normalized flagellar photoresponse for different frequencies of sinusoidal stimulation, with minimal and maximal light intensities of 1 and 20 μmol PAR photons m−2 s−1 (Blue Circles). The theoretical response function (Eq. 5, Red Line) shows quantitative agreement, using τr and τa from Fig. 2B for 16 μmol PAR photons m−2 s−1 . (B) The rotation frequency ωr of V. carteri as a function of colony radius R. The highly phototactic organisms for which photoresponses were measured fall within the range of R indicated by the purple box, and the distribution of R can be transformed into an approximate probability distribution function (PDF) of ωr (Inset), by using the noisy curve of ωr ðRÞ. The purple box in A marks the range of ωr in this PDF (green line indicates the mean), showing that the response time scales and colony except here Fig. 3. Photoresponse frequency dependence and colony rotation. (A) The normalized flagellar photoresponse for different frequencies of sinusoidal stimulation, with minimal and maximal light intensities of 1 and 20 μmol PAR photons m−2 s−1 (Blue Circles). The theoretical response function (Eq. 5, Red Line) shows quantitative agreement, using τr and τa from Fig. 2B for 16 μmol PAR photons m−2 s−1 . (B) The rotation frequency ωr of V. carteri as a function of colony radius R. The highly phototactic organisms for which photoresponses were measured fall within the range of R indicated by the essential f of adaptat of unsym organisms of eyespo also requi In Volv near the outlined a sponsive r could onl the shade phototacti Fig. 4. He phototaxis that actuall are gray. (A posterior-a rection I. Th that the flui value and u photorespo After the in flagellar do an illustrati the organis ientation is with I. A re best attaina Optimal response ! Phototactic ability decreases with rotation frequency Fig. S5. (A) The V. carteri somatic cells at the anterior pole have their orange eyespots facing away from the fluid-mechanical anterior pole. (B) The somatic cells and eyespots at polar angle θ ¼ 50° from the anterior. (Scale bars: 20 μm.) (C) Illustration of the eyespot placement in the somatic cells and the relation to the posterior-anterior axis k. In contrast to this schematic drawing, V. carteri colonies consist of thousands of somatic cells, as shown in Fig. 1A of the main text and as measured in ref. 20. Fig. S6. (A) Schematic diagram of the apparatus used for the population assay. B and C show distributions of the swimming angle with the light direction σ as measured for a population at the viscosity of water (B) and at 40 times the viscosity of water (C). dunkel@math.mit.edu Tuesday, January 14, 14 ality of the population swimming behavior (see SI Text), whereas features: self-propulsion, steering. In reality, however, Volvox is bottom-heavy, which is par-The model thus yields insight in the model it is solely a measure of αf . The data from several etics, and photoresponse ticularly important when the light direction is horizontal. In thisthe phototactic torque and illu populations are in Fig. 7 and(33) arethat found be in quantitacase, weshown previously observed the to organisms reach a finaltorque must be significantly tive agreement with the full model for realistic parameters (given and theachieve high-fidelity phototax angle αf set by the balance of the bottom-heaviness torque in SI Text)phototactic and in qualitative agreement thethe reduced model.abilhe low Reynolds number torque. We thereforewith define “phototactic The success the reduced speed modeltoward highlights that spinning and mming speed and angular ity” of A¼ ðswimming the lightÞ∕ðswimming speedÞ.Conclusion adaptation are the key ingredients qualitative ulated if the fluid velocity Both models predict thatfor as athe viscosity ηunderstanding is increased, whileWe have shown how accurate (31). Phototactic steering keeping the internal parameters τr and τa fixed, the phototactic colonial organism lacking a cen specifying the response of ability decreases dramatically (Fig. 7). Qualitatively, an increase autonomous cells on its anter ing for the effects of each in η reduces ωr , which leads to a reduced photoresponse (Fig. 3A) Phototactic ability decreases with rotation frequency 000901107 Tuesday, January 14, 14 tive flagellar photoresponse. T scales of this photoresponse d Drescher et al.the characteristic spinning of tion. Because the organisms r quency, the flagellar orienta seem to have coevolved to m mathematical model of phot the phototactic fidelity decrea does not spin at its natural fre assay in which spinning wa viscosity are in excellent agre This work raises a number Chief among them are the bioc scale and the reason for displ anterior part of the organism Fig. 7. The phototactic ability A decreases dramatically as ωr is reduced by of phototactically active V. car increasing the viscosity. Results from three representative populations are the frequency response functi shown with distinct colors. Each data point represents the average phototacself displays a coincidence o tic ability of the population at a given viscosity. Horizontal error bars are stanperiod (34, 35), it is natural t dard deviations, whereas vertical error bars indicate the range of population same evolutionary lineage, or mean values, when it is computed from 100 random selections of 0.1% of the tic organisms, can be underst data. A blue continuous line indicates the prediction of the full hydrodynamic The allometry of the adaptati model; the red line is obtained from the reduced model. (Inset) αðtÞ from the fordunkel@math.mit.edu study. It is also of cons full and reduced model at the lowest viscosity. tuning 𝜔 via viscosity increase Outlook & open questions • not all somatic cells photo-responsive ... why ? • what determines 𝜏 ? • chemotaxis vs phototaxis • effects of (intrinsic) noise • Chlamydomonas behave similarly ... generic ? • artificial steering devices a dunkel@math.mit.edu Tuesday, January 14, 14