A Comparison of Parthenogenetic and Sexual Embryogenesis ... · The study of aphid embryogenesis has a long history. Aphid embryos were ﬁrst observed by Leeuwenhoek and later studied

A Comparison of Parthenogenetic and SexualEmbryogenesis of the Pea Aphid Acyrthosiphonpisum (Hemiptera: Aphidoidea)

TORU MIURA,1w CHRISTIAN BRAENDLE,2,4w ALEXANDER SHINGLETON,4

GEOFFROY SISK,4 SRINIVAS KAMBHAMPATI,3 and DAVID L. STERNn4

1Department of Biology, Graduate School of Arts and Sciences,University of Tokyo, Meguro-ku, Tokyo 153-8902, Japan2Laboratory for Development and Evolution, University Museum of Zoology,Department of Zoology, Cambridge CB2 3EJ, England3Department of Entomology, Kansas State University, Manhattan, Kansas 665064Department of Ecology and Evolutionary Biology, Princeton University,Princeton, New Jersey 08544

ABSTRACT Aphids exhibit divergent modes of embryogenesis during the sexual and asexualphases of the life cycle. To explore how a single genome can give rise to these alternativedevelopmental modes, we have initiated embryological studies of the pea aphid, Acyrthosiphonpisum. Here we present a detailed description of parthenogenetic, viviparous embryonicdevelopment in the pea aphid. We compare and contrast development of the parthenogeneticembryo with that of the embryo resulting from sexual reproduction. The primary difference betweenthe embryos is the scale on which development occurs: early parthenogenetic development occurs ina volume approximately three orders of magnitude smaller than the sexual egg, largely because ofthe apparent absence of yolk in the parthenogenetic egg. This results in a drastically differentduration of syncytial energid cleavage and, presumably, patterning processes in the two embryosmust act at scales that differ by orders of magnitude. The eggs also develop on time scales that differapproximately by an order of magnitude and the timing of the embryonic movements, collectivelycalled blastokinesis, have temporally shifted relative to growth of the embryo. In addition, theendosymbiotic bacteria are transferred from mother to embryo in different ways in the two embryos.Finally, the function of the serosa has diverged greatly in the two embryos: in the sexual egg theserosa deposits a thick cuticle that protects the egg, whereas the serosa of the parthenogeneticembryo is greatly reduced and its function is unclear. The pea aphid is a useful model system forexamining how a single genome has evolved to allow divergent modes of development. J. Exp. Zool.(Mol. Dev. Evol.) 295B:59–81, 2003. r 2003 Wiley-Liss, Inc.

INTRODUCTION

One goal of evolutionary developmental biology isto understand how developmental mechanismshave evolved to generate embryological andmorphological diversity. Comparative studies haverevealed both that conserved molecules are ofteninvolved in divergent modes of embryogenesis(Patel et al., ’89; Carroll, ’95) and that divergentmodes of embryogenesis need not result indivergent adult morphologies (Grbic and Strand,’98). Aphids display divergent adult phenotypes,depending on environmental conditions experi-enced by the embryo and nymph, and alsodivergent modes of embryonic development atdifferent times in their complex life cycle (Fig. 1).These divergent modes of embryogenesis do not,

themselves, give rise to phenotypic differences innymphs and adults. Instead, the divergent modesof embryonic development appear to be specializedfor different rates of development and for devel-opment in different environments.

wThese authors contributed equally to this paper.Grant sponsor: T. M.; Grant-in-Aid for Scientific Research (No.

13740435) from the Ministry of Education, Culture, Sports, Scienceand Technology of Japan and a Grant-in-Aid from the Research for theFuture Program of the Japan Society for the Promotion of Sciences(JSPS). C.B.; Boehringer Ingelheim Fonds, the Roche ResearchFoundation, and the Janggen-Poehn-Stiftung. S. K.; USDA-NRI(Grant number 97-35302-4243). D.L.S.; BBSRC David Philips Re-search Fellowship and NIH (Grant number GM63622).

nCorrespondence to: David Stern, Department of Ecology andEvolutionary Biology, Princeton University, Princeton, NJ 08544.E-mail: [email protected]

Received 4 February 2002; Accepted 30 October 2002Published online in Wiley InterScience (www.interscience.wiley.

com). DOI: 10.1002/jez.b.00003

r 2003 WILEY-LISS, INC.

JOURNAL OF EXPERIMENTAL ZOOLOGY (MOL DEV EVOL) 295B:59–81 (2003)

The derived parthenogenetic mode of develop-ment, reinforced by viviparity and the telescopingof generations, where a nymph can containembryos within her that also contains embryos,allows rapid reproduction. Therefore aphids canrapidly colonize new host plants. Rapid reproduc-tion is also a permissive condition that has allowedthe evolution of alternative phenotypes, so calledpolyphenisms (Dixon, ’85). These polyphenismscan be extreme, with the production of winged orunwinged adults, or the production of morpholo-gically specialized altruistic soldiers (Stern andFoster, ’96), aestivating forms (Essig and Abern-athy, ’52), or sexual phenotypes. Aphids alsoproduce alternative phenotypes that may bespecialized for different host plants. These pheno-types may sometimes appear so divergent thatspecimens were originally placed in differentaphid genera. A considerable amount of aphidtaxonomic work is therefore concerned withre-uniting alternative phenotypes of the samespecies.

In contrast, development of the embryo withinthe sexual egg (that is, development within the eggproduced by the sexual generation), is slow andclosely resembles the classical hemipteran mode ofdevelopment (Johannsen and Butt, ’41). Develop-ment within the sexual egg appears to bespecialized to allow the eggs to overwinter. Despitedramatic differences in early embryonic develop-ment, the morphology of the nymphs that hatchfrom the sexual eggs is very similar to that of theviviparously produced nymphs.

Within a species, all of these alternative pheno-types are produced by a single genome and aphidstherefore provide an excellent system for studyinghow development integrates genetic and environ-mental information. In this paper we describedevelopment of the parthenogenetic and sexually-produced embryos (hereafter the ‘‘sexual em-bryos’’) of the pea aphid, Acyrthosiphon pisum.The study of aphid embryogenesis has a longhistory. Aphid embryos were first observed byLeeuwenhoek and later studied by many men who

Fig. 1. Life cycle of the pea aphid, Acyrthosiphon pisum.The life cycle begins with the diapausing eggs (surrounded byrectangle on left) that overwinter and give rise to the clonefoundresses in the spring. The diapausing eggs are the onlygeneration that results from sexual reproduction; all othergenerations result from viviparous, thelytokous parthenoge-netic reproduction. The clone foundress gives birth to asexualfemales. Several generations are then passed in which wingedor wingless (surrounded by rectangle on right) females areproduced. In the autumn, in response to shorter amounts ofdaylight, the asexual generations produce a sexual-producinggeneration. This sexual-producing generation gives rise to

sexual females and sexual males. Males are produced,genetically, by the removal of one X-chromosome during thematuration division of meiosis. However, all sperm carry anX-chromosome, so that all fertilized sexual eggs develop asfemales. The classical aphidological terminology for some ofthe stages is indicated in parentheses. Wing production inasexual females is facultative and responsive to manyenvironmental variables, such as crowding and host-plantquality, whereas wing production in males of the pea aphid iscaused by a genetic polymorphism. The stages of the life cyclethat served as the sources for the two types of embryosstudied in this paper are surrounded by rectangles.

T. MIURA ET AL.60

were particularly intent to discover how aphidsreproduced without males.

Previous workers have described embryonicdevelopment of several aphid species (Johannsenand Butt, ’41; Hagan, ’51; Brusle, ’62; Blackman,’87), but there is no complete descriptionof embryonic development in the pea aphid.Here we provide a detailed description and stagingscheme of pea aphid parthenogenetic embryos.Our staging scheme for the parthenogeneticembryos is similar to a staging scheme developedby Will (Will ’89) for what he called Aphispelargonii, a species that is now recognized asAcyrthosiphon malvae (Eastop and Hille RisLambers, ’76). However, at the time of his work,it was not known that aphids harbor intracellularendosymbiotic bacteria and that these bacteriaare transferred during embryonic development.The transfer of endosymbiotic bacteria fromthe mother to the embryo is the most peculiardeviation of the aphid’s embryonic developmentfrom that of other insects. In order to allowmore meaningful comparison with embryonicdevelopment in other insects, and also to allowa clear description of the elements of embryogen-esis that appear to be specialized for transferof the baceria, for both the parthenogeneticand sexually-produced embryos we reservedetailed description of the transfer of bacteria toa section following the description of the rest ofthe embryo.

MATERIAL AND METHODS

Aphid strains

The pea aphid cultures studied were providedby the laboratories of Mike Majerus, Universityof Cambridge and Marina Caillaud, CornellUniversity. Parthenogenetic aphid clones weremaintained on alfalfa (Medicago sativa) and broadbean (Vicia faba) plants in growth chambers witha long-day photoperiod (16L:8D) at 151–201C.Culture of the sexual generation is describedbelow.

Embryo collection and fixation

Parthenogenetic embryos

We dissected apteriform second, third andfourth-instar nymphs and adults in ice-coldphosphate buffered saline (PBS: 130 mM NaCl;7 mM Na2HPO4 � 2H2O; 3 mM NaH2PO4 � 2H2O;pH 7.0). Ovaries were fixed for approximately

30 minutes in ice-cold 4% paraformaldehydein PBS and then washed in PBS. For most ofthe work, entire ovaries were dissected fromthe mother, fixed and then stained as describedbelow.

Sexual eggs

The sexual phenotype was induced by transfer-ring aphids to a growth chamber at 161C withshort-day illumination (13L:11D). Sexual malesand females were produced in the second orthird generation after transfer. Sexual malesand females were collected and one male andthree females were placed in small petri dishescontaining a leaf of Medicago arborea that hadpreviously been inserted into 3 mL of 2% agarcontaining 1 gL�1 of Miracle-Gro. Eggs depositedonto leaves were collected with a fine paintbrush.One to three day-old eggs were fixed by placingthem in a 500 uL drop of methanol in a watchglass, covering the watch glass with a coverslip,and microwaving them for 7 seconds on thehighest setting (Sharp Carousel 1200 Watts). Thistreatment fixed the embryos and released themfrom the chorion and vitelline membrane. Thisfixation technique is less useful for more advancedembryos, after the serosa has deposited a thickcuticle. This cuticle turns black on about day three(Fig. 2; described in more detail below). Embryoswere dissected from black eggs with the followingprotocol. Eggs were dechorionated in 50% bleachfor 2 minutes. They were then transferred to a0.5 mL eppendorf tubes containing 4% parafor-maldehyde in PBS:heptane (1:1) and agitatedfor 45 minutes. The paraformaldehyde was thenreplaced with methanol and the tube was shakento pop the vitelline membrane. The embryos werestored in absolute methanol at �201C untildissection.

Sexual eggs normally begin development inrelatively high autumnal temperatures and thenoverwinter at temperatures near 01C. We mi-micked this temperature transition by maintain-ing eggs at 161C for the first sixteen days and thentransferring eggs to an incubator with a diurnalcycle of 41C for 13 H and 01C for 11 H for theremainder of development (Via, ’92).

Embryo staining

Fixed embryos were stained with various com-binations of the following fluorescent stains inPBS: DNA was stained with TOTO-3 (MolecularProbes) at a concentration of 1:1000 and propi-

PEA APHID DEVELOPMENT 61

dium iodide (Molecular Probes) at a concentrationof 1 ug/mL; F-actin was stained with OregonGreen 514 phalloidin (Molecular Probes) orTRITC phalloidin (Sigma) at a concentration of100 nM.

Some embryos were also stained with the cross-reacting antibody 4D9 (Patel et al., ’89). InDrosophila, 4D9 detects the products of theengrailed gene (en) and the en-related gene,invected. In Drosophila, invected is expressed inthe same pattern as engrailed, but expression isdelayed (Davis and Patel, 2002). It therefore notclear whether we are detecting the expression ofengrailed, invected, or both in aphids. Some sexualeggs were stained with anti-phospho (s10)-acetyl(K14)-Histone H3 (Upstate Biotechnology) toidentify mitotic nuclei. Fixed embryos werewashed for 30 minutes in 0.3% Triton-X 100 inPBS (PBT), and then blocked for 30 minutes with2% Normal Goat Serum (NGS)/0.2% BovineSerum Albumin (BSA) in PBT (BBT/NGS). Em-bryos were incubated for at least 2 hours at roomtemperature or overnight at 41C with the primaryantibody in BBT/NGS, and then washed four

times in BBT. Embryos were blocked with BBT/NGS for 30 minutes prior to incubation with thesecondary antibodies in BBT/NGS for at least 2hours at room temperature or overnight at 41C.Embryos were washed four times with PBS andthen, for fluorescent-conjugated secondary anti-bodies, mounted in Vectashield (Vector Labora-tories). Embryos were observed on a Leica SPconfocal microscope. Embryos labelled with biotin-conjugated secondary antibodies were developedusing the Vectastain ABC kit Elite (Vector) anddiaminobenzidine (Sigma Fast DAB).

Embryo size measurements

Parthenogenetic embryos were measured fromthe image in Figure 3. The lengths of embryoswere measured, but note that these lengths do notreflect the length of the embryo, since the embryosare folded in upon themselves throughout much ofdevelopment. The stages of embryos in Figure 3were estimated from the sizes of embryos exam-ined with confocal microscopy (Figs. 4–15). Sexualembryos were dissected from eggs and their length

Fig. 2. Embryology of the sexual eggs. (a) A freshlydeposited sexual egg is cream colored. (b) By the second day,the egg turns a dull green. (c) By the third day, the egg hasturned black due to the deposition of a serosal cuticle. (c’) Asectioned egg reveals the serosal cuticle (sc), which has abilayered structure, clear on the inside, black on the outside.The serosa (s) and a yolk cell (yc) are also labelled. (d) An eggcollected between 12–18 hours after egg laying (AEL) andstained with propidium iodide (PI) reveals syncytial energidsundergoing synchronous division. (e) Another egg collectedbetween 12–18 hours and stained with propidium iodide(green) and anti-phospho histone H3 (black), which marksmitotic nuclei, reveals an anterior-posterior mitotic gradient.The brightly staining area in the posterior of the egg is thebacterial mass (b). (g,j) An optical section through the centerof a PI stained egg collected between 18–24 h AEL reveals thatthe blastoderm is composed of an uneven density of cells. Thedensity is highest in the middle of the blastoderm, slightlylower in the anterior third and much lower in the mostposterior region near the bacterial mass. The posteriordiscontinuity between the high and low density cells ismarked with arrows. (h,k) A superficial view of a PI stainedegg collected between 18–24 h AEL showing the posteriordiscontinuity (marked with arrows) between high and lowdensity cells. (i,l) At a slightly later stage, another eggcollected between 18–24 h AEL shows that the high-densitycells have moved posteriorly. The most posterior region ofvery low-density cells cannot be observed, probably becausethey are obscured by the strong staining of the bacterial mass.The discontinuity between high-density cells and the moreanterior low-density cells becomes clearer (marked with

arrows). (m) In an early 2–3-day-old egg the germ band (gb)has begun to fall into the yolk (anatrepsis) with the bacterialmass (b) at the most posterior end of the germ band. (n,o) Inlate 2–3-day-old eggs, the germ band has fallen into themiddle of the yolk and the cephalic lobe has enlarged. (o) Thisembryo has been dissected out of the yolk and the short germband (approximately 100 uM long) and large bacterial massapparently surrounded by dark staining aphid nuclei can beobserved. (p–r) Three embryos collected on day 5–6 are shownin progressive stages of development. By this stage, thecephalic and thoracic regions become more distinct and theabdominal region begins to grow. (p) The earliest 5–6 dayembryos do not yet display engrailed staining. (This embryowas not stained for en, but similar embryos at this stage donot yet display en staining.) (q) The earliest embryos todisplay en staining display four weak stripes of en (brownstaining marked with asterisks). These are the three thoracicsegments plus either the first abdominal segment or the labialsegment. Gastrulation has begun on the ventral side of theembryo (pointing up in these images) as an invagination ofcells from the open ventral margins of the embryo. (r) Thelatest embryos from day 5–6 display six to seven engrailedstripes. These are probably the three cephalic and thoracicsegments plus a weak stripe in the first abdominal segment.(s) In a 7–8 day embryo the limb buds can first be observed,the germ cells can be clearly identified and the bacterial massbegins to turn yellow. (t) A day 15–16 embryo has thecomplete complement of fifteen en stripes and the embryocontinues to grow. (u) A day 21–22 embryo. The embryocontinues to grow and the limbs elongate. All scale bars are100 uM long.

Fig. 2. on page 63

T. MIURA ET AL.62


was measured. We measured multiple embryosfrom each time sample. For embryos that werecurled (e.g., Fig. 2q–t) length was estimated as thesum of several linear measurements approximat-ing the curvature of the embryo.

RESULTS

Aphids have telotrophic meroistic ovariesin which the nurse cells are retained in a commonarea at the anterior tip of the ovariole (Fig. 3).Buning (Buning ’85) provides a useful analysisof the ultrastructure and origin of germ cellsin several species of aphids, although not A. pisum.The pattern of ovariole development appears tobe similar in A. pisum to the species that Buningdescribes. The germ cells are descended fromthe same single cell that gives rise to the nursecells and are held just posterior to the nurse cellsuntil they mature as oocytes. In particular,a single embryonic germ cell gives rise to 32cells (resulting from 5 rounds of cell division),some of which serve as nurse cells and the

remainder of which serve as prospective oocytes(Blackman, ’78; Buning, ’85). The proportionof nurse cells to oocytes varies both betweenand within species. We have counted the numberof nurse cells in five parthenogenetic ovariolesby examining the germaria under DIC optics at100X (Fig. 4a). We found three germaria with21 nurse cells and two with 22 nurse cells. Whilethis variation could have been caused bythe difficulty of discerning nurse cells, we notethat Buning (Buning ’85) also found variationin nurse cell number in all three species thathe examined. This result implies that eachovariole contains approximately 11 prospectiveoocytes.

The nurse cells appear to undergo polyploidiza-tion. In other species, Buning has reportedthat the nurse cells in the parthenogeneticovarioles replicate several times (up to 16n),whereas nurse cells in the ovarioles of sexualfemales replicate approximately nine times (512n)(Buning, ’85). However, Blackman concluded thatthe nurse cells of the parthenogenetic ovarioles ofA. pisum do not become polyploid (Blackman, ’78).

Fig. 3. Dissected ovary of the pea aphid. The germaria arelocated at the tips of the radially positioned ovarioles, asindicated. Aphids develop serially within individual ovarioles(one of which is labelled) and the developmental stage of

embryos is staggered between ovarioles such that a range ofdevelopmental stages can be observed in a single ovary.Photograph kindly provided by Jim Truman.

T. MIURA ET AL.64

We have found that the nuclear volumes of thenurse cells do appear to be larger than of othercells (see Fig. 4a–d), which suggests that the nursecells become polyploid.

Below we provide a brief review of develop-ment of the embryo in the sexual egg, since

development of these embryos reflects a well-known hemipteran mode of embryonic develop-ment (Johannsen and Butt, ’41), and then adetailed staging scheme for the parthenoge-netic embryos. We place particular emphasisin both sections on the details of the earliest

Fig. 4. Germarium and stages 0, 1, and 2 of embryonicdevelopment. (a) A DIC image of the germarium (top) and amore mature (Stage 5) embryo (below) presented to allowcomparison of nuclei sizes. To allow clear separation of thenurse cells and oocytes, the ovariole in this preparation wasstretched by gently sliding the coverslip on the fixed specimen.The anterior tip of the ovariole is composed of the terminalfilament (tf), the nurse cells (nc) and prospective oocytes (poc).(b–d) Stage 0: Formation of the oocyte. The follicle cells (fc)form an epithelium surrounding the germarium (g) and theythicken to surround the oocyte (arrowhead in b, marked withan ‘‘o’’ in c). (c) The condensed chromatin of the prospectiveoocytes (poc) can be seen in this image. In addition, thephalloidin staining reveals what appear to be ring canals (rc)connecting the nurse cells to what Buning (’85) calls thetrophic core {not to be confused with the trophic cord (tc)}. (d)DIC image of the germarium and oocyte that together appearpear shaped. The oocyte nucleus, a prospective oocyte (poc)

and a nurse cell (nc) are labelled. (e,f) Stage 1: Separation ofthe oocyte from the germarium. (e) The oocyte (o) is connectedto the nurse cells in the germarium by a trophic cord (tc). Thefollicle cells (fc) begin to flatten out and the oocyte increases insize. (f) DIC image of germarium and oocyte that illustratesthe early stages of constriction of the follicular epithelium atthe base of the germarium. (g, h) Stage 2: Maturation divisionof the oocyte. The oocyte undergoes a single maturationdivision that produces a polar body (pb) that remains at theposterior of the oocyte. At mid-stage 2, the oocyte nucleus (on)moves more anteriorly and is associated with a concentrationof filamentous actin (arrowhead in g) at the periphery of theoocyte. The polar body is surrounded by filamentous actin andthe DNA of the polar body (panel h) cannot be seen in thefocal plane of panel g. All embryos in this and the followingfigures are oriented with the germarium to the left. Phalloidin(green), TOTO-3 (red). Scale bars: 100 uM (a) and 20 mM(b–h).


stages of development, since it is at these stagesthat the two types of development are mostdivergent.

Embryonic development in the sexual egg

The development of the egg deposited by thesexual female (the ‘‘sexual embryo’’) has beendescribed for several other aphid species (Johann-sen and Butt, ’41; Blackman, ’87) and thedevelopment of the sexual embryo in A. pisum issimilar to these descriptions. We therefore providean outline of development and draw attention toseveral facts that have not previously been notedto allow comparison with the parthenogeneticembryo.

The eggs deposited by the sexual female areapproximately 1 mm long and are filled with yolkexcept for an area at the posterior of the egg wherea package of endosymbiotic bacteria is positioned.The eggs posses a thin chorion and vitellinemembrane and are cream colored when firstdeposited (Fig. 2a). The eggs darken from theposterior towards the anterior pole, first to a dull

Fig. 5. Stages 3 and 4 of embryonic development. (a–d)Stage 3: Early syncytial synchronous nuclear divisions. (a) Thefirst mitotic division (arrowhead) of the embryo, at metaphaseII. (b,c) Two planes from the same embryo during metaphaseII of the second mitotic division. (d) An embryo with 16 nuclei.(e) Stage 4: Localization of nuclei to periphery of embryo.After the embryo has 16 nuclei, the embryo has doubled inlength and most of the nuclei are localized to the periphery ofthe embryo. Phalloidin (green), TOTO-3 (red). Scale bars:20 uM.

Fig. 6. Stage 5: Cellularization and blastoderm formation.(a,b) Confocal sections of early stage 5 embryos displaying thebeginnings of cellurization. Extensions of filamentous actincan be observed to extend from the cortex of the embryo(arrowheads in a) and the loss of mitotic synchrony isobserved in the most posterior nuclei. (c) A DIC image of astage 5 embryo. The syncytial nuclei at the posterior of theembryo (arrow) and in the center of the embryo can be seen.In addition, the most posterior cellularized nuclei of theblastoderm, the presumptive germ cells, display granularmaterial, which are probably polar granules. Two presump-

tive germ cells (gc) are labelled. The enlarged follicle cells atthe posterior are identified with arrowheads. (d,e,f) Threeconfocal sections from a single blastoderm embryo revealpresumptive dorsal-ventral asymmetry. (d) The most ventralsection reveals no cells in the most posterior region. (e) Alateral section reveals syncytial nuclei in both the mostposterior (arrows) and the central region of the embryo. (f) Adorsal section reveals cells across the entire anterior-posterioraxis of the embryo. Phalloidin (green), TOTO-3 (red). Scalebars: 20 uM (a,b,d) and 50 uM (c).

T. MIURA ET AL.66

green and then to black, during the first severaldays (Fig. 2b,c) (Blackman, ’87). We have deter-mined that this darkening is due to the depositionand tanning of a cuticle by the serosa (Fig. 2c’).This thick serosal cuticle is made up of twoobvious layers, an outer black layer and an innerclear layer. Unfertilized eggs do not darken (asnoted by previous authors) because they do notform a serosa and therefore do not form theserosal cuticle.

In the first 24 hours, the energids undergosynchronous divisions in a syncytium (Fig. 2d). Wehave found that this synchrony turns into mitoticdomains, or waves, at approximately 12 h after egglaying (AEL) (Fig. 2e,f). Many nuclei migrate tothe periphery of the egg, whilst some remainwithin the egg to become the yolk nuclei. Thenuclei at the periphery of the egg at blastoderm(Fig. 2g, h) are not distributed uniformly. Thehighest density of nuclei are found in approxi-mately the center third of the embryo (Fig. 2g).The anterior third has slightly lower density thanthe central third, but the most posterior domain ofthe egg, adjacent to the bacteria, has a much lowerdensity of nuclei at this stage (Fig. 2g,h,j,k). Thedistribution of nuclei then changes dramatically,as nuclei move towards the posterior pole. Thehighest density of nuclei is then found at theposterior pole (Fig. 2i,l). There are strong dis-continuities of nuclear density during this transi-tion (arrows mark the approximate location ofthese discontinuities in Fig. 2 g–l). The anteriordiscontinuity (clearest in Fig. 2 i and l) may markthe boundary of serosa, to the anterior, andembryonic primordium to the posterior. By ana-logy with the parthenogenetic embryo (see below),the posterior discontinuity (Fig. 2g,h,j,k) maymark the boundary between the embryo, to theanterior, and bacteriocyte nuclei to the posterior.However, we do not yet know whether thesediscontinuities mark the boundaries betweenthese domains and this question should beresolved by the development of molecular markersfor embryonic and extraembryonic regions.

The epithelium at the posterior of the egg formsa tube that drops into the center of the egg, withthe bacterial mass at the leading end of this tube(Fig 2m,n). The ventral side of this tube willbecome the germ band and the dorsal side willbecome the amnion. This process effectivelyinverts the anterior-posterior and dorsal-ventralaxes of the embryo, a process called anatrepsis(Sander, ’76); the posterior of the blastodermembryo sinks first into the yolk resulting in theposterior of the germ band positioned towards theanterior of the egg. In Figure 2, we have orientedthe embryos consistently with the anterior of theegg, and therefore the final anterior position of theembryo, towards the left.

The germ band elongates and segmentation andlimb-bud development proceed (Fig. 2o–u) as hasbeen previously described (Johannsen and Butt,’41; Blackman, ’87). In Figure 2p–u, the embryo isshown in its correct orientation within the egg and

Fig. 7. Stage 6: Morphogenesis of the presumptive germcells. (a) Shortly after cellularization is complete, the mostposterior cells of the blastoderm, the presumptive germ cells(gc) invaginate into the center of the embryo. (b) The germcells divide the remaining syncytial nuclei into a posterior (ps)and a central (cs) syncytium. (c) DIC image of a stage 6embryo. The polar granules can be observed. Phalloidin(green), TOTO-3 (red). Scale bars: 20 mm (a) and 50 mm (c).


it can be seen that the dorsal side of the embryopoints down towards the ventral side of the eggand the head points towards the posterior of theegg. During katatrepsis (not shown), the embryo’shead moves ventrally and then anteriorly (ana-trepsis and katatrepsis are together called blas-tokinesis). The embryo is then positioned with itshead towards the anterior of the egg and its

ventral surface towards the ventral surface of theegg.

Staging scheme for parthenogeneticembryos

Parthenogenetic females have two ovaries, eachnormally consisting of six or seven ovarioles.

Fig. 8. Stage 7: Incorporation of the maternal endosymbio-tic bacteria. (a,b) Early stages of transfer of endosymbioticbacteria. Enlarged follicle cells at the base of the embryo(arrowheads) contain high levels of filamentous actin (also seeFig. 8b). The germ cells (gc) and bacteria are labelled.Phalloidin (green), TOTO-3 (red). (c) DIC image of stage 7

embryo. The bacteria are visible as small round cells. Thepresumptive germ cells lie dorsally to the invading bacteria.One of the enlarged follical cells associated with the invadingbacteria is indicated (arrowhead). (d) A drawing of a cross-section of a stage 7 embryo. Scale bars: 20 uM (a), 50 uM (b,c)and 10 uM (d).

T. MIURA ET AL.68

Embryos develop directly, without fertilization(apomictic parthenogenesis), within the ovarioles,and serial stages of development can therefore beobserved within a single ovariole (Fig. 3). Thestaging scheme we present is based on recon-

structing the presumptive order of events frommany fixed specimens, there may therefore besmall errors involving the order of events and wemay have also failed to observe some transitionsbetween the major stages we identified.

Fig. 9. (a,b,e) Stage 8: Beginning of gastrulation. (a) Mostof the bacteria have been transferred to the embryo and thefirst invagination of the germband can be observed. Thecentral syncytium (cs), germ cells (gc) and enlarged posteriorfollicle cells (arrowhead) are labelled. (b) Under highermagnification, the follicle cells associated with the bacterialtransfer can be seen to contain high levels of filamentousactin. (e) In the DIC image, the bacteria almost fill theblastoderm cavity, but still extend to the posterior of theembryo. (c) Stage 9: Invagination of the germ band. All of thebacteria have entered the embryo and the germ bandcontinues to invaginate. The germ cells and serosa are

labelled. (d,f) Stage 10: Germ band bending and cellmembrane formation of bacteriocytes. The bacteria are dividedinto individual cells called bacteriocytes by the formation ofcell membranes around packages of bacteria each containing asingle aphid nucleus. The bacteriocytes are pushed towardsthe anterior of the embryo as the germ band continues toinvaginate. A discontinuity (arrow) forms between theepithelium of the presumptive serosa and the presumptivecephalic lobe where the bacteria are pressed against theanterior of the embryo. The cephalic lobe, serosa, germ cellsand bacteria are labelled. (a–d) Phalloidin (green), TOTO-3(red). Scale bars: 20 uM (a), 10 uM (b), and 50 uM (c–f).


Stage 0: Formation of oocyte (Fig. 4a–d)

The oocyte is derived from a nucleus origi-nating in the posterior part of the germarium(Fig. 4a). The prospective oocytes at the posteriorof the germarium contain more condensedchromatin than the nurse cells (Fig. 4c),as previously reported by Blackman (Blackman,’78). The nurse cells also appear to containpolyploid nuclei (Fig. 4b,c), as previously re-

ported by Buning (Buning ’85) for other aphidspecies.

In this first stage of oocyte formation, the folliclecells form an epithelium enclosing an area devoidof cells at the base of the germarium (arrowheadin Fig. 4b). A nucleus and its associated cytoplasmthen moves from the germarium into this spaceand the cytoplasm of this cell increases (Fig. 4c).The germarium and oocyte together appear pearshaped (Fig. 4d). Blackman (Blackman, ’78) has

Fig. 10. (a–c,g) Stage 11: S-shaped embryo. (a) The germband folds into an S shape. The bacteriocytes (b) and germcells (gc) are labelled. In early stage 11, the bacteria arepushed to the extreme anterior of the embryo. (b) In mid-stage 11, engrailed staining is first detected in the germ band.Four stripes, the three thoracic (T1,T2,T3) and first abdom-inal (A1) stripe appear at approximately the same time.Shortly afterward, we detect the first faint engrailed stainingin the three cephalic segments {Mandible (Mn), Maxilla (Mx),Labial (Lb)}. The bacteriocytes begin to move dorsally. (c) Bylate stage 11, we detect seven strong stripes of engrailedstaining. The bacteriocytes continue to move dorsally. (d–f, h)Stage 12: Twisting embryo. (d) The bending of the embryo

becomes more pronounced and the bacteriocytes move furtherdorsally. (e,f) The embryo begins to twist as more abdominalsegments are added. The images in panels e and f are differentfocal planes of the same embryo showing the twisting of thegerm band (e) and the addition of four more engrailedstaining abdominal segments (f). The bacteriocytes havemoved completely to the dorsal side of the embryo. (g) ADIC image of a dorsal view of an early stage 11 embryo showsthe bacteriocytes located anteriorly and beginning to movemore dorsally. (h) A DIC image of a dorsal view of a stage 12embryo, showing the bacteriocytes (b) straddling a centraldorsal region of the embryo. (a–f) Phalloidin (green),engrailed (red). Scale bars: 20 uM (a,b), 50 uM (c–h).

T. MIURA ET AL.70

performed a detailed study of these earliest stagesof development in the pea aphid.

Stage 1: Separation of oocyte from thegermarium (Fig. 4e–f)

The follicles cells pinch off the region betweenthe oocyte and the germarium (Fig. 4e) and theoocyte becomes more rounded (Fig. 4f). The oocyteremains attached to the center of the germariumby a trophic cord (Fig. 4e), until at least blas-toderm formation (Blackman, ’78).

Stage 2: Maturation division of the oocyte(Fig. 4g–h)

The oocyte undergoes a single maturationdivision that produces a polar body that remainsat the posterior of the oocyte (Fig. 4g), containshighly condensed chromatin (Fig. 4h) and issurrounded by filamentous actin (Fig. 4g,h). Sincethere is no loss of heterozygosity associated withthis modified meiosis (Blackman, ’79) and noevidence for meiotic recombination (Blackman,’78) this maturation division more closelyresembles meiosis II than meiosis I. Therefore,parthenogenesis is enabled in pea aphids by amodified meiosis, skipping the reduction division

Fig. 11. Stage 13: Limb bud formation. (a) The limb budsbegin to grow during this stage and the bacteriocytes againmove towards the anterior. Phalloidin (green), engrailed (red).(b) DIC image of stage 13 embryo showing the position of thebacteriocytes. Scale bars: 50 mm.

Fig. 12. Stage 14: Extended germ band. The embryo hasformed all of its segments and the abdomen is curled andtwisted around the embryo. (a) A sagittal section of theembryo shows the abdomen curled in on itself (outlined inblue). All fourteen engrailed stripes are observed by this stage.(b) The DIC image displays the same perspective as (a), withthe bacteriocytes located towards the anterior and dorsal. (c)The abdominal tip of the extended germ band embryo istwisted around in a pretzel shape. In this confocal section, theembryo is sectioned through the abdominal region (near thecenter of the embryo in a), and the abdomen can be seen totwist so that the most posterior abdominal segments (outlinedin blue) are perpendicular to the rest of the embryo. (a,c)Phalloidin (green) and engrailed (red). Scale bars: 50mm (a)and 100 mm (b,c)


of meiosis I. At mid-stage 2, the oocyte nucleusmoves anteriorly and is associated with a concen-tration of filamentous actin at the periphery of theoocyte (arrowhead in Fig. 4g). At this stage weconsider this nucleus to be the nucleus of theincipient parthenogenetic embryo.

Stage 3: Syncytial embryoFcleavagedivisions (Fig. 5a–d)

The nucleus divides synchronously four times,to the 16-nucleus stage (Fig. 5a–d). Nuclei appearto be distributed uniformly throughout the vo-lume of the embryo (Fig. 5d).

Stage 4: Localization of nuclei toperiphery (Fig. 5e)

When the embryo has sixteen nuclei, theembryo enlarges and many of the nuclei remainassociated with the periphery of the embryo(Fig. 5e). It is unclear whether the nuclei activelymigrate to the periphery, or whether nucleiremain associated with the periphery upon expan-sion of the volume of the entire embryo. Thenuclei divide synchronously once more to the32-nucleus stage.

Stage 5: Cellular blastoderm formation(Fig. 6)

For the remainder of this description, we usethe patterns of F-actin staining to infer thelocation of cell membranes. At the 32-cell stage,cell membranes begin to form around most nucleithat lie at the periphery of the embryo (Fig. 6a–c).At this stage, loss of mitotic synchrony is alsoobserved in the most posterior nuclei (Fig. 6a,b).Cell membrane formation is asymmetrical inperipheral nuclei of the embryo. Some of theposterior-most nuclei do not cellularize (arrows inFig. 6e) and this pattern of cellularization revealsan apparent dorsal-ventral asymmetry (Fig. 6d–f).We hypothesize that the side of the embryo inwhich the most posterior nuclei cellularize isdorsal (Fig. 6f). We base this suggestion on theposition of the presumptive germ cells andbacterial endosymbionts in stage 7 embryos (seebelow). After cellularization, syncytial nucleialso remain within the blastoderm (=blastocoel)(Fig. 6c,e) and these may form yolk cells. The mostposterior cellularized nuclei of the blastodermdisplay granular material (Fig. 6c) that areprobably polar granules (see Stage 6 below).

Fig. 13. Stage 15: Flip (Katatrepsis). (a,b) The flip of theembryo, also called katatrepsis, can be most easily observed byfollowing the repositioning of the CNS, which stains as aladder of strong phalloidin staining. The head, which has tothis point been positioned with its tip ventrally and pointedanteriorly, moves anteriorly, while the rest of the germ bandfollows. Panel a shows an early stage of this process, with thehead positioned halfway from the anterior end of the embryo,and panel b shows a more advanced stage, with the tip of thehead positioned approximately one-third from the anteriorend. Phalloidin (green) and engrailed (red). (c) DIC image oflate stage 15. The limbs can be observed positioned along theventral and posterior periphery of the embryo. (d) Anillustration of the different stages of the flip. All scale bars:100 mm

T. MIURA ET AL.72

Stage 6: Morphogenesis of the germ cells(Fig. 7)

The most posterior cells of the cellular blasto-derm drop from the periphery towards the insideof the blastoderm. Based on their position,morphology and apparent final destination (forexample, see Fig. 9c and Fig. 10a,b) these arelikely to be the germ cells. Will (Will, ’89) similarlyattributed cells in the same location of A. malvaeembryos as germ cells. The germ cells contact oneanother and apparently separate the syncytialnuclei within the blastoderm, which we call thecentral syncytium, from syncytial nuclei at theposterior, the posterior syncytium.

Stage 7: Incorporation of maternalendosymbiotic bacteria (Fig. 8)

The endosymbiotic bacteria are transferred tothe embryo (Fig. 8). See below for a completedescription of transfer of bacteria from mother toembryos.

Stage 8: Beginning of anatrepsis(Fig. 9a, b, e)

After most of the bacteria have been trans-ferred to the embryo, the germ band begins

to invaginate (Fig. 9a,b,e). This invaginationbegins from the dorsal posterior region of theembryo.

Stage 9: Invagination of the germbandFanatrepsis (Fig. 9c)

The posterior of the germ band in-vaginates from both the dorsal and ventralsides (Fig. 9c). This invagination, called anatrepsis(Sander, ’76), results in the inversion ofthe anterior-posterior and dorsal-ventral axes ofthe embryo. All of the bacteria have beentransferred and the channel through the follicularepithelium has presumably been closed. Thesymbionts are pushed towards the anterior of theembryo, apparently enclosed by a membrane withnuclei closely apposed to the package of bacteria(not shown). The germ cells remain closelyassociated with the bacteria and the invaginatinggerm band. Previous authors have suggestedthat germ band and extraembryonic membranescan be distinguished by this point in development.We have not yet developed molecular markersto confirm these hypotheses and we have beenunable to identify clear cellular differencesbetween these tissues as this stage. We havetherefore labelled these embryonic stages accord-

Fig. 14. Stage 16. Embryo after flip. (a) Lateral view of astage 16 embryo showing the position of the embryo after theflip. (b) DIC image of stage 16 embryo showing the position ofthe bacteriocytes. (c) A stage 16 embryo stained for nuclei

with TOTO-3 (green) showing the posterior abdomen foldedback over the bacteriocytes. (d) A ventral confocal section ofa stage 16 embryo showing the extent of limb growth. (a,d)Phalloidin (green), engrailed (red). All scale bars: 100 mm


ing to the suggestion of Will (Will, ’89). Hesuggested that the ventral epithelium is fated toform the cephalic lobe and that the dorsal-mostepithelium forms the serosa. He further suggestedthat the dorsal cells of the invaginating germ bandform the amnion and the ventral invaginatingcells give rise to the thoracic and abdominalsegments of the germ band. These suggestionsseem reasonable to us in the absence of molecularmarkers.

Stage 10: Germ band bendingFcellmembrane formation of bacteriocytes(Fig. 9d, f)

The invaginating germ band bends dorsally(Fig. 9d). The epithelium on the dorsal side ofthe invaginating epithelium is likely to become theamnion and the ventral epithelium is the embryo-nic primordium. The germ cells remain associatedwith the anterior tip of the invagination. The

Fig. 15. (a,b) Stage 17: Early germ band retraction. Duringstage 17, the germ band begins to retract and early stages ofeye differentiation can be observed. Dorsal (a) and ventral (c)sections are shown. (c,d) Stage 18: Germ band retractioncompleted. As the germ band completes retracting, eye andmuscle differentiation can be observed. Some muscles areindicated with arrows. Ventral (c) and sagittal (d) sections are

shown. The gut, bacteriocytes (b), and central nervous system(CNS) are labelled. (e) Stage 19: Advanced muscle formation.Phalloidin staining reveals a complete muscle system. Thethoracic and abdominal segments and the bacteriocytes (b)are labelled. (h) Stage 20: Mature embryo. (a–e) Phalloidin(green) and engrailed (red). Scale bars: 100 mm (a–d), 200mm(e) and 0.5 mm (f).

T. MIURA ET AL.74

epithelium on the ventral side of the embryo,which will probably develop into the cephalic lobe,thickens while the dorsal epithelium, the serosa,becomes thin. In addition, at this stage weobserved the first stages of cell membrane forma-tion around the new bacteriocytes in the embryo.The bacteriocytes are pushed firmly against theanterior of the embryo, which is associated with anapparent discontinuity between the cells of thecephalic lobe and the serosa (see arrow in Fig. 9dand f).

Stage 11: S-shape embryoFearlysegmentation (Fig. 10a–c,g)

As the germ band elongates it folds into an Sshape (Fig. 10a). The cephalic lobe is positionedventrally, the thorax in the middle and theelongating abdomen dorsally. The bacteriocytesare pushed to the most anterior pole of the embryoearly during this stage and are then moveddorsally as the germ band elongates (Fig. 10a–c,g).Engrailed staining (red) can first be detected atmid stage 11 in four segments. These foursegments are either the three thoracic segmentsand the first abdominal segment, or the threethoracic segments and the labial segment(Fig. 10b). By late stage 11, en staining is detectedin all the cephalic and thoracic segments and thefirst abdominal segment (seven stripes in total;Fig. 10c).

Stage 12: Twisting of embryo (Fig. 10d–f,h)

As the germ band continues to elongate it twists(most obvious in Fig. 10e). The bacteriocytescontinue to be pushed dorsally during this stage(Fig. 10d–f,h). By the end of Stage 12, thebacteriocytes are located mainly within and dorsalto the twisted embryo (Fig. 10f). At the beginningof this stage we detect seven en stripes and by theend of this stage we detect eleven en stripes: threecephalic, three thoracic, and five abdominalstripes.

Stage 13: Limb bud initiation (Fig. 11)

As the germ band elongates further, limb budsbegin to grow (Fig. 11a). The bacteriocytes aremoved slightly anteriorly, in the direction of thegermarium (Fig. 11). We detect thirteen en stripesat this stage: three cephalic, three thoracic, andseven abdominal stripes.

Stage 14: Extended germ band (Fig. 12)

Segmentation is completed during this stage(the embryo contains fifteen en stripes) and theembryo reaches complete germ-band extension(Fig. 12a). The bacteriocytes have again beenpushed to the anterior pole (Fig. 12a,b). Theembryo is curled in on itself in a ‘‘pretzel’’ shape,so that the most posterior segment (A10) is locatedwithin the embryo in a position perpendicular tothe rest of the embryo (Fig. 12c).

Stage 15: Flip of the embryo (katatrepsis)(Fig. 13)

After the completion of segmentation, theembryo undergoes a dramatic repositioning, tra-ditionally called katatrepsis (Sander, ’76). Thehead of the embryo, which has been positionedventrally until this point (Fig. 13a) moves ante-riorly (Fig. 13b) until it reaches the anterior pole.This results in an inversion of the dorsal/ventraland anterior/posterior axes. The bacteriocytes andgerm cells remain in approximately the samelocation throughout the flip (Fig. 13a–c). Themovement of the germband during the flip isillustrated in Fig. 13d.

Stage 16: Post flip (Fig. 14)

After katatrepsis, the embryo is positioned withits head towards the germarium and the posteriorof the germ band is folded dorsally (Fig. 14a–c).The bacteriocytes are positioned dorsally (Fig. 14b),nestled within the folded germ band (Fig. 14a,c).The limbs continue to elongate (Fig. 14d).

Stage 17: Germ band retraction(Fig. 15 a,b)

The germ band begins to retract during stage 17(early stage of retraction, Fig. 15a) and the legscontinue to grow (Fig. 15b).

Stage 18: Eye differentiation (Fig. 15 c,d)

The legs and antennae are extended to theposterior tip of the body and the germ band isalmost completely retracted (Fig. 15c,d). The brainand thoracic ganglia are compacted (Fig. 15d), thecompound eyes differentiate (Fig. 15c,d), andindividual (non-fused) muscle cells can be visua-lized with phalloidin staining (arrows in Fig. 15c).

Stage 19: Muscle formation (Fig. 15e)

The embryo grows rapidly during stage 19 andthe muscle cells fuse (Fig. 15e).


Stage 20: Mature embryo (Fig. 15f)

The mature embryo, just prior to larviposition,has a fully formed cuticle (Fig. 15f). The embryo isapproximately 1 mm long.

Incorporation of endosymbiotic bacteriainto the sexual and parthenogenetic

embryos

The endosymbiotic bacteria of aphids, Buch-nera, are housed within specialized cells calledbacteriocytes. The bacteriocytes are located in thedorsal abdomen, close to the ovaries. The bacteriaare deposited into the sexual egg and the parthe-nogenetic embryo via superficially different me-chanisms.

Observations of the invasion of bacteria intothe sexual oocyte have been made previouslyon sectioned material by J. Profft (Buchner, ’65)for the adelgid Pineus pini and by Lampel(Lampel, ’58) for the aphid Pemphigus spyrothe-cae. Their descriptions are similar to what we have

observed for the pea aphid, although wehave observed details of the invasion that arerevealed by phalloidin staining that were notobserved by earlier authors. The endosymbioticbacteria are deposited when the oocytehas reached approximately 500 uM in length. Justprior to invasion of the bacteria the follicularepithelium at the posterior of the oocyte adoptsa ‘‘bottleneck’’ shape that is caused by thecells lengthening in the apical-basal axis (arrowin Fig. 16a). The invading bacteria are firstobserved around the posterior of the oocyte(Fig. 16b). The bacteria appear to enter throughmultiple openings in the follicular epithelium(Fig. 16b,c,d). Many of the bacteria form tubessurrounded by filamentous actin, whereasother bacteria are transferred as small packagesof bacteria surrounded by filamentous actin(Fig. 16d,e,f). The bacteria remain in thisposition until the egg is laid and we hypothesizethat the high concentrations of filamentous actinsurrounding the bacteria may be involved inlocalizing the bacteria to this pole until the

Fig. 16. Invasion of endosymbiotic bacteria into the sexualoocyte. (a) Just before the bacteria invade, the follicle cellsbecome elongated along their apical-basal axis (arrow). (b)The invading bacteria are first observed as scattered bacteriainvading through multiple interstitial locations in thefollicular epithelium (arrows). (c) Multiple channels ofbacteria flow between the follicle cells (arrows). (d) Many ofthe invading bacteria form multiple channels surrounded by

filamentous actin (arrows). (e) Some of the bacteria entersingly or as a small package of cells (arrowhead) whereasother bacteria are packaged into large tubes surrounded byfilamentous actin. (f) The green channel from (e) showing thedifferent kinds of actin structures containing bacteria. (a–e)Phalloidin (green) and propidium iodide (red). Scale bars:40 mm (a), 50 mm (b,d) and 20 mm (c,e).

T. MIURA ET AL.76

bacteriocyte nuclei of the embryo invade thebacterial mass.

In the developing sexual egg, it is not clearwhich nuclei invade the bacterial mass to form thebacteriocytes. By analogy with the parthenoge-netic embryos (see below), it seems likely thatnuclei at the most posterior of the blastoderm,perhaps the low density nuclei in Fig. 2g,h,j, and k,become the bacteriocyte nuclei.

In the parthenogenetic embryo, in contrast, thebacteria enter a region of the blastoderm-stageembryo in which the nuclei that will probablybecome the bacteriocytes are already located(Fig. 8). From early stages, the embryo andassociated follicle cells appear to carry modifica-tions designed for the transfer of the bacteria. Theearliest event is the appearance of enlarged folliclecells at the posterior side of the embryo. These arepresent at least from stage 2 (Fig. 4g), but they arenot apparent in all of the images in the figuresbecause the confocal sections are often located at afocal plane that does not include these cells. Thesecells are marked with arrowheads in Fig. 6c, 8a,and 8c and labelled in Fig. 9b.

The earliest event we have observed in theembryo apparently related to bacterial transfer isthe failure of cellularization of nuclei in the mostposterior lateral and ventral region of the embryo(Fig. 6e, arrows). We hypothesize that these nucleiremain associated with the transferred bacteria(Fig. 8) and become the bacteriocyte nuclei.

The transfer of the bacteria appears to firstinvolve the fusion of a membrane-bound maternalbacterial package with the follicular epithelium inthe region of the enlarged posterior follicle cells.A channel between these enlarged follicle cellsthen appears (Fig. 8a,b) and the bacteria then flowinto the posterior of the embryo. It is not clearwhether the membrane containing the bacteria isalso transferred into the embryo. The posteriorsyncytial nuclei remain closely associated withthe incoming bacteria. The bacteriocytes cellular-ize at stage 10 and the compartmentalization ofbacteria into cells can be observed in Fig. 9d asgreen spheres of phalloidin staining approximately20–40 uM in diameter encompassing many bacter-ia. These spheres are not apparent earlier andthis is not simply a consequence of the plane of theconfocal section. The bacteriocytes remainassociated with the germ cells in the dorsalabdominal region throughout the remainder ofdevelopment and the bacteriocytes appear to bepushed around passively by the movements of thegerm band. The stereotypical pattern of move-

ment of the bacteria serves as a useful marker forstaging development.

DISCUSSION

Asexual versus sexual embryogenesis

Development of asexual embryos is strikinglydivergent from development of the sexual embryo.Hemipteran embryonic development typically oc-curs within a yolk-rich egg by the formation of alocalized blastoderm (or blastodisk) that either sitson the surface of the yolk during germ-bandformation or sinks into the center of the yolk(Johannsen and Butt, ’41; Blackman, ’87). Sexu-ally-produced aphid eggs undergo a typical hemi-pteran mode of development, with the earlyblastoderm ‘‘falling’’ into the center of the egg tocomplete germ band formation (Johannsen andButt, ’41; Blackman, ’87). Syncytial cleavage inthe sexually-produced eggs appears to persist formany rounds of division (Fig. 2), whereas cellular-ization occurs in the asexual embryos afterapproximately four rounds of division (stage 5,Fig. 6). (It is possible that in both embryossyncytial division continues until there is thesame nucleus to cytoplasm ratio (Newport andKirschner, ’82), so that this difference in thenumber of syncytial divisions is not a ‘‘pro-grammed’’ part of development but instead aplastic response to the size of the egg.) In addition,the mode of introduction of endosymbiotic bacteriainto the embryos is divergent, with bacteria beingpackaged into sexual eggs before fertilization(Blackman, ’87), whereas bacteria are transferredinto parthenogenetic embryos just after cellular-ization (stage 7, Fig. 8). Superficially, then, thedevelopment of the asexual and sexual embryos ofaphids appears highly divergent, although bothdevelopmental modes are controlled by the samegenome.

A comparison of the two modes of developmentsuggests that the early stages of development aremore divergent than the later stages. This pattern,of course, is also seen in comparisons of divergentinsect species (Davis and Patel, 2002). The mostobvious difference, as noted earlier, is the scale atwhich development occurs. The parthenogeneticembryo is approximately 60 uM long, whereas thesexual egg is approximately 1mm long. If theseembryos are using similar mechanisms to patternthe early embryo, then these mechanisms must becapable of regulation over vastly different scales.This is an intriguing possibility, given the flex-


ibility in early patterning mechanisms known inDrosophila (Namba et al., ’97; Houchmandzadehet al., 2002). An alternative hypothesis is that thetwo embryos utilize different or variant earlypatterning systems (it seems likely that themolecular mechanisms of later patterning eventsare very similar). This hypothesis would requirethat novel mechanisms have evolved to allowdevelopment of the parthenogenetic embryo.

The second major difference is that the parthe-nogenetic embryo develops in a confined space,with very little or no yolk. In contrast, the sexualembryo develops in the middle of a large amountof yolk, apparently without the same physicalconstraints on the positioning of the embryo.These physical differences presumably explainwhy the germ band of the parthenogenetic embryotwists and folds back on itself. There is limitedspace within a parthenogenetic mother and spacemay be saved by excluding yolk from the develop-ing embryo and by supplying nutrients directly tothe developing embryo. Blackman (’74) demon-strated that at least some nutrients can flowdirectly across the follicular epithelium of theparthenogenetic embryo. Buning (’85) found thatthe trophic cord of the sexual ovariole is muchlarger than that of the parthenogenetic ovariole.This suggests that nutrients, or perhaps RNA andprotein required for patterning, are supplied tothe parthenogenetic embryo and sexual oocyte indifferent ways or different quantities.

One potential consequence of this difference inthe mode of nutrition is that the stage of embryogrowth relative to katatrepsis differs between thetwo embryos. Recall that both embryos firstundergo anatrepsis, which inverts the embryonicaxes, and this is normally considered a movementto immerse the embryo in the yolk. Katatrepsisextracts the embryo from the middle of the yolk, aswell as re-establishing the body axes. In the sexualegg, katatrepsis is one of the last events to occur, atapproximately 80% of development (Fig. 17b),after the embryo has grown to almost its full

size and absorbed most of the yolk. In contrast, theparthenogenetic embryo flips early, at approxi-mately 50% of development (Fig. 17a), and mostgrowth is completed after the flip. This observa-tion may provide a clue to understanding whymany insect embryos undergo these dramaticreorientations at all. Blastokinesis (the combina-tion of anatrepsis and katatrepsis) may be amechanism to allow the embryo to more efficientlyutilize nutrients in the yolk. In the parthenoge-netic embryos, where there is little or no yolk andnutrients are probably passed directly across thefollicular epithelium, there is presumably no needfor the embryo to remain in anatrepsis during thegrowth phase of the embryo. In fact, the parthe-nogenetic embryo seems to undergo a partialanatrepsis, with the most anterior cephalic regionnever inverting completely (Fig. 10a, 13a,d).Under this interpretation, blastokinesis in theparthenogenetic embryos is vestigial, although itremains possible that the movements retain someother function that is required for proper devel-opment.

Third, the rate of development differs betweenthe parthenogenetic and sexual embryos. Both theparthenogenetic and sexual embryos grow ap-proximately continuously throughout develop-ment (Fig. 17). In the parthenogenetic embryo,the major patterning events occur in rapidsuccession in approximately the first half ofdevelopment (Fig. 17a). In contrast, the sexualegg normally overwinters for several months.Most of the major patterning events are alsocompleted early, but the embryo then undergoesan extended stage of very slow growth anddevelopment (Fig. 17b). It is not yet clear whetherthis difference is caused simply by the differencesin temperature or whether the slow growth isprogrammed into the sexual egg.

Fourth, the function of the serosa has divergedbetween the two types of embryos. In the sexualegg, the serosa encloses the yolk and deposits athick bilayered cuticle (Fig. 2c’). The precise

Fig. 17. The progression of development in parthenoge-netic and sexual embryos. (a) The length of the parthenoge-netic embryo is compared with the percent of development(bottom axis) and the stages described in the paper (top axis).The entire period of parthenogentic development is estimatedto last approximately ten days. Four of the stages areillustrated with confocal images (st. 5 and 9) and drawings(st. 14 and 15) and the time of anatrepsis and katatrepsis arelabelled. (b) The length of the sexual embryo is compared with

age of the embryo (bottom axis) and with the approximatestages for the parthenogenetic embryos. Fortuitously, em-bryonic development in the sexual eggs lasts for approxi-mately 100 days, so the number of days can be consideredapproximately equivalent to percent development. Standarddeviations are shown for sexual development because thesevalues were estimated from samples of times cohorts. No errorbars are shown for parthenogenetic development becauselengths were measured on the embryos in Fig. 3.

Fig. 17. on page 79

T. MIURA ET AL.78


function of this cuticle is unclear although it maybe an adaptation to protect the egg during the coldand dry winter months. The parthenogeneticembryo does not produce a serosal cuticle (nordoes it have a chorion or vitelline membrane,although these are produced by the follicle cells ofthe ovariole) and it is not currently clear whetherthe serosa surrounds the entire parthenogeneticembryo. The parthenogenetic embryo would not,of course, require a serosal cuticle for protectionfrom the environment. It is therefore of greatinterest to learn how the alternative serosal fatesare determined in the two embryos.

Finally, as discussed in more detail in the nextsection, the endosymbiotic bacteria are incorpo-rated into sexual and asexual embryos in differentways.

Bacteriocyte formation

One of the most peculiar characteristics of aphidembryogenesis involves the transfer of endosym-biotic bacteria from the mother’s cells to theembryo. In the parthenogenetic embryos, we haveobserved that this transfer appears to involvespecialized follicle cells that form a channel toallow the movement of bacteria into the blasto-derm embryo. The posterior syncytial nuclei of theembryo appear to receive the bacteria and remainassociated with them. It seems likely that thesenuclei become the nuclei of the bacteriocytes. Thetransfer of bacteria from mother to embryoappears to involve the transfer of a subpopulationof bacteria from a single mother’s bacteriocyte. Webase this conclusion on two observations. First, wehave never observed a maternal nucleus, whichshould be easy to distinguish because the bacter-iocyte nuclei become highly polyploid and verylarge, in the bacteria transferred to the embryo.Second, mothers do not contain enough bacter-iocytes to allow transfer of the contents of anentire bacteriocyte to each embryo. These obser-vations suggest that the bacteria are eitherpackaged into small membrane-enclosed packetsfor transfer to embryos or that the maternalbacteriocytes fuse with the embryos but onlytransfer a small number of bacteria to the embryo.

We hypothesize that the syncytial nuclei at theposterior of the blastoderm embryo become thebacteriocyte nuclei. However, there are fewernuclei in this region than there are bacteriocytesin an adult. Since bacteriocytes are large polyploidcells that apparently do not divide, there are twopossible explanations for this discrepancy. Either

these nuclei undergo further divisions beforecellularization of the bacteriocytes or other nuclei,perhaps those from the central syncytium, alsobecome associated with the bacteria and becomethe nuclei of further bacteriocytes.

The transfer of bacteria to the sexual embryoinvolves a strikingly different mechanism. Thebacteria first invade the developing oocyte throughmultiple openings in the follicular epithelium, incontrast to the single large opening leading intothe parthenogenetic embryo. In addition, theposterior follicle cells do not appear to enlarge,as they do in the parthenogenetic embryo. Afterthe embryo develops to blastoderm stage, theaphid cells appear to invade the bacterial mass toform the bacteriosome. This mode of bacterialinvasion is presumably the ancestral state foraphids, since the presence of bacterial endosym-bionts predates the origin of parthenogeneticdevelopment in the aphids and other hemipteransdisplay a similar mode of incorporation of thesymbionts into the embryo (Sander, ’76).

Although these two modes of bacterial invasionappear superficially divergent, they do share manysimilarities. In both cases, the bacteria aretransferred through openings in the follicularepithelium at the posterior of the embryo/oocyte.In the parthenogenetic embryo, cells are presentto ‘‘receive’’ these bacteria, whereas in the oocyteit is not clear what holds the bacteria in place(other than the surrounding filamentous actin)and what prevents them from dividing uncontrol-lably. This latter problem, how bacterial prolifera-tion is regulated, forms one of the centralmysteries of the aphid-endosymbiont interaction.Finally, in both embryos, nuclei or cells at or nearthe posterior of the blastoderm become associatedwith the bacteria. This is less clear in the sexualembryos, but is implied by the fact that thebacterial mass remains at the posterior of theinvaginating germ band.

The evolutionary origin of bacteriocytes re-mains unclear. Based simply on the position ofbacteriocytes within adults, one might hypothesizethat bacteriocytes are derived from fat cells. Thiswould suggest that bacteriocytes are of mesoder-mal origin. In contrast, the embryological evidencesuggests that bacteriocyte nuclei might be derivedfrom extra-embryonic cells, perhaps yolk nuclei(vitellophages). It does not seem likely, however,that all yolk nuclei are incorporated into thebacterial mass. Differentiation of these hypothesesrequires identification of tissue-specific markers inthese cells, which will allow reconstruction of the

T. MIURA ET AL.80

evolutionary origin of this novel cell type inaphids.

ACKNOWLEDGEMENTS

We thank Jim Truman for providing Figure 3.Eric Wieschaus and two anonymous reviewskindly provided extensive constructive criticismon the manuscript. T.M. thanks T. Matsumoto, H.Ishikawa and T. Fukatsu for their kind supportduring this study. S. K. wishes to thank ChurchillCollege, Cambridge for a By-Fellowship. D.L.S.thanks Churchill College, Cambridge, the Dept. ofZoology, University of Cambridge, and PrincetonUniversity.

LITERATURE CITED

Blackman RL. 1974. Incorporation of thymidine into thechromosomes of aphid (Myzus persicae) embryos. Experi-entia 30:1136–1137.

Blackman RL. 1978. Early development of the parthenoge-netic egg in three species of aphids (Homoptera: Aphididae).International Journal of Insect Morphology and Embryology7:33–44.

Blackman RL. 1979. Stability and variation in aphid clonallineages. Biol. J. Linnean Soc. 11:259–277.

Blackman RL. 1987. Reproduction, cytogenetics and develop-ment. In: Minks AK, Harrewijn P, Minks AK, Harrewijn P.Aphids: their biology, natural enemies & control, Amster-dam: Elsevier p 163–195.

Brusle S. 1962. Chronologie du developpement embryonnairedes femelles parthenogenetiques de Brevicoryne brassicae(Aphididae, Homopteres). Bull Soc Zool France 87:396–410.

Buchner P. 1965. Endosymbiosis of animals with plantmicroorganisms. New York: John Wiley & Sons.

Buning J. 1985. Morphology, ultrastructure, and germ cellcluster formation in ovarioles of aphids. J. Morphology186:209–221.

Carroll SB. 1995. Homeotic genes and the evolution ofarthropods and chordates. Nature 376:479–485.

Davis GK, Patel NH. 2002. Short, long, and beyond: Molecularand embryological approaches to insect segmentation.Annual Review of Entomology 47:669–699.

Dixon AFG. 1985. Aphid ecology. Glasgow: Blackie.Eastop VF, Hille Ris Lambers D. 1976. Survey of the world’s

aphids. The Hague: Junk.Essig EO, Abernathy F. 1952. The aphid genus Periphyllus

(Family Aphidae); a systematic, biological, and ecologicalstudy. Berkeley: University of California Press.

Grbic M, Strand MR. 1998. Shifts in the life history ofparasitic wasps correlate with pronounced alterations inearly development. Proc Nat Acad Sci USA 95:1097–1101.

Hagan HR. 1951. Embryology of the viviparous insects.New York: Ronald Press.

Houchmandzadeh B, Wieschaus E, Leibler S. 2002. Establish-ment of developmental precision and proportions in theearly Drosophila embryo. Nature 415:798–802.

Johannsen OA, Butt FH. 1941. Embryology of insects andmyriapods. New York: McGraw-Hill Book Co., Inc.

Lampel G. 1958. Die symbiontischen einrichtungen im rah-men des generationswechsels monozischer und hetero-zischer Pemphiginen der Schwarz und pyramidenpappel.Z. Morph u Okol. Tiere Bd 47:403–435.

Namba R, Pazdera TM, Cerrone RL, Minden JS. 1997. Droso-phila embryonic pattern prepair: how embryos respond tobicoid dosage alteration. Development 124:1393–1403.

Newport J, Kirschner M. 1982. A major developmentaltransition in early xenopus embryos: i. Characterizationand timing of cellular changes at the midblastula stage. Cell30:675–686.

Patel NH, Martin-Blanco E, Coleman KG, Poole SJ, Ellis MC,Kornberg TB, Goodman CS. 1989. Expression of engrailedproteins in arthropods, annelid, and chordates. Cell 58:955–968.

Sander K. 1976. Morphogenetic movements in insect embry-ogenesis. In: Lawrence PA, Lawrence PA. Insect develop-ment. Symposia of the royal entomological society ofLondon, New York: John Wiley & Sons.

Stern DL, Foster WA. 1996. The evolution of soldiers inaphids. Biological Reviews of the Cambridge PhilosophicalSociety 71:27–79.

Via S. 1992. Inducing the sexual forms and hatching the eggsof pea aphids. Entomol Exp Appl 65:119–127.

Will L. 1889. Entwicklungsgeschichte der viviparen Aphiden.Zool. Jahrb.


Documents

A Comparison of Parthenogenetic and Sexual Embryogenesis ... · The study of aphid embryogenesis has a long history. Aphid embryos were ﬁrst observed by Leeuwenhoek and later studied