Hematopoietic organs of Manduca sexta and hemocyte lineages

Dev Genes Evol (2003) 213:477–491DOI 10.1007/s00427-003-0352-6

O R I G I N A L A R T I C L E

James B. Nardi · Barbara Pilas · Elizabeth Ujhelyi ·Karl Garsha · Michael R. Kanost

Hematopoietic organs of Manduca sexta and hemocyte lineages

Received: 2 April 2003 / Accepted: 8 July 2003 / Published online: 28 August 2003� Springer-Verlag 2003

Abstract Cells of the moth immune system are derivedfrom organs that loosely envelop the four wing imaginaldiscs. The immune response in these insects is believed todepend on the activities of two main classes of hemo-cytes: plasmatocytes and granular cells. The fates of cellsthat arise from these hematopoietic organs have beenfollowed by immunolabeling with plasmatocyte-specificand granular-cell-specific antibodies. Cells within eachhematopoietic organ differ in their coherence and in theirexpression of two plasmatocyte-specific surface proteins,integrin and neuroglian. Within an organ there is nooverlap in the expression of these two surface proteins;

neuroglian is found on the surfaces of the coherent cellswhile integrin is expressed on cells that are losingcoherence, rounding up, and dispersing. A granular-cell-specific marker for the protein lacunin labels the basallamina that delimits each organ but only a small numberof granular cells that lie on or near the periphery of thehematopoietic organ. When organs are cultured in theabsence of hemolymph, all cells derived from hemato-poietic organs turn out to immunolabel with the plasma-tocyte-specific antibody MS13. The circulatingplasmatocytes derived from hematopoietic organs havehigher ploidy levels than the granular cells and represent aseparate lineage of hemocytes.

Keywords Hemocytes · Plasmatocytes · Granular cells ·Hematopoiesis · Insect immunity

Introduction

The immune response of caterpillars purportedly dependson the activities of two main hemocyte populations,granular cells and plasmatocytes, whose lineages haveremained obscure. In addition to the granular cells andplasmatocytes that comprise approximately 85–95% of allhemocytes in last instar larvae of Lepidoptera (Beetz etal., submitted; Loret and Strand 1998), three other classesof hemocytes have been usually recognized on the basisof morphology: prohemocytes, spherule cells, and oeno-cytoids. For other insect orders, the terminology appliedto hemocytes is likewise based on morphological features,but these features often differ from order to order.Morphological traits are also often a function of devel-opmental stage or the media in which hemocytes areexamined, frustrating attempts to compare hemocyteclasses from different insect orders.

A variety of lineages has been proposed for thedifferent classes of insect hemocytes. In some proposedlineages, all classes of hemocytes arise from a singlepopulation of pluripotent stem cells (Lanot et al. 2001;Yamashita and Iwabuchi 2001; Beaulaton 1979; Gupta

Edited by P. Simpson

J. B. Nardi ())Department of Entomology andDepartment of Natural Resources and Environmental Sciences,University of Illinois,320 Morrill Hall, 505 South Goodwin Avenue, Urbana,IL, 61801, USAe-mail: [email protected].: +1-217-3336590Fax: +1-217-2443499

B. PilasFlow Cytometry Facility, Biotechnology Center,University of Illinois,231 Edward R. Madigan Laboratory,1201 West Gregory Drive, Urbana, IL, 61801, USA

E. UjhelyiCenter for Microscopy and Imaging,College of Veterinary Medicine,University of Illinois,2001 South Lincoln Avenue, Urbana, IL, 61801, USA

K. GarshaImaging Technology Group,Beckman Institute for Advanced Science and Technology,University of Illinois,405 N Mathews Avenue, Urbana, IL, 61801, USA

M. R. KanostDepartment of Biochemistry,Kansas State University,104 Willard Hall, Manhattan, KS, 66506, USA

and Sutherland 1966). In other postulated lineages, thedifferent classes of hemocytes are derived from at leasttwo different populations of precursors (Gardiner andStrand 2000, 1999; Lebestky et al. 2000; Rizki and Rizki1984; Shrestha and Gateff 1982; Hinks and Arnold 1977).

During differentiation, granular cells and plasmato-cytes of Lepidoptera can be distinguished by theirultrastructure as well as their labeling patterns withspecific antibodies (Gardiner and Strand 1999; Willott etal. 1994). These two classes of hemocytes represent twoantigenically distinct lineages. The sources of these twodifferent classes of circulating hemocytes in larvalLepidoptera have been traced to (1) hematopoietic organsand (2) the proliferation of other circulating hemocytes(Ratcliffe et al. 1985). Immunolabeling of hematopoieticorgans from Spodoptera frugiperda with granular-cell-specific and plasmatocyte-specific antibodies (Gardinerand Strand 2000) revealed that about 90% of thehemocytes in each organ are plasmatocytes.

Other findings have supported the proposal thatplasmatocytes, but not granular cells, originate fromthese discrete hematopoietic organs associated with thewing discs of Lepidoptera (Hinks and Arnold 1977; Akaiand Sato 1971). Examining populations of hemocytes insitu following cauterization of Bombyx wing imaginaldiscs and their associated hematopoietic organs, Nittono(1964) observed a marked reduction in prohemocytes andplasmatocytes. Nittono interpreted these results as imply-ing that the hematopoietic organs are the source of onlyprohemocytes and plasmatocytes. Hinks and Arnold(1977) infrequently observed the presence of granularcells and spherule cells in hematopoietic organs of Euxoadeclarata caterpillars, but they concluded that theseparticular cells were derived from the hemolymph anddid not arise intrinsically. The granular cells and spherulecells were always found in those regions of the organfrom which other hemocytes had entered the hemolymphby passing through openings in the organ’s basal lamina.Although these findings support the view that hemato-poietic organs represent aggregations of stem cells thatpopulate the larval hemolymph with plasmatocytes (Gar-diner and Strand 2000; Hinks and Arnold 1977), otherauthors have proposed that both granular cells andplasmatocytes arise from hematopoietic organs (Ya-mashita and Iwabuchi 2001; Beaulaton 1979).

For Manduca sexta, the cells of hematopoietic organshave been characterized using a variety of approaches. Atthe ultrastructural level a few cells with the distinctivefeatures of granular cells have been identified at thesurface of the organ. Immunolabeling the cells ofhematopoietic organs with plasmatocyte-specific andgranular-cell-specific markers has also been used toestablish the fate of these cells at a given stage. Bothelectron microscopy and antibody labeling have shownthat granular cells are either (1) located on the outersurface of the basal lamina that delimits cells within theorgan from the surrounding hemolymph or, (2) in someinstances, they are found just beneath this surface atplaces where the basal lamina is disrupted. By culturing

hematopoietic organs in the absence of hemolymph, thefate of individual cells derived from these organs can betraced with specific antibody markers for plasmatocytesand granular cells.

Materials and methods

Rearing and staging of larvae

All insects were reared on standard artificial diet under constanttemperature (26�C) and photoperiod (18 h light:6 h ark). Eachlarval stadium (L) is designated with a number (e.g., L5). Day 0(d0) of a stadium marks the molt from the previous stadium. Eachsubsequent day (n) of a stadium marks 24x(n) hours after the moltfrom the previous stadium. Larvae were staged according to severaleasily recognized developmental landmarks: the molt from the thirdstadium (L3) to the fourth stadium (L4d0); the molt from the fourthstadium (L4) to the fifth stadium (L5d0); and the initiation of thewandering stage (L5d5) with its unique morphological andbehavioral features.

Sections for light and electron microscopy

Wing imaginal discs and surrounding hematopoietic organs wereremoved with overlying larval integument from carefully stagedlarvae and dissected in Grace’s tissue culture medium (Invitrogen).Organs were separated from the adjacent discs with fine tungstenneedles and transferred to primary fixative of 2.5% glutaraldehydeand 0.5% paraformaldehyde dissolved in a 0.1 M cacodylate buffercontaining 0.18 mM CaCl2 and 0.58 mM sucrose (Tolbert andHildebrand 1981). After several rinses in the cacodylate buffercontaining only sucrose and CaCl2, tissues were post-fixed in thesame buffer containing 2% OsO4 in place of aldehydes. Tissueswere once again rinsed in cacodylate buffer and dehydrated in aseries of ethanol concentrations (10–100%) before final infiltrationwith propylene oxide and Medcast resin.

Sections (1–2 �m) for light microscopy were cut with a ReichertUltracut E, arranged on glass slides, and stained with 1% toluidineblue in a 1% borax solution. Sections for electron microscopy wereviewed with a Hitachi 600 at 75 kV.

Immunolabeling

Following a half-hour fixation with 4% paraformaldehyde dis-solved in phosphate-buffered saline (PBS, pH 7.4), tissues wererinsed several times with PBS and then transferred to blockingbuffer (PBS +3% normal horse serum or 10% normal goat serum+0.1% Triton X-100) for at least 30 min. Horse serum was added toblocking buffer when labeling with horseradish peroxidase (HRP);goat serum was added to blocking buffer when tissues were labeledwith Texas Red or fluorescein. Standard immunolabeling of tissueshas been described in earlier publications (Nardi et al. 1999; Nardiand Miklasz 1989). A 1:10,000 dilution was used for each of thefollowing monoclonal antibodies (MAbs): MS13, MS34, MAb3B11, and MAb 15D11. MS13 and MS34 are specific forplasmatocytes and recognize the beta subunit of integrin (Levinet al., in preparation). MAb 3B11 recognizes the cell adhesionprotein neuroglian; in addition to being expressed by a subpopu-lation of plasmatocytes, neuroglian is expressed by a variety ofepithelial, neural and glial cells (Nardi 1994). MAb 15D11recognizes the extracellular matrix protein lacunin that is expressedby granular cells but not plasmatocytes (Nardi et al. 2001). Ascontrols, tissues were treated with normal mouse serum (1:1,000)for 12 h in the cold prior to treatment with secondary antibodies.Whole immunolabeled tissues were mounted in 30% 0.1 M Tris(pH 9.0) in glycerol.

478

To double-label organs with two different mouse monoclonalantibodies and yet ensure that secondary labeling of these mousemonoclonal antibodies did not result in cross reactivity of the twolabels, one monoclonal antibody (MAb 3B11) was first labeled withgoat anti-mouse fluorescein according to the above procedure. Thesecond mouse primary antibody was biotinylated and secondarilylabeled with Texas Red-avidin D (Vector Laboratories). After thefluorescein-labeled secondary antibody was rinsed from tissues, theorgans were incubated overnight at 4�C with biotin-MS13 diluted1:1,000 in blocking buffer. Unbound biotinylated antibody wasrinsed from organs at room temperature, and tissues then wereexposed overnight in the cold to Texas Red-avidin D diluted1:1,000 in blocking buffer. Several final rinses with blocking bufferat room temperature preceded mounting of tissues on glass slides.

Sections of tissues labeled with HRP were prepared from tissuesthat had been refixed with the aldehydes and 2% OsO4 as describedin the preceding section. After dehydration and infiltration withMedcast resin, the tissues were sectioned at 1–2 �m and mountedon glass slides without staining.

Organ cultures and hemocyte cultures

Whole hematopoietic organs were dissected under sterile condi-tions in Grace’s insect culture medium (Invitrogen) whose pH hadbeen adjusted to 6.5 and then filter sterilized. Organs adhered tosterile cover glasses (22 mm2) that were placed on the bottom ofsmall Falcon culture dishes (35�10 mm) containing Grace’smedium (pH 6.5) plus 20% fetal calf serum (Sigma). Cultureswere maintained at 26�C in large culture dishes lined with moistfilter paper.

Circulating hemocytes from last instar larvae were added toculture dishes prepared as described above. As soon as a drop oflarval hemolymph touched the medium, the blood was swirled, andcells were allowed to settle and adhere to the glass coverslip for 1 h.After this culture period, the medium and unattached hemocyteswere removed and adherent cells were fixed with PBS containing4% paraformaldehyde for 30 min at room temperature. These fixedcells on cover glasses were then rinsed several times with PBS andlater processed for immunolabeling as described earlier.

Confocal microscopy

Whole organs that had been double-labeled with fluorescein andTexas Red were observed using laser scanning confocal microscope(LSCM) instrumentation housed and maintained by the ImagingTechnology Group at the University of Illinois’ Beckman Institute.Imaging was performed using the Leica SP-2 spectral confocalinstrumentation equipped with a �20 plan-apochromatic objectiveand a �63 plan-apochromatic oil immersion objective. The 488laser line from an argon laser was selected for excitation offluorescein, while the 543 laser line from a helium-neon laser wasused to excite Texas Red. To minimize overlap of emission spectrafor the two fluorochromes, excitation was performed sequentially.The emission detection ranges for fluorescein and Texas Redlabeling were tuned respectively to the following bandwidths: 500–535 nm and 593–622 nm. Three-dimensional volumetric data setsconsisting of multiple optical sections were processed using LeicaConfocal Software to yield extended focus projections.

Flow cytometry

Larval mesothoracic wing discs along with the adjacent overlyingintegument were dissected and transferred to Grace’s tissue culturemedium. Hematopoietic organs were removed intact from thesurfaces of these forewing imaginal discs using tungsten needlesand watchmakers’ forceps. Each organ was transferred to a separatesiliconized Eppendorf tube containing 500 �l maceration solutionconsisting of glycerin, glacial acetic acid, and water (1:1:13). Thismixture completely disaggregates tissues yet maintains the integrity

of individual cells (David 1973). Organs remained in this macer-ation solution for at least 15 min and were thoroughly vortexedprior to addition of 500 �l 4% paraformaldehyde dissolved in PBS.

Cells of wing imaginal discs from L5d0 larvae were chosen asexamples of known diploid cells. At this larval stage, neithertracheal cells nor hemocytes have colonized the extracellular spacebetween the two monolayers of the wing discs (Nardi et al. 1985).The cells of these wing discs were disaggregated according to theprocedure above.

The disaggregated cells remained in fixative (2% paraformal-dehyde) for 30 min before being centrifuged at 200 g in a swingingbucket rotor. The cell pellet was washed first with 1.0 ml PBS andthen with 1.0 ml blocking buffer (PBS +10% normal goat serum+0.1% Triton X-100). The washed pellet was suspended in 100 �lblocking buffer containing the mitosis marker, rabbit anti-phospho-histone H3 (Upstate), at a concentration of 10 �g/ml. Controls wereexposed to 1:1,000 normal rabbit serum. The fixed cells remainedin the primary rabbit antibody for 3 h at room temperature and thenwere washed twice with 1.0 ml blocking buffer before beingincubated with 15 �g/ml fluorescein goat anti-rabbit IgG (Vector)in 100 �l blocking buffer. After the cells had been exposed tosecondary antibody for 2 h at room temperature, they were washedonce with 1.0 ml blocking buffer and 1.0 ml PBS and stored at 2�Cin the dark until analyzed with flow cytometry no more than 2 dayslater.

Both the total number of cells per hematopoietic organ as wellas the number of fluorescein-labeled mitotic cells per organ werecalculated by simultaneously adding a known number of 6-�mcarmine beads (Molecular Probes, 620 nm emission) to suspensionsof macerated hematopoietic organs.

To establish the relationship between the class of circulatinghemocyte (granular cells or plasmatocytes) and their ploidy levels,blood from larvae (L5d0, L5d4, L5d5) was collected in anticoag-ulant buffer (AC buffer). To each tube containing 750 �l AC buffer,four drops of blood were added, mixed, and then fixed with 750 �l4% paraformaldehyde in PBS for 30 min at room temperature.These fixed, circulating hemocytes were then immunolabeled withplasmatocyte-specific MS13 and granular-cell-specific MAb15D11 following the procedure outlined for immunolabeling ofmacerated cells from hematopoietic organs. Propidium iodide(1 mg/ml in distilled H2O) was added (8% by volume) tosuspensions of permeabilized cells in PBS to stain for DNA.

For cell counting and analysis of labeled cells, a Coulter EPICSXL-MCL cytometer, equipped with an air-cooled,15-mW argonlaser operating at 488 nm was used. To separate fluorescenceemission of fluorescein isothiocyanate (FITC), carmine beads andpropidium iodide (PI), the following set of band pass filters wasused: 525, 620 and 675 nm, respectively. To discriminate doubletswhen DNA distribution was measured, peak versus integralfluorescence of PI was recorded at the same time.

Results

To establish the fate of cells in the hematopoietic organsof M. sexta, a combination of approaches has been used:(a) high-resolution imaging of cells that permits visual-ization of features distinctive for granular cells; (b)immunolabeling of well-permeabilized organs with anti-bodies that are specific for particular classes of hemo-cytes; (c) culture of isolated organs in vitro.

Morphology and fine structure of hematopoietic organs

Each hematopoietic organ is draped across a wingimaginal disc. The inner surface of the organ lies adjacentto the wing disc, and the outer surface of the organ faces

479

the hemocoel (Figs. 1, 2, 3, 4). A basal lamina delimitsthe organ with its surface intact on the inner surface of theorgan (facing the wing disc) but disrupted in places on itsouter surface that faces away from wing disc (Figs. 6,19,21). Numerous thin sections of organs at different timesduring the penultimate (4th) and last (5th) larval stadiawere cut and examined. The architecture of hematopoieticorgans as well as the fine structure of individual cellswithin these organs and on their surfaces are presented ina series of electron and light micrographs (Figs. 1, 2, 3, 4,5, 6, 7, 8, 9, 10, 11, 12, 13, 14).

After the initiation of wandering (L5d5) the well-defined differences between inner and outer surfaces oforgans become less evident as the delimiting basal laminabecomes folded and convoluted. Many cells of the organlose their coherence and disaggregate. Some cells engulfand phagocytose one another (Fig. 9). In this respect,these cells resemble the phagocytic cells also observed inDrosophila lymph glands at the wandering stage (Lanot etal. 2001) or the reticular cells of Gryllus, Calliphora, andLocusta hematopoietic organs that have dual functions ofhematopoiesis and phagocytosis (Hoffman et al. 1979).

As the hematopoietic organ degenerates at the end oflarval life, the remaining cells adopt interlocking, atten-uated forms and extend numerous fine processes fromtheir surfaces (Figs. 10, 11). Several days later at

Figs. 1–4 In this series of four images, the position of thehematopoietic organ is shown relative to the nearby wing disc,consisting of a peripodial epithelium (p) that surrounds each wingepithelium (W). Both groups of coherent cells and clusters ofnoncoherent cells are found in each hematopoietic organ. Most ofthe coherent cells are located near the inner surface of the organ(i.e., adjacent to the wing disc). The magnification is the same foreach image. Tracheoles are indicated with arrows

Fig. 1 On L4d3, a day prior to the molt from the 4th instar to the5th instar, this lobe of the hematopoietic organ lies between the

larval thoracic integument—epithelium (E) + cuticle (C)—and theperipodial epithelium (p) of the wing disc

Fig. 2 Immediately after the molt from the 4th instar to the 5thinstar (L5d0), an inner-outer polarity of the organ is evidentFig. 3 This section of an L5d1 hematopoietic organ was cut at theperiphery of the wing disc

Fig. 4 By the wandering stage of the 5th instar (L5d5), theorganization of the hematopoietic organ has changed with the basallamina of the organ becoming folded and convoluted (arrowheads)

Figs. 5–8 At higher resolution, most noncoherent and all coherentcells of hematopoietic organs can be distinguished from morpho-logically distinctive granular cells. These latter cells are alwayslocated near the periphery of each organ

Fig. 5 Noncoherent cells from an L4d3 larva are interspersed withthin strands of basal lamina (arrows)

Fig. 6 A thick basal lamina (arrow) separates coherent cells of anL4d3 organ (lower right) from more peripheral cells, one of whichis a granular cell (G) with its characteristic inclusions. Note that thisgranular cell lies within a break in the basal lamina (singlearrowheads)

Fig. 7 On L5d0 the thicker outer basal lamina (large arrows) canbe compared with thinner basal laminae (small arrows) that arefound in the interior of each organ. A granular cell (G) withdistinctive inclusions lies to the right on the periphery of the organ.T Tracheole

Fig. 8 On L5d3 this coherent mass of cells is delimited byrelatively thick basal laminae (arrows) on both inner (top) and outer(bottom) surfaces of the hematopoietic organ

480

481

Figs. 9–14 Views of hematopoietic organs between the wanderingstage of the 5th instar (L5d5) and pupation 5 days later

Fig. 9 On L5d5 convoluted basal laminae (arrows) are foundthroughout the hematopoietic organ. Basal laminae (b) as well ascells (c) are being engulfed by phagocytic cells believed to begranular cells (G)

Fig. 10 On L5d6 remaining cells of the organ interlock with longprocesses and form large, spherical extracellular spaces (S). Cellsextend numerous fine processes (arrows) into these spaces

Fig. 11 The global architecture of the L5d6 organ and its largenumber of spherical extracellular spaces are more clearly visualizedat lower magnification

482

pupation, the cells of the organ have lost most of theirconspicuous intercellular spaces, and their nuclei undergocondensation and adopt highly ramified forms—classicalfeatures of apoptotic cells (Figs. 11, 12, 13, 14).

Expression of surface proteins by cellsof the hematopoietic organ

The cells of the organ’s inner surface are generally morecoherent and densely packed; on the outer surface, thecells lose their coherence and disperse into the hemo-lymph (Figs. 1, 2, 19, 21). Most of the closely coherentcells are concentrated near the inner surface of the organ,although a few clusters of coherent cells lie near the outersurface.

All plasmatocytes that have dispersed from the hema-topoietic organ—circulating as well as adherent—im-munolabel with MS13 and MS34 (anti-integrins; Levin etal., in preparation). Cells that express neuroglian arealways coherent cells that usually lie near the innersurface of the organ, whereas cells that express integrinare found on both inner and outer surfaces of hemato-poietic organs (Figs. 15, 16, 17, 18, 19). Upon becomingnoncoherent hemocytes, the surfaces of these particularcells uniformly express integrin but lose their uniformsurface expression of neuroglian.

Cells of the organ form a coherent mass segregatedfrom granular cells of the hemolymph by a basal lamina.Only a few cells of the hematopoietic organ immunolabelwith MAb 15D11, a specific marker for granular cells andbasal laminae. Granular cells are localized to the periph-ery of the organ where they apparently can transversebreaks in the basal lamina (Figs. 6, 7, 21). In cross-sections of hematopoietic organs labeled with MAb15D11 the basal lamina clearly labels; however, cellswithin the organ, with the exception of a few at theperiphery, are not immunoreactive (Figs. 20, 21). Insections examined at high resolution, basal laminae arefound throughout the interior of the organs that arethinner than the exterior, enveloping basal laminae(Figs. 5, 7). None of these thinner basal laminae labelwith MAb 15D11. The protein lacunin that is recognizedby this MAb (Nardi et al. 1999) is produced by granularcells but not by plasmatocytes (Nardi et al. 2001).

Although all circulating plasmatocytes immunolabelwith the anti-integrins MS13 and MS34, only a smallfraction of the cells within the hematopoietic organimmunolabel with anti-integrin. Those cells of hemato-

poietic organs that label with anti-integrins, however, donot label with anti-neuroglian (Figs. 22, 23).

Like developing T-cells in the mammalian thymus, theplasmatocytes that develop within the moth hematopoieticorgans probably pass through phases marked by changesin expression of surface proteins (Janeway et al. 2001).Development within the lepidopteran hematopoietic or-gan also seems to be compartmentalized, with differencesin surface protein expression and cell cohesion observedalong the inner-outer axis of the organ.

Culture and mitotic activity of cellsfrom hematopoietic organs

Hematopoietic organs of Lepidoptera are known to besources of dividing hemocytes (Gardiner and Strand2000; Akai and Sato 1971). The fates of cells that arisefrom isolated cultures of hematopoietic organs can betracked by marking these cells with antibodies that arespecific for particular classes of hemocytes. The markerscan establish the diversity of hemocyte classes that aredescended from the undifferentiated cells of a givenorgan.

Cells derived from cultured organs in vitro either (1)adhered and spread or (2) remained suspended in culturemedium after dispersing from the hematopoietic organ.All cells that had dispersed from the organ immunola-beled with plasmatocyte-specific MS13 and MS 34(Figs. 24, 25), but not with granular-cell-specific MAb15D11. Cells adhering to the glass and plastic substratawithin the culture dish as well as cells that were collectedfrom the medium were fixed and immunolabeled. Thespecific immunolabeling of all these cultured cells(adherent and nonadherent) with MS13 and MS34 (anti-integrins) provides some of the best evidence that onlyplasmatocytes are derived from the thoracic hematopoi-etic organs.

Mitotic cells are abundant throughout each organ(Figs. 26, 27). Some mitotic cells express the surfaceprotein neuroglian (Fig. 28); however, cells of thehematopoietic organ that express integrin and are dis-persing were never observed to express the mitotic markerin organs from each of the four L5d3 and four L5d5larvae that were doubly immunolabeled (Fig. 29).

The number of cells in each organ clearly increasesduring the first 5 days of the last larval stadium and thendecreases after the inception of wandering and the rise inecdysteroid levels in the hemolymph (Nardi et al.1985;Riddiford et al. 1984). The fraction of mitotic cells withinthe organ as a function of development parallels the timecourse of change in ecdysteroid levels as well as totalnumber of cells within the organ (Figs. 30, 31). Thechanges in size and form of whole hematopoietic organsbetween day 0 and day 6 of the last larval stadium areillustrated in Fig. 32. The corresponding growth of thewing imaginal discs associated with each of thesehematopoietic organs are included for comparison.

Fig. 12 Four days later at pupation the conspicuous extracellularspaces (S) are less numerous and cells are showing ultrastructuralfeatures of programmed cell death

Fig. 13 A close up of a nucleus (N) from a cell of a hematopoieticorgan at pupation having the highly ramified form and condensa-tion typical of nuclei in apoptotic cells

Fig. 14 The global architecture of the organ at pupation showingits tenuous connection to the pupal epidermis at E and the remnantsof the spherical extracellular spaces (arrows)

483

484

Ploidy differences between circulating plasmatocytes andgranular cells

According to Arnold and Hinks (1976), circulatingplasmatocytes are a class of hemocytes that do not dividein the noctuid moth Euxoa declarata. These authors didobserve, however, that plasmatocytes of the hemolymphincrease in size dramatically during the last two larvalstadia of E. declarata. Cell size is known to beproportional to ploidy level. Although these authors didnot measure the DNA content of these plasmatocytes, theobserved increase in plasmatocyte size is consistent withthis class of hemocytes being endomitotic and polyploid.A clear disparity in size exists between granular cells andplasmatocytes in the hemolymph of last instar M. sexta(Fig. 33).

The antibody, anti-phosphohistone H3, that was usedto label mitotic cells cannot distinguish between mitoticcells destined to divide and endomitotic cells that do notdivide. The relative amount of DNA per cell, however, isproportional to intensity of staining with propidiumiodide. Labeling nuclei of circulating plasmatocytes withpropidium iodide provided evidence for endomitotic

activity rather than mitotic proliferation of these cells inM. sexta during the last larval stadium. Hemolymph aswell as cells of hematopoietic organs from four larvae oneach of three different days during this stadium (L5d0,L5d4, L5d5) were doubly labeled with plasmatocyte-specific MS13 and propidium iodide (Fig. 34).

Differences in ploidy levels of different classes ofcirculating hemocytes were noted when plasmatocytesand granular cells were analyzed with flow cytometry.Circulating hemocytes from last instar Manduca larvaewere doubly labeled with propidium iodide and FITC-conjugated MS13, the antibody marker specific forplasmatocytes. The cells that are negative for MS13labeling (Fig. 34A, R1) have the smallest amount of DNAand presumably are diploid cells with mean relativefluorescence peaks at 27 for G0/G1 and 51 for G2/M(Fig. 34B; G1/G2 ratio =1.89). These cells are almostentirely granular cells. The main population of cells thatare positive for MS13 (Fig. 34A, R2) presumablyrepresents a polyploid population with mean relativefluorescence peaks at 102 for G1/G0 and 200 for G2/M(Fig. 34C; G1/G2 ratio =1.96). Plasmatocytes and gran-ular cells of caterpillar hemolymph not only arise fromdifferent lineages, but they also have distinct ploidylevels.

Cells of wing imaginal discs are known to have diploidnuclei and were used as a diploid marker for propidiumiodide staining. These wing epithelial cells were takenfrom discs at a stage (L5d0) when the hemocoel of thewing disc is free of hemocytes (Nardi et al. 1985). Theintensity of propidium iodide staining for granular cells aswell as cells of hematopoietic organs matches that fordiploid cells of the wing imaginal discs (Fig. 34D).Endomitosis and additional DNA synthesis of plasmato-cytes apparently occur after cells have dispersed asdiploid cells from hematopoietic organs.

Discussion

Hemocyte cell lineages in Lepidoptera

In mammals hematopoietic stem cells are the precursorsof all blood cell lineages; in insects, the cells derived fromhead mesoderm may likewise be the precursors for allhemocyte lineages (Lebestky et al. 2000; Tepass et al.1994). As in Drosophila, hemocytes of Manduca firstarise as involution of head mesoderm (Nardi, submitted).Based on the surface marker(s) expressed by these cellsthat are recognized by peanut agglutinin lectin (PNA), thehemocytes that appear in early embryogenesis representgranular cells. Moth granular cells not only are recog-nized by specific lectins and antibodies, but they alsohave characteristic granules within their cytoplasm thatare evident with both the light and/or electron microscope(Figs. 6, 7).

A clear divergence in lineages of granular cells andplasmatocytes occurs during embryogenesis (Nardi, sub-mitted). The plasmatocyte lineage that first appears late in

Figs. 15–21 Immunolabeling of hematopoietic organs (L5d4) withtwo plasmatocyte-specific MAbs and one granular cell and basallamina-specific MAb. In all images HRP was used as a marker

Fig. 15 In this whole-mount of an organ labeled with anti-neuroglian (MAb 3B11), groups of coherent cells on the innersurface of the organ label on their cell surfaces. Epithelial cells oftracheoles label with anti-neuroglian (arrows; Nardi 1994). Theinset shows cells on the outer surface of the same organ. Fewercells label with anti-neuroglian on the outer surface, and onlycoherent cells label. Bar of inset 50 �m

Fig. 16 A cross-section of an organ labeled with MAb 3B11 showsthe clusters of coherent cells that are labeled. The inner surface ofthe organ faces down; it is on this surface that label is concentrated.The surfaces of tracheal epithelial cells also label with MAb 3B11(arrow)

Fig. 17 In this whole-mount of an organ viewed from its innersurface, aggregates of cells as well as single cells label with anti-integrin (MS13)

Fig. 18 An organ labeled with MS13 and viewed from its outersurface again shows labeled cells that form loose aggregates(arrow) rather than the closely coherent aggregates showing MAb3B11 immunoreactivity

Fig. 19 A cross-section of an organ that was labeled with MS13.Groups of labeled cells as well as labeled individual cells areevident. The inner surface of the organ faces down

Fig. 20 The outer surface of an organ labeled with anti-lacunin(MAb 15D11), an antibody that is specific for basal laminae as wellas granular cells. Folds and creases in the basal lamina (smallarrows) are darker than the smooth basal lamina. A few granularcells label near the periphery of the organ (large arrows)

Fig. 21 A cross-section of an organ labeled with anti-lacunin. Theouter surface faces down and the inner surface lies adjacent to theperipodial epithelium (p) and wing epithelium (w) of the wing disc.Cells can be seen dispersing from breaks in the basal lamina of theouter surface (arrows). The inset shows a section of another organwhose inner surface faces up and is covered by a continuous basallamina; its outer surface is covered by a discontinuous basal lamina.One labeled granular cell (arrow) lies near the tracheole (T). Bar ofinset 50 �m

485

embryogenesis either (1) loses the surface antigen(s)recognized by PNA or (2) represents a lineage ofembryonic hemocytes that arises from precursors thatare distinct from the granular cell precursors of headmesoderm. All the findings presented in this paper on thepostembryonic hematopoietic organs are consistent withgranular cells and plasmatocytes representing two differ-ent lineages that diverged during embryogenesis.

1. In hematopoietic organs granular cells can be distin-guished from plasmatocytes at the ultrastructural levelas well as with specific immunolabels; granular cellsare confined to the surfaces of hematopoietic organsfacing the hemocoel.

2. The only adherent hemocytes derived from cultures ofisolated hematopoietic organs are plasmatocytes.

3. Nonadherent cells that morphologically match thedescription of prohemocytes and express integrin arederived from cultured hematopoietic organs and prob-ably represent precursors of differentiated plasmato-cytes.

4. Circulating granular cells are diploid while circulatingplasmatocytes are polyploid.

Both single lineage and dual lineage models have beenoffered to account for the origin of circulating hemocytesand the fate of cells derived from hematopoietic organs ofLepidoptera. With a panel of monoclonal antibodiesgenerated against hemocytes of Pseudoplusia includens,

Gardiner and Strand (1999) clearly showed that granularcells and plasmatocytes represent two antigenicallydistinct lineages. Their findings with immunolabelingsupported earlier claims (Hinks and Arnold 1977; Nittono1964) that hematopoietic organs are sources of plasma-tocytes but not granular cells; the findings are consistentwith the two classes of hemocytes having separatelineages.

Beaulaton (1979) noted that the hematopoietic organsof the moths Bombyx and Antheraea are partitioned intoislets of cells, with compact islets of undifferentiated cellsoccupying the inner surface of the organ closest to thewing disc and with loose islets of cells on the outersurface. Based at least in part on electron micrographs ofhematopoietic organs from several Lepidoptera (Mon-peyssin and Beaulaton 1978), the loose or heterogenicislets were interpreted as representing hemocytes atvarious stages of differentiation and with features of allhemocyte classes. This interpretation that hemocytes ofall classes differentiate within the loose islets of lepi-dopteran hematopoietic organs led Beaulaton (1979) topostulate a single lineage for caterpillar hemocytes inwhich plasmatocytes derived from prohemocytes serve aspluripotent stem cells that give rise to granular cells,oenocytoids, and spherule cells. The equally valid inter-pretation that these hemocytes loosely associated withhematopoietic organs on their outer surfaces had actuallydifferentiated as circulating cells of the hemolymph wasnot considered.

Figs. 22, 23 Laser scanning confocal images of hematopoieticorgans from L5d4 larvae that have been labeled with anti-neuroglian (FITC) and anti-integrin (Texas Red). Note the absenceof co-expression of these two cell surface proteins by the cells ofthese organs. Tracheal epithelial cells label with anti-neuroglian(arrows)

Fig. 22 A low magnification view of an organ showing predom-inantly neuroglian-positive (green) cells

Fig. 23 A higher magnification view of an organ showing theintegrin-positive (red) and neuroglian-positive (green) cell popu-lations. Tracheal epithelial cells label with anti-neuroglian (arrows)

486

Studies of hemocyte transformations in culture havealso been interpreted as support for granular cells andplasmatocytes forming one lineage. Examining transfor-mations of hemocytes in vitro, Gupta and Sutherland(1966) claimed that plasmatocytes are polymorphic aswell as pluripotent and can change either directly orindirectly into all other types of hemocytes. By isolatingindividual prohemocytes from hemolymph cultures ofBombyx mori, Yamashita and Iwabuchi (2001) inferredthat prohemocytes can differentiate into either granularcells or plasmatocytes. These latter authors concluded thatthe prohemocytes released from hematopoietic organs are

the pluripotent cells of the hemolymph and can give riseto both granular cells and plasmatocytes. These twostudies, however, were based strictly on subjectivemorphological classifications of hemocyte types that haveoften proved misleading (Gillespie et al. 1997).

Comparing Drosophila hemocytes with hemocytesof Lepidoptera

Both single lineage and dual lineage models havelikewise been offered to account for the origin of

Figs. 24, 25 Many cells that disperse from isolated, culturedhematopoietic organs (L5d3) adhere and spread on a glasssubstratum. Some cells do not adhere, but all cells immunolabelwith MS13 and MS34 (anti-integrins). Each scale bar represents100 �m

Fig. 24 Cells have been cultured for 24 h and labeled with MS13

Fig. 25 Cells have been cultured for 2 weeks and labeled withMS13

Figs. 26, 27 Whole organs (L5d4) have been fixed and labeld withthe mitosis marker, anti-phospho-histone H3. A large percentage ofcells label with the marker, and only some of the labeled cells liewithin the plane of focus. Each scale bar equals 100 �m

Fig. 26 An overview of mitotic labeling in two lobes of an organ

Fig. 27 A higher magnificaton view of another organ whosemitotic cells have been labeled

487

Figs. 28, 29 In each figure, a whole mount of a hematopoieticorgan (L5d4) has been fixed and doubly labeled with a marker formitosis (FITC) and another marker for a cell surface protein (TexasRed). Each confocal image represents a 1.0 micron slice throughthe hematopoietic organ

Fig. 28 The mitosis marker and anti-neuroglian label some of thesame cells

Fig. 29 The mitosis marker and anti-integrin are never localized tothe same cells

Figs. 30, 31 Flow cytometry established the total number of cellsin each hematopoietic organ at different days during the last larvalstadium (L5d0-L5d6) as well as the percentage of these cellslabeled with the mitosis marker. For each time point, cells from atleast six organs were counted. Bars represent standard errors

Fig. 30 The mean (€ SE) for day 0 is significantly different fromthe mean ( € SE) for days 3 and 5 (p <0.05, unpaired t-test)

Fig. 31 The means (€ SE) for days 0, 3, and 5 are significantlydifferent (p <0.05, unpaired t-test)

Fig. 32 Whole-mounts of hematopoietic organs and their associ-ated wing discs from larvae of the last larval stadium showing theoverall morphology and growth of these organs. Organs and theirassociated wing discs all have the same orientation and magnifi-cation. The edge of the organ that drapes over the dorsal surface of

the disc lies to the right of each figure. The proximal end of eachwing disc lies to the left of each figure. By day 6, hematopoieticorgans are beginning to degenerate and lose their earlier form.Scale 2 mm

488

Drosophila hemocytes. In their investigation of hemato-poietic lymph glands in Drosophila larvae and prepupae,Lanot et al. (2001) listed four cell types as being foundwithin the glands: (1) prohemocytes, (2) crystal cells,plasmatocytes acting as (3) phagocytes and (4) secretorycells. Lamellocytes were never observed within the lymphglands, and these authors hypothesized that lamellocytesarise not from plasmatocytes but from prohemocytes.Prohemocytes of lymph glands serve as pluripotent stemcells that differentiate into at least three hemocyte classes:(1) lamellocytes, (2) plasmatocytes (phagocytes andsecretory cells), (3) crystal cells. This single lineagemodel for differentiation of hemocyte types contrasts withthe dual lineage model for embryonic hematopoiesispresented by Lebestky et al. (2000). A dual lineage modelfor hemocyte lineages is also based on the extensiveobservations of Rizki and Rizki (1984) of circulatinghemocytes as well as Shrestha and Gateff’s (1982)examination of larval hematopoietic organs (lymphglands).

The dual lineage model involves specification of twohemocyte lineages in Drosophila: a plasmatocyte lineagespecified by transcription factor glial cell missing (gcm)and a crystal cell lineage specified by transcription factorlozenge (lz; Lebestky et al. 2000). Lamellocytes, thehemocytes of Diptera that encapsulate foreign objects,

Fig. 33 Hemocytes from an L5d4 larva that adhere to a cover glasssubstratum after 1 h in culture. These cells were immunolabeledwith plasmatocyte-specific MS13 and photographed with differen-tial interference contrast optics. Note the disparity in sizes forplasmatocytes (large arrows) and granular cells (small arrows).One of the large neuroglian-positive plasmatocytes is indicatedwith a double arrow. Scale 50 �m

Fig. 34A–D Hemocytes from an L5d4 larva have been analyzedaccording to their ploidy levels and their surface labeling with theplasmatocyte-specific antibody MS13. Macerated and fixed cells ofwing imaginal discs from L5d0 larvae are known to be diploid. Theywere labeled with propidium iodide in D and used as a diploid

marker. A The FITC fluorescence intensity of hemocytes labeledwith MS13. The population of plasmatocytes specifically labeled bythis antibody is represented by the peak in region 2 (R2). Region 1(R1) represents the population of hemocytes not labeled with MS13(mostly granular cells). B The DNA distribution in unlabeled cellsfrom R1. The first peak on the left represents the G0/G1 peak with amean fluorescence of 27. The second G2/M peak has a meanfluorescence of 51. C The DNA distribution in the MS13-positivecells from region 2 of A. The first peak on the left represents GO/G1with a mean fluorescence of 102. The second peak is presumably theG2/M peak with a mean fluorescence of 200. D The intensity ofpropidium iodide staining for cells of wing imaginal disc epithelium(unshaded peak) whose nuclei are known to be diploid. The labelingpeak for the wing disc cells (unshaded peak) aligns with the shadedpeak for cells of hematopoietic organs (between L5d0 and L5d5) aswell as with the peak for granular cells in B

489

were first postulated to arise from circulating plasmato-cytes. Rizki (1957) noted that lamellocyte numbers inhemolymph increase as plasmatocyte numbers concur-rently decrease; this change in hemocyte populationsoccurs without an accompanying increase in cell divisionor apoptosis. By tracing the progression of cellularphenotypes in the lymph glands of Drosophila, Shresthaand Gateff (1982) hypothesized that plasmatocytes,podocytes, and lamellocytes represent a single lineagebased respectively on their progressive increase innumber of (1) primary lysosomes, (2) phagocytic vac-uoles, and (3) cytoplasmic processes.

Whereas Drosophila lamellocytes and lepidopteranplasmatocytes both function as encapsulating hemocytesin response to foreign invasion, granular cells clearly arethe hemocytes of Lepidoptera that are involved insecretion of basal laminae and phagocytosis (Nardi etal. 2001; Nardi and Miklasz 1989). Lanot et al.(2001)note that no counterpart of lepidopteran granular cells ispresent in the hemolymph of flies (Diptera). In Lepidop-tera these granular cells degranulate as a first line ofdefense in the presence of foreign objects (Schmit andRatcliffe 1977). The functions of moth granular cells–secretion of extracellular matrix and phagocytosis–haveapparently been assumed in Drosophila by the plasma-tocytes (Lanot et al. 2001).

The relationship between plasmatocytesand prohemocytes

Among the circulating hemocytes of Drosophila, mitoticactivity has been observed in prohemocytes and plasma-tocytes but not in crystal cells or in lamellocytes (Lanot etal. 2001). In lepidopteran larvae, Arnold and Hinks (1976,1983) rarely observed division of circulating plasmato-cytes but found that prohemocytes, granular cells, andspherule cells were the only circulating hemocytes thatfrequently divided.

Prohemocytes have been described in Lepidoptera asoval or rounded cells with high nuclear to cytoplasmratios (Gardiner and Strand 1999, 2000). Such cells havenot been described for M. sexta (Beetz et al., submitted),and only recently has a subpopulation of plasmatocytes inP. includens that fit this morphological description beenidentified by their special immunoreactivity (Gardinerand Strand 2000). As these authors point out, Arnold andHinks (1976) had earlier suggested that prohemocytes areactually precursors of plasmatocytes.

In the ultrastructural images of Manduca hematopoi-etic organs (Figs. 5, 6, 7, 8, 9), the rounded cells clearlyhave high nuclear to cytoplasmic ratios. As these cellsdisperse from the hematopoietic organs and beginexpressing integrin (Figs. 18, 19), they have the samemorphological features used to describe prohemocytes(Gardiner and Strand 1999; Jones 1962). In cultures ofManduca hematopoietic organs, many plasmatocytesadhere to the glass substrate (Figs. 24, 25); however,numerous rounded cells remain in suspension. Both

adherent and nonadherent cells in these cultures stainwith anti-integrin. The nonadherent, integrin-positivecells observed in cultures of Manduca hematopoieticorgans are derived from the same organ as the adherent,integrin-positive plasmatocytes and probably representthe class of hemocytes that other investigators have calledprohemocytes.

These observations of cells derived from culturedhematopoietic organs of M. sexta are consistent withGardiner and Strand’s (2000) finding that two subpopu-lations of plasmatocytes in P. includens can be distin-guished on the basis of their labeling with the particularmonoclonal antibody 43E9A10. They suggest that thesubpopulation of nonadherent, 43E9A10-negative plas-matocytes is the equivalent of the prohemocytes describedby other researchers.

Following ligation of larvae into anterior and posteriorregions, granular cell and spherule cell populations ofhemocytes show only a slight increase in the anteriorregion of the larva. However, prohemocytes and plasma-tocytes show a marked, several-fold increase in numbersat the anterior end of the larva (Hinks and Arnold 1977).While this phenomenon was evident in both Spodopterafrugiperda and E. declarata, another noctuid caterpillar,P. includens, showed no regional differences in hemocytedensities following ligations. This latter species is knownto have greatly reduced hematopoietic organs andpresumably plasmatocyte populations in P. includensare maintained by divisions of cells within the hemocoel(Gardiner and Strand 2000). The BrdU labeling patternsof circulating plasmatocytes in S. frugiperda and P.includens also indicate that these cells are synthesizingDNA and possibly proliferating as diploid cells.

Rather than finding evidence that circulating plasma-tocytes of the last larval instar of M. sexta proliferate asdiploid cells, however, their plasmatocytes were found toundergo endomitosis and to have higher ploidy levels thangranular cells of the hemolymph. Differences in ploidylevels for plasmatocytes and granular cells have not beenpreviously noted; however, both Arnold and Hicks (1976)as well as Shrestha and Gateff (1982), respectively,suggested that plasmatocytes increase in size within thehemolymph of caterpillars and within the lymph glands offly larvae. This difference in ploidy levels betweengranular cells and plasmatocytes is another structuraldifference that distinguishes these two major classes ofhemocytes and that probably reflects the different func-tional roles of granular cells and plasmatocytes in theimmune response.

Acknowledgements This research was supported by a grant fromthe National Institutes of Health (1 R01 HL 64657). Charles MarkBee helped with the scanning and final preparation of the figures.Andy Anderson and Stephanie Shockey carefully formatted thefinal manuscript. Two anonymous reviewers provided helpful,constructive suggestions for improving this manuscript.

490

References

Akai H, Sato S (1971) An ultrastructural study of the haemopoieticorgans of the silkworm, Bombyx mori. J Insect Physiol17:1665–1676

Arnold JW, Hinks CF (1976) Haemopoiesis in Lepidoptera. I. Themultiplication of circulating haemocytes. Can J Zool 54:1003–1012

Arnold JW, Hinks CF (1983) Haemopoiesis in Lepidoptera. III. Anote on the multiplication of spherule cells and granularhaemocytes. Can J Zool 61:275–277

Beaulaton J (1979) Hemocytes and hemocytopoiesis in silkworms.Biochimie 61:157–164

David C (1973) A quantitative method for maceration of hydratissue. Roux’s Arch Dev Biol 171:259–268

Gardiner EMM, Strand MR (1999) Monoclonal antibodies binddistinct classes of hemocytes in the moth Pseudoplusiaincludens. J Insect Physiol 45:113–126

Gardiner EMM, Strand MR (2000) Hematopoiesis in larvalPseudoplusia includens and Spodoptera frugiperda. ArchInsect Biochem Physiol 43:147–164

Gillespie JP, Kanost MR, Trenczek T (1997) Biological mediatorsof insect immunity. Annu Rev Entomol 42:611–643

Gupta AP, Sutherland DJ (1966) In vitro transformations of theinsect plasmatocyte in some insects. J Insect Physiol 12:1369–1375

Hinks CF, Arnold JW (1977) Haemopoiesis in Lepidoptera. II. Therole of hematopoietic organs. Can J Zool 55:1740–1755

Hoffman JA, Zachary D, Hoffman D, Brehelin M (1979) Postem-bryonic development and differentiation: hemopoietic tissuesand their functions in some insects. In: Gupta AP (ed) Insecthemocytes. Cambridge University Press, Cambridge, pp 29–66

Janeway CA, Travers P, Walport M, Shlomchik MJ ( 2001)Immunobiology: the immune system in health and disease.Garland, New York

Jones JC (1962) Current concepts concerning insect hemocytes.Am Zool 2:209–246

Lanot R, Zachary D, Holder F, Meister M (2001) Postembryonichematopoiesis in Drosophila. Dev Biol 230:243–257

Lebestky T, Chang T, Hartenstein V, Banerjee U (2000) Specifi-cation of Drosophila hematopoietic lineage by conservedtranscription factors. Science 288:146–149

Loret SM, Strand MR (1998) Follow-up of protein release fromPseudoplusia includens hemocytes: a first step toward identi-fication of factors mediating encapsulation in insects. Eur J CellBiol 76:146–155

Monpeyssin M, Beaulaton J (1978) Hemocytopoiesis in the oaksilkworm Antheraea pernyi and some other Lepidoptera. I.Ultrastructural study of normal processes. J. Ultrastruct Res64:35–45

Nardi JB (1994) Rearrangement of epithelial cell types in an insectwing monolayer is accompanied by differential expression of acell surface protein. Dev Dyn 199:315–325

Nardi JB, Miklasz SD (1989) Hemocytes contribute to both theformation and breakdown of the basal lamina in developingwings of Manduca sexta. Tissue Cell 21:559–567

Nardi JB, Hardt TA, Magee-Adams SM, Osterbur DL (1985)Morphogenesis in wing imaginal discs: its relationship tochanges in the extracellular matrix. Tissue Cell 17:473–490

Nardi JB, Martos R, Walden KKO, Lampe DJ, Robertson HM(1999) Expression of lacunin, a large multidomain extracellularmatrix protein, accompanies morphogenesis of epithelialmonolayers in Manduca sexta. Insect Biochem Mol Biol29:883–897

Nardi JB, Gao C, Kanost MR (2001) The extracellular matrixprotein lacunin is expressed by a subset of hemocytes involvedin basal lamina morphogenesis. J Insect Physiol 47:997–1006

Nittono Y (1964) Formation of hemocytes near the imaginal wingdisc in the silkworm, Bombyx mori L. J Sericult Sci Jpn 33:43–45

Ratcliffe NA, Rowley AF, Fitzgerald SW, Rhodes CP (1985)Invertebrate immunity: basic concepts and recent advances. IntRev Cytol 97:186–350

Riddiford LM, Wolfgang WJ, Green J, Fittinghoff C, Karasek K(1984) The significance of intermolt hormone levels in thecontrol of sequential gene expression in insect epidermis. J CellBiol 99:201a

Rizki TM (1957) Alterations in the haemocyte population ofDrosophila melanogaster. J Morphol 100:437–458

Rizki TM, Rizki RM (1984) The cellular defense system ofDrosophila melanogaster. In: King RC, Akai H (eds) Insectultrastructure, vol 2. Plenum, New York, pp 579–604

Schmit AR, Ratcliffe NA (1977) The encapsulation of foreigntissue implants in Galleria mellonella larvae. J Insect Physiol23:175–184

Shrestha R, Gateff E (1982) Ultrastructure and cytochemistry of thecell types in the larval hematopoietic organs and hemolymph ofDrosophila melanogaster. Dev Growth Differ 24:65–82

Tepass U, Fessler LI, Aziz A, Hartenstein V (1994) Embryonicorigin of hemocytes and their relationship to cell death inDrosophila. Development 120:1829–1837

Tolbert LP, Hildebrand JG (1981) Organization and synapticultrastructure of glomeruli in the antennal lobes of the mothManduca sexta: a study using thin sections and freeze-fracture.Proc R Soc Lond B 213:279–301

Willott E, Trenczek T, Thrower LW, Kanost MR (1994) Immu-nochemical identification of insect hemocyte populations:monoclonal antibodies distinguish four major hemocyte typesin Manduca sexta. Eur J Cell Biol 65:417–423

Yamashita M, Iwabuchi K ( 2001) Bombyx mori prohemocytedivision and differentiation in individual microcultures. J InsectPhysiol 47:325–331

491