demonstration of cytoplasmic processes in millipore ... · 3 % uranyl acetate in 50 % ethanol 1 an...

/ . Embryol. exp. Morph. Vol. 33, 1, pp. 187-203, 1975 \ 87

Printed in Great Britain

DEDICATED TO PROFESSOR ETIENNE WOLFF ON THE OCCASION OF HIS RETIREMENT

Demonstration of cytoplasmic processes in Milliporefilters permitting kidney tubule induction

By E. LEHTONEN,1 J. WARTIOVAARA, S. NORDLINGAND L. SAXEN

From the Third Department of Pathology,the Electron Microscope Laboratory,

and the Department of Zoology, University of Helsinki

SUMMARYThe presence of cytoplasmic material inside thin Millipore filters between interacting

mouse metanephric mesenchyme and spinal cord was investigated using different fixationmethods.

The transmission of induction was studied from sections of Zenker-fixed paraffin-embeddedexplants. Formation of kidney tubules was taken as evidence for induction. Filters with0-8 and 0-22 /*m pores permitted induction, whereas only 6 out of 31 filters with 01 fim.pores did so.

Glutaraldehyde-fixed Epon-embedded explants were used for study of cytoplasmicpenetration into filters. In thick sections, filters with large pores were seen to contain cyto-plasmic material at all levels. Filters with 01 ftm pores usually showed only shallow ingrowth,but those which had permitted passage of induction contained material at least half wayfrom the spinal cord and shallow ingrowth from the mesenchyme. With 0-8 /*m filters theingrowths from both sides met first after 18 h of transfilter cultivation. This has previouslybeen shown to be the minimum time needed for induction to take place in this system.

In electron microscopy cytoplasmic processes were seen deep inside the 0-8 and 0-22 ftmfilters regularly permitting induction. In small pores such material was only preserved bycertain glutaraldehyde fixatives.

Diffusion studies did not reveal major differences between induction-permitting 0-22 [imfilters and induction-preventing 01 ftm filters.

Thus in the kidney tubule induction system this and our previous work speak in favourof a mechanism based on close apposition of cells rather than on long-range diffusion ofinductive substances or on matrix interaction.

INTRODUCTION

Practically all of the inductive tissue interactions studied so far can takeplace through thin Millipore filters (cf. Saxen & Kohonen, 1969). This has beeninterpreted as evidence for extracellular transmissible signal substances (cf.Grobstein, 1967; Saxen, 1972). However, the transmission of the signal for the

1 Author's address: Third Department of Pathology, Haartmaninkatu 3, SF-00290Helsinki 29, Finland.

188 E. LEHTONEN AND OTHERS

induction of kidney tubules by spinal cord through such filters is too slow tobe explained by long-range diffusion (Nordling, Miettinen, Wartiovaara, &Saxen, 1971). Recently, a correlation has been demonstrated between transfiltertubule induction, cytoplasmic ingrowth and close cell apposition in Nuclepore®filters (Wartiovaara, Nordling, Lehtonen & Saxen, 1974). This conflicts withthe observation by Grobstein & Dalton (1957) that induction took place throughthin Millipore filters with a pore size of 0-1 /*m although no cytoplasmicmaterial was seen in the filters. The possibility exists that their chrome osmiumfixation did not preserve cytoplasmic material inside the filter and, especiallysince the methods of fixation for electron microscopy have been improved, thematter was reinvestigated.

MATERIALS AND METHODS

Transfilter induction experiments

Eleven-day mouse metanephrogenic mesenchymes were brought into trans-filter apposition with pieces of spinal cord of the same age, and cultured for6-72 h (Saxen & Saksela, 1971). Tubule formation was evaluated from a seriesof 78 explants that were fixed in Zenker's solution and serially sectioned fromparaffin blocks.

For evaluation of cytoplasmic ingrowth either chrome osmium or doublefixation and Epon embedding were used. The double fixation included pre-fixation with 1-3 % glutaraldehyde in a solution of 0-0-3 M sucrose and 0-1 Msodium cacodylate buffer, pH 7-2-7-8, for 90 min at +4°C or +25°C, andpostfixation with 1 % osmium tetroxide in the same buffer at + 4 °C for 60 min.1-5 % glutaraldehyde in 0-3 M sucrose and 0-1 M sodium cacodylate buffer,pH 7-8 at +4 °C, was used when not otherwise stated. The osmolarity of thisfixative was 730 m-osmole/kg measured by the freezing point depression method.Some explants were fixed at + 4 °C for 60 min in chrome osmium (485 m-osmole/kg; 1 % osmium tetroxide, 1 % potassium dichromate and 0-85 %NaCl; Dalton & Felix, 1956) at pH 7-2 and washed in 0-85 % NaCl (Grobstein &Dalton, 1957).

After fixation the assemblies were dehydrated in ethanol and embedded inEpon 812 in a Fullam mould for flat embedding (Ernest F. Fullam, Inc, Schenec-tady, N.Y. 12301). Thick Epon sections were stained with toluidine blue andexamined by light microscopy. Thin Epon sections were stained for 90 min with3 % uranyl acetate in 50 % ethanol and for 1 min with Reynolds lead citratediluted to 20 % with 0-01 N-NaOH and examined in a Jeol JEM-100B trans-mission electron microscope.

Filters

Thin nitrocellulose (Millipore) filters (about 25 fim thick) with pore sizes of0-8, 0-22 and 0-1 /an (according to the manufacturer) were obtained from theMillipore Co., Bedford, Mass., U.S.A.

Cell processes in kidney induction 189

1. Thickness measurements

The thickness of the dry filters was measured with a micrometer, in whichthe displacement of an inductive sensor was recorded electronically. Thereproducibility of the measurements was better than 1 /*m. At least 10 differentmeasurements were made on each filter type. The thickness of the actual pieceof filter separating the explants on 0-1 /im filters was measured with an ocularmicrometer from stained paraffin sections. If the sections were not cut perpen-dicular to the surface of the filter, the thickness would appear greater. But evenan error of 20° in cutting would only cause an increase of 7 % in the measure-ment. A more serious source of error is the fact that ethanol causes changes inthe thickness of the filter (Lehtonen, Nordling & Wartiovaara, 1973).

2. Scanning electron microscopy

The filters were immersed in 70 % ethanol (Lehtonen et ah 1973) as in thetransfilter induction experiments, rinsed in water, dried in vacuum, coated witha thin layer of carbon and gold in a Balzers' Micro-BA3 evaporator, andexamined in a Jeol JSM-U3 scanning electron microscope.

3. Diffusion studies

The permeability of thin Millipore filters to 3H-thymidine (3H-TdR; mole-cular weight MW 242; New England Nuclear, Dreieichenhahn, Germany) andto blue dextran (MW2x 106; Pharmacia, Uppsala, Sweden) was measured ina diffusion apparatus. The diffusion results were expressed as rate constants ofdiffusion in the formula

l n AC

where Co is the initial concentration of the test substance, and AC is thedifference in the concentrations of the substance in the two compartments ofthe chamber at time / (Nordling et ah 1971).

RESULTS

Transfilter induction experiments

1. Correlation between induction and cytoplasmic ingrowth in various filters

For evaluation of induction 78 explants grown on filters with different poresizes were fixed in Zenker's solution, embedded in paraffin and serially sectioned.Tubule formation was seen after 72 h cultivation in all (30) explants on0-8 /<m filters and in nearly all (16 out of 17) on 0-22 jum filters. With 0-1 /.imfilters 6 out of 31 explants displayed tubule induction (Table 1). After Zenkerfixation and paraffin embedding only some fuzzy material was seen in the filters.


Table 1. Effect of the pore size of Millipore filters on tubule induction

Pore size Number of explants*(/tm) with tubules/total

Fraction of positiveexplants

010-220-8

6/3116/1730/30

0-20-910

* Explants were cultivated for 72 h, Zenker-fixed and serially sectioned for light micro-scopy from paraffin blocks.

Fig. 1. Thick Epon section of 0-22 fixn Millipore filter with mouse metanephrogenicmesenchyme above and dorsal spinal cord below. Total cultivation time 48 h.Note abundant material at all levels of the filter. Toluidine blue staining, x 1700.Fig. 2. 01 pirn Millipore filter in explant with tubule induction. Experimentalconditions as in Fig. 1. Note ingrowth of material from the spinal cord side tomore than half way across the filter, x 1700.

therefore cytoplasmic ingrowth was studied from glutaraldehyde-fixed andEpon-embedded material.

Thick sections of double-fixed, Epon-embedded specimens of both 0-8 and0-22 /.tin filters showed ingrowth of cytoplasmic material (Figs. 1 and 14). Evenin 0-22 jLim filters it extended more than half way, and in some places all theway, through the filter. The 0-1 /an filters usually did not reveal much cyto-plasmic ingrowth. A few of these, however, showed cytoplasmic materialextending at least half way through the filter from the spinal cord (Fig. 2) andshallow ingrowth from the mesenchyme. They were serially sectioned and werefound to be tubule positive.


•i '

Figs. 3-6. Thick Epon sections of 0-8 /tm Millipore filters after different times oftransfilter cultivation of mesenchyme above and spinal cord below, x 1300.

Fig. 3. 6 h. Slight ingrowth of material into the filter from both sides.Fig. 4. 12 h. Ingrowth from both sides of the filter. Spinal cord processes extend

one-third of the distance across the filter.Fig. 5. 18 h. Spinal cord processes extend two-thirds and mesenchymal processes

up to one-third of the distance across the filter.Fig. 6. 24 h. The filter appears to be filled with material.

2. Time sequence of cytoplasmic ingrowth

Thick sections of glutaraldehyde-nxed, Epon-embedded explants with 0-8filters were used for evaluation of the time sequence of cytoplasmic ingrowth.After 6 h of transfilter cultivation there was slight ingrowth of cytoplasm intothe filter (Fig. 3). After 12 h the cytoplasmic material extended about one-thirdacross the filter on the spinal cord side, and there was also some ingrowth onthe mesenchymal side of the filter (Fig. 4). After 18 h cytoplasmic material had

Insert: thick Epon section of the same block, demonstrating abundant ingrowth ofcytoplasm, x 400.


Figs. 8-10. Higher magnifications of Fig. 7. A well preserved membrane sur-rounds the processes (arrows). Double fixation preserves the structures equally wellat all levels. No continuous basement lamina-like structure is present on the cellsurface, but at places small patches of extracellular material can be seen (doublearrows). A thin layer of electron-dense material is seen on the surface of filterchannels (P). x 50000.

13 EM B 33


penetrated about two-thirds through the filter from the spinal cord and aboutone-third from the mesenchyme (Fig. 5). After 24 h the filters were nearlyfilled with cytoplasmic material (Fig. 6).

3. Nature of the material within the filter

In electron microscopy cytoplasmic material was seen at all levels in 0-8 and0-22 ju,m filters (Fig. 7). Most of the filter pores were, however, empty. Even athigh magnification (Figs. 8-10) no extracellular matrix could be seen betweenthe two interacting tissues. Elongated cytoplasmic material, located at all levelsof the filter, apparently represented oblique sections of processes rather thanindependent vesicles in the filter pores (Fig. 7). Due to the three-dimensionalstructure of the filter the processes could not be followed through the 30 /*mdistance, but in many places cellular processes extended a considerable distancefrom the tissues into the filter (Figs. 7, 8, 10). The cells of both the mesenchymeand the spinal cord tissue did not have any continuous surface-adherent, basementlamina-like extracellular material. At some places the membrane of the cellprocesses had a fuzzy appearance, probably due to tangential sectioning. Alsoin some places small amounts of surface-adherent extracellular material couldbe seen on the cell surface (Figs. 8, 10). Patches of electron-dense material werealso seen on the surface of the filter channels (Figs. 8-10).

Even in 0-1 jum filters (Fig. 2) processes were occasionally visible on thespinal cord side in light microscopy. In electron microscopy the processes in thefilter were similar, although more scanty than in filters with large pores (Fig. 11).Most of them were seen on the spinal cord side. In tubule-positive cases with0-1 /.cm filters the processes extended deeper into the filter from the spinal cordside, and cytoplasmic material was also found near the mesenchyme (Fig. 12).

4. The effect of fixative on preservation of cytoplasmic material in the filters

Changes in pH (from 7-2 to 7-8), in temperature ( + 4 or +25 °C), or in theconcentration of glutaraldehyde (from 1-0 to 3-0%) or sucrose (from 0 to0-3 M) resulted in great differences in the preservation of cytoplasmic materialwithin the filters viewed by both light and electron microscopy. The materialdeepest in the filter was the most sensitive to the fixation method and ingrowththus appeared shallower after poor fixation. After exposure to certain fixativesthe pores were seemingly empty, although there was good preservation of thetissues on both sides of the filter. The best results were obtained with 1-5 %glutaraldehyde in 0-3 M sucrose and 0-1 M sodium cacodylate buffer, pH 7-8,at +4°C.

Processes in 0-1 and 0-22 jum pores disappeared more easily than those in0-8 jLim pores. Modifications of the glutaraldehyde fixation which gave goodpreservation of processes in 0-8 jam pores could result in poor preservation ofcytoplasmic processes in 0-22 /im. pores.

There was no difference in the extent of cytoplasmic ingrowth into 0-8 /«n


Fig. 11. 01 /<m Millipore filter close to the spinal cord (SC) after 48 h of culture.The cytoplasmic processes within the filter often have a surrounding membrane(arrow). The preservation of the processes is not as good as in 0-22 /im filters,x 24000.Fig. 12. 0-1 /tm Millipore filter close to the mesenchyme (M) after 48 h of culture.Cytoplasmic processes within the filter. Filter pore (P), filter material (FM).x 40000.

13-2


Figs. 13-14. Thick Epon sections of 0-8/irn filters. The two explants cultivated onthe same screen for 48 h. Chrome osmium fixation (13), double fixation (14). Nodifference in the amount of cytoplasmic material in the pores, x 1700.

filters visible by light microscopy after fixation with chrome osmium accordingto Grobstein & Dalton (1957) or with double fixation including glutaraldehydeprefixation (Figs. 13, 14). Explants on 0-22 /*m filters fixed with chrome osmiumshowed only very shallow cytoplasmic ingrowth into the filters (Fig. 15) incontrast to the marked ingrowth visible after double fixation (Fig. 16). Electronmicrographs showed that chrome osmium is not such a good fixative as glutar-aldehyde. With chrome osmium fixation cytoplasmic material was visible allthrough the 0-8 /im filters, whereas in 0-22 /*m filters only occasional shortprocesses or remnants of processes were seen near the tissue on both sides(Figs. 17-19).

Properties of the filters1. Thickness

The mean thickness differed between filter types (Table 2), but also betweensheets of the same filter type. Even within the same filter sheet the thicknessvaried, but the changes were gradual. The mean thickness of 0-1 /im filters(measured from paraffin-embedded specimens) in the six induction-positiveexplants was 31 ± 2 /im and that of the 25 tubule-negative explants 32+1 jum.

2. Structure

Filters with a pore size of 0-1 /on have a matt and a glossy side. The ultra-structure of the two sides, as shown by scanning electron microscopy, wasdifferent, with larger holes on the matt surface (Figs. 20, 21). The two otherfilter types are similar on both sides and resemble each other (Figs. 22, 23). The


Figs. 15-16. 0-22/tm filters. The two explants cultivated on the same screen for48 h. x 1700.

Fig. 15. Chrome osmium fixation. Only shallow ingrowth (arrows). Kidneytubule (T) in the mesenchyme.

Fig. 16. Double fixation. Clear ingrowth of cytoplasm on both sides. Kidneytubule (T) in the mesenchyme.

filters consist of tortuous filaments crossing each other randomly and leavingspaces of various shapes and sizes.

3. Permeability

The diffusion rate constants for both blue dextran and 3H-thymidine weremuch the same with the different types of filters (Table 3). The concentration inchamber B reached 10 % of the value in chamber A after about 6 min in thecase of 3H-thymidine and after 3h (0-1 and 0-22 jum pores) or 2 h (0-8pores) in the case of blue dextran.


• . - i^ ' , 2 / < m ' '

•' *̂ *«,.sgv*

riK

^

:'

,*. - .

'h\ZTd*•*>•>' * ; - / 1 8

\

/? - , j c " j

FIGURES 17-19

Electron micrographs of 0-22 /tm filter. Cultivation time 48 h. Chrome osmiumfixation.Fig. 17. Short process (arrows) near the spinal cord, x 8000.Fig. 18. Mesenchymal cells not well preserved, cell membranes discontinuous(arrow). Remnant (/?) of mesenchymal process, x 50000.Fig. 19. Poor preservation of spinal cord. Vesicles (V), probably degenerated cellmembranes. Remnants (R) of spinal cord processes, x 50000.


Table 2. Thickness of Millipore filters with various pore sizes

Pore size Thickness* ± S.D. Range(/tin)

010-220-8

27-2 ±2-9024-1 ±2-5628-611-99

23-3218-2726-33

* Dry filters were measured with a micrometer.

Figs. 20-23. Scanning electron micrographs of surface of Millipore filters, x 15000.Fig. 20. Pore size 01 /tm, glossy side of filter.Fig. 21. Pore size 01/tm, matt side of filter. The proportion of holes seems

higher than on the glossy side.Fig. 22. Pore size 0-22 /tm.Fig. 23. Pore size 08/tm.

DISCUSSION

Little is known about substances carrying morphogenetic messages in inter-active processes. The search for such hypothetical compounds is hampered bythe fact that their localization and mode of transmission are unknown. Threedifferent hypotheses have been put forward: diffusion of signal molecules(Holtfreter, 1955), induction through cell-to-cell contacts (Weiss, 1947), andinteraction through compounds in the extracellular matrix (Grobstein, 1955).


Table 3. Permeability of Millipore filters

Rate constant of diffusion! ± S.D.Pore size of the , A

filter Om) 3H-TdR (MW* 242) Blue dextran (MW 2 x 106)

0-1 2-5±0-2 h-1 007±004hr1

0-22 2-0±0-2 h-1 006±0004 h-1

0-8 2-1 ±004h-1 0-11 ±0003 hr1

* MW = molecular weight in Daltons.t The rate constant of diffusion was determined in a diffusion chamber. The concentration

of the test substances in the two chambers was determined at least five different times and alinear regression line was calculated by the method of least squares.

Long-range diffusion is considered to occur when a substance due to thermalmotion flows from one cell with high concentration to another cell with a lowconcentration over intercellular spaces of the order of micrometers. The termcell contact has been used here not only for intimate membrane contacts butalso for contact of molecules on the cell surface (cell periphery). Extracellularmatrix refers to structural material between cells; it is secreted by cells, but isnot an integral part of them.

Already Grobstein (1956) pointed out that diffusion is most unlikely to be auniversal mechanism for morphogenetic interactions. Recent observations inthe tubule-induction system have provided further evidence against diffusionas a model: efforts to extract inductively active compounds from the inductortissue have failed (Gossens & Unsworth, 1972); the transmission velocity of thesignal, measured in transfilter experiments, is too slow to be explained throughmechanisms based on free diffusion (Nordling et al. 1971); and filters whichallow diffusion of large molecules can prevent tubule induction (Wartiovaaraet al. 1974). Furthermore, the present results show that the permeability ofinduction-preventing 0-1 /im and induction-permitting 0-22 /im filters for twomodel substances was the same (Table 3). Therefore, the discussion will con-centrate on the remaining alternatives, contact-mediated induction and matrixinteraction.

An interspace of 10-20 nm recently observed between the interacting tissuecomponents in the branching rat salivary gland (Cutler & Chaudry, 1973) anddeveloping rat duodenal mucosa (Mathan, Hermos & Trier, 1972) does not, byour definition, exclude functional cell contacts. Also cells separated by basementlamina are considered to be in contact. Matrix interaction, as termed byGrobstein, takes place over a cytoplasm-free interzone which is several magni-tudes greater, as the theory was based on experiments with filters 20 /<m or morein thickness (Grobstein, 1957; Grobstein & Dalton, 1957). Their hypothesisassumes that material is passing over this interspace, whereas the contacthypothesis suggests that the interactive molecules are carried by the cell surface.


In the present study, induction of kidney tubules by spinal cord occurredregularly through 0-8 and 0-22 /im filters, which contained cytoplasmic processesat all levels. At high magnification patchy extracellular material was seen on thesurface of cell processes and outlining the filter channels (Figs. 8-10). Compar-able material has been noted in filters when used for in vitro cultivation oftissues (England, 1969). The patchy material did not constitute organizedstructures with, for example, extracellular fibres and does not fall into thecategory of extracellular matrix material. Induction rarely occurred through0-1 //m filters, which only occasionally contained cytoplasmic processes deepin them. These results suggest a correlation between cytoplasmic ingrowth intothe filters and induction. The hypothesis that cytoplasmic contacts are neededfor induction is supported by the finding that the cytoplasmic frontiers growingin from both tissues reached a common level, thus permitting contacts at aboutthe minimum induction time, 18 h for 0-8 /*m filter (Nordling et al. 1971).

These results conform with the view that interpositioned filters do notnecessarily prevent cell contacts (Wartiovaara, Lehtonen, Nordling & Saxen,1972) and are in accord with the observation that close cell apposition existedbetween the interacting tissues in situations which permitted induction througha Nuclepore filter (Wartiovaara et al. 1974).

Our results agree with Grobstein & Dalton's (1957) results on the effect ofthe pore size of Millipore filters on tubule induction. They found that filterswith large pores regularly permitted the passage of the inductive signal, whereasthose with 0-1 /<m pores rarely did so. In filters with large pores (0-8 ju,m) theyobserved cytoplasmic material in the filter channels, whereas they never ob-served such material deep inside 0-1 /*m filters. Their natural conclusion wastherefore that the cytoplasmic ingrowth into filters with large pores was coinci-dental to induction and that 'inductive activity in this system is not dependentupon cytoplasmic contact and hence is resident in materials which are at leastpotentially extracytoplasmic' (Grobstein & Dalton, 1957). However, ourelectron micrographs of those few 0-1 jam filters which permitted the passageof the inductive signal clearly showed cytoplasmic material deep inside thefilter. The seeming discrepancy is probably explained by the improvement infixation methods for electron microscopy since 1957, when glutaraldehyde wasnot yet used. Even with modern fixatives it was difficult to preserve processeswithin filters with smaller pores. We also found that the chrome osmiumfixation method used by Grobstein & Dalton (1957) preserved cytoplasmicmaterial in 08 /on filters, but not in 0-22 or 0-1 [im filters. This is consistentwith their findings that cytoplasmic material is abundant in 0-8 pcm, sparse in0-45 fim and non-existent in 0-1 jum pores after chrome osmium fixation.

A pore size of 0-1-0-2 jam seems to be the limiting size both for induction andfor ingrowth of cytoplasmic processes. This agrees with previous results ob-tained with Nuclepore filters (Wartiovaara et al. 1974). This has also beenfound to be the limiting pore size for ingrowth and tissue interactions in other


systems, such as in diffusion chamber experiments (Capalbo, Albright &Bennett, 1964; Nettesheim, Makinodan & Chadwick, 1966), and in experimentson tumorigenesis with Millipore filters implanted in mice (Karp et al. 1973).

Grobstein & Dalton (1957) reported that 0-1 /im Millipore filters vary greatlyin thickness, and that the majority of explants on thin (16-19 ^m) filters werepositive. Contrary to this, only one culture out of ten on the thick (32-38 jam)filters was definitely positive with only one tubule rudiment. In the present studyall experiments with 0-1 fim. filters were performed with relatively thick filters.Measured from Zenker-fixed sections (as in the work of Grobstein & Dalton,1957), the mean thickness was 31-8 /on with a range of 27-33 fim. This seemsto explain our relatively low incidence of induction, 20 %.

Scanning electron micrographs revealed considerable differences between thestructure of the generally induction-permitting 0-22 fim filters and the generallyinduction-preventing 0-1 fim filters. Such structural differences were not seenbetween 0-2 and 0-1 /̂ m Nuclepore filters, which revealed the same differencesin the transmission of inductive signals (Wartiovaara et al. 1974). Hence struc-tural features other than small pore size appear unlikely to be responsible forthe usual barrier to transmission of the inductive signal presented by 0-1 fimMillipore filters.

In conclusion, we would emphasize that in the interactive system studiedhere, all observations speak against the diffusion hypothesis. Moreover, ourresults from experiments with different filters and improved fixation methodshave demonstrated a good correlation between intimate cell contacts andtransmission of inductive signals, making the hypothesis of matrix interactionunlikely. It must be stressed that these observations should not be generalizedto other interactive systems in which diffusion and matrix interaction may bethe mechanisms for transmission of inductive signals.

The technical assistance of Ms Pirkko Leikas, Monica Schoultz, Ann-Kristin Thors,Anja Tuomi, and Elina Waris is acknowledged.

Supported by the Sigrid Juselius Foundation, the Foundation for Cancer Research, andthe National Research Council for Medical Sciences, Finland.

REFERENCES

CAPALBO, E. E., ALBRIGHT, J. F. & BENNETT, W. E. (1964). Evaluation of the diffusionchamber culture technique for study of the morphological and functional characteristicsof lymphoid cells during antibody production. / . Immun. 92, 243-251.

CUTLER, L. S. & CHAUDRY, A. P. (1973). Intercellular contacts at the epithelial-mesenchymalinterface during the prenatal development of the rat submandibular gland. Devi Biol. 33,229-240.

DALTON, A. J. & Felix, M. D. (1956). A study of the Golgi substance and ergastoplasm in aseries of mammalian cell types. In Fine Structure of Cells, pp. 274-293. Groningen, TheNetherlands: P. Moordhoff Ltd.

ENGLAND, M. A. (1969). Millipore filters studied in isolation and in vitro by transmissionelectron microscopy and stereo-scanning electron microscopy. Expl Cell Res. 54, 222-230.

Cell processes in kidney induction 203GOSSENS, C. L. & UNSWORTH, B. R. (1972). Evidence for a two-step mechanism operating

during in vitro mouse kidney tubulogenesis. /. Embryol. exp. Morph. 28, 615-631.GROBSTEIN, C. (1955). X. Tissue interaction in the morphogenesis of mouse embryonic

rudiments in vitro. In Aspects of Synthesis and Order in Growth (ed. D. Rudnick), pp.233-256. Princeton, New Jersey: Princeton University Press.

GROBSTEIN, C. (1956). Inductive tissue interaction in development. In Adv. Cancer Res. 4,187-236.

GROBSTEIN, C. (1957). Some transmission characteristics of the tubule inducing influence onmouse metanephrogenic mesenchyme. Expl Cell Res. 13, 575-587.

GROBSTEIN, C. (1967). Mechanisms of organogenetic tissue interaction. Natn. Cancer Inst.Monogr. 26, 279-299.

GROBSTEIN, C. & DALTON, A. J. (1957). Kidney tubule induction in mouse metanephrogenicmesenchyme without cytoplasmic contact. /. exp. Zool. 135, 57-73.

HOLTFRETER, J. (1955). Studies on the diffusibility, toxicity and pathogenic properties of the'inductive' agents derived from dead tissues. Expl Cell Res. Suppl. 3, 188-209.

KARP, R. D., JOHNSON, K. H., BUOEN, L. C, GHOBRIAL, H. K. G., BRAND, I. & BRAND,K. G. (1973). Tumorigenesis by Millipore filters in mice: Histology and ultrastructure oftissue reactions as related to pore size. /. natn Cancer Inst. Monogr. 51, 1275-1285.

LEHTONEN, E., NORDLING, S. & WARTIOVAARA, J. (1973). Permeability and structure ofethanol-sterilized nitrocellulose (Millipore) filters. Expl Cell Res. 81, 169-174.

MATHAN, M., HERMOS, J. A. & TRIER, J. S. (1972). Structural features of the epithelio-mesenchymal interface of rat duodenal mucosa during development. / . Cell Biol. 52,577-588.

NETTESHEIM, P., MAKINODAN, T. & CHADWICK, C. J. (1966). Improved diffusion chambercultures for cytokinetic analysis of antibody response. Immunology 11, 427-439.

NORDLING, S., MIETTINEN, H., WARTIOVAARA, J. & SAXEN, L. (1971). Transmission andspread of embryonic induction. I. Temporal relationships in transfilter induction of kidneytubules in vitro. / . Embryol. exp. Morph. 26, 231-252.

SAXEN, L. (1972). Interactive mechanisms in morphogenesis. In Tissue Interactions inCarcinogenesis (ed. D. Tarin), pp. 49-80. London: Academic Press.

SAX£N, L. & KOHONEN, J. (1969). Inductive tissue interactions in vertebrate morphogenesis.In International Review of Experimental Pathology vol. 8 (ed. G. W. Richter & M. A.Epstein), pp. 57-128. New York and London: Academic Press.

SAX£N, L. & SAKSELA, E. (1971). Transmission and spread of embryonic induction. II.Exclusion of an assimilatory transmission mechanism in kidney tubule induction. Expl CellRes. 66, 369-377.

WARTIOVAARA, J., LEHTONEN, E., NORDLING, S. & SAXEN, L. (1972). Do membrane filtersprevent cell contacts? Nature, Lond. 238, 407-408.

WARTIOVAARA, J., NORDLING, S., LEHTONEN, E. & SAXEN, L. (1974). Transfilter induction ofkidney tubules: Correlation with cytoplasmic penetration into Nucleopore filters. / .Embryol. exp. Morph. 31, 667-682.

WEISS, P. (1947). The problem of specificity in growth and development. Yale J. Biol. Med.19, 235-278.

(Received 1 July 1974)

demonstration of cytoplasmic processes in millipore ... · 3 % uranyl acetate in 50 % ethanol 1 an...

Documents