J. Sci. 3, 207-230 (1968) 207 Printed in Great Britain

RECONSTRUCTIONS OF FORMATION AND CILIOGENESIS IN MAMMALIAN LUNGS

S. P. SOROKIN Department of Anatomy, Harvard Medical School, Boston, Massachusetts 02115, U.S.A.

SUMMARY This study presents reconstructions of the processes of centriolar formation and ciliogenesis based on evidence found in electron micrographs of tissues and organ cultures obtained chiefly from the lungs of foetal rats. A few observations on living cultures supplement the major findings. In this material, are generated by two pathways. Those centrioles that are destined to participate in forming the achromatic figure, or to sprout transitory, rudimentary (primary) cilia, arise directly off the walls of pre-existing centrioles. In pulmonary cells of all types this direct pathway operates during interphase. The daughter centrioles are first recognizable as annular structures (procentrioles) which lengthen into cylinders through acropetal deposition of osmiophilic material in the procentriolar walls. Triplet fibres develop in these walls from singlet and doublet fibres that first appear near the procentriolar bases and thereafter extend apically. When little more than half grown, the daughter centrioles are released into the cyto- plasm, where they complete their maturation. A parent centriole usually produces one daughter at a time. Exceptionally, up to 8 have been observed to develop simultaneously about 1 parent centriole. Primary cilia arise from directly produced centrioles in differentiating pulmonary cells of all types throughout the foetal period. In the bronchial epithelium they appear before the time when the ciliated border is generated. Fairly late in foetal life, centrioles destined to become kinetosomes in ciliated cells of the epithelium become assembled from masses of fibrogranular material located in the apical . Formation of these centrioles may be under the remote influence of the diplosomal centrioles. More certainly, the precursor material accumulates in close proximity to Golgi elements. Within the fibrogranular areas, osmiophilic granules (400-800 A) increase in size and eventually become consolidated into dense spheroidal bodies (deuterosomes), which organize the growth of procentrioles around them. When mature, the newly formed centrioles become aligned in rows beneath the apical plasma membrane. There each centriole produces satellites from its sides, a root from its base, and a from its apex. Early stages in the formation of both primary cilia and those of the ciliated border are similar. In developing cilia of the ciliated border, however, the outer ciliary fibres rapidly reach the tips of the elongating shafts, and a central pair of fibresi s formed (9 + 2 arrangement). In primary cilia, development of the fibres seems to lag behind the elongation of the shafts, and only the outer ciliary fibres appear (9 + 0 arrangement). The strengths and weaknesses of the proposed reconstructions of centriolar formation and ciliogenesis are discussed, and the occurrence in other living forms of similar pathways for centriolar formation is noted. Further discussion leads to an interpretation of the centriole as a semi-autonomous whose replicative capacity is separable from the characteristic triplet fibre structure of its wall. 208 S. P. Sorokin

INTRODUCTION During the development of the lung, two distinct types of cilia are formed: rudi- mentary cilia and those of the ciliated border. The rudimentary cilium is the first to appear. It is produced as a solitary appendage by virtually all of the cells present, including those that later develop a ciliated border. The cilium grows out directly from one centriole of the pair that participates in organizing the achromatic figure, and it seems to have only a transitory existence. In being a direct derivative of the achro- matic centrioles and the first cilium to emerge from the cells, the rudimentary cilium may be termed 'primary' and the process of its formation 'primary ciliogenesis'. Beginning at a later stage of pulmonary development, a second and more familiar kind of cilium is produced by those epithelial cells that are destined to possess a ciliated border. Numerous basal bodies (kinetosomes) at first become visible in the cytoplasm of these cells. Then a long, motile cilium grows out from each . This paper reconstructs from electron micrographs the cytological events that are connected with the formation of the basal bodies and the generation of both types of cilia. One of the most interesting problems related to the genesis of a ciliated epithelial border concerns the formation of the basal bodies, which in most higher animals are morphologically identical to the centrioles. Indeed, it has long been widely accepted that centrioles divide to form the basal bodies (Henneguy, 1898; von Lenhossek, 1898; Lwoff, 1950). None the less, some investigators have shown that centrioles (Faure- Fremiet, Rouiller & Gauchery, 1956; Bernhard & de Harven, i960) or basal bodies (Gall, 1961) may arise from a position off the wall of a parent centriole, where they almost certainly are not formed by division of the pre-existing organelle. Others have suspected, in the case of ciliated cells, that the basal bodies are formed de novo in the cytoplasm (Randall et al. 1963; Stockinger & Cirelli, 1965). In this paper it will be shown that centrioles or basal bodies can be produced either directly from a pre- existing centriole or indirectly through a range of precursors that do not resemble centrioles. Furthermore, the results permit one to see that the reproductive capacity of a typical centriole is separable from its characteristic triplet fibre structure. Where this study is concerned with centrioles, however, it is limited to a consideration of the 9-membered cylindrical organelle that performs the centriolar functions in most animal cells. It does not encompass a view of other more unusual centrioles that occur in certain flagellates (Cleveland, 1957), in testicular cells of fungus gnats (Phillips, 1966), and in other subjects of the animal kingdom. The reconstructions presented in this paper are based on study of an extensive series of electron micrographs. There is some reason to remark that the reconstruction of a biological process from a series of stages can express a hypothesis, but that it does not establish its truth. In the cases of centriolar formation and ciliogenesis, however, no more than a superficial understanding of these processes so far has been achieved. Some merit may therefore be found in the reconstructions offered if they are helpful in establishing the conceptual framework that so often precedes the design of subtle and telling experiments. Centriole formation and ciliogenesis 209

MATERIALS AND METHODS The developing lungs of foetal rats were the principal materials used for study. They were examined by light and electron microscopy both as they matured in utero and in organ culture (Sorokin, 1961). The normally developing lungs ranged in age from 14 to 21 days of gestation if the morning when spermatozoa were found in the vagina of the mother rat was designated day 1. Organ cultures were explanted on all days between the 14th and 19th and were cultured for varying periods up to 10 days. In general, the ciliated border first appeared in the epithelium of trachea and primary bronchi during the 20th day of gestation or at a corresponding age in vitro, as judged both by examination of sections and by study of living cultures. Consequently attention was focused on lungs that fell in the 19- to 22-day range. In contrast to the preceding, primary cilia could be observed to undergo formation during the entire period surveyed. In addition to the material from rats, tissues from newly hatched chicks were examined for centrioles and primary cilia. Specimens for microscopy were fixed in barbiturate- or phosphate-buffered osmium tetroxide, as well as in glutaraldehyde plus osmium tetroxide or formaldehyde- glutaraldehyde plus osmium tetroxide (Karnovsky, 1965), and all were embedded in Epon. One-micron sections taken for light microscopy were stained with toluidine blue, while thin sections cut with glass knives for electron microscopy were stained sometimes with lead plumbite or lead citrate and sometimes with uranyl acetate. The grids were examined in RCA microscopes EMU 3 E, 3 F, or 3 G.

OBSERVATIONS Primary cilia The primary cilia of mammalian lungs resemble closely the rudimentary, or abortive, cilia that are known to occur in a wide variety of cells present in other organs (Zimmer- mann, 1898; Barnes, 1961; Latta, Maunsbach & Madden, 1961; Sorokin, 1962; Grillo & Palay, 1963; Dahl, 1963; Adams & Hertig, 1964; Schuster, 1964; Motta, 1965; Deane & Wurzelmann, 1965; Wheatley, 1967; Breton-Gorius & Stralin, 1967). In the lung such cilia are produced by virtually all of the cell types present during formation of the lung and its glands. During normal development the cilia are in peak production while cells are actively undergoing differentiation. That is, while most of the cells are sufficiently differentiated to be recognizable as to type—fibroblast, chondrocyte, myocyte, epithelial cell (Fig. 18), or mesothelial cell (Fig. 19)—they retain certain characteristics of immaturity, such as an open chromatin pattern in the nucleus, and the presence of many free , of glycogen, or of triglyceride (Sorokin, Padykula & Herman, 1959) in the cytoplasm. Epithelial cells in developing bronchi produce primary cilia some days before they begin to produce a ciliated border, and a single cilium may protrude from an immature goblet cell (Fig. 13). Primary cilia occur occasionally in cells of adult lungs, but they occur far more frequently in developing tissues, where they easily are spotted by electron microscopy in scanning grids of thin sections. Owing to their scarcity in adult tissues, these cilia are considered to have 210 S. P. Sorokin only a transitory existence. The cells of adult lungs that possess such cilia are interpre- ted to be undergoing maturation from an indifferent state, after being released to differentiate into replacements for worn elements of the lung. A cell usually produces one and rarely two primary cilia. When fully formed these cilia may be shorter than typical motile cilia. In cross-section the absence or incom- plete development of the central pair of ciliary fibres is noted (Fig. 14), which places primary cilia among those whose fibres are disposed in a 9 + 0 arrangement. While many modified 9 + 0 cilia function as sensory appendages, the function of primary cilia remains unknown. They seem to be ineffective as motile processes; at best they are said to beat erratically (Stubblefield & Brinkley, 1966).

Primary ciliogenesis The major events that occur during primary ciliogenesis are summarized in a drawing (Fig. 1). Formation of the basal body. Since the basal body of the primary cilium is derived from a centriole of the achromatic figure, or else its direct descendant, the ultimate origin of the basal body and the centrioles of the achromatic figure is the same. In both the chick and the rat, a new centriole begins to form during interphase. It arises off the wall of a parent centriole close to the base, which is the end of the parent that faces the second centriole of the diplosome (Fig. 4). In this position the long axes of parent and daughter are perpendicular to each other. The newly forming organelle begins to become visible once the parent centriole extends a feltwork of fine filaments out from its wall (Fig. 3). Thereafter, osmiophilic material for the base of the daughter is deposited at the ends of these filaments (Fig. 4), beginning at a distance of 50-70 m/t from the wall of the parent. Occasionally a section is obtained whose plane passes through both centrioles of the diplosome and one or two of these short daughter-centrioles, or procentrioles (Gall, 1961). In such cases longitudinal sections of all centrioles are obtained (Figs. 4-6). The daughter once again is sliced longitudinally when a section transects a parent centriole (Fig. 7); hence, the procentriole is an annulus. It increases in height and becomes a cylinder (Figs. 3, 4, 7, 6) through acropetal deposition of material on to its wall. In chicks (Figs. 3-5) and in rats (Figs. 6-8), all of the centrioles needed for daughter-cells and for basal bodies of primary cilia are supplied in this direct manner. In cells that are destined to possess a ciliated border, only the basal bodies of the component motile cilia are produced by a different morphogenetic scheme. The two parent- and two daughter-centrioles shown in the apical cytoplasm of a bronchial cell from a 16-day foetal rat (Fig. 6) assuredly were not intended for its ciliated border because the basal bodies for the border normally would not begin to appear in that cell until some days later. In epithelial cells centriolar formation occurs in the apical cytoplasm (Figs. 5, 6), while in fibroblasts it occurs near the nucleus where the centrioles normally reside (Fig. 8). Consequently, centrioles are able to produce procentrioles irrespective of their location in the cytoplasm. The procentriole evidently remains close to its parent until after it has reached about half of its maximum length. Thereafter the still immature centriole becomes free to move away. In a Centriole formation and ciliogenesis 211 recently divided daughter-cell it may drop down below the parent centriole to become the second centriole of the diplosome. For this reason, perhaps, the two centrioles of the diplosome sometimes are unequal in length (Fig. 4). The centriole is fully grown in rat cells when its length is about twice its diameter (Figs. 9, 10). It may then become transformed into the basal body of a primary cilium. In becoming a basal body, a centriole extends one or more satellites out from its wall (Fig. 9). The satellites join the triplet fibres of the wall at a point a little more than half-way up from the base of the centriole (Figs. 9-11, 15, 18). They help to fix the position of the centriole beneath the cell surface (Szollosi, 1964). Rootlets are more rarely produced by these basal bodies than by those of the ciliated border. The proximal half of the basal body, which is the half nearer the base, often is more densely osmiophilic than the distal half (Figs. 13, 16, 19) and consists of older material, which was laid down when the basal body was a procentriole. Others have reached a similar conclusion concerning the proximal ends of centrioles and basal bodies in other animal and plant cells (Gall, 1961; Renaud & Swift, 1964; Mizukami & Gall, 1966). In other words, the basal body retains the polarity of the centriole from which it developed. Ciliogenesis. The stages of primary ciliogenesis in the rat lung are the same as those that occur in fibroblasts and smooth muscle of the gut (Sorokin, 1962), but they differ in several respects from the stages observed during the formation of motile cilia for the ciliated epithelial border. (1) The basal bodies for primary cilia are assembled directly by the achromatic centrioles, while those of the ciliated border are produced indirectly in a manner to be described later. (2) Primary cilia arise in large numbers during much of the foetal period, while almost all of the motile cilia appear over a short span of time that begins comparatively late in foetal life. (3) Primary cilia exhibit greater morphological variation than motile cilia, both during their growth period and afterwards. In particular, the ciliary fibres of primaries seem to develop slowly. As a result, the shafts of primaries are often well lengthened before they contain well- formed fibres. In addition, fully developed primary cilia are rarely as uniform in length as the cilia of the epithelial border. Several figures summarize various phases of primary ciliogenesis: the formation of a vesicle at the distal end of a basal body (Figs. 9, 11, 12) and the appearance of the ciliary bud beneath it (Fig. 10); the elongation of the shaft and the emergence of the cilium from the cell (Figs. 13, 15, 18, 19). In connective-tissue cells, whose centrioles normally remain near the nucleus, the ciliary shaft often grows out from one of the deeply embedded centrioles. It develops within a conspicuous, trilaminar sheath (Fig. 1) that sooner or later becomes continuous with the plasmalemma. Phase- contrast observations on living fibroblasts that possess juxtanuclear cilia indicate that fully formed primary cilia can retain for some time their deep-seated position (Stubblefield & Brinkley, 1966). In epithelial cells the ciliary sheaths are either much reduced in length (Figs. 15, 55) or absent (Fig. 13) from developing primary cilia or from those of the ciliated border. In these cells, however, the centrioles or basal bodies normally lie close to the apical surface (Figs. 5, 6, 20, 59). In an earlier paper based on electron microscopy (Sorokin, 1962) it was proposed that the sheath develops from a cytoplasmic vesicle that hovers about the distal end of a ciliating basal body (Figs. 11, 212 S. P. Sorokin

Fig. i. A reconstruction of the major events of primary ciliogenesis presented sequen- tially from left to right. At the left two centrioles are sketched in a diplosomal con- figuration, much as they occur in an interphase cell. Near its basal end, each centriole begins to form a daughter centriole by extending fibrillar material out from its wall. Before occurs the centrioles separate (dashed arrows) and move to opposite poles of the cell. The subsequent fate of only one of these centrioles is depicted. Its daughter centriole, or procentriole, begins to take on form near the tips of the fibrils that extend for a distance of 50-70 mfi from the centriolar wall. Osmiophilic matter accumulates about the fibrils, and the procentriole first becomes recognizable as an annular structure with a diameter of about 100 m/i. Subsequently the procentriole lengthens into a cylinder whose long axis parallels the fibrils. At the same time it widens to approach the diameter of a mature centriole (150 m/t). When half grown the procentriole loses its right-angle orientation to the centriolar wall and may slip down, to adopt a new position facing away from the base of the parent centriole. When fully grown the procentriole resembles its parent, and the two constitute the new diplosome. A primary cilium is generated from the distal end of the more mature centriole, which becomes modified into a basal body by producing satellite arms from its wall. These appear to aid in fixing the position of the organelle. If the diplosome normally resides in the apical cytoplasm, as in many epithelial cells (upper arrow), the basal body becomes closely attached to the apical plasmalemma, and the ciliary bud grows out into the epithelium-lined cavity. If the diplosome normally rests in the neighbourhood of the nucleus (lower arrow), then the ciliary shaft lengthens within the confines of a ciliary sheath. In most cases the sheath appears to originate from cytoplasmic vesicles, which eventually fuse with the apical plasmalemma. Ciliary fibres develop in a centrifugal direction within the lengthening shaft and may reach the tip. The second centriole of the pair meanwhile moves between the various preferred positions indicated. Centriole formation and ciliogenesis 213 12, 54). The vesicle was thought to originate in the cytoplasm as a result of interaction between the basal body and the surrounding Golgi lamellae, and later to lengthen and to fuse with the plasmalemma. A less likely interpretation of the same material is that the cytoplasmic ' vesicle' is in reality no more than an invagination of the cell surface and that it remains continuous with the surface at all times. In the first interpreta- tion, the centriole rather than the plasmalemma assumes primacy as initiator of ciliary development. In the second interpretation, no conclusion can be drawn either way. The basal body of a primary cilium rarely is engaged in organizing the formation of daughter centrioles. In the example illustrated, several daughters are being generated simultaneously in a ring that encircles the base of the kinetosome (Figs. 16, 17). In this configuration the daughter centrioles are not perpendicular to the wall of the parent but are displaced downward, possibly because of crowding by the satellites of the basal body. The long axes of the daughters nevertheless point to the centre of the proximal end of the kinetosome. A grand-daughter centriole is the most exceptional feature of the generative complex illustrated (Fig. 17).

Table 1. Onset of ciliary motion observed in cultures of foetal rat lungs

Latest date no Earliest ciliary Enhanced ciliary Gestational age cilia observed motion observed activity Culture series at explantation* (days in vitro) (days in vitro) (days in vitro)

1 IS — 7 10 2 is 2 5 7 3 IS 4 6 7 4 15 — S — 5 IS — 7 — 6 IS 3 — — 7 IS S — — 8 18 3 St 7 * The first day of gestation was designated as beginning on the morning when spermatozoa were found in vaginal smears. f Ciliary motion was conspicuous.

The ciliated border of the epithelium A general survey of basal-body formation and ciliogenesis. The ciliated border first appears in the epithelium of trachea and bronchi of developing rats' lungs on approxi- mately the 20th day of gestation and in large bronchi of 15-day pulmonary explants after about 5 days in culture (Table 1). Its appearance follows swiftly on completion of the formation of basal bodies in the apical cytoplasm. Ciliary motion is first evident in cultures as a fine vibration over the apical surface. During the next day or two, ciliary motion becomes more prominent, principally because in that period the cilia reach their full length, and partly because more are added to the border. In this period the lashing of individual cilia is observed, and co-ordinated ciliary action becomes wavelike. Ciliogenesis in the lung therefore seems to take longer to complete than it does in certain fungi (Renaud & Swift, 1964) and amoeboflagellates (Schuster, 1963; Dingle & Fulton, 1966), where it may require little more than 1 h. 214 S. P. Sorokin After first being seen in the epithelium of the trachea and the largest bronchi, ciliary vibration progressively appears in the epithelium of smaller bronchial branches, in harmony with the pattern of epithelial differentiation in the foetal lung, which follows a centrifugal direction (Sorokin, 1965). Consequently, in any specimen of foetal lung, early stages of ciliogenesis will be found in bronchi located distal to those possessing cilia. Such an arrangement is of value to one who seeks to reconstruct the process of ciliogenesis from a series of fixed stages. Nonetheless, the guidance it affords on the order of events in ciliogenesis is limited by the occurrence of several presumptive stages in the same cell. In spite of the disadvantages of this approach, the following information concerning the correct order of events was obtained by using it. (1) In very recently formed, peripheral bronchial branches, low columnar epithelial cells contain a centriole and fibrogranular material in the apical cytoplasm (Fig. 20). (2) Other cells, judged older than the preceding by their position in the bronchial tree and by their tall columnar appearance, contain both the fibrogranular areas and independent electron-dense bodies, here termed ' deuterosomes' (Fig. 35). (3) Still other bronchial cells contain the foregoing plus several clusters of procentrioles undergoing development around the deuterosomes (Fig. 36). (4) Cells that contain free basal bodies in addition to all these

•;; •=•••:• b2rsv&«', ;V : : ?>' ^ / - ;-..&.Sv-. -:f?ft-f -MMf>"£

Fig. 2. See facing page for legend. Centriole formation and ciliogenesis 215 precursors (Fig. 46) are considered to be more mature; and (5) those that bear cilia, along with reduced numbers of the elements mentioned (Fig. 59), are thought to be the most mature. From these examples it was deduced that the morphogenetic sequence begins with fibrogranular elements and continues on, in order, through stages with deuterosomes, procentrioles, basal bodies, and cilia. Additional evidence from electron micrographs served to confirm the preceding deduction and to permit the reconstruction of a plausible, although not definitive, sequence of events that occur during the morphogenesis of the ciliated border (Fig. 2). This evidence is presented in the following sections. Indirect formation of basal bodies. The kinetosomes of the developing ciliated border are assembled from simpler precursors that begin to appear in the supranuclear cytoplasm of maturing epithelial cells somewhat less than a day before cilia are seen. Although it is clear that the basal bodies do not arise directly from pre-existing cen-

Fig. 2. A pictorial summary of developmental stages encountered in the epithelium of the bronchial tree during the formation of basal bodies and cilia. The stages have been arranged into a plausible sequence (A-F). The objects illustrated are all drawn roughly to scale. The onset of basal-body formation in immature, non-ciliated cells is marked by the appearance of fibrogranular aggregates in the apical cytoplasm near the site of the cell's previously existing pair of centrioles (A). At first the filamentousmatri x appears to predominate, but later highly osmiophilic, granular elements come into prominence. In some sections these components seem thrown together in disarray (A, above); in others they appear to be fashioned into a cored structure (A, below). Annulate lamellae and Golgi elements frequently border on the aggregates. At a more advanced stage of development the aggregates become rearranged into a bundle (B) of strands composed of parallel coursing matrical filaments surrounded by the osmio- philic granules. In cross-sections 8-12 granules rim the individual strands. Tubular extensions from the freely ramify in the space between strands. Eventually the bundle becomes subdivided into smaller units, and individual strands are released to the cytoplasm. Through continued deposition of osmiophilic material at the periphery of the strands, the ring of granules becomes consolidated into fewer elements (C, above). These round up into radial structures whose axis of symmetry lies perpendicular to the long axis of the strand (C, below). The structures are released to the cytoplasm and develop into deuterosomes (D), which are represented in surface view and in section. Fine filaments extend radially outward to form a corona about the deuterosome. Elaboration of procentrioles (E) occurs at the outer limits of the corona. As many as 14 evenly spaced procentrioles may develop simultaneously about a deutero- some; for this reason, sections of procentriolar clusters frequently pass through symmetrical arrays of 2, 3 (illustrated), 4, 5 or 6 (illustrated) procentrioles, whose long axes point to the centre of the deuterosome. Procentrioles develop similarly, whether about a deuterosome or against a centriole (see Fig. 1). The procentrioles elongate centrifugally, and their distal ends curve slightly inward. Cross-sections (illustrated) reveal the relative immaturity of the distal ends. Triplet fibres of the wall seem to be developed from singlet and doublet fibres, which precede them in order of appearance. In short procentrioles the doublets and triplets occur in nearly radial arrangement. When about half grown, the procentrioles are released from the deuterosomes. They continue to elongate and to increase their diameters to the dimensions of centrioles. The mature then migrate to the cell surface, where they form cilia apically, satellites laterally, and rootlets basally (F). Material not dissimilar to the fibrogranular aggregates appears to be involved in the formation of the rootlets. When the ciliary shafts have reached somewhat more than one-sixth of their full length, the internal fibres extend to the tips and the cilia become motile. 14 Cell Sci. 3 216 S. P. Sorokin trioles, there is some reason to believe that the simple precursors are synthesized or are assembled in a close spatial relationship with the centrioles of the achromatic figure. Renyi (1924) stated that in tracheal epithelium a previously apical and centrally located pair of centrioles migrates to one side of the apical cytoplasm close to the cell surface. Thereupon it proceeds to form basal bodies, which he thought occurred by a process of centriolar division. This study verifies by electron microscopy that in preciliated cells of newly formed bronchi the centrioles lie in lateral regions of the apical cytoplasm. In the micrograph selected to illustrate this point (Fig. 20), the section passes through two centrioles located in lateroapical regions of adjacent cells. An area of aggregated fibrogranular material (Figs. 20, 21) is found about 250 m/i away from one of these centrioles. This material is described below as the first recognizable precursor of the basal bodies. The absence of cilia and precursors from neighbouring cells helps to confirm the impression that a very early stage in basal-body formation is depicted. The micrograph therefore is evidence that (1) at least one of the centrioles of the achromatic figure is present in a cell that is beginning to produce basal bodies, and that (2) it is located close by the precursor material. Careful examination of other parts of similar preciliated cells, including the nucleus, has so far provided no evidence that the precursor material first localizes elsewhere. After first being seen in the apical cytoplasm of preciliated cells, the fibrogranular aggregates increase in size and become conspicuous features of the apical and supra- nuclear cytoplasm of preciliated and ciliated cells. In various sections the bodies of fibrogranular material appear round, ovoid, or spindle shaped. A typical body may measure 1 /i in diameter by somewhat more in length, and it is not separated from the cytoplasm by a limiting membrane. It consists of numerous 400-800 A electron-dense granules embedded in a matrix of filaments some 30-50 A thick (Figs. 21-24). The electron-dense granules themselves consist of interwoven strands of both smaller (30-50 A) and larger (about 100 A) filaments. In some of these heterogeneous bodies the granules are small and inconspicuously mixed in with the matrix (Figs. 23, 24). As the granules increase in size, they tend to collect at the periphery (Fig. 22). It is uncertain whether more than one of these fibrogranular aggregates exists in the pre- ciliated epithelial cell. Although they appear to originate close by the pre-existing centrioles, they exhibit close associations with Golgi elements and annulate lamellae (Figs. 23, 24, 26), from which they may receive partially synthesized material. It is just conceivable that the fibrogranular body represents a centriole of the diplosome in a dispersed state; in some sections a core and shell organization is apparent (Fig. 22). As the granular component of the fibrogranular aggregate becomes more prominent, the entire aggregate body becomes rearranged into a bundle of strands composed of parallel coursing filaments of the matrix surrounded by the granules (Figs. 25-27, 46). In cross-sections 8-12 granules surround the strands and form common walls between adjacent strands. The main body of central filaments occupies a space about 150 m/t in diameter, but radial connectives link the centre to the peripheral granules, which form a ring about 200-250 mju, in diameter (Figs. 27, 29). Some well-developed fibrogranular bundles measure well over 1 (i in diameter. The strands within these bodies either are numerous and separate, or else are part of one continuous thread Centriole formation and ciliogenesis 217 wound up like a ball of yarn. Eventually the large bundles either unravel or become subdivided into smaller bundles (Figs. 25, 26, 46), and individual strands lie free in the cytoplasm (Figs. 28, 30). After the granules have become prominent in the fibrogranular areas, they begin to become consolidated into a number of larger units, with the result that fewer granules than formerly are counted at the periphery of the individual fibrogranular strands (compare Figs. 27-30 with Fig. 33) and that some of the granules are conspicuously elongated (Fig. 34). The elongated granules then become transformed into rounded bodies whose osmiophilic material is arranged about a centre of lower electron density (Figs. 31, 32). Like procentrioles, the bodies have no limiting membrane and measure about 100 m/i across. Unlike procentrioles, they do not originate from a basal disk and develop a wall of triplet fibres. Some of these bodies are attached to the fibrogranular strands by means of fine filaments (Fig. 29), but others occur as independent bodies (Fig. 34). In a few sections it is possible to find evidence that a maturation sequence occurs along a single fibrogranular strand and terminates in the release of the rounded, osmiophilic bodies at the end (Fig. 35). These independent bodies are thought to mature into deuterosomes. The second certain precursor of the basal bodies, the deuterosome, appears in the apical cytoplasm of preciliated cells (Fig. 35) once the fibrogranular areas have arisen and undergone the transformation just described. The deuterosome is an extremely osmiophilic body that in some sections appears solid and in others annular (Figs. 37- 39). It functions as an organizing centre for the growth of procentrioles that are destined to mature into kinetosomes of the ciliated border (Fig. 36). In sections the deuterosome measures some 100-150 m/t in diameter. Whenever a section passes through the centre of a deuterosome, it transects an electron-dense cortex and an electron-transparent medulla (Fig. 38). Moreover, sections never pass through the medulla without passing through the encircling cortex. Hence, the deuterosome is spheroidal. As far as they can be identified in electron micrographs, the structural components of the deuterosomes resemble those of the fibrogranular areas, except that in the deuterosomes they are tightly compacted. The medulla rarely contains more than a few filaments (30-50 A) of intermediate electron density, but the cortex consists of a rich admixture of filaments (30-50 A) and amorphous osmiophilic matter (Figs. 37- 45). In addition, some clear areas stand out against the osmiophilic background of the cortex (Figs. 38, 39, 45). These measure about 100 A across and represent a second fibrous component of the body. Outside the cortex, fine filaments radiate into the surrounding cytoplasm to form a corona about the deuterosome (Figs. 37, 38, 42). Within the apical cytoplasm the deuterosomes may lie free (Figs. 35, 37-39) or in association with the proximal end of a newly formed kinetosome (Fig. 46). Most frequently they are associated with procentrioles (Fig. 36). Under these circumstances, a deuterosome usually occupies the centre of a cluster of procentrioles, which approxi- mates 5500 m/t in diameter. In sections through these clusters the following con- figurations commonly occur: 2 procentrioles lie in one plane on opposite sides of the deuterosome (Figs. 42, 46), 3 are arranged in an equilateral triangle about the central 14-2 2i8 S. P. Sorokin body (Fig. 43), 4 form a square (Fig. 36), and 6 outline a hexagon (Fig. 44). A sym- metrical arrangement of 5 procentrioles about the centre is infrequently seen (Fig. 36). In addition, irregular groupings of up to 6 procentrioles are recognized (Figs. 36, 47). All of the above configurations are consistent with the interpretation that procentrioles typically develop in spherical arrays of up to 14 regularly spaced units clustered about one central deuterosome. A section may nevertheless reveal only one procentriole adjacent to a deuterosome (Figs. 35,40,41,46). As this configuration is not encountered in sections through a cluster of 14 procentrioles, it must be that, like a centriole, a deuterosome can organize the formation of either one or more procentrioles at a time. In brief, by serving as a centre for the growth of procentrioles, the deuterosome sub- stitutes for a centriole. In this sense it may be viewed as a condensed centriole with closed ends. The procentrioles that arise from the deuterosomes are similar in all respects to the ones that develop directly off the wall of a parent centriole prior to mitosis or to primary ciliogenesis (Fig. 45). In the course of the elaboration of the basal bodies all the procentrioles of a given cluster mature simultaneously, but those of different clus- ters in the same cell do not necessarily develop in synchrony with each other (Figs. 36, 46). Their bases are laid down at the outer limits of the corona of filaments that extend radially from the deuterosome (Fig. 40). Each procentriole is oriented so that its base is tangential to the surface of the deuterosome, and its long axis roughly parallels the coronal filaments (Fig. 47). Growth is plainly visible as an increase in height, which proceeds in a direction away from the central body (Figs. 40-45). The procentriole develops into a cylinder, except near its growing tip, which curves inward (Figs. 41, 43). While engaged in a cluster, a procentriole typically is separated from the deuterosome by a distance of 50-70 m/t. Transverse sections through various levels of procentrioles and immature centrioles show that the wall is always more highly de- veloped nearer the base than nearer the apex (Figs. 43, 47). A procentriole at first is radially symmetrical. It has an axial filament throughout most of its length and radial struts that connect the axis to the wall (Fig. 47). At first its fine structure consists of interwoven filaments (30-50 A), but shortly afterwards the centriolar fibres (about 200 A) begin to appear in the wall near the base and to extend apically. These fibres are present initially as singlets and doublets, or as intermediate forms, either singlets with arms or doublets with arms (Figs. 2, 47); and they tend to be arranged in a radial pattern. Elements of the wall thus occupy much of the volume of a procentriole and leave little room for material of the core. Later on a set of 9 triplet fibres is developed from these precursors, possibly through closure of the arms on the intermediate forms, and the definitive centriolar wall becomes established throughout the organelle, progressively from base to apex. In becoming triplets, the centriolar fibres evidently rotate, so that the innermost subfibre (subfibre A of Gibbons & Grimstone, i960) is swung outward and each triplet comes to lie at the periphery of the centriole, where it forms an angle of 300 with the tangent to the circumference. In this manner the radial symmetry present in the procentriole is lost during subsequent development. Simul- taneously with the displacement of the triplet fibres, a central core is created within the immature centriole, and the organelle increases slightly in diameter. Once they are a Centriole formation and ciliogenesis 219 little more than half grown, the immature centrioles are released from the deutero- somes. They continue to elongate until they reach maturity, which occurs when their height ranges between i-8 and 2-5 times their diameter (Fig. 48). After attaining its majority, a centriole undergoes the following changes in becoming transformed into a kinetosome. (1) It migrates, or is drawn, to a position just beneath the apical plasmalemma, where it joins other maturing basal bodies in forming a single layer made up of rows of kinetosomes. In this layer the apical (distal) ends of the kinetosomes point outward. In general, the organelles space themselves evenly, and, as a new crop of basal bodies matures, individuals are fitted into available spaces until the cell possesses its full share. (2) Each maturing basal body acquires a rootlet of osmio- philic material. Like the basal body, the rootlet has fibres, filaments, and amorphous components; but in the rootlet they are less regularly arranged than in the basal body. Consequently, rootlets are easily recognized in sections (Figs. 55,60, 61, 64). Relatively little is known about the formation of these structures. They are not present when procentrioles are attached to deuterosomes, although the coronal filaments that link the deuterosome to the procentriolar base occupy the position of the rootlet. Sometimes a deuterosome persists below the base of a maturing kinetosome (Fig. 46). More frequently, a cloud of granules like those in a fibrogranular area occupies the space beneath a basal body. The granules appear to withdraw as the rootlet lengthens down- ward (Figs. 52, 54, 60, 62). The rootlet therefore may be produced by mechanisms similar to those employed in fashioning a centriole. (3) At approximately the same time one or more satellites are extended from the wall of the basal body out into the cytoplasm. A small satellite will join 2 adjacent triplet fibres, and a larger one will join 3 consecutive triplets (Fig. 64). According to Gibbons's (1961) study of cilia in Anodonta, if reference is made to an overlying cilium, the triplet fibres that bear the satellite (basal foot) will be seen to correspond to ciliary fibres 5 and 6 and the foot will be seen to point in the direction of the effective ciliary stroke. If this were true of kinetosomes in the rat, then it would be clear that in the subapical layer of the bronchial epithelium the kinetosomes do not become aligned until some time after they have begun to produce cilia (Fig. 64). None the less, Gibbons's finding concerning the basal foot (satellite) in Anodonta is not strictly applicable to the rat because the mammalian basal body can produce several satellites (Figs. 9, 59). The fate of the deuterosome has not been traced satisfactorily beyond the point where it gives up its daughter centrioles to the cytoplasm. During the period while the procentrioles are clustered about them, the deuterosomes seem to undergo little change in structure. Specifically, the dimensions of cortical and medullary regions remain fairly constant whether the deuterosomes are surrounded by relatively short or relatively tall procentrioles. Moreover, the textures and electron densities of these regions do not seem to change during the period, in so far as this can be determined by inspection and measurement of deuterosomes in electron micrographs (Figs. 35-47). The differences that are observed among the deuterosomes pictured are attributable with more certainty to variations in level of section or quality of fixation than to dynamic changes. For the present, one further observation may be borne in mind: relatively rarely a number of procentrioles are seen to cluster about a more fully 220 S. P. Sorokin developed centriole or basal body (Figs. 49, 50). When the organizing centrioles are in diplosomal formation and the procentrioles are in a configuration of 8 around them (Fig. 49), the pre-existing centrioles of the cell are depicted. On the other hand, when the organizing centriole is solitary and is ringed by several procentrioles in one of the configurations expected around a deuterosome (Fig. 50), then the central body may well represent a deuterosome transformed into a centriole. Formation of the cilia. In general, cilia are not produced until the kinetosomes have completed all of the maturation changes described. Nevertheless, some kinetosomes begin to produce cilia a little early. In each epithelial cell a crop of cilia begins to appear immediately after a population of kinetosomes matures. Within one such ciliary wave individual cilia are formed at slightly different rates. Consequently, while most of the cilia in a given ciliary wave develop synchronously, a few lead and a few trail the others. In addition, the ciliating cell contains precursors for subsequent generations of basal bodies (Fig. 46) which will produce other crops of cilia until no more are required. In the ciliated border of the epithelium, the ciliary shafts develop for the most part outside a confining ciliary sheath. None the less, frequently it is possible to observe several small vesicles at the distal end, or a single vesicle covering that end, of a mature basal body (Figs. 51, 53-55); and this may be interpreted to be either a rudimentary ciliary sheath (the primary ciliary vesicle) or an invagination from the plasma mem- brane. The ciliary bud always develops in continuity with the distal end of the basal body. When it is just large enough to protrude from the cell, it is a tiny club of homo- geneous and electron-permeable cytoplasm that is surrounded by a continuation of the plasma membrane (Figs. 56, 57). At such an early stage the developing shaft of a motile cilium does not differ greatly from the shaft of a developing primary cilium, a budding rod outer segment of the retina (De Robertis, 1956; Tokuyasu & Yamada, 1959), or a sprouting (Renaud & Swift, 1964; Dingle & Fulton, 1966). Very shortly thereafter, the ciliary fibres appear within the shafts, beginning at the end nearer to the basal body (Figs. 58, 60, 61). Growth of the fibres seems to occur at their tips, inasmuch as the fibres of developing motile cilia, as well as of primary cilia, are least well organized at their tips. There they sometimes are associated with small (400-600 A) vesicles (Fig. 61), which are seen as well in the cores ot the basal bodies (Fig. 60). By the time when the ciliated border has grown to a sixth of its full height, the fibres of the cilia nearly reach the ends of the shafts (Fig. 58). When the cilia have lengthened to a third of their full height, the internal fibres extend throughout the shafts (Fig. 62). At this stage of development the incompletely lengthened cilia resemble mature cilia in details of fine structure (Figs. 62-64), and tne shafts are motile, although co-ordinated ciliary activity is not yet established. Action of these short cilia evidently accounts for the vibratile motion that can be seen by light micro- scopy of living lung cultures and is the first reliable sign of ciliation to be detected by that method. Concerning the elongation of the shaft, as opposed to the fibres within, recent studies on regeneration of flagella (Rosenbaum & Child, 1964; Tamm, 1967) have been more informative than electron micrographs. The micrographs also do not make Centriole formation and ciliogenesis 221 clear by what mechanism the ciliary are made, whether by ribosomes of the cytoplasm or by synthetic machinery within the basal body. The ciliated border continues to develop beyond the time when the first functional cilia appear. These first-formed cilia grow out to their full height while new waves of kinetosomes are added to the border, and the cilia become precisely aligned with respect to one another. The maturation period closes when the ciliated border contains its full number of motile units, and their actions have become co-ordinated.

DISCUSSION The indirect synthesis of basal bodies Fibrogranular areas. It is fairly clear that the fibrogranular areas are the first precursors of basal bodies to be seen in electron micrographs of ciliating cells. After comparative study of many immature, potentially ciliated cells, one can conclude that the electron-dense granular components of these fibrogranular areas originate within the matrix. As there are about 250-300 cilia at the surface of each mature pulmonary epithelial cell in the rat (Rhodin & Dalhamn, 1956), only the most casual examination of the fibrogranular areas should be sufficient to show that too many granules are present for each to be the sole precursor of a basal body. At a later stage of development, when the fibrogranular aggregates have become organized into bundles of strands that are encircled by the granules, cross-sections of the strands sometimes reveal 9 granules in the peripheral ring. It is possible to imagine that each is destined to become 1 of the 9 triplet fibres in the wall of the basal body. None the less, it is also possible to count 8, or n, or 12 peripheral granules in other cross-sections; and, after the large bundles have broken up into smaller ones, one can find rare examples of individual strands ringed by fewer than 6 granules, as shown in the illustrations. Clearly, critical analysis favours the concept that the granules result from deposition of osmiophilic material in the fibrillar matrix and increase in size as a consequence of further deposi- tion. At the same time, the matrix becomes organized into strands, and the granules become displaced to the periphery. Eventually several granules fuse together and become transformed into the walls of the deuterosomes. The relative scarcity of consolidating granules, as compared to fibrogranular aggregates and bundles in preciliated cells, may be a reflexion of their rapid transformation into deuterosomes. While it is evident that fibrogranular areas are remote predecessors of the basal bodies and immediate precursors of the deuterosomes, it is not yet clear exactly what the filaments and granules represent, or how they arise in the cytoplasm. In relation to these uncertainties, however, it is worth keeping two observations in mind, namely (1) that components of these aggregates seem to accumulate near one of the pre- existing centrioles in the immature, non-ciliated cell; and (2) that annulate lamellae and adjacent Golgi elements are seen to have a particularly extensive and intimate association with the fibrogranular areas, as compared with the other precursors of basal bodies. In describing the occurrence of the granular components of these areas in cells of adult human bronchial epithelial cells, Frasca, Auerbach, Parks & Stoeck- enius (1967) also noted occasional close associations between the granules ('filosomes') 222 S. P. Sorokin and annulate lamellae (Palade, 1955; Swift, 1956; Kessel, 1965; Gross, 1966). In short, it appears as if the fibrogranular areas result from the contributions of both the principal synthetic centres of the cytoplasm and the centriolar compartment. The former may supply materials until sufficient is on hand for use in constructing both deuterosomes and procentrioles. The latter may introduce a template and polymerizing into the areas to organize into deuterosomes the partially assembled structural elements received from the Golgi apparatus. At present it is speculative whether or not the filaments in the matrix of these areas are comparable in properties other than dimensions and electron density to the filaments that extend from the wall of a centriole to reach a daughter procentriole, or to the corona of filaments that radiates from a deuterosome. Similarly, it is not known whether or not the amorphous fibrogranular aggregate represents a dispersal of one of the achromatic centrioles, somewhat like a chromosome at interphase. As for the granules, nothing is known for certain of their chemical composition, although they assuredly are not aggregates of glycogen, nor clusters of ribosomes, as these substances are known to appear in electron micrographs. In other circumstances, granules of similar size, filamentous substructure, and osmiophilia seem also to be associated with centrioles or basal bodies. In ciliating cells, granules and filaments similar to those under discussion have been observed below the basal bodies while rootlets were under development; and granules like those of the fibrogranular areas may form one or two rings about a centriole (Bessis, Breton-Gorius & Thiery, 1958) prior to the appearance of satellites for the attachment of spindle fibres, or for the attachment of basal bodies to the cell surface. In all these cases the granules seem to represent incompletely organized structural material for these centriolar affiliates. Deuterosomes and procentrioles. There is reason to believe that deuterosomes are composed of the same structural subunits as procentrioles, and to the extent that both are built on a radial plan and both are composed of tightly knit osmiophilic filaments, the bodies are similar. Nevertheless, the two bodies differ in shape, and this difference is attributable to their dissimilar manners of origin. A deuterosome is spherical because it arises by condensation of preformed granules about a central point. A procentriole is cylindrical because it forms against an organizing body. Before a procentriole begins to appear, the parent structure extends from its wall numerous radially directed filaments. Osmiophilic material is then deposited at the tips of these filaments. As the material polymerizes, certain macromolecular chains may form and link end-to-end into a ring. The ring possesses an axis and radial struts and becomes the base for a newly emerging procentriole. The base lies in a plane tangential to the surface of the organizing body. New material for the procentriole is deposited on the side of the base that faces away from the organizer along the long axes of the orienting filaments. In this way the procentriole elongates into a cylinder. Centriolar fibres first begin to develop as singlets and doublets at the base of the procentriole and gradually spread upward. By the time the procentriole leaves its site of formation, its wall has become definitively organized throughout most of its length. The triplets contribute to the stability of the cylinder, possibly because without rearrangement they cannot all join ends to close it off. Where they are absent, as at the growing end of the procentriole, the wall turns in. Centriole formation and ciliogenesis 223 Thus, deuterosomal formation is analogous to rapid crystallization from a super- saturated solution. Procentriolar growth is like that of a large crystal in a solution that has been seeded and allowed to evaporate slowly. Deuterosomes and centrioles. In possessing the capacity to form procentrioles, or, in an alternative formulation, to direct the orderly assembly of procentrioles from partially assembled subunits, the deuterosomes share a most fundamental property with centrioles and basal bodies. In this respect all three must be considered to be equivalent structures. Deuterosomes differ from the other bodies principally in that they are nearly spherical and do not contain well-defined triplet fibres in their deeply osmio- philic cortices. In spite of these differences, the cortex of the deuterosome can be compared to the basal portions of the walls of centrioles and kinetosomes, because both cortex and walls issue the fine filaments that orient and perhaps organize the pro- centrioles that develop around these bodies. In simpler terms, the wall is the reproduc- tive surface of the centriole and the cortex is that of the deuterosome. In that body, moreover, the cortical surface is the only surface available for interaction with the cytoplasm, since the deuterosome has no open ends. Consequently, the deuterosome produces no polarized appendages, such as cilia at the top and rootlets at the base. It also produces no satellites, which typically arise from the middle or upper portions of centriolar walls; and for this reason it cannot participate in mitotic events. As the cilia, rootlets, and satellites are all in direct continuity with the centriolar triplet fibres, it may be argued that the triplets are a necessary prerequisite for the appearance or the maintenance of these appendages. Since the more simply constructed deuterosome possesses the centriole's reproductive capacity, it follows that neither the presence of triplet fibres nor a cylindrical parent body is essential to the fabrication of centrioles.

Alternative pathways to the basal bodies In the rat lung, basal bodies can be produced by either or both of two pathways: (1) through successive generations of centrioles, or (2) through fibrogranular aggregates and deuterosomes. If 250-300 kinetosomes were to be produced by successive genera- tions of centrioles producing a maximum number of descendants each time, it is just possible that the required number could be produced in about two generations, assuming few failures along the way. Gall (1961) has shown in atypical spermatocytes of Viviparus that 11 procentrioles can form at once about a centriole. He speculates that if all the procentrioles had been seen in his micrographs, the true number might have fallen between 15 and 20. Were such fecund centrioles present in tissues of the rat, two generations would furnish all the basal bodies needed. It is true that under certain circumstances in rats the centrioles of the diplosome produce a cluster of procentrioles about their bases; and, if the maximum number of procentrioles so far observed around one of these centrioles is only 8, it is also true that both centrioles can generate them. Consequently, 250-300 basal bodies could be produced by some- what more than 2 generations, or in 3, if in the first generation the centrioles only duplicated themselves, as they normally do in the period before nuclear division. None the less, despite such a possibility, the kinetosomes of the bronchial epithelium normally proceed from deuterosomes. If these spheroids can produce a crop of 224 S. P. Sorokin 14 procentrioles and possibly become transformed themselves into basal bodies, a minimum of only 18-20 deuterosomes would be sufficient to furnish in one generation the required number of basal bodies. Inasmuch as several deuterosomes frequently are included in one 100-150 m/i section, there can be little doubt but that at least 20 deuterosomes are present in the whole preciliated cell. Because electron micrographs of preciliated bronchial cells most frequently include views of lengthening procentrioles, it may well be that this step occupies much of the time taken for the synthesis of basal bodies. If this is true, then there is reason to believe that a large number of basal bodies can be produced in less time by the indirect pathway than by direct budding off the centrioles. (1) A deuterosome is the only organizer of procentrioles in rats that does not form directly off a centriole; hence, a cell can make more of them at one time than it could produce organizers at the limited number of sites available on the centrioles of the diplosome. (2) Once formed, deuterosomes apparently can begin to organize the growth of procentrioles without further maturation. Procentrioles cannot do this and as a rule mature to the point where they are released from their organizer before they begin to produce daughters. Accordingly, in the indirect pathway, sufficient numbers of deuterosomes can be supplied at one time to furnish growth sites for all of the basal bodies needed, and these can begin to form without delay. Consequently, only one period of pro- centriolar maturation need elapse before the cell possesses all of its basal bodies. In the direct route, two or three maturation periods would have to elapse before the cell could produce a similar number. On the other hand, where only a few replacement centrioles are needed before cell division or primary ciliogenesis, it is more efficient to use the direct pathway as virtually all cells do.

Centriole formation in diverse species In a wide variety of living forms, centrioles or basal bodies most frequently are produced through the agency of synthetic pathways similar to the two present in cells of the rat's lung. (1) That centrioles in many animal cells can arise directly off the wall of a parent centriole has been known since the turn of the century (Sleigh, 1962; Went, 1966) and more recently has been verified by electron microscopy. In most cases the daughter centriole eventually grows into a body morphologically indistinguishable from its parent, even when the parent centriole possesses characteristics not found in typical centrioles (Cleveland, 1957). Nevertheless, Phillips (1967) has shown in larval sperma- togonia of Sciara that giant centrioles composed of 20-50 tubules arise from a position off the wall of a typical 9-membered centriole, and that still larger centrioles may arise from the giant centrioles. In this case the daughter centrioles clearly are morphologi- cally distinct from their parents. It is also known, particularly from studies on Protozoa (Grasse, 1961; Vickerman, 1962; Bradbury & Pitelka, 1965; Randall et al. 1963), that basal bodies can produce single centrioles. Like free centrioles, the basal bodies of rats in addition are capable of producing several daughters at once, even if normally they are restrained from doing so. (2) In the few types of ciliating epithelium so far studied by electron microscopy, the kinetosomes all seem to arise indirectly from precursors similar to those described in the rat lung. Since these precursors are too small to be Centriole formation and ciliogenesis 225 resolved in the light microscope, it is not surprising that little knowledge about this indirect pathway has been contributed by light microscopists of the past. All of the following papers, like my own, present somewhat conjectural accounts of the indirect process of basal body formation. They all are based on study of a series of fixed stages and differ from each other more in the completeness with which they reconstruct the indirect pathway than in points of interpretation. There is agreement that in rats and mice procentrioles develop in clusters about a deuterosomal core and that fibrogranular aggregates somehow are involved in the formation of basal bodies (Stockinger & Cirelli, 1965; Cirelli, 1966; Dirksen & Crocker, 1966; Sorokin & Adelstein, 1967; Frisch, 1967). Indeed, in the salamander Xenopus laevis (Steinman, 1968) as in rats, tracheal cells produce essentially the same array of precursors in their cytoplasm before basal bodies appear, and cilia are formed in a similar way. Two of these accounts (Dirksen & Crocker, 1966; Sorokin & Adelstein, 1967) agree further that fibrogranular aggregates are precursors of the dense bodies termed deuterosomes in this paper. In commenting on the fate of the deuterosome ('condensation form'), Dirksen & Crocker (1966) advanced the idea that it became shrunken after transferring its substance to the procentrioles undergoing development around it. At present, however, there is little evidence to support this view. (3) A morphologically distinct type of indirect pathway for the manufacture of kinetosomes may be found in male gametophytes of certain cycads and pteriodophytes (Sharp, 1914; Mizukami & Gall, 1966). In these plants mass production of basal bodies from simpler precursors occurs within a large cyto- plasmic body termed the blepharoplast. The procentrioles appear at the periphery of the body with their bases facing the centre. When partially mature, they are released into the cytoplasm, where they complete their maturation into kinetosomes, migrate to the surface, and produce the cilia of the sperm.

The centriolar organizer A substance capable of organizing material into centrioles or basal bodies would seem to be a common constituent of centrioles or their precursors in the two indirect pathways. Nevertheless, nothing definite is known about the chemical composition of the hypothetical substance, nor anything about the morphogenetic roles it plays. If it synthesizes the centriolar subunits as well as directs their assembly into centriolar •walls, such an organizer may well contain DNA or RNA (Randall & Disbrey, 1965; Hoffman, 1965). Both have been sought with some success in basal body fractions from Tetrahymena using biochemical techniques (Seaman, i960, 1962; Argetsinger, 1965; Hoffman, 1965) and by observing both staining with acridine orange and incorporation of tritiated thymidine in these organelles (Randall & Disbrey, 1965). If nucleoproteins are present in basal bodies, it is certain that they are present in low concentrations and that the DNA value in particular is near the sensitivity limit for the assay (Hoffman, 1965). The blepharoplasts oiZamia are negative for nucleic acids, as determined by the azure A/Feulgen technique for DNA and azure B for RNA (Mizukami & Gall, 1966). The negative results nevertheless leave open the possibility that minute quantities of nucleoproteins are present. On the other hand, if nucleo- proteins are absent from centrioles, non- organizers might be present. They 226 5. P. Sorokin would function simply by linking into centrioles the macromolecules synthesized elsewhere (Hoffman, 1965). In that contingency, the classical notion of centriolar autonomy would have to be discarded. Whatever may be the organizer substance, it is very resistant to damage inflicted by X-rays (Sorokin & Adelstein, 1967).

Growth of cilia Three points are emphasized concerning the growth of cilia. (1) A lag was noted in the formation of ciliary fibres in shafts of developing primary cilia as compared to those of motile cilia in the bronchial epithelium. The observation suggests that for- mation of the ciliary shaft and of the fibres are to some extent separable phenomena. (2) The outer ciliary fibres first appear at the distal end of the basal body as extensions of the fibres in its wall. They serve to maintain the basic organizational plan of the basal body in the ciliary shaft. Because the filamentous substructure of a developing ciliary fibre is more loosely knit at its growing tip than at the junction with the basal body, the fibre can be envisioned to form by steps, in agreement with the concept of Roth & Shigenaka (1964), and possibly by mechanisms similar to those used in producing centriolar fibres. Breton-Gorius & Stralin (1967) have provided evidence that the formation of doublet fibres in the shafts of primary cilia is preceded first by the appearance of singlet fibres and later by singlets with lateral arms. Little is known about the formation of the central pair of fibres in the shafts of motile cilia or the manner in which the transition zone between basal body and cilium becomes definitively organized. Because this zone is arranged differently in various animals (Gibbons & Grimstone, i960; Gibbons, 1961; Fawcett, 1961; Rhodin & Dalhamn, 1956), its final form in a given species may result from reworking of a general pattern. (3) In the cytoplasm of elongating ciliary shafts, the absence of ribosomes and of identifiable substrates like glycogen or fat is some evidence against the idea that the contents of the shaft are synthesized in situ. Partially assembled materials may be transported to the tips of the ciliary fibres in the vesicles that occur in the basal bodies and in the elon- gating shafts. The remaining cytoplasm may flow into the shaft behind the advancing ciliary membrane, whose origin, like that of the plasma membrane, remains a matter for conjecture.

Regulation of centriole formation and ciliogenesis In the developing rat lung, the following situations can occur at different times in the same type of epithelial cell: direct formation of one centriole at a time; growth of a cilium from one of the centrioles in the diplosome; direct formation of several cen- trioles from a parent centriole or basal body; and indirect formation of many basal bodies followed by ciliogenesis. In the material studied so far, only the first two and the last two situations have been observed to occur at the same time in the same cell. A few conclusions can be drawn from these observations. (1) The control over centriolar formation is distinct from the control over ciliogenesis. One process can occur in the absence of the other, or both can occur together. As discussed earlier, different parts of the centriole may be concerned with these separate functions. (2) It is exceptional to find basal bodies that bear a cluster of procentrioles. In other words, basal bodies Centriole formation and ciliogenesis 227 possess the capacity to form centrioles but normally are inhibited from doing so. (3) The direct pathway can be regulated to permit the generation of only one, or of a crop of centrioles. (4) In cells that normally possess a ciliated border, production of centrioles through the direct pathway can be supplemented through the indirect pathway. In this sense the indirect route is activated separately from the direct route, but whether or not by the action of the same initiator is unknown. Moreover, since both the diplosomal centrioles and the deuterosomes can produce procentrioles in these cells at the same time, it seems probable that the initiating influence comes from outside the centriole. If this is true, then the centriole acts only to direct the synthesis of its kind and becomes restricted in its morphogenetic role to semi-autonomy.

I wish to thank Miss Gillian Pederson-Krag for culturing the lungs and to thank Mr R. F. Hoyt, Jr, and Dr D. M. Phillips for offering valuable discussion and criticism during the course of this study. I am also grateful to Drs J. M. Frasca, O. Auerbach, V. R. Parks and W. Stoeckenius for making their manuscript available to me prior to publication. This work has been supported by a Research Career Development Award GM 19441 and Research Grant GM 10949 to the author from the U.S. Public Health Service.

REFERENCES ADAMS, E. C. & HERTIG, A. T. (1964). Studies on guinea-pig oocytes. Electron microscopic observations on the development of cytoplasmic organelles in oocytes of primordial and primary follicles, jf. Cell Biol. 21, 397-427. ARGETSINGER, J. (1965). The isolation of ciliary basal bodies (kinetosomes) from Tetrahymena pyriformis. J. Cell Biol. 24, 154-156. BARNES, B. G. (1961). Ciliated secretory cells in the pars distalis of the mouse hypophysis. J. Ultrastruct. Res. 5, 453-467. BERNHARD, W. & HARVEN, E. de (i960). L'ultrastructure du centriole et d'autres elements de l'appareil achromatique. In Vierter internationaler Kongress fur Elektroncnmikroskopie, vol. 2 (ed. W. Bargmann, G. Mollenstedt, H. Niehrs, D. Peters, E. Ruska & C. Wolpers), pp. 217- 227. Berlin: Springer. BESSIS, M., BRETON-GORIUS, J. & THIERY, J. P. (1958). Centriole, corps de Golgi et astre des leucocytes, fitude au microscope 61ectronique. Revue Hemat. 13, 363-386. BRADBURY, P. & PITELKA, D. R. (1965). Observations on kinetosome formation in an apostome ciliate. J. Microscopie 4, 805-810. BRETON-GORIUS, J. & STRALIN, H. (1967). Formation de cils rudimentaires dans les cellules sanguines primitives du sac vitellin d' embryons de rat et de poulet. Nouv. Revue fr. Hemat. 7. 79-94- CIRELLI, E. (1966). Elektronenmikroskopische Analyze der pra- und postnatalen Differen- zierung des Epithels der oberen Luftwege der Ratte. Z. mikrosk.-anat. Forsch. 74, 132-178. CLEVELAND, L. R. (1957). Types and life cycles of centrioles of flagellates. J. Protozool. 4, 230-241. DAHL, H. A. (1963). Fine structure of cilia in rat cerebral cortex. Z. Zellforsch. mikrosk. Anat. 60, 369-386. DEANE, H. W. & WURZELMANN, S. (1965). Electron microscopic observations on the postnatal differentiation of the seminal vesicle epithelium of the laboratory mouse. Am. J. Anat. 117, 9I-I34- DE ROBERTIS, E. (1956). Morphogenesis of the retinal rods. jf. biophys. biochem. Cytol. (Suppl. 2). pp. 209-218. DINGLE, A. D. & FULTON, C. (1966). Development of the flagellar apparatus of Naegleria. J. Cell Biol. 31, 43-54- DIRKSEN, E. R. & CROCKER, T. T. (1966). Centriole replication in differentiating ciliated cells of mammalian respiratory epithelium. An electron microscopic study. J. Microscopie 5, 629- 644. 228 S. P. Sorokin FAURE-FREMIET, E., ROUILLEH, C. & GAUCHERY, M. (1956). La structure fine des cilies. Presentation de micrographies electroniques. Bull. Soc. sool.fr. 81, 168-170. FAWCETT, D. W. (1961). Cilia and fiagella. In The Cell, vol. 2 (ed. J. Brachet & A. E. Mirsky), pp. 217-297. New York and London: Academic Press. FRASCA, J. M., AUERBACH, O., PARKS, V. R. & STOECKENIUS, W. (1967). Electron microscopic observations of bronchial epithelium. II. Filosomes. Exp. molec. Path. 7, 92-104. FRISCH, D. (1967). Fine structure of the early differentiation of ciliary basal bodies. Anat. Rec. 157, 245- GALL, J. G. (1961). Centriole replication. A study of spermatogenesis in the snail Viviparus. J. biophys. biochem. Cytol. 10, 163-193. GIBBONS, I. R. (1961). The relationship between the fine structure and direction of beat in gill cilia of a lamellibranch mollusc. .7. biophys. biochem. Cytol. 11, 179-205. GIBBONS, I. R. & GRIMSTONE, A. V. (i960). On flagellarstructur e in certain flagellates. jf. biophys. biochem. Cytol. 7, 697^716. GRASSE, P. P. (1961). La reproduction par induction du btepharoplaste et du flagelle de Trypanosoma equiperdum. (Flagell6 protomonadine.) C. r. hebd. Seanc. Acad. Sci., Paris 252, 3917-3921. GRILLO, M. & PALAY, S. L. (1963). Ciliated Schwann cells in the autonomic nervous system of the adult rat. jf. Cell Biol. 16, 430-436. GROSS, B. G. (1966). Annulate lamellae in the axillary apocrine glands of adult man. jf. Ultra- struct. Res. 14, 64-73. HENNEGUY, L. F. (1898). Sur les rapports des cils vibratiles avec les . Archs Anat. microsc. Morph. exp. 1, 481-496. HOFFMANN, E. J. (1965). The nucleic acids of basal bodies isolated from Tetrahymena pyriformis. J. Cell Biol. 25, 217-228. KARNOVSKY, M. J. (1965). A formaldehyde-glutaraldehyde fixative of high osmolality for use in electron microscopy, jf. Cell Biol. 27, 137A-138A. KESSEL, R. G. (1965). Intranuclear and cytoplasmic annulate lamellae in tunicate oocytes. jf. Cell Biol. 24, 471-487. LATTA, H., MAUNSBACH, A. B. & MADDEN, S. C. (1961). Cilia in different segments of the rat nephron. jf. biophys. biochem. Cytol. 11, 248-252. LENHOSSEK, M. von (1898). Uber Flimmerzellen. Verh. anat. Ges., Kiel 12, 106-128. LWOFF, A. (1950). Problems of Morphogenesis in Ciliates. New York: Wiley and Sons. MIZUKAMI, I. & GALL, J. (1966). Centriole replication. II. Sperm formation in the fern, Marsilea, and the cycad, Zamia.J. Cell Biol. 29, 97-1 n. MOTTA, P. (1965). Sulla presenza al microscopio elettronico di ciglia e centrioli in cellule della granulosa ovarica di Lepus cuniculus (Linn.). Boll. soc. ital. Biol. sper. 41, 31-33. PALADE, G. E. (1955). Studies on the . II. Simple disposition in cells in situ. jf. biophys. biochem. Cytol. 1, 567-582. PHILLIPS, D. M. (1966). Observations on spermiogenesis in the fungus gnat Sciara coprophila. jf. Cell Biol. 30, 477-497. PHILLIPS, D. M. (1967). Giant centriole formation in Sciara. jf. Cell Biol. 33, 73-92. RANDALL, J. T. & DISBREY, C. (1965). Evidence for the presence of DNA at basal body sites in Tetrahymena pyriformis. Proc. R. Soc. B 162, 473-491. RANDALL, J. T., WATSON, M. R., SILVESTER, N. R., ALEXANDER, J. B. & HOPKINS, J. M. (1963). Studies of cilia, basal bodies and some related organelles. II. Genesis and macromolecular properties. Proc. Linn. Soc. Lond. 174, 37-40. RENAUD, F. L. & SWIFT, H. (1964). The development of basal bodies and flagella in Allomyces arbusculus. jf. Cell Biol. 23, 339-354. RENYI, G. (1924). Untersuchungen ilber Flimmerzellen. Z. Anat. EntwGesch. 73, 338-357. RHODIN, J. & DALHAMN, T. (1956). Electron microscopy of the tracheal ciliated mucosa in rat. Z. Zellforsch. mikrosk. Anat. 44, 345-412. ROSENBAUM, J. L. & CHILD, F. M. (1964). Incorporation of tritiated precursors into regenerating flagella of two flagellates, jf. Protozool. 11 (Suppl.), 22. ROTH, L. E. & SHIGENAKA, Y. (1964). The structure and formation of cilia and filaments in rumen protozoa, jf. Cell Biol. 20, 249-270. SCHUSTER, F. L. (1963). An electron microscope study of the amoebo-flagellate, Naegleria gruberi (Schardinger). I. The amoeboid and flagellate stages. .7. Protozool. 10, 297-312. Centriole formation and ciliogenesis 229 SCHUSTER, F. L. (1964). Ciliated fibroblasts from a human brain tumor. Anat. Rec. 150, 417-421. SEAMAN, G. R. (i960). Large-scale isolation of kinetosomes from the ciliated protozoan Tetrahymena pyriformis. Expl Cell Res. 21, 292-302. SEAMAN, G. R. (1962). Protein synthesis by kinetosomes isolated from the protozoan Tetra- hymena. Biochim. biophys. Ada 55, 889-899. SHARP, L. W. (1914). Spermatogenesis in Marsilia. Bot. Gaz. 58, 419-431. SLEIGH, M. A. (1962). The of Cilia and Flagella. Oxford: Pergamon. SOROKIN, S. P. (1961). A study of development in organ cultures of mammalian lungs. Devi Biol. 3, 60-83. SOROKIN, S. P. (1962). Centrioles and the formation of rudimentary cilia by fibroblasts and smooth muscle cells. J. Cell Biol. 15, 363-377. SOROKIN, S. P. (1965). Recent work on developing lungs. In Organogenesis (ed. R. L. De Haan & H. Ursprung), pp. 467-491. New York: Holt, Rinehart and Winston. SOROKIN, S. P. & ADELSTEIN, S. J. (1967). Failure of 1100 rads of X-radiation to affect cilio- genesis and centriolar formation in cultured rat lungs. Radiat. Res. 31, 748-759. SOROKIN, S. P., PADYKULA, H. A. & HERMAN, E. (1959). Comparative histochemical patterns in developing mammalian lungs. Devi Biol. 1, 125-151. STEINMAN, R. (1968). Fine structure of ciliogenesis in Xenopus laevis. Am.jf. Anat. 122 (in the Press). STOCKINGER, L. & CIRELLI, E. (1965). Eine bisher unbekannte Art der Zentriolenvermehrung. Z. Zellforsch. mikrosk. Anat. 68, 733-740. STUBBLEFIELD, E. & BRINKLEY, B. R. (1966). Cilia formation in Chinese hamster fibroblasts in vitro as a response to Colcemid treatment. J. Cell Biol. 30, 645-652. SWIFT, H. (1956). The fine structure of annulate lamellae. J. biophys. biochem. Cytol. (Suppl. 2), PP- 415-418. SZOLLOSI, D. (1964). The structure and function of centrioles and their satellites in the jellyfish Phialidium gregarium. J. Cell Biol. 21, 465-479. TAMM, S. L. (1967). Flagellar development in the protozoan Peranema trichophorum. J. exp. Zool. 164, 163-186. TOKUYASU, K. & YAMADA, E. (1959). The fine structure of the retina studied with the electron microscope. IV. Morphogenesis of outer segments of retinal rods. J. biophys. biochem. Cytol. 6, 225-230. VICKERMAN, K. (1962). Cited in Sleigh, M. A. (1962). WENT, H. A. (1966). The behavior of centrioles and the structure and formation of the achro- matic figure. In Protoplasmatologia, vol. 6G1 (ed. M. Alfert, H. Bauer, C. V. Harding, W. Sandritter & P. Sitte), pp. 1-109. Vienna and New York: Springer. WHEATLEY, D. N. (1967). Cilia and centrioles of the rat adrenal cortex. J. Anat. 101, 223-237. ZIMMERMANN, K. W. (1898). Beitrage zur Kenntniss einiger Driisen und Epithelien. Arch, mikrosk. Anat. EntwMech. 52, 552-706. (Received 21 June 1967) 230 S. P. Sorokin

Fig. 3. Two centrioles (clt c2) from a duodenal epithelial cell of a chick. An early stage in the formation of a daughter procentriole (pc) is indicated by the extension of fila- mentous material from the wall of one of the centrioles (c2). Barbiturate-buffered osmium, x 56000.

Fig. 4. The diplosome (c1( c2) from a chick duodenal cell illustrated in Fig. 5. A short but definite procentriole (pc) is made visible by the osmiophilia of its wall. It forms at right angles to the wall of the parent centriole (c2), being connected to it by an array of filaments. As they first form, procentrioles are annular structures. Compare Figs. 4 and 7, where procentrioles are viewed from positions 900 apart. Barbiturate-buffered osmium, x 56000. Fig. 5. A section through the duodenal epithelium of a newly hatched chick. The diplosome is located near the apical surface as in many epithelial cell types. Barbiturate- buffered osmium, x 12000. Fig. 6. An interphase bronchial epithelial cell from a 16-day foetal rat. Two centrioles (c1; c2) and two daughters (pc) lie in the apical cytoplasm. Although nearly half grown, the procentrioles still occupy their position of origin, with their bases facing the centriolar walls and their long axes at right angles to the long axes of their parents. Phosphate-buffered osmium, x 48000. Fig. 7. A centrosomal region in an interphase fibroblast from a 7-day postnatal rat. A procentriole (pc) forms next to a centriole. Barbiturate-buffered osmium, x 105000.

Fig. 8. Two centrioles (clt c2) and a daughter (pc) in the perinuclear cytoplasm of an interphase fibroblast from the lungs of a 21-day foetal rat. The procentriole is about half grown. Phosphate-buffered osmium, x 51000. Journal of Cell Science, Vol. 3, No. 2

S. P. SOROKIN Fig. 9. Two centrioles (clt c2) from an immature epithelial cell in a 21-day foetal rat lung. One (c2) has become committed to produce a primary cilium, as indicated by the appearance of satellite arms at the sides and vesicles (v) at the distal end of the cen- triole. Barbiturate-buffered osmium. X 56000. Fig. 10. A ciliary bud (b) has appeared at the distal end of a centriole in the bronchial epithelium of a 21-day foetal rat. The bud invaginates a ciliary vesicle located at the distal end of the centriole, which bears a satellite (s). Barbiturate-buffered osmium, x 59000.

Fig. 11. Two epithelial centrioles (ct) c2) from a small branch of the bronchial tree in a 21-day foetal rat. The centrioles lie close to the cell surface. The presence of a large ciliary vesicle (cv) at the end of one centriole (c2) indicates that primary ciliogenesis is under way. Barbiturate-buffered osmium, x 56000. Fig. 12. An early phase of primary ciliogenesis in a fibroblast from an organ culture of 16-day foetal rat lung after 5 days in vitro. The ciliary vesicle either is closely proxi- mate or continuous with the extracellular space. Phosphate-buffered osmium, x 50000. Fig. 13. At the surface of a bronchial goblet cell a ciliary bud has begun to elongate into a ciliary shaft. Organ culture of a 15-day foetal rat lung after 5 days in vitro. Phosphate- buffered osmium, x 56000. Fig. 14. Cross-section of a cilium from condensing mesenchyme in a 16-day foetal rat lung. The peripheral doublet fibres are present, but the central pair is imperfectly developed. The shaft occupies a well bounded by membranes of the ciliary sheath. Formaldehyde-glutaraldehyde plus osmium tetroxide. x 48000. Fig. 15. A primary cilium emerging from an epithelial cell in the developing respiratory region of a 21-day foetal rat lung. The basal body lies at the bottom of a shallow well that engulfs part of the ciliary shaft and opens into the bronchial lumen. The basal body possesses satellites, while the centriole below does not. Barbiturate-buffered osmium, x 56000. Journal of Cell Science, Vol. 3, No. 2

4^

S. P. SOROKIN Fig. 16. Basal body of a primary cilium from the cicatricial epithelium of an organ culture of rat lung explanted on the 15th prenatal day and grown 5 days. The basal body possesses a rootlet (r) and satellites (s), as well as a ring of procentrioles (pc) round its proximal end. Compare with Fig. 17. Phosphate-buffered osmium, x 42000. Fig. 17. An adjacent section of the basal body illustrated in Fig. 16. Additional members of the procentriolar ring (pc) are seen. One of these immature centrioles is engaged in producing a third-generation procentriole (pc'). When undergoing formation about a cilium-bearing centriole, a ring of daughter centrioles is displaced downward from its usual position perpendicular to the long axis of the parent centriole. In this modified position the procentrioles none the less face a point within the basal portion of the centriole. Phosphate-buffered osmium, x 41000. Fig. 18. A section of presumptive bronchiolar epithelium in the lung of a 21-day foetal rat. A ciliated border is present in one of two adjacent cells (ci, cilia) a primary cilium (pci) in the other. Barbiturate-buffered osmium, x 27000. Fig. 19. A fully developed primary cilium extending from a mesothelial cell in the vis- ceral pleura of a 16-day foetal rat lung. The ciliary fibres have reached the tip of the shaft. Formaldehyde-glutaraldehyde plus osmium tetroxide. x 25 000. Journal of Cell Science, Vol. 3, No. 2 Fig. 20. A low columnar epithelial cell from a peripheral branch of the airway in a 15-day foetal rat lung grown in organ culture for 5 days. No cilia have yet appeared in the cell or in its neighbours. In these immature epithelial cells the centrioles (c) occupy the lateroapical cytoplasm. Among the various structures known to be pre- cursors of basal bodies, only fibrogranular aggregates are present. They are separated from the centriole by a distance of about 250 m/4. Below, an abundance of Golgi lamellae fills the supranuclear cytoplasm. Phosphate-buffered osmium, x 13000. Fig. 21. An enlargement of the pericentriolar region marked in Fig. 20. The section is understood to graze a fibrogranular area similar to that shown in Fig. 22. The fibrogran- ular aggregates consist of a 400-800 A electron-dense granular component embedded in a matrix of filaments some 30-50 A thick. The electron-dense granules themselves are composed of interwoven strands of both small (30-50 A) and larger (about 100 A) filaments. A few Golgi vesicles touch the area, x 48000. Fig. 22. In the bronchial epithelium of a 21-day foetal rat a fibrogranular aggregate exhibits a cortex of mixed filaments and granules and a medulla of electron-permeable cytoplasm. The smaller cortical granules are held fast by the matrix. The larger ones are less tightly enmeshed and tend to occupy the periphery. Phosphate-buffered osmium, x 44000. Journal of Cell Science, Vol. 3, No. 2

S. P. SOROKIN Figs. 23-28 are from various epithelial cells of 16-day foetal rat lungs grown in organ culture for 5 days. They represent both amorphous and bundled forms of the fibro- granular aggregates. Fig. 23. A relatively amorphous fibrogranular aggregate in undifferentiated bronchial epithelium. The fibrillar matrix predominates; the granules are for the most part small and inconspicuous. Golgi elements encroach upon the aggregate body (lower left). Phosphate-buffered osmium, x 41000. Fig. 24. Another amorphous fibrogranular aggregate. The granules are mixed in with the filaments of the matrix; in some granules a threadlike substructure can be discerned. Golgi vesicles (above) and annulate lamellae (below) closely border the aggregate mass on the right. Phosphate-buffered osmium, x 37000. Fig. 25. A bundle of strands that is interpreted to represent a rearrangement of material earlier present in the amorphous fibrogranular aggregates. The strands are about 200-250 m/t in diameter. They are composed of matrical filaments centrally and granules peripherally. The granules are larger than when present in the amorphous bodies. Phosphate-buffered osmium, x 28000. Fig. 26. A detail of a bundle that is penetrated by a ramus from the Golgi apparatus. The bundle is loosely compacted, possibly prior to breakup. Phosphate-buffered osmium, x 39000. Fig. 27. Detail of a fibrogranular bundle showing a nearly cross-section of one of the component strands. The filaments of the matrix are most densely packed about the centre of the strand; among them are some thicker fibrils that have the diameter of . The granules are arranged in a peripheral ring and are connected to the centre by radial extensions. Phosphate-buffered osmium, x 86000. Fig. 28. An individual strand, apparently released to the cytoplasm following sub- division of a fibrogranularbundle . Compare with Fig. 27. Phosphate-buffered osmium, x 70000. Journal of Cell Science, Vol. 3, No. 2

S. P. SOROKIN Figs. 29-34 represent a stage at which the osmiophilic granules of the fibrogranular arrays become consolidated into radial structures that are interpreted to be immature forms of the deuterosomes. They are taken from sections of bronchial epithelium in organ cultures of foetal rat lungs that were explanted on days 15 or 16 and cultivated for 5 days. Fig. 29. A view of several large osmiophilic granules located at the periphery of a strand in a loosely compacted fibrogranular bundle. In the left centre two granules are seen in side view. From them numerous filaments run inward to join the mass of matrical filaments at the centre of the strand. Phosphate-buffered osmium, x 72000. Fig. 30. A single strand of fibrogranular material. The peripheral ring of osmiophilic granules is composed of fewer elements of a larger average size than are shown in Figs. 27 and 28. Phosphate-buffered osmium, x 68000. Fig. 31. A section that grazes the surface of a small fibrogranular bundle. Several of the peripheral osmiophilic bodies have a rounded conformation. A curved line has been inserted to indicate the direction of the strand intersected. Phosphate-buffered osmium, x 40000. Fig. 32. A detail of Fig. 31. The encircled peripheral granule exhibits a filamentous substructure as well as a radial plan of organization not unlike that seen in deuterosomes and in relatively unformed procentrioles. Compare with Figs 34, 40, and 47. Phosphate- buffered osmium, x 80000. Fig. 33. Two independent fibrogranular strands. On each strand one peripheral granule (encircled) is larger than the others. Both approximate the size of a deutero- some. Phosphate-buffered osmium, x 59000. Fig. 34. In several fibrogranular strands the peripheral granules are undergoing consolidation. The solitary body adjacent (arrow) is interpreted to be a newly formed deuterosome. Phosphate-buffered osmium, x 56000. Journal of Cell Science, Vol. 3, No. 2

S. P. SOROKIN Fig. 35. A section through several immature columnar epithelial cells in the bronchial epithelium of an organ culture of foetal rat lung explanted on day 16 and grown for S days. No cilia or basal bodies are yet present. The apical cytoplasm contains fibro- granular aggregates (left arrow) and deuterosomes (right arrow). Phosphate-buffered osmium, x 11000. Journal of Cell Science, Vol. 3, No. 2

7

•) - \

S. P. SOROKIN Figs. 36-39 are sections of bronchial epithelium in organ cultures of foetal rat lung cultivated for 5 days. They illustrate major features of deuterosomes. Fig. 36. Apical cytoplasm of two non-ciliated cells. About 7 groups of centriolar precursors are shown, either solitary deuterosomes (numbers /, 2), or a cluster of procentrioles (pc) about a deuterosome (j—7). More procentrioles are included in a section the closer its plane approaches the centre of the deuterosome. The micrograph includes clusters sectioned to reveal nearly 4 (number 3), 3 (4), and 5 procentrioles (5). When a deuterosome lies amidst several developing procentrioles, its wall often appears drawn out opposite the procentrioles. Explanted on day 17. Phosphate- buffered osmium, x 24000. Fig. 37. A deuterosome sectioned off centre in a partially ciliated cell. At this level it seems smaller than a basal body (bb), but its cortex is more osmiophilic. Radial filaments form a corona about the deuterosome. Explanted on day 16. Phosphate- buffered osmium, x 70000. Fig. 38. A deuterosome sectioned through its centre, where its diameter is seen to equal that of the centriole (c) nearby. Explanted on day 16. Phosphate-buffered osmium, x 69000. Fig. 39. Three deuterosomes sectioned through their cortices. Some clear areas stand out against the densely osmiophilic background. They are approximately 100 A in diameter. Explanted on day 16. Phosphate-buffered osmium, x 71000. Journal of Cell Science, Vol. 3, No. 2

S. P. SOROKIN Figs. 40-46 reconstruct steps in the development of procentrioles. All but Fig. 43 are taken from organ cultures of foetal rat lung explanted on days 15 or 16 and grown 5 days. Fig. 43 is from the lung of a 21-day foetal rat. Fig. 40. A procentriole (pc) undergoing formation closely adjacent to a deuterosome (d) in the supranuclear cytoplasm of a non-ciliated cell. At this stage the procentriole is annular rather than cylindrical in shape. Phosphate-buffered osmium, x 65000. Fig. 41. A procentriole at a more advanced state of development than the one illustrated in Fig. 40. Its base is narrowly separated from the deuterosome. At that end the wall is cylindrical and is constructed of discernibly parallel elements. It tapers towards the distal end (arrows). Phosphate-buffered osmium, x 66000. Fig. 42. A deuterosome, its corona of filaments, and two procentrioles (pc) under- going development nearby. Phosphate-buffered osmium, x 65000. Fig. 43. Longitudinal sections of two procentrioles and a part of a third arranged symmetrically about a deuterosome that is sectioned through its cortex. The pro- centrioles are nearly cylindrical, although the upper part of their walls curve inward. Below, a neighbouring procentriole (pc) is transected near its apex, at a level approxi- mately indicated by the line. Components of its wall are sectioned obliquely, owing to the incurving of the region. Barbiturate-buffered osmium, x 65000. Fig. 44. Parts of 6 procentrioles disposed symmetrically about a deuterosome that is partly obscured by an overlying procentriole. Phosphate-buffered osmium, x 50000. Fig. 45. A deuterosome and two well-developed procentrioles. Compared with the less mature examples in Figs. 41 and 43, the procentrioles here are taller, and the structure of their wall is more clearly denned. Phosphate-buffered osmium, x 49000. Fig. 46. Bronchial epithelial cells that contain various precursors of basal bodies. Among them are fibrogranular bundles (fgb), consolidating granules (g), deuterosomes surrounded by procentrioles (pc), and immature basal bodies (bb). On the left one basal body retains its association with a deuterosome. Above, another (ci) is beginning to produce a cilium. Phosphate-buffered osmium, x 15000. Journal of Cell Science, Vol. 3, No. 2

S. P. SOROKIN Fig. 47. The apical cytoplasm of a non-ciliated bronchial epithelial cell in a 21-day foetal rat lung. Several procentriolar clusters are present. At the left (encircled) a pro- centriole is sectioned transversely near its base. The wall is composed of 9 fibres ranged in a circle. Most of these appear to be doublets; but a triplet (above) and a few singlets are present, as well as intermediary forms with projecting arms. The major axes of the sectioned fibres are nearly radial to the circle. At the right (encircled) a procentriole is sectioned near its apex. Wall components are less clearly arranged than they are nearer to the base. At both levels illustrated, faintly discernible radial filaments connect points of the wall to the axis of the procentriole. Barbiturate-buffered osmium, x 57000. Fig. 48. Maturing basal bodies beneath the surface of the bronchial epithelium in a 21-day foetal rat lung. The organelles are fully grown; in longitudinal and cross- section they are indistinguishable from centrioles. Compare with Fig. 47. A few vesicles seem to be associated with the apical ends of the basal bodies (arrows). Some fibrogran- ular material (fg) is nearby. Barbiturate-buffered osmium, x 46000. Fig. 49. An alternative source of procentrioles in pulmonary epithelial cells. An apically located diplosome (cu c2) in a non-ciliated cell. Both centrioles bear procentrioles (pc) laterally. The procentrioles are positioned in a ring of 8 (pci-8) round the lower centriole (cj), and parts of 6 are included in the section. Organ culture of 16-day foetal rat lung grown for 5 days. Phosphate-buffered osmium, x 44000. Fig. 50. A centriole (c) surrounded by procentrioles (pc 1-4) in an epithelial cell from an organ culture of 15-day foetal rat lung grown 5 days. The section passed through two additional procentrioles (not shown) in the same cell. Phosphate-buffered osmium, x 53000. Journal of Cell Science, Vol. 3, No. 2

- 50.

S. P. SOROKIN Figs. 51-54 illustrate changes that occur in basal bodies prior to the emergence of cilia. Numerous small vesicles gather at the apical end (arrows) and appear to fuse to form a ciliary vesicle (cv), or an invagination from the apical surface. At the same time fibrogranular material (fg) accumulates around the base. It remains there until a ciliary rootlet is formed. Illustrations are from organ cultures of foetal rat lungs explanted on day 16 and grown 5 days. Phosphate-buffered osmium. Figs. 51-53, x 58000; Fig. 54, x 62000. Fig. 55. A basal body at the surface of a ciliating epithelial cell in the lung of a 21-day foetal rat. At the distal end (apex) of the basal body the wall becomes attached to the membrane of the flattened vesicle above (inside the arrows), (r, rootlet.) Barbiturate- buffered osmium, x 59000. Fig. 56. An emerging ciliary bud in the bronchus of a 21-day foetal rat lung. Barbitu- rate buffered osmium, x 58000. Fig. 57. The surface of a ciliating cell in an organ culture of foetal rat lung explanted on day 15 and grown 5 days. Ciliary fibres have not yet formed inside the swelling ciliary bud. The mosaic pattern seen atop the bud is likely to be a surface view of the ciliary (plasma) membrane. Compare with the microvillus in Fig. 56. Phosphate- buffered osmium, x 68000. Fig. 58. Two diminutive cilia in the bronchial epithelium of a 16-day foetal rat lung grown in organ culture 5 days. The ciliary fibres have nearly reached the tops of the shafts. Phosphate-buffered osmium, x 59000. Journal of Cell Science, Vol. 3, No. 2

S. P. SOROKIN Fig. 59. A view of the apical cytoplasm of several ciliating cells. In the cell at the lower left, two unequal cilia are lengthening. At the right, two basal bodies are budding. Organ culture of 16-day foetal rat lung grown for 5 days. Phosphate-buffered osmium, x 15000. Fig. 60. A developing cilium whose fibres (/) extend about two-thirds of the way up the shaft. Some vesicles occupy the centre of the basal body. Below, the rootlet is partially formed, but fibrogranular material (fg) persists at its tip. Organ culture of 16-day foetal rat lung grown for 5 days. Phosphate-buffered osmium, x 59000. Fig. 61. Two developing cilia bearing incompletely developed sets of internal fibres. A few vesicles occupy the shafts distal to the terminations of the ciliary fibres. Organ culture of 16-day foetal rat lungs grown for 5 days. Phosphate-buffered osmium, x 58000. Journal of Cell Science, Vol. 3, No. 2

S. P. SOROKIN Fig. 62. Cilia grown to about one-third of their full length. In all, the internal fibres have reached the tips of the ciliary shafts, and the cilia are capable of motion. Sub- sequently, the fibres elongate in pace with the shaft. Bronchial epithelium in an organ culture of 16-day foetal rat lung grown for 5 days. Phosphate-buffered osmium, x 33000. Fig. 63. A fully grown cilium in the bronchial epithelium of a 16-day rat grown in organ culture 5 days. The shaft differs little in fine structure from the shorter shafts illustrated in Fig. 62. Basal bodies possess satellites (s) and fully developed rootlets. Phosphate-buffered osmium, x 36000. Fig. 64. Cross-sections of rootlets, basal bodies, and the transition to the ciliary shaft in a recently ciliated cell from the lung of a 21-day foetal rat. Single microtubular fibres from the wall of the basal body turn inward as they continue into the rootlet; for this reason they usually appear indistinct in sections of rootlets. Satellites (s) link three adjacent triplet fibres of the basal body (lower left) to a point which serves as an attachment for cytoplasmic microtubules (not shown). In the ciliary shaft (lower right), several single tubular fibres occupy the central space above the level where the inner subfibre of the basal body's triplet disappears, on the one hand, and below the level where the central pair of the shaft originate, on the other. Barbiturate-buffered osmium, x 45000. Journal of Cell Science, Vol. 3, No. 2

S. P. SOROKIN