Summary. Sensilla trichodea of the silk moths, Antheraea pernyi and Bombyx mori, were reconstructed from serial sections after freeze substitution. The volume ...
Cell and Tissue Research
Cell Tissue Res (1984) 235:35-42
9 Springer-Verlag 1984
Pheromone receptors in Bombyx mori and Antheraea pernyi II. Morphometric analysis W. Gnatzy 2, W. Mohren 2, and R.A. Steinbrecht 2 i Zoologisches Institut, Universit/it Frankfurt/Main, Bundesrepublik Deutschland; 2 Max-Planck-Institut fiir Verhaltensphysiologie, Seewiesen, Bundesrepublik Deutschland Summary. Sensilla trichodea of the silk moths, Antheraea pernyi and Bombyx mori, were reconstructed from serial sections after freeze substitution. The volume and surface area of the different sensillar cells were calculated from the area and circumference of consecutive section profiles. A. pernyi and B. mori differ largely in the size of the sensory hair and the larger outer dendritic segments as well as in the volume of the receptor lymph within the hair, while there are only small differences regarding inner dendritic segments, receptor-cell somata, trichogen and tormogen cells and the volume of the receptor lymph below the hair base. In each sensillum the two (or three) receptor-cell somata, dendrites, and initial axonal segments differ significantly in volume and surface. The apical cell membranes of the trichogen and tormogen cells, which border the receptor-lymph cavity and which are the presumed site of electrogenie cation pumps, are deeply invaginated and enlarged by microlamellae and microvilli, so that their area is twice that of the remaining basolateral cell membrane. In contrast to mechanoreceptors, the trichogen cell is the largest auxiliary cell and has the largest apical membrane surface. The morphometric data are discussed with regard to recent electrophysiological observations. Key words: Sensillum trichodeum- Silk moth - Cell volume - Membrane surface area - Freeze substitution
The pheromone receptors of male silkmoths (Bombyx mori, Antheraea pernyi) are among the most intensively studied exteroceptors of insects. Recently, an electric circuit model has been proposed for the antennal sensilla trichodea (s. trichodea) of Antheraea pernyi (Kaissling and Thorson 1980) based partly on preliminary morphological data (W. Gnatzy, unpublished results). Following the reconstruction of s. trichodea in B. mori and A. pernyi (Steinbrecht and Gnatzy 1984) we are now able to provide quantitative morphometric data on the volume of the various cells, the extracellular receptor-lymph space, and especially on the surface area of those membrane regions that are of functional significance.
Send offprint requests to: Dr. R.A. Steinbrecht, Max-Planck-Institut fiir Verhaltensphysiologie D-8131 Seewiesen, Federal Republic of Germany
The complex spatial arrangement of the different cells forming a sensillum does not allow an exact identification of the various cell types in random sections as required for statistical stereological methods in morphometry (Weibel 1979, 1980). We have, therefore, tried a different approach, taking advantage of having complete section series through several sensilla. Image analyzers are now available allowing the direct measurement of area and circumference of section profiles by tracing their outline. We propose a procedure to use these planimetric data for estimating the volume and surface of individual, complex cell configurations. In the present study we used for specimen preparation the freeze-substitution technique that combines the advantages of cryofixation with those of conventional serial sectioning (Steinbrecht 1982). Comparison with freeze etching has proven the reliability of freeze substitution for morphometric studies (Steinbrecht 1976, 1980). The absence of swelling and shrinking artifacts is an essential prerequisite for this analysis and could not be achieved in our specimens by conventional fixation techniques. The morphometric data obtained will be useful for comparing different types of sensilla, for improving the electric circuit model, as well as for assigning electrophysiologically measured features to defined structural elements of these sensilla. A preliminary report of the present study has been given elsewhere (Steinbrecht and Gnatzy 1982).
Materials and methods Adult males of Antheraea pernyi (Gu6rin-M6n~ville) and Bombyx mori L. were used in the present study. For cryofixation and subsequent treatment of the antennae, see Steinbrecht and Gnatzy (1984). Individual sensilla were reconstructed from section series. In each species three identified s. trichodea were chosen for the measurement of cell volumes and membrane surfaces. At constant section intervals (see below), the contours of all cell profiles pertaining to these sensilla were traced with a semi-automatic image-analyzer (Leitz ASM; Kontron MOP II) to determine their area and circumference. Cell volume was calculated by summing the volume of consecutive section profiles [Appendix, Eq. (1)]. Cell surface was calculated in a step approximation [Appendix, Eq. (3)] using the circumference, thickness, and difference in area of consecutive section profiles. The accuracy and
A
e -,4~-~
~I
D ~ E
0
N TH
5~m
TO
TR
RC
|
q51Jm
, lOpm
Fig. I a, b. Antennal s. trichodeum of Bombyx mori. a Schematic illustration of cross sections through the hair shaft (pore tubules are not shown); the corresponding levels (A-C) are indicated. Outer dendritic segments (oDSA; ODSB) of the two receptor cells; Cu cuticle of hair shaft; HL hair lumen, b Three-dimensional reconstruction of a single s. trichodeum from serial cross sections; the corresponding levels of sections (D-Q) are indicated (compare with Fig. 2). Parts of cuticle (Cu), auxiliary and receptor cells are cut away to show inner details of hair lumen (HL), receptor-lymph spaces and cellular sheath around the two receptor cells. A axons; BL basal lamina; TH thecogen, TO tormogen, TR trichogen, RC receptor cell, respectively
~
TH oRC
|
D
J
Q
P
~
..
L
Fig. 2. Series of cross sections (redrawn from electron micrographs) through the s. trichodeum of B. mori shown in Fig. 1 b (compare also with Fig. 2 in Steinbrecht and Gnatzy 1984). Inner (iDS) and outer dendritic segments (oDS); the two receptor-cell somata (RCA, RC~); distally the outer receptor-lymph cavity is bordered by the tormogen cell (oRL), proximally by the trichogen cell (oRE'); iRL inner receptor-lymph space; other symbols as in Fig. 1
37 length, reconstruction is necessary for the receptor-cell somata a n d even more so for the highly complex shape of the auxiliary cells (Figs. 1, 2).
conditions of applicability of this approximation are discussed in the Appendix. In A. pernyi, the orientation of the section series was at right angles to the hair bases of the s. trichodea; the constant depth interval was 1.2 lam. I n B. mori, the orientation of the section series in the sensory epithelium was at right angles to the antennal branches, the constant depth interval was 1.2 or 2.4 ~tm. Thus, 10-17 section planes were evaluated in both species. The caliber a n d length of the dendrites and of the initial axonal segments were measured in at least 10 sensilla.
1. Receptor cells (Table 1) The dendrites in Antheraea are three times as long as in Bombyx due to the longer sensory hairs. In both species there is a thicker a n d a thinner dendrite (Figs. 1 a, 2)~; in
A. pernyi often one thick dendrite and two thin dendrites occur; sometimes only one thick dendrite is found. The thicker dendrite in A. pernyi exceeds its counterpart in B. mori considerably. Therefore, this outer dendritic segment in A. pernyi exceeds that of B. mori a b o u t five times in m e m b r a n e area a n d nearly ten times in cytoplasmatic volume. The thin dendrite, on the other hand, in A. pernyi has an even smaller diameter than that of B. mori, so t h a t surface and volume values are similar. In contrast to the outer dendritic segments, there are only m i n o r differences between Bombyx and Antheraea regarding the dimensions of the inner segments and receptor-cell bodies. In both species, the difference in caliber observed in the two outer dendrites of each sensillum persists at the level of the inner segment, the cell soma, and the first 10 ~tm of the axons. In B. mori, the diameter of the axons in the initial axonal segments is 0.6 a n d 0.4 Ixm, respectively, while in the antennal nerve the fibers are much thinner ( < 0.3 ~tm; Steinbrecht 1969). Due to the larger a n t e n n a of A. pernyi (cf. Fig. 1 a, b in Steinbrecht a n d G n a t z y 1984) most of the axons necessarily are longer in this species than in B. mori.
Results
While the volume and the surface of the cylindrical sensory processes can be easily calculated from mean diameter and
Table 1. Antennal sensilla trichodea of Antheraea pernyi (A.p.) and
Bombyx mori (B.m.) : Volume and surface of receptor cells
Receptor cell A soma" inner dendritic segment b outer dendritic segment below hair base b within hair lumen Receptor cell B soma a inner dendritic segment b outer dendritic segment below hair base b within hair lumen Axon e a b c d e
Volume (lam3)
Surface (I.tm2)
A.p.
B.m.
A.p.
B.m.
178 10
156 10
214 43
186 42
9.9 619 b, o
7.7 125 b, d
154 36
163 33
1.2 113 b, c
0.7 12 b, d
129 6
126 6.4
0.2 6 b, c
0.3 4 b, d
20-280
15-90
4.4 137 b, c
2. Auxiliary cells (Tables 2, 3) The trichogen cell (TRC) is by far the largest of all auxiliary cells, being two to three times as large as the tormogen cell (TOC) and over seven times as large as the thecogen cell (THC). The respective volumes of T R C a n d T O C are almost equal in both species. Regarding the surface area of the cell m e m b r a n e we distinguish between the apical cell surface facing the receptor-lymph cavity and the remaining bast-lateral surface. Apical and basolateral cell m e m b r a n e s are morphologically separated by a belt of septate junctions. While the T O C and T R C face different compartments of the outer receptor-lymph cavity, the T H C faces the inner receptor-lymph cavity (Figs. 1 b, 2). I n both species, due to extensive microlamellae and microvilli, the apical membrane surfaces of T O C and T R C are a b o u t two times larger than the corresponding bast-lateral surfaces. The absolute values of surface area of these cells in A. pernyi, however, are smaller than in B. mori. The T H C differs from T R C
5.1 62 b, d
1180-14,100 780M,700
From reconstruction of three identified sensilla Calculated from average length and diameter Data from K.E. Kaissling (personal communication) Data from Steinbrecht (1973, and unpublished) Minimal and maximal values depending on location of sensillum on antenna; axon diameter from cross sections of antennal nerve; no distinction possible between cell A and B
Table 2. Volume of auxiliary cells (].tm3)
Antheraea pernyi
Bombyx mori
Sensillum No.
SensillumNo.
1 Thecogen cell Trichogen cell Tormogen cell Not measured
. 892 408
2 .
. _b 347
.
3
Mean
Mean
1
2
3
. 921 378
907 378
129 959 350
128 946 _b
139 919 347
119 1021 352
b NO value due to incomplete sections
38 Table 3. Surface of auxiliary cells (].tm2)
Antheraea pernyi
Bombyx mori
SensillumNo.
Sensillum No.
1
Theocogen cell apical surface facing inner receptor-lymph space baso-lateral surface
2
63 .
3
28 .
.
22
Mean
Mean
38
29
29
18
41
971
1030
1077
807
.
1
Trichogen cell apical surface facing outer receptor-lymph space baso-lateral surface
1768
1286
1041
1365
3036
3153
2649
3324
877
440
584
634
1202
1184
1087
1334
Tormogen cell apical surface facing outer receptor-lymph space baso-lateral surface
727
835
944
835
1395
1406
1384
625
477
433
512
703
773
633
a Not measured;
b b
b NO value due to incomplete sections
Table 4. Volume of receptor-lymph spaces (lam3)
Antheraea pernyi
Bombyx mori Sensillum No.
Sensillum No. 1 Inner receptor-lymph space Outer receptor-lymph space within sensory epithelium within sensory hair a total
2 6.0
401
3 3.6
404
4.0 323
Mean
Mean
4.5
4.8
376 603 b 979
202 50 c 252
1 4.8 182
4.1 177
5.6 240
a Calculated from volume of hair lumen minus volume of outer dendritic segments b Data from K.E. Kaissling (personal communication) " Data from Steinbrecht (1973) a n d T O C in having a small apical cell m e m b r a n e with little m e m b r a n e folding.
3. Receptor-lymph cavities (Table 4) The inner receptor-lymph space is a b o u t equal in both species and is very small. The volume o f the outer receptor l y m p h in Antheraea is nearly four times larger than in Bombyx. This is mainly due to the extreme volume difference o f the hair lumina o f the two species, whereas the receptorl y m p h cavities within the sensory epithelium differ only by a factor o f two (compare also Figs. 1, 2). Discussion
1. Accuracy of measurements T a k i n g into account that (i) the magnification o f the micrographs was calibrated by a cross-grating replica, and (ii) the measurement o f area and circumference with the imageanalyzer was accurate to a b o u t 1%, the main uncertainty o f our m o r p h o m e t r i c d a t a concerns the a p p r o x i m a t i o n o f the volume and surface at relatively thick section intervals. This is especially true for the estimation o f the surface areas,
which strongly depends on the m o d u l a t i o n o f the surface contours (see Appendix). Considering the limitations o f the applied procedure as discussed in the Appendix, our d a t a for cell-surface areas should, however, be correct within the range o f _+25%, except for the receptor-cell s o m a t a (Table 1), which m a y be overestimated by up to 40%. Volume d a t a certainly are more accurate ( < _+ 10%). The d a t a o f A. pernyi possibly are somewhat less accurate than those o f B. mori due to larger freezing damage in deeper tissue regions (cf. Steinbrecht and G n a t z y 1984) and the less suitable plane o f section (cf. Methods). Three reconstructed sensilla per species are not a valid statistical sample, especially with regard to the possibility of systematic variation o f the measured d a t a according to the location o f sensilla on the antenna (Steinbrecht 1970). However, taking into consideration the current state o f electrophysiological investigations (see below), the expense of a larger sample does not seem to be justified at present. On the other hand, the variation o f d a t a among the different sensilla studied is surprisingly low. It is intriguing that the cell volumes o f corresponding cell types are almost equal in b o t h species. On the other hand, cell volumes o f different cell types correlate well with the volumes o f their nuclei (R.A. Steinbrecht, unpublished).
39
2. Receptor cells Morphometric data on the dimensions of the receptor-cell processes are important for estimating their cable properties. Implications of length and cross sections of dendrites for the sensitivity and conductance of these processes have been extensively discussed by Kaissling (1971). The fact that the two receptor cells of a given sensillum are different in size is well correlated with the different amplitude of nerve impulses consistently obtained by extracellular recordings from these sensilla. As discussed by Steinbrecht (1973, see also Wiese and Schmidt 1974, for a comparable situation with mechanoreceptors), in B. mori the large impulses triggered by bombykol probably are produced by the larger receptor cell A, hence the bombykal cell firing the smaller impulses (Kaissling et al. 1978) would be the smaller receptor cell B. Likewise, in A. pernyi the larger cell A probably is the receptor for E-6, Z-11-hexadecadienyl-l-aldehyde, and the smaller cell B the receptor for E-6, Z-11-hexadecadienyl-l-acetate, both being components of the female pheromone gland of this species (Kaissling 1979). The size differences are not restricted to the outer segments, but pertain to the other parts of cell A and B as well, including the initial axonal segments; hence, not even vague suggestions as to the location of the spikegenerating site are possible from these morphometric data alone.
3. Auxiliary cells This paper reports for the first time direct measurements of the cell-surface areas of insect sensilla, which is especially interesting with regard to the conspicuously enlarged apical cell membranes of the tormogen and trichogen cells bordering the outer receptor-lymph space. As shown in Table 3, these membranes surpass the remaining basolateral membranes by a factor of two in both species. However, any estimate of the surface enlargement produced by the microlamellae and microvilli requires definition of the course of a reference surface. Depending on whether the microlamellae are regarded as folds standing up from a smooth basis or as invaginations from a smooth outer surface, in the trichogen cell of Bombyx we arrive at different ratios of 14:1 and 47:1, respectively. However, if we take as reference a flat plane between the apical septate junctions and calculate the enlargement brought about by the microlamellae and the invaginated pouch, we obtain the enormous ratio of 380:1. According to Keil (1978), the enlargement of the apical surface of the tormogen cell of two mechanoreceptors in Musea and Calliphora is 15 and 40 times, respectively. Oschmann and Berridge (1970) estimated that similar structures provide a 250-fold increase of the apical surface of salivary gland cells in CalIiphora. However, since in none of these papers indications are given on how the values are derived, these widely diverging figures cannot be usefully compared with our data. On the other hand, little is known about the function of the auxiliary cells and the possible implications of the enlargement of the apical membranes. Although it has been suggested for a long time that these cells secrete the receptor lymph (Ernst 1969; Slifer 1970), direct evidence for the special involvement of the different auxiliary cells is still scanty (however, see Phillips and Vande Berg 1976). It should be emphasized that the receptor lymph is a complex secretion product containing mucopolysaccharides such as hyaluronic acid and/or chon-
droitin sulfate (Gnatzy and Weber 1978; W. Gnatzy, unpublished), esterases and pheromone-binding proteins (Vogt and Riddiford 1981). It has also been shown that the receptor lymph contains potassium in concentrations three to four times higher than in the haemolymph (Kfippers 1974; Kaissling and Thorson 1980; Steinbrecht and Zierold 1982; W. Gnatzy, unpublished). The possible role of electrogenic cation pumps in the function of insect sensilla was first pointed out by Thurm (1970; for further references see Steinbrecht and Gnatzy, 1984). He proposed the apical membrane of the tormogen cell as the most probable site of the cation pump, because in the mechanoreceptors which he studied only this membrane was enlarged by folds and showed the membrane particles characteristic for such pumps ("portasomes"; Harvey 1980). However, in the olfactory sensilla of Antheraea and Bombyx the trichogen cell might be more important than the tormogen cell due to its larger apical membrane area; this notion is supported by the fact that regular arrays of portasomes in Bombyx were only found on the apical membrane of the trichogen, but not of the tormogen cell (Steinbrecht and Gnatzy 1984). Nevertheless, it should be kept in mind that there is still no experimental evidence of a quantitatively or qualitatively different secretory activity of the two auxiliary cells. At present, due to the complex cellular organization of sensilla, it is very difficult to assign electrophysiologically recorded features to defined sensillar structures. However, recent experiments under voltage and current clamp conditions allow the determination of resistors and capacitors which play an important role in the passive electrical network of these sensilla (de Kramer 1982 and unpublished): For example, a capacitance of -,~30 pF was observed in s. trichodea of Antheraea polyphemus. This capacitance can only be explained with a membrane area of 3000 pm 2, if we assume the usual specific capacitance of biological membranes (1 ~tF/cm 2 Cole 1972). This postulated large-area membrane corresponds best with the summated surface area of the apical cell membrane of the trichogen and tormogen cells (Table 3). Furthermore, a resistance of 300 Ms (at 10 ~ C) was found in parallel with the above-mentioned capacitance and was attributed to the same apical cell membranes on morphological as well as electrophysiological grounds. The resistance of the basolateral membranes of these auxiliary cells was estimated to be only 10-40 Mg2. Using our morphometric data we can now calculate the specific membrane resistance and arrive at values in the order of 104Qcm 2 for the apical membranes of the trichogen and tormogen cells, while the basolateral membranes of the same cells appear rather leaky (lOZ~cm2). These examples may suffice to illustrate the usefulness of quantitative fine-structural data in combination with adequate physiological experiments on insect sensilla. Further progress in bridging the gap between structure and function can be expected.
Appendix Calculation of cell volume and surface from the area and circumference of serial section profiles 1. Volume. If a hemisphere is sectioned into n slices of equal thickness t, the section planes being parallel to the equator
40 Z
Z
Z
a b c Fig. 3a, b. Models illustrating the calculation of surface areas from serial section profiles. a, b Hemisphere sectioned into planes perpendicular to the z-axis, demonstrating the cylinder and extended cylinder approximation of the surface area in a and the cone approximation in b (see text); the indices of the section planes are shown; C W cylinder wall, H S horizontal step, t section thickness, c Hemisphere with modulation of the surface in the x, y-plane (as, e.g., on a lemon squeezer). The additional z-modulation provided by Eq. (6) is not drawn (see text). Only two section planes (i; i-1) are shown. With increasing x, y-modulation, the area of the cylinder walls (CW) grows considerably, while the area of the horizontal steps (HS) is less affected; thus, the cylinder approximations become more accurate (Fig. 3 a), its volume can be easily a p p r o x i m a t e d as n
(1)
V= ~A~.t
sphere is obtained, if we calculate first the convex areas Ji o f the truncated cones (Fig. 3 b) created by two consecutive section profiles (i-1, i):
i=l
Ji = re" (ri_, + r i ) ' r 2 + (ri_1 -- rl) 2
with A~ being the areas o f the section profiles ( i = 1, n). W i t h decreasing t, V will assymptotically a p p r o a c h the real value. This a p p r o x i m a t i o n is also valid for irregular bodies (see step integration procedures in textbooks o f geometry). As long as t is sufficiently small, the shape o f the profiles yielding Ai m a y be quite irregular without seriously afflicting the accuracy o f V. 2. Surface. In contrast to the volume approximation, where surface irregularities deviate in a positive and negative direction and, therefore, tend to cancel each other, the problem is different with the a p p r o x i m a t i o n o f surface areas, because here any surface irregularity will always increase the area. F o r estimating the convex surface of the hemisphere, we m a y first sum up all the cylinder walls o f the step model (Fig. 3 a) :
$1 = ~ C~. t
(2)
(4)
with rl and q-1 being the radii o f the section profiles in these planes, and then sum up the surface as n
s3 = Z J,-
(5)
i=1
F o r the hemisphere, this " c o n e a p p r o x i m a t i o n " is very accurate: With only 10 sections the error o f S~ is < 1%. W e now tested the accuracy o f the cylinder, extended cylinder, and cone a p p r o x i m a t i o n with a m o r e complex model where we could induce defined surface modulations and calculate on a computer the real surface as well as the surface area resulting from the three mentioned approximations. The test model somewhat resembles a lemon squeezer (Fig. 3 c) and is defined in cylinder coordinates: x = R. cos~o, y = R. sin~0, z = z with R as the radius and ~0 as the angle o f the p o l a r coordinate system in the x, y-plane. To induce surface modulations we define:
i=1
with C i being the circumferences o f the section profiles (i = 1, n), and t the section thickness. This " c y l i n d e r a p p r o x i m a t i o n " obviously will produce values of S 1 smaller than the real surface, even with infinitely thin sections. If, in a d d i t i o n to the cylinder walls, we consider the horizontal areas o f the steps, our a p p r o x i m a t i o n will take the form: $2=
s C,. t + s i=1
[AAI,
(3)
i=1
or
S 2 = C i. t + A
....
(3a)
since the s u m m a t e d consecutive area differences A A of all n slices equal the a r e a A m a x o f the equatorial cross section o f the hemisphere. This " e x t e n d e d cylinder a p p r o x i m a t i o n " will usually p r o d u c e values o f $2 larger than the real surface. The best a p p r o x i m a t i o n o f the surface area o f a hemi-
R = R o 9cos (z2) "[1 + 2-sin z (2 n z). cos (m~0)] and
0
