ABSTRACT
Neural connections between the eye and optic tectum in Xenopus laevis were anatomically traced by observing the tectal location of Wallerian degeneration after discrete retinal lesion. These retinotectal connections were mapped in post metamorphic frogs and tadpoles at stage 51, the stage at which retinal axons have grown into about the rostral one-half of the tectum. The course of the experimental degeneration was the same in frogs and tadpoles, but degeneration proceeded faster in the younger animals.
In the frogs, connections were ordered, with nasal retina mapping to the caudal part of the tectum and temporal retina mapping to the rostral tectum. In the tadpoles, within the innervated area at the rostral tectum, the retino-tectal connections were generally organized as in the adults, with the temporal retina mapping to the rostral part of the innervated tectum and nasal retina mapping primarily to the caudal part. But a portion of the nasal fibers consistently mapped to the far rostral tectum as well. Electron microscopic observations showed degenerating synaptic terminals at both rostral and caudal portions of the innervated tectum after lesion of just the nasal retina. Degeneration was not seen in control animals. These results indicate that some fibers (particularly from nasal retina) may shift their terminals caudally on the tectum to match tectal growth and produce the adult pattern of connections. If there is such connection readjustment, the ‘aberrant’ connections from nasal retina in tadpoles may be an indication of this process.
INTRODUCTION
An explanation of the mechanisms by which nerve cells make and maintain connections is fundamental to an understanding of the organization of the nervous system. Several hypotheses have been proposed to explain the ability of some neurons and sets of neurons to make specific contracts. The mechanisms proposed include mechanical guidance (Weiss, 1955), differential timing of axon growth (Jacobson, 1970) and chemospecific affinity of complementary neurons (Sperry, 1963, 1965).
Of all the systems in which there is evidence for specificity in the formation of connections between nerve cells, the visual system of lower vertebrates is one of the most thoroughly studied. One reason is that there is a clear topographical relationship between the position of retinal neurons and their connections in the optic lobe, or tectum, making analysis of experimental perturbations of the system relatively simple. Although embryologically part of the central nervous system, the retina is uniquely accessible to experimental manipulation, and in animals such as the amphibians Rana and Xenopus, the system can be manipulated even in the earliest embryonic stages, allowing the study of developmental processes.
Several recent studies have been made of the retino-tectal connections as they form during normal development. Connections of the entire eye have been mapped histologically during development in Xenopus laevis (Scott, 1974; Longley, 1974; Scott & Lâzàr, 1976; Jacobson, 1977); and Rana (Currie & Cowan, 1975), and the visuo-tectal projection of Xenopus tadpoles has been mapped electrophysiologically (Gaze, Chung & Keating, 1972; Gaze, Keating & Chung, 1974; Chung, Keating & Bliss, 1974). It is evident from these studies that during development in the frog, axons of the retinal ganglion cells from all parts of the existing retina grow down the optic nerve to reach the antero-lateral tectum at about stage 40. The retinal axons from the growing eye spread pos-teromedially to cover the entire tectum by about stage 64, shortly before metamorphosis at stage 66. Only electrophysiological methods have thus far been used to map connections of parts of the retina (as opposed to the entire eye) during development. Some of the potential difficulties in interpretation of electrophysiological data have been pointed out (Hunt & Jacobson, 1974). In the work described here, the location of degeneration in the tectum is determined histologically after lesion of part of the retina. The retino-tectal connections of larval and adult Xenopus are examined. Although there are inherent limitations in both histological and electrophysiological methods, the retino-tectal relationships obtained using these methods are complementary in elucidating the pattern of connections which are actually made as the axons of the retinal ganglion cells innervate the optic tectum.
MATERIALS AND METHODS
Animals and operations
Adult Xenopus laevis for breeding were obtained from Jay E. Cook, importer, Cockeysville, Md., and from MogulEd, Oshkosh, Wis. Methods of staging animals, obtaining eggs and rearing larvae were as described in Nieuwkoop & Faber (1956). Animals were kept at 18 °C except as otherwise indicated. Juvenile frogs (stage 66) were at least 1 month post-metamorphosis and 2-33-5 cm in length (nose-rump).
Retino-tectal connections were mapped anatomically by observing the location of degeneration following lesion of a portion of the retina. The events and time-course of this Wallerian (anterograde, orthograde) degeneration were followed with light and electron microscopy in the stages to be mapped, and the optimum degeneration times for mapping after retinal lesion were determined. The stages used were tadpoles at the early hind limb-bud stage 51, and postmetamorphic juvenile frogs at stage 66.
For purposes of the degeneration study, the entire eye was removed after cutting the optic nerve just behind the eye. To make focal lesions for mapping, the skin near the eye was slit to expose the sclera and the eye was turned in its orbit. Using fine forceps and iridectomy scissors, and cutting through all layers of the eye, a portion of the eye was removed central to the iris. The eye was then released and allowed to return to its normal position within the orbit. In several animals, both eyes were lesioned and the two tecta could be directly compared in the same animal. The anesthetic for operations was MS-222 (Sandoz) at a dilution of 1 /3000 for larvae and 1 /2000 for juveniles. After eye removal or lesion, the animals were allowed to recover from the operation and were kept at room temperature (24 °C) for a specified period of degeneration ; they were again anesthetized, and the brains were removed and fixed in 1 % (w/v) OsO4 in amphibian Ringer’s (Rugh, 1962). Tadpole brains were fixed 90 min at 0–4 °C; postmetamorphic brains were fixed for 30 min, then split at the midline and fixed 90 min longer. They were then embedded in Epon (Luft, 1961 ; Pease, 1964) and cut in cross-section on an ultramicrotome.
Microscopy
Thin sections for electron microscopy were stained with uranyl acetate (5 % in H2O) and with lead citrate (Reynolds, 1963). Staining procedures were as described in Pease (1964) except for mass staining of grids, which utilized the method described by Sjoström, Thornell & Hellstrom (1973). For light microscopy, the same plastic embedded tissue was used as for electron microscopy, but sections were cut 1 μm thick.
Mapping procedure
Serial 1 μm cross-sections were taken through the optic tecta and examined with the phase optics of a Zeiss microscope at 200 ×.
An eyepiece micrometer with a square grid containing 20 × 20 squares was used to construct sketches of the location of black dots indicating degeneration in relation to the tectal outline (Fig. 1B). The grid squares were 16–3μm on a side. In scoring degeneration (seen as black dots using phase microscopy at 200 × magnification) a criterion of at least three dots per small square for tadpoles and four dots per square for adults was set. These criteria were set at a level low enough to include most areas of known degeneration (i.e. innervated tectum after whole eye removal), and large enough to exclude most areas of control tecta and thus avoid spurious counts from blood vessels, etc. Before scoring for degeneration, slides were shuffled randomly and assigned code numbers in a single-blind manner to eliminate any possible bias in scoring.
Two-dimensional maps of degeneration were constructed from the sketches of the tectal cross-sections (Fig. 1).
Procedure for making two-dimensional maps of tectal degeneration from tectal cross-sections. Sketches of the location of spots of degeneration were made (B) from tectal cross-sections (A). The x s marking degeneration were located in two dimensions; latero-medial and rostro-caudal. The lateral dimension is expressed as degrees of arc along the curved surface of the tectum in cross-section, with 0” at the midline (B). For these measurements, the origin of the protractor is set on the midline at a distance 2 d from the tectal surface, where d is the distance along the midline from the tectal surface to the point where the lateral extent of the ventricle is greatest (D). The longitudinal dimension is the position of the section along the length of the tectum (C), expressed as percentage of the total tectal distance with 0 % at the rostral end. Reference points for location along the longitudinal axis are, in frogs, the point of greatest width of the ventricle (at 70 % of the total tectal distance, (D), and in tadpoles, at the point where the ventricle becomes three-lobed (at 90 % of the total distance; E). (F) shows a completed map of areas of tectal degeneration in which the degeneration areas of (B) form a horizontal row of small rectangles to the left of the arrow. Each rectangle on the map represents one or more of the small squares of the eyepiece grid (A) which contained degeneration sufficient to meet the criterion (i.e. three black spots/square for this stage-51 tadpole).
Procedure for making two-dimensional maps of tectal degeneration from tectal cross-sections. Sketches of the location of spots of degeneration were made (B) from tectal cross-sections (A). The x s marking degeneration were located in two dimensions; latero-medial and rostro-caudal. The lateral dimension is expressed as degrees of arc along the curved surface of the tectum in cross-section, with 0” at the midline (B). For these measurements, the origin of the protractor is set on the midline at a distance 2 d from the tectal surface, where d is the distance along the midline from the tectal surface to the point where the lateral extent of the ventricle is greatest (D). The longitudinal dimension is the position of the section along the length of the tectum (C), expressed as percentage of the total tectal distance with 0 % at the rostral end. Reference points for location along the longitudinal axis are, in frogs, the point of greatest width of the ventricle (at 70 % of the total tectal distance, (D), and in tadpoles, at the point where the ventricle becomes three-lobed (at 90 % of the total distance; E). (F) shows a completed map of areas of tectal degeneration in which the degeneration areas of (B) form a horizontal row of small rectangles to the left of the arrow. Each rectangle on the map represents one or more of the small squares of the eyepiece grid (A) which contained degeneration sufficient to meet the criterion (i.e. three black spots/square for this stage-51 tadpole).
Preliminary experiments indicated that in tadpoles, eye lesions which were smaller than about 0-6 mm in diameter gave degeneration that was insufficient for mapping. For that reason, the tadpoles were fixed and embedded in paraffin after the brains had been removed. Serial frontal sections (10 μm) were taken through the anterior end of the animal to include the eyes. The sections were stained with Ehrlich’s acid alum hematoxylin (Gray, 1954). Eye lesion diameter was determined from these sections. Tadpoles were used for mapping in which the lesion was confined to the appropriate quadrant with no apparent injury to the rest of the eye and in which the lesion was between 0-6 and 0-95 mm in diameter. The larger size of the juvenile eyes made the lesion operation easier, and all of the juveniles were used for mapping.
RESULTS
Selection of degeneration times
The optimum times for mapping degeneration in the tectum following retinal lesion were determined by following the time-course of degeneration with light and electron microscopy. The course of degeneration is very similar in stage-51 tadpoles and metamorphosed frogs, but proceeds much more quickly in the tadpoles. The first signs of degeneration seen ultrastructurally are osmiophilic bodies the size of mitochondria in synaptic terminals at 4 and 8 h in tadpoles and 1 day in juveniles (Fig. 2 A). The debris appears to arise from clumped synaptic vesicles and mitochondria (Fig. 2B, C). By 16 h in tadpoles and 4 days in juvenile frogs, ependymal glia begin to envelop the terminal debris (Fig. 3 A, B), and this ependymal debris is prevalent at 20 and 24 h in tadpoles; 4 and 5 days in frogs. Numerous small black dots in the optic layer of the tectum can be seen with light microscopy beginning at about 20 h after nerve cut in the tadpoles and 4 days in the juvenile frogs. Their maximum number is reached at about 24 h and 5 days (Fig. 4). Comparison of neighboring thin and thick sections with light and electron microscopy demonstrated correspondence of the ependymal debris with black dots seen with light microscopy (Fig. 5 A, B). The times following eye lesion which were selected for mapping the tectal debris of degeneration were 1 day for tadpoles and 5 days for juvenile frogs.
(A) Electron micrograph from a stage-51 tadpole tectum 8 h after section of the optic nerve. Two nerve processes with synapses (S) and synaptic vesicles (SV) also contain osmiophilic debris (DB) bounded by a membrane. Bar represents 1 μm.
(B) Clumped vesicles bounded by a membrane (CSV) 8 h after nerve section. The clump contains both spherical and dense-core vesicles. Stage-51 tadpole. (C) An abnormal mitochondrion (M) near a synaptic cleft with vesicles (SV) 8 h after optic nerve section. Note darkening of the cristae. Stage-51 tadpole.
(A) Electron micrograph from a stage-51 tadpole tectum 8 h after section of the optic nerve. Two nerve processes with synapses (S) and synaptic vesicles (SV) also contain osmiophilic debris (DB) bounded by a membrane. Bar represents 1 μm.
(B) Clumped vesicles bounded by a membrane (CSV) 8 h after nerve section. The clump contains both spherical and dense-core vesicles. Stage-51 tadpole. (C) An abnormal mitochondrion (M) near a synaptic cleft with vesicles (SV) 8 h after optic nerve section. Note darkening of the cristae. Stage-51 tadpole.
(A) An ependymal glial cell containing debris (ED). Ultrastructural view at the tectal surface (S) of a stage-51 tadpole 24 h after optic nerve section. (B) Ependymal glial debris (ED) in the juvenile tectum after 4 days of degeneration. The debris is patchy, dark, and variable in size.
(A) Phase micrograph of a tadpole tectum in transverse section at stage 51 one day after section of the optic nerve. The tectal surface is at the top. Numerous small black dots (D) are located outside of cell bodies in the outer layer of the tectum. Nucleoli (not counted in maps) can be distinguished from these spots by their location within cell bodies (CB). (B) Transverse section from the tectum of a control stage-51 tadpole. Note the relative absence of black dots. (C) Stage-51 tadpole tectum in transverse section 3 days after optic nerve section. Note the change in distribution and the increased size and density of the black dots.
(A) Phase micrograph of a tadpole tectum in transverse section at stage 51 one day after section of the optic nerve. The tectal surface is at the top. Numerous small black dots (D) are located outside of cell bodies in the outer layer of the tectum. Nucleoli (not counted in maps) can be distinguished from these spots by their location within cell bodies (CB). (B) Transverse section from the tectum of a control stage-51 tadpole. Note the relative absence of black dots. (C) Stage-51 tadpole tectum in transverse section 3 days after optic nerve section. Note the change in distribution and the increased size and density of the black dots.
(A, B) Subadjacent phase and electron micrographs of the frog tectum after 4 days of degeneration. A blood vessel (BV) and the tectal surface (S) serve as land-marks. The osmiophilic debris in B is contained in electron-lucent cytoplasm characteristic of ependymal glia. (Shown at greater magnification in Fig. 3B.) Some of the phase-dark dots and ultrastructural debris which appear to be in correspondence are numbered.
(A, B) Subadjacent phase and electron micrographs of the frog tectum after 4 days of degeneration. A blood vessel (BV) and the tectal surface (S) serve as land-marks. The osmiophilic debris in B is contained in electron-lucent cytoplasm characteristic of ependymal glia. (Shown at greater magnification in Fig. 3B.) Some of the phase-dark dots and ultrastructural debris which appear to be in correspondence are numbered.
Postmetamorphic juveniles
Removing the entire retina in juveniles (eight animals) results in the appearance of degeneration over the entire contralateral tectum (Figs. 6 and 7). There is some variation between individuals in density of degeneration, which is artificially enhanced in the maps because of the use of a density criterion for scoring degeneration. The one unoperated control animal showed no degeneration (Fig. 8).
Map of tectal degeneration following retinal lesion in a juvenile frog. Direct retinal connections end in the contralateral tectum. The type of retinal lesion is indicated by a blacked-in area; in this case, the right eye was removed completely, and a lesion was made within the temporal quadrant of the left eye. The location of the lesion-induced degeneration is indicated by the rectangles on the tectum. The maps were constructed as described in Fig. 1 and the text.
Map of tectal degeneration following retinal lesion in a juvenile frog. Direct retinal connections end in the contralateral tectum. The type of retinal lesion is indicated by a blacked-in area; in this case, the right eye was removed completely, and a lesion was made within the temporal quadrant of the left eye. The location of the lesion-induced degeneration is indicated by the rectangles on the tectum. The maps were constructed as described in Fig. 1 and the text.
Map of retino-tectal connections in a frog. No lesion was made in the right eye. Conventions as in Fig. 6.
Map of retino-tectal connections in a frog. No lesion was made in the right eye. Conventions as in Fig. 6.
The temporal retina maps to the rostral tectum in all of the seven animals examined. The nasal retina maps to the caudal tectum in six of the eight frogs, with a small scattering of degeneration also appearing at the rostral portion of the tectum in three of the six. Two frogs showed very little degeneration, perhaps because the eye lesion was too small. Figures 6–8 show representative retinotectal projections of nasal and temporal retina in juvenile frogs.
The mapping results with temporal lesions and entire retinal removal are consistent with previously published electrophysiological studies on adult Xenopus (Gaze, 1958; Gaze, Jacobson & Szekeley 1963; Jacobson, 1968) and histological studies of Triturus and Rana (Stroër, 1939; Lázár, 1971 ; Scalia & Fite, 1974). The previously published data also suggested that the nasal retina projects solely to the caudal tectum, but in the present experiments, a portion of the nasal fibers projected to the rostral tectum, though most did in fact project caudally. The apparent degeneration at the rostral tectum may be simply due to spurious counts, or this difference of projection may be a function of the maturity of the animals. The frogs used in the present experiments were 2·3–3·5 cm in length and from 1 to about 5 months postmetamorphosis, while those used in other studies have generally been much larger and older.
Tadpoles
After removal of the entire eye in six tadpoles, the ensuing degeneration appeared at the rostral part of the tectum (Fig. 9). The caudal extent of the degeneration varied with the individual animals from about 45 % to 85 % of the total tectal length. Degeneration was more extensive at the rostral part of the innervated tectum. Areas of the tectum which were thin and not yet well laminated (the far caudal tectum) showed practically no degeneration (Fig. 9). It was consistently found that unoperated control animals showed very little or no tectal degeneration. Of the four unoperated controls, two showed no degeneration and two scored for degeneration as shown in Fig. 10. Degeneration in unoperated animals has been reported in stage-45 Xenopus, decreasing by stage 47 (Scott, 1974). Periods of degeneration in developing nerve centres are known to be a part of normal development in several systems (Jacobson, 1970; Cowan, 1973; Rogers & Cowan, 1973), but by stage 51 in Xenopus, degeneration is virtually nil in control maps.
Tectal degeneration following retinal lesion in a stage-51 tadpole. Con-ventions as in Fig. 6.
Tectal degeneration following retinal lesion in a stage-51 tadpole. Con-ventions as in Fig. 6.
An unlesioned control tadpole of stage 51. Cross-hatched area indicates an eye and tectum which were not mapped. Conventions as in Fig. 6.
An unlesioned control tadpole of stage 51. Cross-hatched area indicates an eye and tectum which were not mapped. Conventions as in Fig. 6.
Of the eight animals in which a lesion of the temporal retina was made, six mapped completely, or nearly so, to the far rostral tectum (Figs. 11, 12). One showed no degeneration, and one showed degeneration at both rostral and caudal parts of the tectum (Fig. 13). The temporal retina of the tadpoles mapped closer to the rostral pole of the tectum than it did in juveniles with comparable retinal lesions. In contrast to the pattern of tectal degeneration that is evident after lesion of the temporal retina, and despite similar eye lesions, degeneration in the tectum after lesion of the nasal retina in tadpoles was widespread. Of the nine animals in this category, one had debris localized almost entirely at the caudal part of the innervated tectum (Fig. 1F), one showed virtually no degeneration, and seven showed extensive degeneration at both rostral and caudal parts of the innervated tectum. In those seven animals, the greater amount of degeneration was at the caudal part (Figs. 9, 11, 12, 13).
Map of retino-tectal connections at stage 51. Conventions as in Fig. 6.
Map of retino-tectal connections at stage 51. Conventions as in Fig. 6.
Map of retino-tectal connections at stage 51. The caudal tectal degeneration following temporal retinal lesion is unusual. Conventions as in Fig. 6.
Map of retino-tectal connections at stage 51. The caudal tectal degeneration following temporal retinal lesion is unusual. Conventions as in Fig. 6.
Because debris of degeneration at the rostral tectum after nasal eye lesion could conceivably be due to axonal rather than terminal degeneration (the nasal retinal ganglion cell axons cross over the rostral tectum to terminate at the caudal tectum in adults), a study of the ultrastructural location of the degeneration was undertaken. Nasal retinal lesions were made in both eyes of several tadpoles. The lesion of the left eye was made 24 h before fixation (the time used for mapping the black dots with light microscopy), and that of the right, 8 h before (the time at which debris is observed ultrastructurally in synaptic terminals). This procedure produced animals in which early and late stages of degeneration were present on different sides of the same brain. The brains were fixed, and the eyes were fixed and sectioned as described earlier. Only animals in which both eyes met the lesion criteria mentioned previously were examined further. Thick sections of the optic tectum were examined with the light micro-scope for degeneration of the 24 h side at the rostral portion of the innervated tectum (15-35 % of the total tectal distance). Sections which showed black dots in that region were remounted for sectioning (Schabtach & Parkening, 1974), trimmed to cut away the 24 h side (the time at which the degeneration is seen with light microscopy) leaving only the 8 h side (the time at which degeneration can be seen ultrastructurally in synaptic terminals) and sectioned for electron microscopy.
Thin sections from the rostral 15 % to 35 % of the tecta of two animals contained synaptic processes with debris (Fig. 14, 15). Quantification of the debris of degeneration was not attempted, but the debris did appear to be sufficiently prevalent to account for the degeneration seen with the light microscope at the rostral tectum after nasal retinal lesion. Sections from the caudal innervated tectum at 50–70 % of the entire tectal distance were also examined for degenerating synapses and they were found, as expected. These results show that fibres from the nasal retina have synapses at both rostral and caudal parts of the innervated tectum at stage 51.
Electron micrographs of degeneration in synaptic processes. These sections were taken from the rostral portion of the innervated tectum in tadpoles at stage 51 following lesion of the nasal retina. Synapses are marked by arrows; debris characteristic of degeneration (D) is close by. Silver to grey sections post-stained with lead citrate and uranyl acetate.
Electron micrographs of degeneration in synaptic processes. These sections were taken from the rostral portion of the innervated tectum in tadpoles at stage 51 following lesion of the nasal retina. Synapses are marked by arrows; debris characteristic of degeneration (D) is close by. Silver to grey sections post-stained with lead citrate and uranyl acetate.
DISCUSSION
Wallerian degeneration can follow one of a number of courses, depending on the tract and the species (Raisman & Matthews, 1972). The degeneration found here in retinal fibres after they have been severed from their cell bodies is similar in its first stages (the production of dark debris in otherwise healthy-looking terminals) to that reported in cats (Hámori, Lang & Simon, 1968). The subsequent engulfment of this debris by ependymal glia has also been reported in other amphibian visual tracts (Scott, 1973, 1974; Reier & Webster, 1974; Turner & Singer, 1975). The glial engulfment produces debris large enough to be mapped with light microscopy.
Wallerian degeneration in some systems is known to begin at the terminal end of the severed axon, and axonal degeneration proceeds later (Hámori et al. 1968). Relatively short degeneration times were chosen here to minimize the possibility of axonal degeneration, but in practice it is not possible to be certain that all of the debris mapped with light microscopy is terminal. Where necessary for interpretation of the maps, the source of the degeneration in particular areas can be checked with electron microscopy, as was done here for degeneration appearing in both rostral and caudal tectum after lesion of the nasal retina in tadpoles. The finding of terminal degeneration in both tectal areas demonstrates that ganglion cell bodies located in the nasal retina have at least some terminations in rostral as well as caudal tectum during their period of axon outgrowth.
To summarize, fibres from both temporal and nasal retina terminate in the optic tectum even at an early stage of innervation. The retinotectal map at this stage is organized roughly like an adult map in miniature on the innervated portion of the tectum. This is in agreement with the electrophysiologically obtained results of Gaze et al. (1972, 1974), and indicates that some fibres (particularly from nasal retina) may shift their terminals caudally on the tectum to match tectal growth in order to produce the adult pattern of connections (Gaze, 1974). Substantial numbers of connections which are ‘aberrant’ in the sense that they are not located in accordance with the adult projection are present in the stage-51 tadpole. Most of the aberrant terminals come from the nasal retina, which forms terminals in the rostral, as well as the expected caudal, tectum. The incorrect connections appear to be considerably reduced in number in postmetamorphic juveniles, although it is possible that the aberrant connections are retained, but are ‘diluted’ by the addition of large numbers of correctly made synapses. Gaze et al. (1972, 1974) observed that in stage-47 tadpoles, the receptive field of electrodes placed in the rostral tectum is much larger than that of more caudally placed electrodes. This distinctive difference in receptive field disappears as development proceeds. As noted by Gaze (1975), these observations both suggest that there is a reduction in the relative number of incorrect connections with development. If there is readjustment of retinal connections as the tectum grows, the ‘aberrant’ connections observed here in the tadpoles at stage 51 may be an indication of this process.
ACKNOWLEDGEMENTS
1 would like to thank Dr Philip Grant and Dr Edith Maynard for stimulating discussions and helpful criticisms during the course of this work, and Dr J. S. Edwards and Dr R. D. Lund for comments on the manuscript. This research fulfills, in part, the requirements for the degree, Doctor of Philosophy, and was supported by contracts AT (45-l)-2011 and AT (45-1)-2230 to Dr Philip Grant and Health Science Advancement Award FR-06027 from the National Institutes of Health.