Three-dimensional architecture of inner medullary vasa recta

【摘要】  The manner in which vasa recta function and contribute to the concentrating mechanism depends on their three-dimensional relationships to each other and to tubular elements. We have examined the three-dimensional architecture of vasculature relative to tubular structures in the central region of rat kidney inner medulla from the base through the first 3 mm by combining immunohistochemistry and semiautomated image acquisition techniques with graphical modeling software. Segments of descending vasa recta (DVR), ascending vasa recta (AVR), descending thin limb (DTL), ascending thin limb (ATL), and collecting duct (CD) were identified with antibodies against segment-specific proteins associated with solute and water transport (urea channel B, PV-1, aquaporin-1, ClC-K1, aquaporin-2, respectively) by immunofluorescence. Results indicate: 1 ) DVR, like DTLs, are excluded from CD clusters that we have previously shown to be the organizing element for the inner medulla; 2 ) AVR, like ATLs, are nearly uniformly distributed transversely across the entire inner medulla outside of and within CD clusters; 3 ) DVR and AVR outside CD clusters appear to be sufficiently juxtaposed to permit good countercurrent exchange; 4 ) within CD clusters, about four AVR closely abut each CD, surrounding it in a highly symmetrical fashion; and 5 ) AVR abutting each CD and ATLs within CD clusters form repeating nodal interstitial spaces bordered by a CD on one side, one or more ATLs on the opposite side, and one AVR on each of the other two sides. These relationships may be highly significant for both establishing and maintaining the inner medullary osmotic gradient.

【关键词】  threedimensional reconstruction PV urea channel B aquaporin ClCK Bcrystallin countercurrent multiplier concentrating mechanism vasa recta

AMONG ALL FUNCTIONAL DOMAINS within the kidney, the concentrating mechanism in the inner medulla (IM) is perhaps the most complicated and, certainly, the least well understood. Recent characterization of proteins associated with epithelial and endothelial membrane transport in defined vascular and tubular segments of the IM has provided new information for models of the concentrating mechanism. However, interactions of the transport of fluid and small solutes that these proteins enable depend profoundly on the three-dimensional (3-D) architecture of the nephrons, collecting ducts (CDs), blood vessels, and interstitial cells and matrix. We have previously described some of the 3-D relationships of the inner medullary thin limbs of Henle?s loops and CDs and the location of some of the transport or channel proteins along them ( 19, 20 ). This information helped us to propose one possible model for the concentrating mechanism in the IM ( 9 ). However, these studies did not include information on the inner medullary vasculature. The vasa recta are generally considered to function in the IM as countercurrent exchangers to delay the washout of NaCl and urea and prevent the excess accumulation of water during the concentrating process. However, their actual function and the manner in which they contribute to the concentrating mechanism depends on their 3-D relationships to each other and to the tubular elements. Although some aspects of the relationships of these vascular elements to the tubules have been described by Kriz and co-workers (e.g., 4, 10), the exact transverse and vertical relationships were not completely clear.

In the present study, we have continued our previous studies by examining the 3-D architecture of the vasculature relative to the tubular structures in the central region of rat kidney IM from the base through the first 3 mm. As in our previous studies ( 19, 20 ), we combined immunohistochemistry and semiautomated image acquisition techniques with graphical, volumetric modeling software to compile multiple, serial tissue sections in 3-D surface and volumetric representations of the inner medullary descending vasa recta (DVR) and ascending vasa recta (AVR) and their relationships to CDs, descending thin limbs (DTLs), and ascending thin limbs (ATLs). We used antibodies that specifically labeled the urea channel B (UTB), a protein of fenestral diaphragms (PV-1), the water channels aquaporin-1 (AQP1) and aquaporin-2 (AQP2), and the inner medullary chloride channel ClC-K1 to identify DVR, AVR, DTLs, CDs, and ATLs, respectively. We also used a specific antibody against the heat shock-related protein B-crystallin to identify portions of tubules not labeled by the other antibodies.

Our primary results indicate that 1 ) DVR, like DTLs, are excluded from the CD clusters that we have previously shown to be the organizing element for the IM ( 20 ); 2 ) AVR, like ATLs, are nearly uniformly distributed transversely across the entire IM outside of and within the CD clusters; 3 ) DVR and AVR outside the CD clusters appear to be arranged close enough together to permit good counter current exchange; 4 ) within the CD clusters, about four AVR closely abut each CD, surrounding it in a highly symmetrical fashion; and 5 ) the AVR abutting each CD and the ATLs within the CD clusters form repeating nodal interstitial spaces bordered by a CD on one side, one or more ATLs on the opposite side, and one AVR on each of the other two sides. These relationships may be highly significant for both establishing and maintaining the inner medullary osmotic gradient.

METHODS

Animals. Young male Munich-Wistar rats (average wt 90 g) were purchased from Harlan (Indianapolis, IN). The animals were anesthetized with pentobarbital sodium (0.2 ml/100 g body wt). All animal experiments were approved by the University of Arizona institutional care and use committee and were conducted in accord with the Guide for the Care and Use of Laboratory Animals.

Tissue preparation for immunohistochemistry. Kidneys from male Munich-Wistar rats were prepared for immunohistochemistry by retrograde perfusion through the aorta with PBS (pH 7.4) for 5 min, followed by periodate-lysine-paraformaldehyde (0.01 M, 0.075 M, 2%) in PBS (pH 7.4) for 5 min before removal from the animal. The whole medulla was dissected free, the outer medulla (OM) was discarded, and the IM was immersed in fixative for 3 h at 4°C and washed in PBS. Tissue was then prepared for either cryosectioning or epoxy sectioning for immunohistochemistry. For cryosectioning, tissue was immersed in 30% sucrose in PBS for 2 h and then frozen in plastic forms containing Tissue-Tek Optimal Cutting Temperature Compound (Sakura-Finetek, Torrance, CA) that were placed on the surface of an isopentane bath cooled with liquid nitrogen. For epoxy sectioning, tissue was dehydrated through an ethanol series and embedded in Spurr?s epoxy resin (Ted Pella). Serial transverse sections were cut at a thickness of 5 µm (cryosections) or 1 µm (epoxy) either exhaustively or partially from medullas beginning near the base of the IM and continuing in a papillary direction. To reconstruct uninterrupted segments of either nephrons or vessels from the complete kidney, two sets of 1-µm serial sections were prepared from epoxy-embedded tissue. The initial 1-µm sections for each set were offset from each other by 2 µm, and each set had 5-µm steps between sections. Consecutive sections were placed on glass microscope slides for immunohistochemistry (4 sections/slide). The boundary between the IM and OM was identified on the basis of structural criteria ( 7 ).

Tissue preparation for electron microscopy. Kidneys from male Munich-Wistar rats were prepared for electron microscopy by retrograde perfusion through the aorta with 0.08 M cacodylate buffer, pH 7.2, containing 0.6% NaCl and 3% dextran (mean mol mass 38 kDa; see Ref. 1 ). This was followed by perfusion with the same buffer containing 3% glutaraldehyde. The kidney was removed, and the whole medulla was dissected free. The OM was discarded, and the IM was cut into five sections transverse to the cortico-papillary axis. These were immersed in fixative for 18 h at 4°C, washed in buffer, postfixed with 1% osmium tetroxide in 0.08 M cacodylate buffer at 4°C, dehydrated through a graded series of ethanol solutions, and embedded in epoxy resin. Grids were stained with uranyl acetate and lead citrate. Thin sections were cut on a Leica Ultracut microtome (Leica, Deerfield, IL) and examined and photographed using a Philips CM-12S electron microscope (Philips Electronic Instruments, Mahwah, NJ).

Immunohistochemistry. Generally, two sets of serial sections, as described above, were prepared for each kidney; one set was labeled for nephrons and CDs, and one set was labeled for vasa recta and CDs. Nephron segments were labeled by indirect immunocytochemistry as described previously ( 20 ) using affinity-purified polyclonal antibodies against the COOH-terminal regions of the human water channel AQP1 (diluted 1:200, raised in chicken; provided by John Regan and W. Daniel Stamer, University of Arizona), the rat kidney-specific chloride channel (ClC-K, diluted 1:200, raised in rabbit; Chemicon), the human water channel AQP2 (diluted 1:200, raised in goat; Santa Cruz), and a bovine monoclonal antibody raised against purified B-crystallin (diluted 1:50, raised in mouse; Stressgen). AQP1 and AQP2 antibodies serve as markers for DTLs and CDs, respectively, the ClC-K antibody serves as a marker for ATLs, and B-crystallin serves as a common marker for all tubules ( 20 ).

The second set of sections was labeled for vasa recta and CDs. AVR and capillaries were labeled with a polyclonal antibody raised in chicken against rat PV-1, a plasmalemmal vesicle protein formerly known as gp68 (diluted 1:500; provided by Radu Stan, Dartmouth College; see Ref. 23 ). PV-1 is a component of the fenestral diaphragm the physiological function of which is presently poorly understood. In the rat IM, only AVR and capillaries are fenestrated, and these all are believed to have diaphragms. DVR were labeled with a polyclonal antibody raised in rabbits against rat UTB (diluted 1:200; provided by Jeff Sands, Emory University). CDs were labeled for AQP2 as described above.

Image analysis. Separate stacks of digitized, serial images were generated by capturing UTB, PV-1, AQP1, AQP2, ClC-K1, or B-crystallin immunofluorescence from each tissue section. Final images measured 900 x 1,600 pixels (608 x 1,081 µm within the central region of the IM) and encompassed a vertical depth of up to 3.3 mm beginning at the base of the IM. Continuous surface and volume representations for each vessel and tubule were constructed as described previously ( 19, 20 ) with Amira 2.3 visualization and volume modeling software (Mercury, Chelmsford, MA). Continuous 3-D surface views of each vessel or tubule were created from serial sections no greater than 5 µm apart. In the reconstructed 3-D surface representations, the vessel and tubule positions and lengths are drawn to scale in the x -, y -, and z -axes. Diameters of reconstructed vessels and tubules are approximately those that exist near the base of the IM in vivo.

Pattern differentiation of DVR and AVR distributions in two-dimensional images was determined by comparing the mean and variance of distances between nearest neighbors using the method of Schwarz and Exner ( 22 ). The mean and variance of nearest neighbor distances for approximate center points of DVR and AVR in individual two-dimensional images (each image encompassing an area ranging from 0.17 to 0.65 mm 2 ) were determined using PhotoShop (Adobe) and the Image Processing Toolkit (Reindeer Graphics). The expected mean and variance of nearest neighbor distances were calculated from the observed feature density (number of features per unit area) assuming a random distribution as represented by a Poisson distribution ( 22 ). The ratio of the observed mean nearest neighbor distance to the expected distance (Q) and the ratio of the corresponding observed mean variance to the expected variance (R) will each be equal to nearly one for features that are randomly distributed ( 22 ). By contrast, for a uniform pattern of distribution, Q 1 and R << 1 and for a clustered pattern Q < 1 and R < 1 ( 22 ).

RESULTS

Patterns of DVR and AVR distribution through the IM. Representative immunohistochemical transverse sections from the rat renal IM showing the DVR, AVR, DTLs, and ATLs are shown in Fig. 1. These sections are from the central region of the IM and are positioned 0.9 to 1.3 mm below the base of the IM. As we reported previously ( 19, 20 ), individual CDs tend to be grouped or bundled together in clusters. This clustering arises through early development as the ureteric bud branches extensively to form the mature CD system. Our concept of the concentrating mechanism in the mature kidney, in contrast, views the CD system as clusters of CD segments that coalesce as they descend along the cortico-papillary axis. The AQP1-expressing DTLs ( Fig. 1 A, red tubules) encircle CD clusters as asymmetric rings ( 20 ). DVR ( Fig. 1 A, green vessels) are positioned in a pattern that resembles this DTL distribution. Near the IM base, DVR are clustered relatively tightly together but less so than in the OM. As they descend to deeper levels along the cortico-papillary axis, DVR become even less clustered. The ATLs ( Fig. 1 B, green tubules) are more uniformly spaced with an apparently more uniform density ( 20 ). AVR and fenestrated connecting capillaries ( Fig. 1 C, red vessels) are positioned in a nearly uniform pattern that resembles the ATL distribution. These distinct two-dimensional patterns of vessel and nephron distribution extend across the entire transverse area of the IM.

Fig. 1. A: transverse section showing reticulated pattern formed by random or nearly random distribution of descending thin limb (DTL)/aquaporin-1 (AQP1; red) and descending vasa recta (DVR)/urea channel B (UTB; green) across a single plane of the inner medulla (IM). Each void, or black space encompassed by DTLs and DVR, is filled with a single cluster of collecting ducts (CDs) that coalesce as a unit as the segments descend from the IM base toward the papilla. Section in A is from 900 µm below the IM base. B and C show near uniform distribution of ascending thin limb (ATL)/ClC-K1 (green) and ascending vasa recta (AVR)/PV-1 (red) in adjacent transverse sections from the renal IM. Sections in B and C are from 1,300 µm below the IM base. Unequal numbers of DTLs and ATLs reflect the prebend region and AQP1-null DTLs. Scale bars, 100 µm.

We quantitatively compared distribution patterns for UTB-expressing DVR and PV-1-expressing AVR in transverse sections from the IM of kidneys from five animals. Although, as noted above ( Fig. 1 A ), DVR and DTLs exhibit a similar visual pattern in transverse sections, DVR have a nearly random distribution (Q and R each approximately equal to 1), whereas, as shown in our previous study ( 20 ), DTLs have a somewhat nonuniform but not truly random 1; R < 1; Table 1 ). Both AVR and ATLs have a similar, very nearly uniform distribution (Q 1; R << 1) ( Table 1 ), as suggested by the visual images ( Fig. 1, B and C ).

Table 1. Pattern differentiation for vasa recta and thin limbs in the inner medulla as determined by the method of Schwarz and Exner ( 22 )

A signature feature of DVR in the OM is their formation of vascular bundles as they descend through the inner stripe ( 10 ) before dispersing in the nearly random distribution at the perimeters of CD clusters (described above; Fig. 1 ) as they descend through the IM. To determine the consistency of these patterns among individual animals, we quantitatively analyzed the DVR distribution in sections from the OM (or base of the IM before dispersion) from three kidneys. Q and R values were both <1 as expected for a clustered distribution in vascular bundles in this region of the kidney ( Table 1 ).

3-D patterns of DVR throughout the IM. DVR descend predominantly as unbranched vessels from the IM base into the deep papilla. Infrequently, DVR do form as many as two to three relatively short branches as they descend, and these branches continue descending to deeper levels (data not shown). DVR remain positioned alongside the perimeters of CD clusters as they make their descent and do not comingle with the CD segments that collectively compose each CD cluster ( Figs. 1 and 2, A-D ). This pattern suggests that DVR are restricted to a compartment that lies outside compartments occupied by CD clusters containing ATLs and AVR within them. As published previously ( 20 ) and as shown in Figs. 1 A and 2, A-D, DTLs are also restricted to the same compartment as DVR.

Fig. 2. Three-dimensional reconstruction showing spatial relationships of DVR (green tubules) and DTLs (red tubules) to CDs (blue tubules) for a single CD cluster. DTL segments that do not express AQP1 were identified by their expression of B-crystallin and are shown in gray. DVR and DTLs are spatially separate from the CDs, and A-D show that this relationship continues along the entire axial length of the CD cluster. Tubules in A-D have been rotated forward 20 degrees. Axial positions of A-D are shown in E with lowercase letters. Tubules are oriented in a corticopapillary direction, with the upper edge of the image near the base of the IM. Scale bar, 500 µm.

The expression of UTB along DVR ends fairly abruptly, apparently constituting the end of the DVR and the beginning of the capillary network or connection with AVR. However, the terminal portion of the DVR shows overlap of PV-1 with UTB, apparently indicating the beginning of fenestrations before the end of the DVR ( Fig. 3 ). For 35 reconstructed DVR from one single kidney, the overlap occurred for an average length of 139.7 ± 20.9 (SE) µm. The frequency distribution of these overlap lengths is shown in Fig. 4. All DVR break up into at least two capillary branches before giving rise to AVR ( Figs. 3, 5, and 6 ). As reported previously by Jamison and Kriz ( 4 ), we observed no direct connection of a DVR to an AVR. More frequently, a complex capillary network forms immediately after the DVR terminus. AVR arise from this network at variable distances from the DVR terminus ( Figs. 3 and 6 ).

Fig. 3. Three separate vessels (white arrowheads) exhibit colocalization of UTB ( A, green) and PV-1 ( B, red). Merged images are shown in C. Section is from 1 mm below the IM base. D shows a 3-dimensional reconstruction of terminal segments of three additional DVR. UTB expression for each descending vas rectum (green vessels) overlaps PV-1 expression for each ascending vas rectum (red vessels) for variable lengths (white arrows). These AVR continue a descent toward the tip of the papilla in this region. Fenestrated vessels arising from each DVR join together to form a capillary network. Scale bars, 10 µm in A-C and 500 µm in D.

Fig. 4. Length of overlap between DVR and AVR. Chart shows frequency of overlap ranges for 35 reconstructed DVR that lie within 3.3 mm from the base of the IM.

Fig. 5. Three-dimensional reconstruction of a single DVR (green) connecting to an AVR (red) that exhibits simple branching pattern. Green and red overlap identifies colocalization of UTB and PV-1 (arrows). Upper ends of reconstructed vessels lie at 350 µm below the IM base. Scale bar, 500 µm.

Fig. 6. Three-dimensional reconstruction of a single DVR (green) that terminates in two branches and connects to several fenestrated vessels (red) that exhibit complex branching patterns as they ascend the IM. Green and red overlap identifies colocalization of UTB and PV-1 (arrows). Upper ends of reconstructed vessels lie at 350 µm below the IM base. Scale bar, 500 µm.

3-D patterns of AVR and capillaries throughout the IM. Both AVR and capillaries express PV-1, so we were unable to differentiate between the two with this label alone. However, in the 3-D reconstructions, we assumed that PV-1-expressing vessels running 1-3 mm along the cortico-papillary axis were AVR ( Fig. 5 ). As shown in the 3-D reconstructions in Fig. 7 and as discussed above for the transverse sections ( Fig. 1 ), AVR and ATLs are distributed nearly uniformly within and around the CDs throughout the length of the IM examined. Moreover, in these 3-D reconstructions, we were able to follow a number of DVR through short apparent capillary connections to single PV-1-expressing vessels that we identified as AVR ( Fig. 6 ). In this fashion, we were able to model complete loops ( Figs. 5 - 7 ). These AVR generally continued to descend for variable lengths after termination of UTB expression. These lengths ranged from 500 µm. PV-1-labeled vessels include relatively unbranched and very highly branched vessels; these latter likely include primarily interconnecting capillaries. Figures 5 and 6 show examples of unbranched and highly branched AVR, respectively. The unbranched AVR that have been reconstructed lie predominantly outside the CD clusters.

Fig. 7. Three-dimensional reconstruction showing spatial relationships of AVR (red tubules) and ATLs (green tubules) to CDs (blue tubules) for the same CD cluster shown in Fig. 2. AVR and ATLs are distributed relatively uniformly outside of and within the CD cluster, and A-D show that this relationship continues along the entire axial length of the CD cluster. Tubules in A-D have been rotated forward 20 degrees. Axial positions of A-D are shown in E with lowercase letters. Tubules are oriented in a corticopapillary direction, with the upper edge of the image near the base of the IM. Scale bar, 500 µm.

Spatial arrangements of fenestrated vessels abutting CDs. As noted above, AVR are distributed relatively uniformlythroughout the IM. Those AVR outside CD clusters lie adjacent to or near DVR, whereas those AVR within CD clusters are physically separated from DVR by some significant distance. Some, but not all, AVR within the CD clusters abut CDs and lie parallel to each CD along the cortical-papillary axis for significant distances ( Figs. 8 and 9 ). The close relationship between these AVR and each CD can be clearly seen in electron micrographs ( Fig. 10 ) that show that the AVR basal plasma membrane is typically positioned within 0.5-1.0 µm of the CD basal plasma membrane ( Fig. 10 B ). The electron micrographs ( Fig. 10, C and D ) indicate that physical contact between the AVR and CDs is limited to basal processes arising from the AVR and extending between the two structures. These "microvilli" arising from renal endothelia have been described previously by Bulger and Trump ( 2 ) and Takahashi-Iwanaga ( 25 ). Microvilli arising from endothelia of nonrenal tissue have been described by Friederici ( 3 ). Takahashi-Iwanaga ( 25 ) suggested that they anchor the AVR to the tubule, thereby helping to hold these venous structures open when the hydrostatic pressure of the interstitial fluid rises above that within the vessels as a result of pelvic contractions or increased interstitial fluid volume ( 12, 13, 25 ).

Fig. 8. Three-dimensional reconstruction of single CD segment (blue) with multiple AVR (red) butted up against it. Top : 90° axial rotation of segments shown in adjacent panels. Scale bar, 100 µm.

Fig. 9. Transverse section from near the base of the IM. A : inner medullary AVR and capillaries (red) are symmetrically positioned around CDs (blue) and make intimate contact with adjacent CDs. This relationship extends from the base of the IM to the tip of the papilla for most CDs. Inner medullary UTB-expressing DVR (green) tend to be distant from CDs. B : diagrammatic representation of tubules and vessels to facilitate pattern analysis. Scale bar, 30 µm.

Fig. 10. Electron micrographs showing transverse sections of CDs and AVR from 1.5 mm ( A, B, and D ) and 4 mm ( C ) below the base of the IM. A : CD surrounded by 4 AVR (*). Other tubular structures surrounding the CD are ATLs. Scale bar, 10 µm. B : AVR abuts CD with minimal direct contact. Scale bar, 1 µm. C : AVR abuts CD with microvillus (arrow). IS, interstitium. Scale bar, 1 µm. D : AVR abuts CD with microvilli (arrows). Scale bar, 1 µm.

Those AVR that abut CDs are typically arranged in a near symmetrical pattern around the perimeter of each CD ( Figs. 8 - 10 A ). The number of AVR surrounding each CD varies, depending on diameter of the CD, and ranges approximately from three for smaller-diameter CDs to five for larger-diameter CDs. For each of five kidneys from five individual animals, we determined the average number of inner medullary AVR surrounding five to seven CDs within 1,400 µm from the IM base. The mean number of AVR surrounding CDs for these five kidneys was 4.12 ± (SE) 0.15. Each AVR associated with any group of abutting AVR may disengage, or branch off from one CD, and adhere to a neighboring CD at shallower or deeper levels. However, the general pattern of three to five symmetrical AVR adhering to each CD occurs at any level along the cortico-papillary axis.

For the same five kidneys from five individual animals referred to above, we determined the average amount of CD surface area in close association with AVR. As indicated in Table 2, 53.7% of the CD surface area abuts AVR.

Table 2. Fractional CD surface area in contact with AVR

Interstitial compartments delimited by CD, AVR, and ATL. As we reported previously ( 20 ) and as shown in Figs. 1 A and 7, a uniform expression of ATLs occurs within the CD clusters. In addition, as described above, a close association of AVR with CDs occurs both transversely across the IM and along the cortico-papillary axis ( Figs. 7 - 9 ). The combination of these arrangements suggests that the ATLs, CDs, and AVR may form interstitial compartments within the CD clusters. One example of this arrangement and the resulting compartments is shown in Fig. 11 (although this arrangement can also be seen in Fig. 10 A ). These compartments are physically separated from the outlying DTLs and DVR and suggest that solute diffusion from ATLs and CDs into AVR may be preferentially restricted to this compartment.

Fig. 11. One single CD in transverse section showing interstitial nodes (marked with X) between CD, AVR, and ATLs in a composite image of 2 sections 3 µm apart, from near the inner medullary base. Blue, CD; red, AVR; green, ATL. Open space in wall of central CD is the location of an intercalated cell, which does not label for aquaporin-2. Scale bar, 10 µm.

Colocalization of UTB and AQP1 in DVR. The presence of AQP1 in basolateral and apical membranes and caveolae of nonfenestrated endothelia and absence in fenestrated endothelia of the IM has been previously established with immunohistochemistry ( 14, 21 ). Functional studies indicate that significant water permeability through mercurial-sensitive water channels exists in outer medullary DVR ( 14, 16 ). We previously reported a steep decline in AQP1 immunoreactivity in DTLs of the innermost half of the IM compared with levels occurring nearer to the base, along with an overall weak AQP1 expression in transverse sections from this region. Here, we evaluated AQP1 colocalization with UTB in DVR that lie between the base and the middle IM in cryosections from four kidneys. Many DVR were seen to express both UTB and AQP1, and, as reported previously ( 14 ), the labeling intensity of AQP1 was less than that seen for inner medullary DTLs. Nonetheless, some DVR (as determined by UTB expression) had no apparent expression of AQP1. One example of different expression levels for DVR at 3.4 mm below the IM base is shown in Fig. 12.

Fig. 12. UTB ( A ) and AQP1 ( B ) consistently colocalize in DVR (merged images in C, yellow). However, variable levels of immunoreactivity were observed for some adjacent DVR. Arrows identify two DVR that express little or no apparent AQP1. Cryosection from 3.4 mm below the inner medullary base. Scale bar, 20 µm.

DISCUSSION

The present 3-D functional reconstructions of vasa recta and 3 mm within the rat renal IM reveal a number of spatial relationships between vessels and tubules that may be significant for water and solute exchange and the concentrating mechanism. First, the DVR, like the DTLs, are excluded from the CD clusters that form the organizing motif for the IM (summarized diagrammatically in Fig. 13 B ). Second, the AVR, like the ATLs, are nearly uniformly distributed transversely across the entire IM and are found within the CD clusters (summarized diagrammatically in Fig. 13, A, C, and D ). Third, DVR and AVR outside the CD clusters appear to be arranged close enough together to promote good countercurrent exchange. However, it must be emphasized that all such countercurrent exchange between DVR and AVR must occur in a limited region outside the CD clusters. Fourth, the proximity of DVR to DTLs outside the CD clusters suggests some possible exchange, although the exact nature and physiological function of such exchange is unclear. Fifth, and, we think, of particular significance, about four AVR closely abut each CD, surrounding it in a highly symmetrical fashion and covering 54% of its surface area. The distance of 1 µm or less between the external surfaces of these AVR and CDs would certainly promote diffusive exchange between them. This arrangement could facilitate rapid removal of water absorbed from CDs in the IM via AVR.

Fig. 13. Diagrammatic transverse representation of ATL (green) and AVR (red) distribution around a single CD cluster (blue, A ), DTL (purple) and DVR (aqua, B ), AVR alone ( C ), and ATL alone ( D ). Additional CD clusters exist but are not shown in A. These would be positioned within the open spaces created by DTLs and DVR in B. Unequal numbers of DTLs and ATLs reflect the prebend region and AQP1-null DTLs. Symmetry is artificial.

In addition, the arrangement of CDs and of AVR and ATLs inside the CD clusters, as seen in transverse sections, produces a repeating pattern of nodal interstitial spaces. 1 In transverse sections, each of these spaces is bordered on one side by a CD (in which axial tubular flow is away from the medullary base), on the opposite side by one or occasionally more ATLs (in which axial tubular flow is toward the medullary base), and on each of the other two sides by an AVR (in which blood flow is toward the medullary base; Figs. 10 A and 11 ). On the cortico-papillary axis, these spaces are probably not simple columns. Studies showing a ladder-like arrangement of interstitial cells on this axis suggest that the spaces may be 1 µm thick, bordered above and below by interstitial cells and their processes ( 5, 11, 25 ). Thus there would be stacks of these nodal spaces, rather than columns. These nodal spaces may play a role in some version of the "solute-separation, solute-mixing" mechanism for concentrating the urine proposed recently ( 9 ).

UTB and PV-1 expression overlapped in the terminal portion of all DVR studied. We assumed that the DVR ended at the point where UTB ceased to be expressed, an assumption supported by the apparent connection to AVR via capillaries beyond this point. However, the presence of PV-1 expression before the cessation of UTB expression indicates that fenestrations begin to appear in the endothelium of the DVR before they terminate in a capillary network. The presence of a spotty distribution of fenestrations in the terminal portions of DVR (similar to those shown with our antibodies to PV-1; Figs. 3, 5, and 6 ) has been demonstrated previously by the ultrastructural studies of Jamison and Kriz ( 4 ). However, the present study indicates that there is considerable variation in the length of the terminal fenestrated portion of the DVR. Moreover, the overlap of UTB and PV-1 expression suggests that the fenestrations in the terminal portion of the DVR are insufficient for appropriate movement of urea across the capillary wall without the presence of UTB. An increased urea movement may be important for exchange between DVR and AVR in this region of the IM. The overlap of UTB and PV-1 at the junctions of DVR and AVR in the last 2 mm of the papilla (not examined in this study) may vary from that which occurs in the outer part of the IM because of different patterns of UTB expression and different patterns of arterial vessel architecture.

In the present study, we found no expression of AQP1 in numerous DVR in the IM, even when frozen sections rather than epoxy sections were used ( Fig. 12 ). However, other DVR clearly expressed AQP1 ( Fig. 12 ), as described in previous studies ( 14, 21 ). Although the antibodies we are using appear to label AQP1 very well in epoxy sections, we also used some frozen sections to be certain that we were not missing small amounts of AQP1 because of some lack of sensitivity to the antibodies in epoxy sections. Therefore, it appears that numerous DVR in the central portion of the IM (at least from near the base to a depth of 3.3 mm) lack AQP1 water channels. This is significant because a number of studies have demonstrated that water is absorbed from the DVR along the length of the IM and that this absorption is largely driven by the small solute (NaCl and urea) gradient across the vessel walls ( 15, 18 ). If AQP1 water channels are present, they can provide an exclusive pathway for water movement through the endothelium with a high reflection coefficient for NaCl and urea. Indeed, this appears to be the case for DVR in the OM, where the osmotic water permeability and NaCl-induced water flux in the DVR are reduced to near zero when the AQP1 water channels are blocked with p -chloromercuribenzosulfonic acid in rats ( 17 ) or when they are absent in AQP1-null mice ( 16 ). However, UTB has also been shown to function as an osmotic water channel under some circumstances ( 26, 27 ), although its quantitative importance in DVR is unknown. Nevertheless, UTB might function as a significant pathway through which water could be driven by the NaCl osmotic gradient in those inner medullary DVR lacking AQP1.

GRANTS

This study was supported in part by National Institutes of Health Grants DK-16294 and ES-06694 for the Southwest Environmental Health Sciences Center and National Science Foundation Grant IBN9814448.

ACKNOWLEDGMENTS

We are grateful to Diane Abbott for technical assistance. We thank Drs. John Regan and Dan Stamer of the University of Arizona for providing AQP1 antibodies, Radu Stan of Dartmouth College for providing PV-1 antibodies, and Jeff Sands of Emory University for providing UTB antibodies.

【参考文献】
  Bohman SD. The ultrastructure of the rat renal medulla as observed after improved fixation methods. J Ultrastruct Res 47: 329-360, 1974.

Bulger RE and Trump BF. Fine structure of the rat renal papilla. Am J Anat 118: 685-722, 1966.

Friederici HHR. Extrusion of basal endothelial projections through the capillary basement membrane. Angiology 16: 163-16, 1965.

Jamison RL and Kriz W. Urinary Concentrating Mechanisms: Structure and Function. New York, Oxford, 1982.

Kneen MM, Harkin DG, Walker LL, Alcorn D, and Harris PJ. Imaging of renal medullary interstitial cells in situ by confocal fluorescence microscopy. Anat Embryol 200 117-121, 1999.

Kokko JP and Rector FC. Countercurrent multiplication system without active transport in inner medulla. Kidney Int 2: 214-223, 1972.

Kriz W and Bankir L. A standard nomenclature for structures of the kidney. Kidney Int 33: 1-7, 1988.

Kriz W and Kaissling B. Structural organization of the mammalian kidney. In: The Kidney: Physiology and Pathophysiology, edited by Seldin DW and Giebisch G. New York: Raven, 1992, p. 707-778.

Layton AT, Pannabecker TL, Dantzler WH, and Layton HE. Two modes for concentrating urine in rat inner medulla. Am J Physiol Renal Physiol 287: F816-F839, 2004.

Lemley KV and Kriz W. Cycles and separations: the histotopography of the urinary concentrating process. Kidney Int 31: 538-548, 1987.

Lemley KV and Kriz W. Anatomy of the renal interstitium. Kidney Int 39: 370-381, 1991.

MacPhee PJ and Michel CC. Subatmospheric closing pressures in individual microvessels of rats and frogs. J Physiol 484: 183-187, 1995.

Michel CC. Renal medullary microcirculation: Architecture and exchange. Microcirculation 2: 125-113, 1995.

Nielsen S, Pallone T, Smith BL, Christensen EI, Agre P, and Maunsbach AB. Aquaporin-1 water channels in short and long loop descending thin limbs and in descending vasa recta in rat kidney. Am J Physiol Renal Fluid Electrolyte Physiol 268: F1023-F1037, 1995.

Pallone TL. Effect of sodium chloride gradients on water flux in rat descending vasa recta. J Clin Invest 87: 12-19, 1991.

Pallone TL, Edwards A, Ma T, Silldorff EP, and Verkman AS. Requirement of aquaporin-1 for NaCl-driven water transport across descending vasa recta. J Clin Invest 105: 215-222, 2000.

Pallone TL, Kishore BK, Nielsen S, Agre P, and Knepper MA. Evidence that aquaporin-1 mediates NaCl-induced water flux across descending vasa recta. Am J Physiol Renal Physiol 272: F587-F596, 1997.

Pallone TL, Yagil Y, and Jamison RL. Effect of small-solute gradients on transcapillary fluid movement in renal inner medulla. Am J Physiol Renal Fluid Electrolyte Physiol 257: F547-F553, 1989.

Pannabecker TL, Abbott DE, and Dantzler WH. Three-dimensional functional reconstruction of inner medullary thin limbs of Henle?s loop. Am J Physiol Renal Physiol 286: F38-F45, 2004.

Pannabecker TL and Dantlzer WH. Three-dimensional lateral and vertical relationships of inner medullary loops of Henle and collecting ducts. Am J Physiol Renal Physiol 287: F767-F774, 2004.

Sabolic I, Valenti G, Verbavatz JM, Van Hoek AN, Verkman AS, and Ausiello DA. Localization of the CHIP28 water channel in rat kidney. Am J Physiol Cell Physiol 263: C1225-C1233, 1992.

Schwarz H and Exner HE. The characterization of the arrangement of feature centroids in planes and volumes. J Microsc 129: 155-169, 1983.

Stan RV, Ghitescu L, Jacobson BS, and Palade GE. Isolation, cloning, and localization of rat PV-1, a novel endothelial caeolar protein. J Cell Biol 145: 1189-1198, 1999.

Stephenson JL. Concentration of urine in a central core model of the renal counterflow system. Kidney Int 2: 85-94, 1972.

Takahashi-Iwanaga H. The three-dimensional cytoarchitecture of the interstitial tissue in the rat kidney. Cell Tissue Res 264: 269-281, 1991.

Yang B and Verkman AS. Urea transporter UT3 functions as an efficient water channel. J Biol Chem 273: 9369-9372, 1998.

Yang B and Verkman AS. Analysis of double knockout mice lacking aquaporin 1 and urea transporter UT-B. J Biol Chem 277: 36782-36786, 2002.

作者单位：Department of Physiology, College of Medicine, University of Arizona, Tucson, Arizona