Novel hemodynamic structures in the human glomerulus.

37 To investigate human glomerular structure under conditions of physiological 38 perfusion we have analysed fresh and perfusion fixed normal human glomeruli at 39 physiological hydrostatic and oncotic pressures using serial resin section reconstruction, 40 confocal, multiphoton and electron microscope imaging. 41 Afferent and efferent arterioles (21.5±1.2µm and 15.9±1.2µm diameter), 42 recognised from vascular origins, lead into previously undescribed wider regions 43 (43.2±2.8 µm and 38.4±4.9 µm diameter) we have termed vascular chambers (VCs) 44 embedded in the mesangium of the vascular pole. Afferent VC(AVC) volume was 1.6 45 fold greater than Efferent VC(EVC) volume. From the AVC long non-branching high 46 capacity conduit vessels (n=7) (Con; 15.9±0.7µm diameter) led to the glomerular edge 47 where branching was more frequent. Conduit vessels have fewer podocytes than filtration 48 capillaries. VCs were confirmed in fixed and unfixed specimens with a layer of banded 49 collagen identified in AVC walls by multiphoton and electron microscopy. Thirteen 50 highly branched efferent first order vessels (E1;9.9±0.4µm diam.) converge on the EVC 51 draining into the efferent arteriole (15.9±1.2µm diam.). Banded collagen was scarce 52 around EVC. 53 This previously undescribed branching topology does not conform to the 54 branching of minimum energy expenditure (Murray’s law), suggesting even distribution 55 of pressure/flow to the filtration capillaries is more important than maintaining the 56 minimum work required for blood flow. We propose that AVCs act as plenum manifolds 57 possibly aided by vortical flow in distributing and balancing blood flow/pressure to 58 conduit vessels supplying glomerular lobules. These major adaptations to glomerular capillary structure could regulate haemodynamic pressure and flow in human glomerular capillaries.

Glomerular capillaries operate at relatively high pressure in life which in turn sets 82 urinary driving pressure in the Bowman's capsule and tubules producing tubular flow. 83 For instance, the human glomerular capillary hydrostatic pressure of 60 to 65 mmHg at 84 the afferent end (43) falls only 2-3mmHg to the efferent end. Countering this filtration 85 pressure is an afferent plasma colloid osmotic pressure of 25mmHg rising to 32mmHg at 86 the efferent end (1). As a result of filtration, urinary space hydrostatic pressure is 20-87 25mmHg (61) pressurizing the proximal convoluted tubule producing flow through to the 88 collecting duct and the renal hilus. Thus, the function and structure of the whole nephron 89 relies upon the glomerular perfusion of an oncotically appropriate fluid at the correct 90 hydrostatic pressure to raise the right physiological pressures and flows in the tubules. In 91 biopsy/necropsy kidney specimens the absence of pressure during immersion fixation 92 results in the collapse of both the glomeruli and tubules. Fixing at the correct 93 physiological pressures (oncotic and hydrostatic) is therefore essential in investigating the 94 true 'functionally inflated' architecture of the glomerulus. 95 We have previously shown that 3D ultrastructural reconstruction of animal and 96 human glomeruli fixed under hydrostatic and oncotic physiological conditions allow the 97 detailed analysis of the GFB and the identification of novel structural features such as the 98 subpodocyte space (SPS)(39) One unexpected feature of light microscopic sections from 99 these resin embedded human glomeruli was the frequency of wide vessel regions at the 100 vascular pole when compared with rodent vascular poles implying different vascular 101 structure. No mention of any such difference could be found in any recent study of 102 human glomerular structure. 103 The haemodynamic requirements of rat and human glomeruli could shed light on 104 any differing evolved morphologies. For instance, if glomerular volume is assumed to 105 estimate perfused glomerular volume, this parameter does not scale in size with the 106 increase in afferent arteriolar conductivity between rodents and humans. The human 107 afferent arteriole has a conductivity 13 fold greater than that of the mouse (14000µm 4 vs 108 1100µm 4 ) but supplies a 23 fold larger glomerular volume [see Footnote 1]. Similarly, it 109 is 3 times as conductive as that of the rat (4600µm 4 ), while supplying a 5 fold larger 110 glomerular volume. If human glomerular morphology was simply scaled up from a small 111 rodent pattern, then afferent arterioles should be closer to 26µm in diameter instead of 112 21µm. 113 This study therefore aimed to investigate these novel wide vascular regions of 114 human glomeruli. How big were these regions? What was the wall structure and 115 dimensions and were there any other associated features? Did the region constitute a 116 wider region at the base of the afferent arteriole or a region of a thin walled capillary? 117 Could these structural differences be involved in compensating for a high glomerular 118 volume relative to the vascular input in human glomeruli? To address such questions, 119 human kidneys were perfuse fixed (at physiological hydrostatic and oncotic pressures) 120 and processed in such a way to reduce any accompanying tissue volume changes. 121 Glomerular vasculature was observed and reconstructions made from fresh or fixed 122 human kidney cortex using conventional light microscopy, confocal microscopy, 123 multiphoton microscopy and transmission electron microscopy. 124 125 126

Fixation techniques 128
Human kidney tissue was sourced (with full ethical approval and consent of next 129 of kin) from transplant kidneys (n=9) unused for technical reasons (eg poor major vessel 130 condition, damage at retrieval, tumour in the contralateral kidney). The transport solution 131 perfused through the kidney was Soltran (Potassium Citrate 0.86% w/v, Sodium Citrate 132 0.82% w/v, Mannitol 3.38% w/v, Magnesium Sulphate 1.0% w/v; Baxter Healthcare, 133 UK). Approximately 2-3 litres of the solution was perfused through the kidney 134 (200ml/minute,4°C) and then stored on ice. All other chemicals were 135 sourced from Sigma-Aldrich, UK. 136 Kidneys were transported in ice-cold flush media. Centimetre diameter fresh 137 cortical tissue was sampled from one pole for confocal and multiphoton microscopy and 138 stored in chilled (4ºC) HEPES buffered Ringers solution. Smaller 1mm diameter tissue 139 pieces were taken from the cut surface and fixed in 2.5% glutaraldehyde in HEPES buffer 140 to serve as immersion fixed samples for TEM. At 4-10ºC kidneys were debrided of 141 excess fat preserving the hilar components (renal artery, vein and ureter) and the sampled 142 polar area of the kidney was clamped off with a large locking forceps. The renal artery 143 was cannulated and the renal vein was cleared of any debris to allow outflow of perfusion 144 fluid. 145 To offset any hyperfiltration and hyperperfusion during fixation normal 146 hydrostatic and oncotic pressures were re-established by perfusing with an oncotically 147 balanced (25mmHg oncotic pressure) flush solution (50ml, 20ºC). Colloid osmotic 148 pressures were measured using a modified Hanson osmometer. The flush solution 149 temperature was kept low to minimise autolytic/proteolytic activity. The hydrostatic 150 pressure in the renal artery was set at 100mmHg (similar to human mean arterial 151 pressure). After the flush bolus, 400ml of fixative was perfused through the kidney at the 152 same pressures and temperature. Flush solution concentration was (mM); NaCl(132), KCl 153 (4.6), MgSO 4 (1.3), CaCl 2 (2), HEPES (5), NaHCO 3 (25), D-glucose (5.5), 6.5% (w/v) 154 Ficoll 400. Fixative was the same as the flush solution but with 1.25% (w/v) 155 glutaraldehyde. The glycocalyx stain 0.5% lanthanum nitrate and 0.5% dysprosium 156 chloride was incorporated into the solutions in 2 kidneys. 157 1mm diameter samples of perfusion fixed kidney were taken from a medial sub-158 capsular position and together with subcapsular immersion fixed samples were post-fixed 159 in osmium tetroxide, dehydrated with ethanol and processed into Araldite resin using 160 standard procedures. 161 To promote consistency in structural comparisons, measurement and observations 162 were limited on the glomeruli of the outer (subcapsular) cortex of kidneys in a medial 163 location half way between the poles (unless otherwise stated). 164 165 Reconstruction of vascular poles from perfusion fixed kidneys 166 Seven areas of resin embedded kidneys (n=4) which contained a high density of 167 glomeruli were identified in Toluidine Blue stained sections. These areas were serially 168 sectioned on a Reichert Ultracut microtome at 1µm thickness (2,095 sections 169 approximately 300 sections per area). From these serial section runs, 3 or 4 fully 170 sectioned glomeruli from each kidney were selected that clearly showed a vascular pole. 171 The afferent arterioles of each of the 14 glomeruli were identified by tracing to a larger 172 artery and/or the efferent arteriole traced to a peritubular position. 173 The glomerular diameters (2r x 2r y 2r z ) measured during the calibration of section 194 thickness were used to calculate glomerular volume (V G = 1.33 π r x r y r z ).

196
Glomerular and vascular orientation in resin section reconstruction 197 Vascular pole recognition was most easily achieved in 1µm serial resin sections 198 where the section plane was par-axial with the vascular pole -urinary pole axis of the 199 glomerulus, as a result reconstructed glomeruli were sectioned close to a paraxial plane. 200 The true diameters of any vessel profile was measured by searching the sequential images 201 for the appropriate vessel section and measuring vessel width (x,y). Section depth 202 diameter was taken from the limits of vessel walls in the sectioning direction (z). Vessel 203 lengths (between branch points for example) through the image stack were measured on 204 section if possible or by triangulating through the stack using sectioning depth and 205 horizontal 'on section' distance. 206 The three diameters of VCs (x,y and z) used to calculate the means in table 1 and  207 2 were further used to calculate afferent and efferent vascular chamber volume (V AVC = 208 1.33 π r AVC r AVC r AVC ; V EVC = 1.33 π r EVC r EVC r EVC ). 209 Bends between arterioles and VCs were assessed in resin section image stacks of 210 10 glomeruli by assessing the afferent and efferent arteriole axis vector and measuring 211 the change in angle into the VC axis vector (Fig.3A). This included measurements on 212 section and in the sectioning direction and triangulation in vessels moving at angles to the 213 section plane. 214 215 Afferent first order (conduit) vessel ballooning in resin sections 216 Any ballooning or hyperinflation of first order afferent (conduit) vessels was 217 estimated initially by comparing conduit diameters in areas of potentially high transmural 218 pressure gradient (conduit vessels with large areas of GFB, 0-60% mesangial cover) with 219 conduit diameters in areas of potentially low transmural pressure gradient (conduit 220 vessels with 80-100% mesangial cover). These data were further dissected in each 221 conduit vessel by subdividing the initial 0-60% mesangial cover group into 4 groups and 222 using 488nm wavelength was used to image and obtain z stacks from glomerular vascular poles 248 of up to 100µm depth from the cut surface. 249 Using a multiphoton microscope, two fresh and two fixed unstained slices of renal 250 cortex, were imaged as previously described (2). Two imaging modes were applied, 251 fibrous collagen was visualised using second harmonic generation (SHG) and elastin 252 from its intrinsic two photon fluorescence (TPF) along with any background 253 fluorescence. TPF and SHG images were obtained using a modified confocal microscope 254 (FluoView IX71 and F300, Olympus). Signal was produced using the 800 nm output of a 255 mode-  Fig.2B). 320 Vascular width and connectivity is illustrated in a scale diagram in figure 3A, 321 (measurements from tables 1&2). To summarize, the 21µm diameter afferent arteriole 322 (AA) leads into an ellipsoidal afferent vascular chamber (AVC; 49x48x32µm) which 323 branches into on average 7 first order afferent vessels of 16µm diameter we have termed 324 conduit vessels (Con; Fig.2A In 10 of the 14 glomeruli where the orientation of afferent and efferent arterioles 340 on entry into the VCs could be easily assessed, the AA bent 60º off its straight track into 341 the AVC (AA/AVC angle = 120±6˚), similarly, the EA bent 71º off track into EVC 342 (EA/EVC angle = 109±7º Fig.3A). 343 344 VC and glomerular size 345 AVC volume (V AVC = 41±5x10 3 µm 3 ) was 1.6 fold greater than EVC volume 346 (V EVC =28±7 x 10 3 µm 3 ), with no correlation between them (R 2 = 0.164 P=0.152). V AVC 347 varied over a greater size range (15-70x10 3 µm 3 ) with V EVC more conserved (12 out of 14 348 between 10-40x10 3 µm 3 ). Both V AVC and V EVC correlated significantly with V G (Fig.4C, 349 D, Tab.4), V G being 100 fold larger than V AVC and 150 fold larger than V EVC 150. This 350 implies a relationship of both the input the output manifold with the magnitude of the 351 perfused volume. 352 If the glomerular and VC volume (Fig.4C, D) correlation is extrapolated back 353 from larger glomeruli then a minimal VC volume can be reached where the volume 354 describes a mere continuation of the attached arteriole (Fig.3B). Accordingly, a 355 cylindrical minimum VC volume was calculated using average VC length (L) and 356 arteriole radius (r), a minimum AVC volume of 1.57x10 4 µm 3 would occur at a V G of 357 2.2x10 6 µm 3 (Fig.4c). Similarly, a minimum EVC volume of 0.75x10 4 µm 3 would occur at 358 a V G of 2.9x10 6 µm 3 (Fig.4D). Translating V G into glomerular diameter, VCs would be 359 minimal (a continuation of the arteriole) in human glomeruli below 160-180µm diameter 360 (i.e.V G = 2 -3x10 6 µm 3 ). 361 362 Conduit podocytes 363 In resin section image stacks spanning a conduit vessel, we noted a significant 364 lack of coverage of podocyte cell bodies (PCB) over the GFB surface (e.g. Fig 425 Using a combination of fixation induced autofluorescence (FIA), two photon 426 fluorescence (TPF) and second harmonic generation (SHG) modes, AVC could be seen 427 with attached wide conduit vessels and AA in both fixed and fresh kidney slices (Fig.7). 428

VC imaged by confocal and multiphoton microscopy
EVC was more difficult to observe with narrower blood vessels (E1) emerging from 429 them. Measurements of recognised structures show similar dimensions using these 430 optical sectioning methods and resin section reconstruction methods (Tab.4). 431 In addition to morphology SHG can detect collagen without the need for fixation 432 or labelling. Coherent emission in SHG mode in unfixed glomeruli revealed a signal 433 consistent with banded collagen which when overlaid with co-registered TPF images was 434 positioned in the AVC walls (Fig.7 In human glomeruli both arterioles exhibit vascular widenings more frequently 464 associated with low pressure veins (venous sinuses of the brain) or with large arteries 465 (carotid sinus). However, the glomerular VCs are high pressure arteriolar afferent and 466 efferent chambers with multiple openings, the closest definition in physical terms is a 467 plenum manifold (plenum -a chamber containing pressurized fluid to control 468 distribution; manifold -a pipe or chamber branching into several openings). 469 Plenums and manifolds in industry stabilize, distribute or balance fluid flow 470 through multiple inlets and outlets (i.e. inlet and exhaust manifolds on internal 471 combustion engines). Therefore, our initial hypothesis for glomerular vascular chambers 472 is that they function to balance the pressure and/or flow through the intervening filtration 473 regions without the need for conventional branching within the confined space of the 474 glomerulus. These haemodynamic considerations are not relevant in smaller rodent 475 glomeruli with smaller perfusion volumes relative to arteriolar conductivity (see 476 introduction). 477 These VC manifolds persist in the glomerulus despite pressure changes, VC walls 478 are resistant to collapse during immersion fixation or when observed fresh at zero 479 pressure. The VC position at the vascular pole allows mesangial structural support and 480 Collagen I/III appears to provide (additional) structural integrity. The physiological 481 significance of this collapse resistance is not yet clear. 482 Collagen III has been observed in glomeruli of collagen nephropathies (7, 14)  483 with collagen III in mesangium and/or capillary walls. No report could be found of 484 Collagen I or III in mesangium of normal glomeruli and this report is the first to find 485 banded Collagen (I and/or III) in normal glomeruli close to the vascular pole. Banded 486 collagen has previously been found in kidney cortex, where 30nm fibres showed hybrid 487 labelling with Collagen I and III (13), however, the identity of VC wall banded collagen 488 remains to be confirmed by immunohistochemistry. 489 490 VCs appear to be ubiquitous in the adult kidney. We confined resin section 491 reconstructions in this study to subcapsular glomeruli to surmount any size difference 492 between subcapsular and juxtaglomerular glomeruli seen in humans and other species 493 (17,34,51,55,58) Evenso, the resin single section work shows a surprisingly similar 494 occurrence of vascular widening in 50-60% of vascular pole glomerular profiles (Fig.6), 495 implying that VCs exist in both cortical locations with similar sized VCs in both juxta-496 medullary and subcapsular glomeruli. 497 498 Afferent and efferent arterioles 499 No previous study has measured the diameter of fully opened human glomerular 500 arterioles perfusion fixed at their operating pressures. Previous human AA diameters vary 501 from 13-16µm (18) to diabetic biopsy diameters of 29µm for AA and 19µm for EA(44). 502 Other than biological variability, this range of arteriolar diameter is likely due to: volume 503 changes in tissue processing, oblique sections of vessel or low pressure fixation 504 producing collapsed profiles (for example; Tab.4 fresh AA -13.8µm; Fig.1 in ref. (45)). 505 These problems appear minimized with the fixation and resin embedding techniques of 506 this paper. 507 A correlation between afferent arteriolar diameter and mean glomerular capillary 508 area has previously been seen as consistent with loss of autoregulation (18). Here a 509 correlate of AA resistance per unit length (R AA ) did not scale with any other glomerular 510 parameter measured including R EA (Tab.3) preserving the independent autoregulatory 511 control of AA. In contrast EA resistance per unit length (R EA ) was inversely correlated 512 with V G (Fig.5D; Tab.3), and correlating remarkably with R Con at the afferent end 513 (Tab.3). Unlike AA, EA is linked in fluid dynamic terms with the Glomerulus it drains. 514 515 Conduit vessels 516 The first order afferent vessels or conduits were noted by Bowman in 1842, with 517 2 to 8 branches which visibly 'subdivide only once or twice as they advance over the 518 surface of the ball' (5). The few buried deep inside the glomerulus unseen by Bowman 519 may explains the result of 2 to 11 seen in this current study. We also confirm the luminal 520 width of these first order afferent vessels as being as wide as the efferent arteriole (21). 521 Conduit vessels show fewer branches than their efferent counterparts but branch 522 frequency increases at the start of perfusion regions often at some point on the glomerular 523 periphery (Fig.2). No previous branch data exists for these vessels however, the 524 interbranch length for all rat glomerular vessels at 26.3±24.9µm(SD) (48) is between the 525 medians, 32.8µm (conduit afferent) and 15µm (efferent) of the skewed distributions 526 found here. 527 Conduit vessels close to the AVC are embedded in mesangium, those distal to the 528 AVC have a GFB. While detailed conduit ultrastructure remains to be confirmed, no 529 aberrant GFB capillary morphology has been noted in all our studies of normal human 530 glomeruli (data not shown). It appears that conduit GFB is similar to filtration capillary 531 GFB except for the scarcity of podocyte cell bodies on the conduit GFB surface. It 532 remains to be determined if conduit podocytes are just responding to local conditions or 533 are a sub-population of conduit podocytes with the extra-long major processes necessary 534 to cover the GFB area in foot processes. 535 The GFB is known to remain intact and expand under excess pressure (25, 27) 536 and conduit vessels with a 86-100% GFB -or a sparse 0-14% mesangial attachment 537 around the circumference showed diameter expansion by 7% compared to conduit vessels 538 surrounded by and embedded in mesangium (Fig. 5A) -not enough GFB expansion to 539 explain podocyte cell body free areas on the conduit vessels but below the damaged 540 'giant capillary' inflation levels previously reported (25). Conduit inflation might be 541 expected considering the reduced podocyte coverage, thin walls and wide diameter and 542 estimates of wall forces show conduit vessels with a high proportion of GFB and low 543 mesangial attachment are the most susceptible to hoop stress of all glomerular vessels 544 (Appendix 2). This marks conduits as a target in hypertensive disease and hoop stress 545 failure has been observed in rat primary afferents (equivalent to conduits) due to 546 glomerular hypertension (with marking albuminuria and glomerulosclerosis) (26). 547 The subpodocyte space, identified under podocytes (39)  Microscopy (data not shown) and also found no evidence of VC. 558 Mammalian arterioles can widen pathologically (32), for instance, mesangiolysis 559 can remove mesangial support causing glomerular vessel aneurysms (35) but such 560 features would not be as highly conserved in shape or have an organized collagenous 561 support as seen in VC found here. Bowman also noted in the larger horse glomerulus that 562 afferent arterioles dilate on the surface prior to dividing but not in human glomeruli (5)  563 we show here that human glomerular vascular dilations are subsurface and would have 564 been invisible to Bowman. The modern conventional description merely reports that the 565 afferent arteriole branches into the glomerular capillary network (22). 566 VCs may not be present in all human glomeruli, during development, glomerular 567 capillaries arise from one dilated vessel (11) and neonate vascular widening has been 568 shown prior to the five first order afferent branches (21) although this has been ascribed 569 to a vessel remnant from the developing nephron (11). Interestingly, the glomerular 570 diameter increase in children from 112µm (birth) to 167µm (15years) (34) and VC 571 scaling with V G shows that VCs may not exist in child glomeruli which are below 160-572 180µm diameter, providing these glomeruli follow the adult glomerular correlation (Fig.  573 3B & 4C,D). Conduit vessel resistance (R Con ) also scales with V G , whether this 574 correlation continues in smaller (child) glomeruli or whether the primary afferents in 575 children even constitute 'conduit' vessels needs evaluation. Where n V is the number of vessels and r is the radius. Using r AA , r AVC , r Con , r E1 , 610 r EVC , r EA and appropriate n to calculate K, the Murray relationship breaks at the VCs and 611 the first order vessels (conduit and E1 vessels; Fig.6B), where daughter vessels do not 612 have the same Murray constant as parent vessels. 613 This is an exception to Murray's Lawa plenum/manifold exception, where flow 614 distribution from a single arteriole provides a high pressure distributive flow into many 615 glomerular lobes in a short distance. An estimate of K values for second order afferent 616 vessels (A2 in 2 glomeruli) showed that K may return to the value predicted by the 617 afferent arteriolar radius after skipping the VC and conduit vessels (Fig.6B). Other 618 Murray's law exceptions occur where a higher surface area is required in the exchange 619 vessels of an organ, for instance alveolar capillary networks (59). 620 The possible mechanisms producing a set of vessels following Murray's law 621 includes an endothelial transducer triggering remodelling after a shear force threshold 622 was exceeded(46). Altering the threshold could induce the vessel diameter changes seen 623 here. However, the Murray relationship requires laminar flow through vessels and the 624 haemodynamic flow will be complex from an afferent arteriole into an ellipsoidal 625 vascular chamber with several outlets. 626 627 VC haemodynamics 628 If glomerular volume is used as a measure of perfusion capacity, it rises and falls 629 along with the size of the AVC and the EVC (Fig.4 C&D). Larger AVCs feed more 630 blood to larger glomerular filtration regions and thence to larger EVCs. As the size 631 increases the resistance of the conduit vessels, E1 and EA (not AA) falls to accommodate 632 the flow (vessels get wider in proportion to Poiseuille flow) (Fig.5 C&D). All of the 633 major vessels of the human glomerulus past the afferent arteriole are linked in some way 634 in terms of flow and capacity (Tab.3). How would flow progress from laminar flow in an 635 afferent arteriole through the AVC to the conduit vessels? And similarly from efferent E1 636 vessels through EVC to the efferent arterioles? 637 A clue to VC flow characteristics comes from the kinks and bends in AAs. One 638 constant feature of the glomeruli analysed is the bend as the afferent arteriole enters the 639 AVC. These bends can be readily seen in the glomeruli of figures 1, 2A and 2B 640 (supplemental videos 2a and 2b) and showed an average 60º deviation from a straight 641 path. The fluid flow at a bend in a channel is known to induce vortices (49), we 642 hypothesize that the summation of all bends in an afferent arteriole (i.e. see bend from 643 interlobular -AA junction in Fig.1) could induce a single major vortex in the AVC 644 possibly aiding distributive flow centrifugally into conduit vessels. 645 If such a vortex with its axis in the midline of the AVC adopts the properties of a 646 "rigid-body" or "rotational" vortex, then the pressure at the AVC edge at the conduit 647 vessel openings would depend both on the hydrostatic pressure and the dynamic pressure 648 (set by the angular momentum of the moving fluid -½ρω 2, where ρ=density; ω = angular 649 velocity). Crucially however the dynamic pressures within this form of vortex are 650 uniform (3). 651 We speculate that in health the AVC and the complex (vortical) fluid movement 652 within it, may ensure a uniform driving pressure into the conduit vesselsmaximising a 653 uniform distribution of flow to each of the glomerular lobules. The mesangial backed conduit vessels (Fig.3a The conduit vessels away from the AVC are connected to mesangium only on a 797 small part of their circumference the rest being normal GFB and GBM (Fig.3a, Figure 2A&B. Serial resin sections through a glomerulus. Selected light micrographs 996 from 2 complete 1µm serial section series to show the route blood takes from an afferent

Footnote 1 1145
[Afferent arteriole conductance estimated from the 4 th power of vessel radii (mice, r = 5-1146 6.5µm,(16, 28, 29); rats, r = 7-9.5µm(12, 24, 54, 57) Figure 2A&B. Serial resin sections through a glomerulus. Selected light micrographs from 2 complete 1µm serial section series to show the route blood takes from an afferent arteriole (AA) into an afferent vascular chamber (AVC) leading into conduit vessels (Con) of high capacity and few branches. At the other end of the microcirculation many branching efferent 1 st order vessels (E1) drain into a smaller efferent vascular chamber (EVC) leading to an efferent arteriole (EA). Serial section numbers at bottom left. Scale bar 100µm in micrograph of section 254 or 198 (see Supplemental video S2A and S2B for glomerular image stacks of Fig.2A (Fig.4c&d). VC shrinkage in the radial direction would reduce the diameter and VC volume until it was a continuation of the attached arteriole (Fig.4 C&D).  where r is radius, n V is vessel number; see text) was calculated for the afferent and efferent arteriolar tree leading through the VCs and thence into the 1 st order vessels (Con and E1). In 2 glomeruli K was calculated for 2 nd order vessels. The Murray relationship of equal K at each vessel level is absent in the AVC, EVC and conduit vessels.   Table 3. Vascular resistance and capacity relationships. Significant correlationships (8 out of 21) between 7 variables measured in human glomerular initial vasculature. Correlates of vascular resistance for afferent arterioles (R AA ), Conduit vessels (R Con ), first order efferent vessels (R E1 ) and efferent arterioles (R EA ) were compared with each other and with AVC volume (V AVC ) glomerular volume (V G ) and EVC volume (V EVC ). + positive correlation,negative correlation; * = P < 0.05, ** = P ≤ 0.01; **** = P ≤ 0.0001; § higher significance with outlier removed.