J Clin Pathol 1981 ;34:1197-1206

New ideas on the anatomy of the kidneyDB MOFFAT

From the Department of Anatomy, University College, Cardiff

A discussion of new ideas on the anatomy of thekidney presents some difficulties, firstly in definingwhat is meant by "anatomy" and secondly in select-ing which particular new ideas to present. Bowman,'perhaps, answered the first question not only bydescribing the structure of the glomerulus andtubules but also by discussing the functionalsignificance of his findings. Nowadays the scope ofsuch functional anatomy needs to be expanded toinclude biochemical anatomy so that the choice ofsubjects becomes even more difficult. I have chosentwo features of the glomerulus-namely the glomeru-lar filter and the mesangium, because these are ofsuch importance in understanding the pathology ofthe kidney-and two features of the renal medulla-its blood supply and its interstitial tissue, becausethese offer interesting subjects for speculation onpossible future developments.

The glomerular filter

From an anatomical point of view, this consists ofthe capillary endothelium, the glomerular basementmembrane (GBM) and the filtration slit between thebases of the foot processes of the epithelial cells orpodocytes. Until recently, the clearance values ofmacromolecules were interpreted, not entirelysuccessfully, in terms of a mechanical filtering effectbut it has become increasingly evident that thebiochemical make-up of the filter, including thecomposition of the GBM and of the cell coat of theendothelial and epithelial cells, is of fundamentalimportance.

CAPILLARY ENDOTHELIUMThe capillary endothelium offers no anatomicalbarrier to the passage of even the largest molecules.It is a fenestrated endothelium but the diameters of thefenestrations are between 50 and 100 nm, very muchlarger than any of the macromolecules in the plasma.Also, unlike the fenestrations in other capillaries,the openings are not closed by a membrane (exceptin the developing glomerulus) so that the capillaryblood comes into direct contact with the GJBM.The endothelial cells have, however, a negativelycharged cell coat which is up to 12 nm in thickness.

GLOMERULAR BASEMENT MEMBRANEThe human GBM is 250-350 nm thick. In electronmicrographs three layers can be distinguished, acentral lamina densa (LD) with a less dense laminarara interna (LRI) on the endothelial side and alamina rara externa (LRE) on the side of the podo-cytes. It has a fibrillar structure2 but no "pores" arevisible, and when large molecules such as ferritinpass through it no tracks can be seen. Space does notpermit a full account of the chemical make-up of theGBM but the most important features are that it hascollagenous and non-collagenous components, isrich in carbohydrates, and contains glycosamino-glycans including sialic acid (2%) and heparansulphate. It is therefore strongly anionic and hascharacteristic staining properties. Thus it stainsintensely with periodic acid-Schiff (PAS) and withcationic stains such as Ruthenium red, alcian blueand colloidal iron. Suitable modifications of thesestaining methods make it possible to demonstratethe actual anionic sites in the GBM. Kanwar andFarquhar3 administered cationised ferritin intra-venously to rats followed by fixation by perfusion,and they also used Ruthenium red added to thefixative as a cationic label. They found clusters ofelectron-dense particles in both laminae rarae andin the mesangial matrix. These formed a fairlyregular lattice-work with the centres of the clustersabout 60 nm apart. When both cationic markerswere used together, they localised at the same pointsin the GBM. Recent work by the same authors hasshown that these anionic sites consist of glycosamino-glycans, mainly heparan sulphate. This was showncytochemically by appropriate enzyme digestion invivo4 and by cellulose acetate electrophoresis of theisolated GBM.5 The importance of this highlycharged polymer in glomerular permeability will bediscussed later.

EPITHELIAL CELLS (PODOCYTES)The epithelial cells are arranged around the capillar-ies, and their foot processes interdigitate to give theglomerular filtration membrane its characteristicappearance. The foot processes are partly embeddedin the GBM, and the space between their bases isthe filtration slit or slit pore, and has a width of

1197

1 **

copyright. on N

ovember 7, 2020 by guest. P

rotected byhttp://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.34.11.1197 on 1 Novem

ber 1981. Dow

nloaded from

1198

20-30 nm. It is closed by a membrane (the slitdiaphragm) which is clearly seen when sectionedtransversely; sections tangential to the GBM, how-ever, show that the diaphragm has a complexstructure.6 7 After fixation with a tannic acid-glutaraldehyde fixative, a central bar can be seenrunning along the slit, with alternating light anddark areas on either side of it, so that the whole slitresembles a zipper. The light areas may representopenings and measure approximately 4 x 14 nm-about the size of the albumin molecule.The outer surface of the podocytes is covered by a

strongly anionic cell coat which is sufficiently thick(15-80 nm) to fill the filtration slit completely so thatit covers the outer surface of the slit diaphragm. Thecell coat, like the GBM, stains strongly with variouscationic dyes. The foot processes, when cut trans-versely, have a characteristic "elephant's foot"shape, being much wider at the base than at thesummit but after some types of fixation their lateralsides are parallel along their whole length so that it ispossible that in vivo the cell coat fills the wholespace between the foot processes.2 8 The presence ofthe polyanionic cell coat may well be responsible formaintaining the shape and interrelationship of thefoot processes by their mutual repulsion, as will bediscussed later.The podocytes themselves are phagocytic, as has

been shown by a number of workers, and theircytoplasm contains fine filaments that resemblemyofilaments. They have been shown to containactin and heavy meromysin and may therefore becontractile.9 The cytoplasm also contains a largenumber of microtubules. These, in other cells, havevarious functions including that of forming askeleton to maintain the shape of the cells. Tyson'0and Andrews" destroyed the microtubules in ratpodocytes with vinblastine and found changes inpodocyte shape, although the foot processes them-selves were not affected.

FILTRATION PROPERTIES OFI THEG LOMERULUS

During the last few years, knowledge of the processof glomerular filtration has increased rapidly as newmethods of study have become available. Untilrelatively recently, the filtration properties of theglomerulus were thought to be mainly a function ofmolecular size and, to some extent, of shape, andwere studied by clearance methods and electronmicroscopy, using a whole spectrum of moleculeswith molecular weights ranging from 12 000 (equinecytochrome C) to 480 000 daltons (ferritin). Theseexperiments, reviewed by Rennke and Venkata-chalam in 1977,12 led to the idea that while some

Moffat

large molecules are held up by the GBM, smallermolecules are able to penetrate as far as the slitdiaphragm or beyond. The strongly anionic natureof the components of the filtering membranesuggested that molecular charge as well as sizemight have an effect on the filtration of macro-molecules and this was shown to be true in 1975 byChang et al,13 who compared clearance values fordextran and dextran sulphate, and by Rennke et al,'4who studied the filtration of ferritins with differentisoelectric points by electron microscopy. Theresults of such studies by these and other workersshowed that while molecular size is obviously animportant factor, cationic molecules penetrate thefiltration membrane very much more freely thanneutral or anionic molecules of a similar molecularweight (see review by Karnovsky).8 The mostrecent contribution to this field is that of Kanwaret all5 who followed up their demonstration ofheparan sulphate in the GBM by studying the effectof its removal by heparinase, using native ferritin asa tracer. In control animals the ferritin did notpenetrate beyond the lamina rara interna to anygreat extent but after heparinase treatment ittraversed the GBM freely and was found in theurinary space.A review of the filtration properties of the glom-

erulus is not complete without reference to haemo-dynamic factors although this is not strictly ananatomical problem. However, the laboratory ratsin which this has been studied do have an interestingand useful feature which might well be classified as a"new idea in anatomy." Munich-Wistar rats, whichoriginated in Munich but are now found in everylaboratory in which glomeruli are studied, have asmall number of glomeruli on the surface of thekidney where they can be seen quite easily and areavailable for micropuncture. Ryan and Karnovsky"'have studied the filtration of endogenous albuminby an immunoperoxidase technique. The visibleglomeruli were fixed during normal activity bydripping the fixative on to the surface of the kidney.Under these conditions it was found that albuminwas held up at, or just beyond, the endothelialfenestrations. If, however, the kidneys were fixed byimmersion, or if the circulation was interruptedbefore fixation, albumin was found in the GBM andin the urinary space. Similar results were found usingcatalase and endogenous IgG (MW 240 000 and150 000 daltons respectively), although thesematerials did not reach the urinary space."7

Finally, for the sake of completeness, it must bementioned that other haemodynamic factors alsoare involved and the fractional clearance of largemolecules is partly dependent upon the determinantsof the glomerular filtration rate."'

copyright. on N

ovember 7, 2020 by guest. P

rotected byhttp://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.34.11.1197 on 1 Novem

ber 1981. Dow

nloaded from

New ideas on the anatomy of the kidney

OTHER FUNCTIONS OF GLOMERULARPOLYANIONFurther aspects of the importance of glomerularpolyanion can be studied by removing or blocking it,thus causing the loss of the fixed negative charge.This has been done, for example, by perfusion withcations such as protamine sulphate,'9 by treatment ofkidney slices with neuraminidase which removes

sialic acid20 and by the culture of isolated glomeruli inthe presence of polycations.21 These processes allhave effects which mimic those of human andexperimental nephroses including cell swelling, theloss of separate and distinct foot processes, theformation of junctions between foot processes andchanges in the slit diaphragm. Numerous otherworkers have studied the glomeruli in human andexperimental glomerulopathy and have shown a

correlation between loss or fusion of foot processes,decrease of polyanion, and proteinuria.8 Theseobservations suggest that the foot processesmay be held apart by the electrostatic repulsion oftheir fixed negative charges and that the loss ofpolyanion will lead to the "fusion" of foot processesand will increase the permeability of the glomerularfilter, thus leading to proteinuria. Reeves et a!22 haverecently produced some interesting evidence infavour of this hypothesis in their study of glomerulardevelopment. Examination of 2 to 5 day-old ratkidneys, in which glomeruli at all stages of develop-ment can be studied in the nephrogenic zone, showedthat in the early stages of development the poly-anionic cell coat of the epithelial cells only stainsfaintly and only extends down to the occludingjunctions that unite the foot processes at this time.The development of normal, separated foot pro-cesses coincides with the development of a full cellcoat around the entire lateral surfaces of the pro-cesses.In addition to its probable function in main-

taining normal filtration slit architecture, the glomer-ular polyanion may also be of importance in attach-ing both endothelial and epithelial cells to the GBM.Kanwar and Farquhar23 perfused rats with neura-minidase for 30-60 min before fixation for electronmicroscopy. The endothelial cells and the footprocesses became detached from the GBM and theirnormal staining with colloidal iron was lost. Freesialic acid was found in the bladder urine. After along perfusion, large areas of theGBM were denudedof cells and the mesangial cells became detachedfrom the matrix. The GBM, however, showednormal cationic probe binding, and removal of itsheparan sulphate had no effect on the attachment ofcells. It was surmised that the attaching agent mightbe fibronectin, a large glycoprotein molecule con-taining sialic acid, or possibly laminin, which also

contains sialic acid. In a footnote, the authors statethat they have succeeded in locating fibronectin inthe glomerulus, and it has also been produced in vitroby cultured glomerular cells.24 Recently, however,Madri et at25 found fibronectin only in the mesangialregion of mouse kidneys while laminin occurred inthe mesangium as well as in the lamina rara internaand the adjacent area of the lamina densa of theglomerular basement membrane.

The mesangium

The mesangial region, the central core of tissue inthe glomerulus, consists of an extensive matrix inwhich irregular cells with numerous processes areembedded. At one time the existence of these cellswas doubted, but with the advent of electronmicroscopy and, more recently, as a result of experi-mental studies, it has become evident that themesangium is a dynamic and important componentof the glomerulus which is the focus for manydisease processes. In order to explain the relationbetween the mesangium and the GBM it will firstbe necessary to say a little more about the GBMitself. It is a compound membrane, being formed bythe basement membranes of both the endotheliumand the epithelium which fuse to produce the three-layered GBM when the developing glomerulus is atthe S-shaped body stage.22 Huang26 has recentlyshown that these two components can be distin-guished by treatment with 5M guanidine hydro-chloride before fixation. The endothelial componentand the mesangial matrix are then found to be lesselectron-dense than the epithelial component. Afterthis treatment it can be seen that the epithelialbasement membrane is a continuous layer thatpartially covers each capillary tuft by being wrappedaround it and being invaginated between individualcapillaries. It does not, however, completely en-circle any capillaries, being absent over the regionof the capillary wall that is adjacent to the mesan-gium. In this region, therefore, the endothelialbasement membrane is continuous with the mesan-gial matrix so that here there is a low permeabilityroute from the capillary lumen to the mesangium.The mesangial cells are scattered in the matrix,

leaving channels of matrix between them which mayserve as one of the routes for the passage of materialsthrough the mesangium. The cells appear to teactive, for they contain a fairly prominent Golgizone, a good deal of endoplasmic reticulum, glyco-gen and mitochondria. They also contain micro-filaments which are probably contractile (see below).They are, in fact, very similar to smooth musclecells and this similarity is emphasised by theircontiguity with the lacis cells of the juxtaglomerular

1199

copyright. on N

ovember 7, 2020 by guest. P

rotected byhttp://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.34.11.1197 on 1 Novem

ber 1981. Dow

nloaded from

1200

apparatus (JGA) and the smooth muscle cells of theafferent and efferent arterioles.27 28 Christensen29examined the JGA from 49 human glomeruli andfound contiguity with the mesangium and the laciscells, with a gradual transition of cell types from oneto the other. A number of investigators30 have foundthat the cells of the mesangium and of the JGA areinterconnected by gap junctions so that they mayact together as a syncytium, their contractionperhaps affecting the glomerular blood flow.

CONTRACTILITYThere seems little doubt that the mesangial cells arecontractile. Their microfilaments have been men-tioned already and immunoelectronmicroscopicstudies have demonstrated the presence of acto-myosin in the cells and also in the matrix of glomeruliboth in situ and in cultures.31-33 Whole glomeruli incultures have been seen to contract34 while morerecently Mahieu et al35 and Ausiello et at36 have beenable to show contraction of presumed mesangialcells from cultured rat glomeruli, after stimulationwith angiotensin II, noradrenaline and argininevasopressin. Osborne et al37 have shown that afterthe injection of 3H angiotensin II into rats, radio-activity was localised mainly in the mesangial cells.It seems likely that the angiotensin interacts withsome specific site within the cells, although otherexplanations are possible.

FATE OF LARGE MOLECULESThe mesangial region is of particular interest inrenal disease because of its ability to take up anddispose of large molecules, particularly immunedeposits, so that the rates of entry and exit of suchmolecules have been the subject of much investiga-tion. The molecules are presumed to enter themesangial region mainly via the "bare area" ofthe capillary wall where the main GBM is absent.The factors that control this afferent limb have beenadmirably reviewed by Michael et at38 and only theanatomical aspects of the processes will be discussedhere.Many workers have studied the passage of large

molecules into the mesangium by light and electronmicroscopy, using such tracers as colloidal gold,dextrans, ferritin and colloidal carbon. The fate ofthese substances depends on the molecular size butin all cases the particles pass into the mesangialmatrix and some, at least, are taken up by themesangial cells. Recent works have shown theimportance of the contiguity of the mesangium withthe JGA. Elema et al,39 for example, used colloidalcarbon and showed that the uptake of particles bythe mesangium was maximal at 32 h after the injec-tion and that the carbon first entered the channels

Moffat

between the cells. Thence, it was taken up by themesangial cells and was apparently passed on fromcell to cell to the lacis region where it was found 2-7wk after the injection.

There is a good case for classifying the mesangiumas part of the reticuloendothelial (RE) system butwith a time-lag. Thus, the uptake of injected materialby the mesangium lags behind that of the RE systemand, for this reason, is dose-dependent. A small doseis taken up relatively avidly by the RE system andaccumulation in the mesangium will not occurunless the RE system is partly saturated and theblood level rises. Similarly, after the material hasreached the mesangium, it subsequent disappearancelags behind the fall in blood level.The size of the molecules also affects the fate of

particles in the mesangium. Many workers havenoted that very large molecules such as immunecomplexes and aggregated proteins tend to be foundin the matrix with relatively little phagocytosis. Thisis not invariably so, however, and, since space doesnot permit a full discussion of the subject, two of themost recent papers will serve to illustrate the prob-lems. Takamiya et al40 compared the effects ofinjections of native ferritin and ferritin complexedwith IgG and albumin. There was a very clearlydose-dependent accumulation of ferritin in themesangium, the conjugates requiring a much smallerdose than native ferritin to produce maximalmesangial concentration. Native ferritin was clearedalmost completely in 14 days but the conjugate wasstill very evident in the mesangium, mainly withinthe cells. They also showed that, when the conjugatehad disappeared from the blood but was still presentin the mesangium, injected rabbit antiserum toferritin became localised in the mesangium, showingthat the mesangial cells are accessible to circulatingantigens. Lee and Vernier4' used aggregated humanalbumin as a markerand found it in the matrixwithin40 min of administration, with maximal accumula-tion at 8 h. There was some uptake by mesangialcells but there always appeared to be more in thematrix channels than within the cells. The aggre-gated albumin was also found between the laciscells and within the walls of arterioles as early as4 h after injection.The fate of mesangial deposits after reaching the

JGA region or after phagocytosis by the cells is notknown but it is worth mentioning a recent investi-gation by Albertine and O'Morchoe42 who studiedthe cortical lymphatics in the dog kidney. In con-trast to a number of previous workers, they wereable to demonstrate an intralobular distribution oflymphatics, including some in close relation to theJGA (which may, incidentally, explain the highrenin concentration in therenal lymph). It is possible,

copyright. on N

ovember 7, 2020 by guest. P

rotected byhttp://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.34.11.1197 on 1 Novem

ber 1981. Dow

nloaded from

New ideas on the anatomy of the kidney

therefore, that the lymphatic system may receive theend products of mesangial processing.

The intrarenal blood vessels

The distribution of the main arteries has been welldocumented, as has that of the smaller vessels invarious animals,43-45 but it is only recently that adetailed study of the small vessels in the humankidney has been made using modern techniques.46 47The origin of the afferent arterioles and the distri-bution of the efferent arterioles are very similar tothose of animals, and Beeuwkes classified theefferents into 10 types according to their position inthe cortex and their distribution. He has also per-fected an elegant technique in which both vesselsand tubules are injected with coloured siliconerubber (Microfil, Canton Biomedical Products) sothat the vascular-tubular relationships can be studied.In the subcapsular cortex, the proximal and distaltubules ofany one nephron are usually supplied in partby the efferent vessel of the corresponding glomerulusbut this association is not as marked as in thecanine kidney, which Beeuwkes has also studied. Inthe middle and inner zones of the cortex the con-voluted tubules and the loops of Henle are almostcompletely supplied by the efferents of otherglomeruli and, in general, each nephron receivesblood from many glomeruli, each efferent supplyinga short segment. This lack of a 1 :1 efferent arteriole:

nephron relationship throws some doubt on certaintheories of glomerulotubular balance based onhydrostatic and oncotic capillary pressures.

BLOOD SUPPLY OF THE MEDULLA

The blood supply of the human medulla has beenstudied in four human kidneys.46 The main bloodsupply is derived from the efferent arterioles of thejuxtamedullary glomeruli. These descend into themedulla by looping around the arcuate vessels,giving small branches to the loose capillary plexus inthe subcortical region (also known as the outerstripe of the outer medulla). Near the arcuates, eachefferent arteriole breaks up abruptly into a largenumber of descending vasa recta (DVR) which are ofalmost the same size as the parent vessel (Fig. 1).The number of DVR from each efferent arteriolevaries between 12 and 25 but, since each bundle ofDVR receives branches from a number of efferents,the number in each bundle may be as many as 50.As the bundles traverse the outer medulla, vesselsleave the periphery of the bundle to supply the localcapillary plexus, while rather more DVR leavethe bundles as they enter the inner medulla. Thebundles therefore soon lose their identity in the innermedulla but some of the DVR continue their course,unbranching, to various levels in the papilla beforebreaking up into capillaries.From the elongated capillary plexus of the inner

medulla, venous vessels collect tributaries and form

n:er obutarvessel_

Fig. 1 Human kidney,Microfil injection. Only

--q -- _ _ the descending vasa-~ recta have been injected,

the ascending vasa||VascuLar ,- w. ;recta remainingi undLes unfilled.

1201

copyright. on N

ovember 7, 2020 by guest. P

rotected byhttp://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.34.11.1197 on 1 Novem

ber 1981. Dow

nloaded from

1202

ascending vasa recta (AVR) and in the outer medullathese become closely intermingled with the DVR toform vascular bundles. These are a very prominentfeature of the outer medulla and since the numbersof descending and ascending VR are approximatelyequal in the human kidney, each bundle maycontain up to 100 vessels. There is no evidence thatthe descending limbs of the loops of Henle areincorporated into the vascular bundles as they are inmany animals. The bundles receive additional AVRthat drain the capillary plexus of the outer medullaand they then drain into the arcuate or interlobularveins.

FINE STRUCTURE OF THE VASCULAR

BUNDLESIn the upper part of the vascular bundles the DVR,like the juxtamedullary efferent arterioles, have alayer of smooth muscle in their walls (Fig. 2) andthey are accompanied by non-medullated nervesthat lie in close proximity to the vessel walls. Inbetween these relatively thick-walled DVR are thewide AVR whose wall is composed only of anextremely thin fenestrated endothelium. In thedeeper parts of tho outer medulla the descending

Moffat

vessels lose their smooth muscle and with it theiraccompanying nerves. They are surrounded byperivascular cells; these are thin cells that arewrapped around the descending vessels and have alayer of microfilaments on the side facing theDVR and a row of micropinocytotic vesicles on theside adjacent to the AVR. In many places the peri-vascular cells are absent so that the descending andascending vessels come into very close contact.The structure of the vasa recta thus suggests aregulatory function in the upper part of the vascularbundles and a countercurrent exchange system lowerdown.An accessory blood supply to the papilla is

derived from a number of caliceal vessels that archover the fornix to supply the most superficiallayers of the papilla. The functional significance ofthis blood supply is difficult to assess but it is interest-ing that Heaton and Bourke48 have described a casein which acute arteritis of these vessels appeared tobe a cause of papillary necrosis. They are certainlyvulnerable to ruptures of the calicine fornices suchas occur when the calices are strongly distended forany reason.

There is a great deal of experimental evidence for

DeSCerci3;r1,4 4.eredfrcz,;lf

-1Fig. 2 Human kidneyvascular bundle in cross

AscenFj section. One of the descendingr'ec ti. vasa recta (arro wed) shows

'~>2~ pathological changes ofunknown aetiology x 1220.

copyright. on N

ovember 7, 2020 by guest. P

rotected byhttp://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.34.11.1197 on 1 Novem

ber 1981. Dow

nloaded from

1203New ideas on the anatomy of the kidney

the presence of a countercurrent exchange system inthe vascular bundles which can result in the trappingof solutes in the medulla or their exclusion from it.The exclusion of water is obviously important in themaintenance of the osmotic gradient in the medullabut perhaps of more clinical interest is the possibleeffect of such exchanges on the concentration ofdrugs in the medulla. Phenazone (antipyrine), forexample, is said to be excluded from the medulla, acurious feature in the light of the ability of this drugto produce papillary necrosis.

The interstitial tissue of the medulla

Finally, I would like to discuss a much neglectedsubject, namely the interstitial tissue of the medulla.This consists of a copious matrix, rich in glyco-saminoglycans, in which are embedded all thetubules and vessels, together with a number ofinterstitial cells which are associated with the metab-olism of various prostaglandins. The matrix itselfis of great interest because it is through this mediumthat all exchanges of water and solute between thetubules and vessels have to take place. One of thepuzzling features of the renal medulla is that it hasno lymphatics.49 It is therefore necessary to considerhow the various materials that may enter theinterstitium are removed. Most of the water thatcomes (mainly) from the collecting ducts probablyenters the ascending vasa recta, but the fate of theextravascular protein50 and of any tubular contentsthat may enter the interstitium in pathologicalconditions such as intrarenal reflux need to beinvestigated. In this respect the work of Schmidt-Nielsen51 is of great potential interest. She injectedalcian blue into the papilla 50 mm from the tip andfound that the stain moved up channels around thecollecting ducts and beneath the papillary epitheliumat about 200 pm/s. The channels could be identifiedin sections both by light and electron microscopy andwere about 0-5 ,um in width. The dye was later foundin the interstitial tissue in the outer medulla and inthe cortex.

In an effort to discover how extravascular materialsmight be removed from the medulla, the clearanceof ferritin and Imferon from the interstitium hasbeen investigated.52 These were introduced into themedulla of rat kidneys by reflux up the ureter, whichcauses forniceal rupture, by direct injection with amicroneedle; or by intravenous injection. Nolymphatics were found by light or electron micros-copy. The injected material was taken up by phago-cytic cells in the interstitium, in which it could berecognised by means of the Prussian blue reaction foriron. In the inner medulla the iron-containingcells were arranged in isolated longitudinal chains,

T

X.0', 4*, *dK*~

Fig. 3 Rat, inner medulla, 20 minbafter the injectionof a small quantity of Imferon into the interstitialtissue near the tip of the papilla. The arrows indicateone of the chains of iron-laden cells. The tip of thepapilla is towards the bottom. Prussian blue reactionx 190.

usually widely separated, with no cells in the inter-vening spaces (Fig. 3). The cells were inconspicuousnear the tip of the papilla but became larger andmore densely packed with iron-containing granulesas the chains of cells were followed towards theouter medulla. In the outer medulla itself, clumps ofvery large cells were found in relation to distendedblood vessels packed with blood cells (Fig. 4),probably venous in origin. Here, some of the cellsappeared to be passing through the vessel walls andmany were found within the vessels (Figs. 4 and 5).Although these cells have been identified by electronmicroscopy it has not yet been possible to confirmthat they pass through the vessel walls or to dis-cover the cause of the vessel distension-it may bethat the vessels are partially obstructed by theclumpsof hugephagocytic cells within the lumen. Itisdifficult, also, to understand the disposition of thechains of cells in the inner medulla but the channels

copyright. on N

ovember 7, 2020 by guest. P

rotected byhttp://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.34.11.1197 on 1 Novem

ber 1981. Dow

nloaded from

1204

C ffiA.,I A

A~~~~~~~~~~~~

IA*$\ D M izte c~~~~'o ese

41,~A

Xi~ ~*-e S i*-i ^

Fig. 4 Rat, outer medulla, 45 min after the injectionof a small quantity offerlritin into the papilla. Cluster.sof cells lie in relation to a distended vessel (arrows) andsome appear to be passing into the vessel. Prus.sian hluereaction x 190.

Moffat

described by Schmidt-Nielsen offer a possibleexplanation. If the ferritin passes from the inter-stitium into one or more of the channels around thecollecting ducts and thence towards the outermedulla, phagocytic cells along this pathway mightpick up the particles along the route, thus causingthe linear distribution. It is well known that phago-cytosis is inhibited by a high osmolarity so that theamount of iron picked up by the cells would increasetowards the base of the papilla.

TAMM-HORSFALL PROTEINAn interesting new tool for investigating tubulardamage with extravasation of tubular contents intothe interstitium is provided by Tamm-Horsfallprotein, a peculiar product of renal tubular epi-thelium. This glycoprotein, which is a constituent ofnormal urine, has recently attracted a good deal ofinterest in relation to renal disease. A number ofprevious authors have studied the distribution inanimals53 and recently the human kidney has beenstudied by Sikri et al.54 The latter authors found theprotein in relation to the plasma membranes of thethick ascending limb of the loop of Henle and of thedistal tubule with the notable exception of the cellsof the macula densa, a distribution very similar tothat of the animals which have been studied. Apartfrom its possible physiological importance, the mostinteresting feature of Tamm-Horsfall protein topathologists is its role in cast formation and itscontribution to extratubular extravasations. Whileits distribution is normally restricted to the cellsmentioned above and to the urine, when the tubules

Fig. 5 Rat, outer medulla,one hour after the injection

.4> .f;.$^r of Imferon into the papilla.0e<5t. A distended vessel contains.* an iron-laden cell. Prussian

blue reaction x 480.

copyright. on N

ovember 7, 2020 by guest. P

rotected byhttp://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.34.11.1197 on 1 Novem

ber 1981. Dow

nloaded from

New ideas on the anatomy of the kidney

are damaged it is found in the surrounding tissuesand thus acts as an indicator of the site of damage.It has been found in the interstitium in medullarycystic disease and in chronic pyelonephritis whilesimilar deposits have been found in experimentalintrarenal reflux in pigs.55 The possible effects ofthis "foreign" protein outside the tubules will, nodoubt, be the subject of much further study.

References

Bowman W. On the structure and use of the Malpighianbodies of the kidney with observations on the circulationthrough that gland. Philos Trans R Soc Lond (Biol) 1842;132 :57-80.

2 Latta H, Johnston WH, Stanley TM. Sialoglycoproteinsand filtration barriers in the glomerular capillary wall.J UltrastructRes 1975;51:354-76.

3Kanwar YS, Farquhar MG. Anionic sites in the glomer-ular basement membrane. J Cell Biol 1979 ;81 :137-53.

4Kanwar YS, Farquhar MG. Presence of heparan sulfatein the glomerular basement membrane. Proc Natl AcadSci USA 1979;76:1303-7.

5 Kanwar YS, Farquhar MG. Isolation of glycosamino-glycans (heparan sulfate) from glomerular basementmembranes. Proc Natl Acad Sci USA 1979;76:4493-7.

6 Rodewald R, Karnovsky MJ. Porous substructure of theglomerular slit diaphragm in the rat and mouse. J CellBiol 1974;60:423-33.

7Karnovsky MJ, Ryan GB. Substructure of the glomerularslit diaphragm in freeze-fractured normal rat kidney.J Cell Biol 1975 ;65 :233-6.

8 Karnovsky MJ. The structural bases for glomerularfiltration. In: Kidney disease-present status. IAPmonograph No 21. Baltimore: Williams and Wilkins Co,1979.

9 Trenchev P, Dorling J, Webb J, Holborow EJ. Localizationof smooth muscle-like contractile proteins in kidney byimmunoelectron microscopy. JAnat 1976 ;121 :85-95.

0 Tyson GE. Scanning electron microscopic study of theeffect of vinblastine on podocytes of rat kidney. Vir-chows Arch B [CellPathol] 1977;25:105-15.

Andrews PM. The effect of vinblastine-induced micro-tubule loss on kidney podocyte morphology. Am JAnat 1977;150:53-62.

12 Rennke HG, Venkatachalam MA. Structural determinantsof glomerular permselectivity. FedProc 1977 ;36:2619-26.

13 Chang RLS, Deen WM, Channing RR, Brenner BM.Permselectivity of the glomerular capillary wall: IIIRestricted transport of polyanions. Kidney Int 1975;8:212-8.

14 Rennke HG, Cotran RS, Venkatachalam MA. Role ofmolecular charge in glomerular permeability. Tracerstudies with cationised ferritins. J Cell Biol 1975;67:638-46.

15 Kanwar YS, Linker A, Farquhar MG. Increased perme-ability of the glomerular basement membrane to ferritinafter removal of glycosaminoglycans (heparan sulfate)by enzyme digestion. J Cell Biol 1980;86:688-93.

6 Ryan GB, Karnovsky MJ. Distribution of endogenousalbumin in the rat glomerulus: role of hemodynamicfactors in glomerular barrier function. Kidney Int 1976 ;9:36-45.

17 Ryan GB, Hein SJ, Karnovsky MJ. Glomerular perme-ability to proteins. Effects of hemodynamic factors on

the distribution of endogenous immunoglobulin G and

exogenous catalase in the rat glomerulus. Lab Invest1976;34:415-27.

18 Brenner BM, Bohrer MP, Baylis C, Deen WM. Determin-ants of glomerular permselectivity: insights derived fromobservations in vivo. Kidney Int 1977;12:229-37.

9 Seiler MW, Rennke HG, Venkatachalam MA, Cotran RS.Pathogenesis ofpolycation-induced alterations ("fusion")of glomerular epithelium. Lab Invest 1977;36:48-61.

2o Andrews PM. Glomerular epithelial alterations resultingfrom sialic acid surface coat removal. Kidney Int 1979;15 :376-85.

21 Norgaard JOR. Experimentally induced ultrastructuralchanges in epithelial cells of isolated glomeruli. LabInvest 1979 ;41 :224-36.

22 Reeves WH, Kanwar YS, Farquhar MG. Assembly of theglomerular filtration surface. Differentiation of anionicsites in glomerular capillaries of newborn rat kidney.J Cell Biol 1980;85:735-53.

23 Kanwar YS, Farquhar MG. Detachment of endotheliumand epithelium from the glomerular basement membraneproduced by kidney perfusion with neuraminidase. LabInvest 1980;42:375-84.

24 Oberley TD, Mosher DF, Mills MD. Localization offibronectin within the renal glomerulus and its produc-tion by cultured glomerular cells. Am J Pathol 1979;96:651-62.

25 Madri JA, Roll FJ, Furthmayr H, Foidart J-M. Ultra-structural localization of fibronectin and laminin in thebasement membranes of the murine kidney. J Cell Biol1980;86:682-7.

26 Huang TW. Basal lamina heterogeneity in the glomerularcapillary tufts of human kidneys. J Exp Med 1979;149:1450-9.

27 Barajas L. The ultrastructure of the juxtaglomerularapparatus as disclosed by three-dimensional recon-structions from serial sections. J Ultrastruct Res 1970 ;33:116-47.

28 Zamboni L, DeMartino C. A re-evaluation of the mesan-gial cells of the renal glomerulus. Zeitschrift fdr Zell-forschung 1968;86:364-83.

29 Christensen JA, Bjaerke HA, Meyer DS, Bohle A. Thenormal juxtaglomerular apparatus in the humankidney. Acta Anat (Basel) 1979;103:374-83.

30 Taugner R, Schiller A, Kaissling B, Kriz W. Gapjunctional coupling between the JGA and the glomerulartuft. Cell Tissue Res 1978 ;186:279-85.

31 Becker CG. Demonstration of actomyosin in mesangialcells of the renal glomerulus.Am JPathol 1972 ;66 :97-110.

32 Accinni L, Natali PG, Vassallo L, Hsu KC, DeMartino C.Immunoelectron microscopic evidence of contractileproteins in the cellular and acellular components ofmouse kidney glomeruli. Cell Tissue Res 1975;162:297-312.

33 Scheinman JI, Fish AJ, Brown DM, Michael AJ. Humanglomerular smooth muscle (mesangial) cells in culture.Lab Invest 1976 ;34:150-8.

34 Bernik MB. Contractile activity of human glomeruliin culture. Nephron 1969;6:1-10.

35 Mahieu PR, Foidart JB, Dubois CH, Dechenne CA,Deheneffe J. Tissue culture of normal rat glomeruli:contractile activity of the cultured mesangial cells.Invest Cell Pathol 1980;3:121-8.

36 Ausiello DA, Kreisberg JI, Roy C, Kamovsky MJ.Contraction of cultured rat glomerular cells of apparentmesangial origin afterstimulationwith angiotensin II andarginine vasopressin. J Clin Invest 1980;65 :754-60.

37 Osborne MJ, Droz B, Meyer P, Morel F. Angiotensin II:renal localisation in glomerular mesangial cells byautoradiography. Kidney Int 1975 ;8 :245-54.

1205

copyright. on N

ovember 7, 2020 by guest. P

rotected byhttp://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.34.11.1197 on 1 Novem

ber 1981. Dow

nloaded from

1206

3S Michael AF, Keane WF, Raij L, Vernier RL, Mauer SM.The glomerular mesangium. Kidney Int 1980;17:141-54.

Elema JD, Hoyer JR, Vernier RL. The glomerularmesangium; uptake and transport of intravenouslyinjected colloidal carbon in rats. KidneY Int 1976;9:395-406.

'u Takamiya H, Batsford S, Kluthe R, Vogt A. Comparison ofthe handling of ferritin and ferritin-protein conjugates bythe glomerular mesangium: genetic studies in the rat.Lab Invest 1979 ;40:18-24.

41 Lee S, Vernier RL. Immunoelectron microscopy of theglomerular mesangial uptake and transport of aggregatedhuman albumin in the mouse. Lab Invest 1980;42:44-58.

42 Albertine KH, O'Morchoe CCC. Distribution anddensity of the canine renal cortical lymphatic system.Kidney Int 1979 ;16 :470-80.

13 Fourman J, Moffat DB. The blood vessels of the kidney.Oxford: Blackwell, 1971.

Kaissling B, Kriz W. Structural analysis of the rabbitkidney. Advances in Anatomly, EmbrYology, and CellBiology 1979 ;56:1-123.

Beeuwkes R. The vascular organization of the kidney.Annu Rev Physiol 1980 ;42:531-42.

6 Moffat DB. The anatomy of the renal circulation. In:Black D, Jones NF, eds. Renal disease. Oxford: Blackwell.1979.

4 Beeuwkes R. Vascular-tubular relationships in the humankidney. In: Leaf A, Giebisch G, eds. International

Moffat

SYmposium of Renal Pathophv siolog% New York: Raven,1980.

8 Heaton JM, Bourke E. Papillary necrosis associated withcalyceal arteritis. Nephron 1976; 16:57-63.

Kriz W, Dieterich HJ. Das Lymphgefasssystem derNiere bei einigen Saugetieren. Licht und elektroneinmikroskopische Untersuchungen. Z Anat EntwGesch1970;131 :111-47.

Moffat DB. Extravascular protein in the renal medulla.Q J Exp Phjysiol 1969;54:60-7.

' Schmidt-Nielsen B. Excretion in mammals: role of therenal pelvis in the modification of the urinary con-centration and composition. Fed Proc 1977;36:2493-503.

.2 Moffat DB. Clearance of extravascular materials fronthe renal medulla. Ann R Coll Slurg Engl 1979;61 :315.

Hoyer JR, Seiler MW. Pathophysiology of Tamnml-Horsfall protein. Kidne), Int 1979 ;16 :279-89.

" Sikri KL, Foster CL, McHugh N, Marshall RD. Localiza-tion of Tamm-Horsfall glycoprotein in the humankidney using immunofluorescence and immunoelectronmicroscopic techniques. J Anat 1981 ;132:597-605.

Cotran RS, Hodson CJ. Extratubular localization ofTamm-Horsfall protein in experimental reflux nephro-pathy in the pig. In: Hodson J, Kincaid-Smith P. eds.Reflixnephropathb. New York: Masson, 1979.

Requests for reprints to: Prof. DB Moffat, Departmentof Anatomy, University College, Cardiff, Wales.

copyright. on N

ovember 7, 2020 by guest. P

rotected byhttp://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.34.11.1197 on 1 Novem

ber 1981. Dow

nloaded from