Download Organization of the Kidney Proximal

565th MEETING, STIRLING
101 1
Conclusion
Although the endoplasmic reticulum, Golgi and plasma membranes contain different
enzymes and proteins (Ehrenreich et al., 1973)) the three major regions of the plasma
membrane of the hepatocyte differ mainly in the concentration of similar classes of
glycoproteins functioning as cell-surface receptors and enzymes. The generation and
maintenance of these concentration gradients of glycosylated macromolecules on the
hepatocyte plasma membrane appears to be a consequence of a biosynthetic route
involving the Golgi apparatus that attaches sugar residues to plasma-membrane precursors and directs their insertion at the sinusoidal region. The final location and
orientation of newly inserted molecules is then dictated by their association with other
membrane or submembraneous components. I n the hepatocyte a limited loss of plasmamembrane ectocomponents at the canalicular region occurs, owing to detergent catalysed
dissolution into the bile fluid (Evans et al., 1976), a process that is intensified in
extrahepatic cholestasis.
Bennet, G. Leblond, C. P. & Haddad, A. (1974) J. Cell Biol. 60,258-284
Bergeron, J. J. M., Evans, W. H. & Geschwind, I. I. (1973a)J. CelZBiol. 59,771-776
Bergeron, J. J. M., Ehrenreich, J. H., Seikevitz, P. & Palade, G. E. (1973b)J. CeNBiol. 59,73-88
Bergeron, J. J. M., Berridge, M. & Evans, W. H. (1975) Biochim. Biophys. Acta 407,325-337
Cuatrecasas, P. (1974) Annu. Rev. Biochem. 43,169-214
Ehrenreich, J. H., Bergeron, J. J. M., Siekevitz, P. & Palade, G. E. (1973) J. Cell Biol. 59,45-72
Essner, E., Novkoff, A. B. & Masek, B. (1958)J. Biophys. Biochem. Cytol. 4,711-716
Evans, W. H. (1970) Biochem. J. 116,833-842
Evans, W. H. & Gurd, J. W. (1971) Biochem. J. 125,615-624
Evans, W. H. & Gurd, J. W. (1973) Biochem. J. 133,189-199
Evans, W. H., Hood, D. 0. & Gurd, J. W. (1973a) Biochem. J. 135,819-826
Evans, W. H., Bergeron, J. J. M. & Geschwind, I. I. (19736) FEBS Lett. 34,259-262
Evans, W. H., Kremmer, T. & Culvenor, J. C. (1976) Biochem. J. 154,589-595
Farquhar, M. G., Bergeron, J. J. M. & Palade, G. E. (1974) J. Cell Biol. 60,8-25
Fisher, K. A. (1976) Proc. Nafl.Acad. Sci. USA. 73,173-177
Frey, L. D. & Edidin, M. (1970)J. Cell Sci. 7,319-355
Jeejeebhoy, K. N., Ho, J., Greenberg,G. R., Phillips, M. J., Bruce-Robertson,A. & Sodtke, U.
(1975) Biochem. J . 146,141-155
Kremmer, T. B., Wisher, M. H. &Evans, W. H. (1976) Biochim. Biophys. Acfain the press
Little, J. S. & Widnell, C. C. (1975) Proc. Natl. Acad. ScL U.S.A. 72,4013-4017
Ong, S . H., Whitley, T. H., Stowe, N. W. & Steiner, A. L. (1975) Proc. Nafl.Acad. Sci. U.S.A.
n,2022-2026
Paphadjopoulas, D. (1974) J. Theor. Biol. 43,329-337
van Hoeven, R. P., Emmelot, P., Krol, J. H. & Oomen-Muelemans, E. P. M. (1975) Biochim.
Biophys. Acfa 380,l-11
Wacker, H. (1974) Biochim. Biophys. Acta 334,417-423
Widnell, C . C. & Unkeless, J. E. (1968) Proc. Natl. Acad. Sci. U.S.A. 61, 1050-1057
Wisher, M. H. &Evans, W. H. (1975) Biochem. J. 146,375-388
Zwaal, R. F. A., Roelefsen, B. & Colley, C. M. (1973) Biochim. Biophys. Acta 300,151-182
Organization of the Kidney Proximal-TubulePIasma Membrane
A. JOHN KENNY and ANDREW G. BOOTH
Department OfBiochernistry,UniversityofLeeds, 9 Hyde Terrace,Lee& LS2 9LS, U.K.
The cells of the kidney proximal tubule display a very obvious polarity. The luminal
membrane is specialized to form a brush border comprised of microvilli, and the basal
pole shows an infolded plasma membrane enclosing many mitochondria within its folds.
The lateral plasma membrane is morphologically unremarkable: adjacent cells appear
closely apposed and the intercellular spaces characteristic of intestinal-mucosal cells are
not a feature of the proximal tubule. Both poles of the cell are therefore specialized so as
to increase their surface areas, especially the luminal surface, which is increased about
100-fold by virtue of its brush border. Although the proximal tubule is capable of some
VOl. 4
1012
BIOCHEMICAL SOCIETY TRANSACTIONS
secretory activity, the primary physiological function is concerned with absorption.
About 80% of the glomerular filtrate is reabsorbed without alteration of its osmolarity.
Reabsorption of amino acids and glucose is localized in the proximal tubule and both
are known to be carrier-mediated Na+-dependent processes. Histochemical methods
have been largely directedtowards the localization of phosphatases,and there is no dispute
that the (Na++K+)-activatedATPase* is concentrated in the basal plasma membrane
and the non-speck alkaline phosphatase in the microvillus membrane. Serious biochemical contributions to our understanding of these membrane functions have awaited
the development of adequate methods for preparing brush border and other proximaltubule plasma membranes. There are now several acceptable procedures for the preparation of brush borders (i.e. clumps of luminal membrane bearing numerous microvilli) and of free microvilli that have been sheared from the remaining membrane during
homogenization (see Booth & Kenny, 1974). Marker enzymesfor microvilli are enriched
about 15-20-fold in such preparations compared with the cortical homogenate. Electron
micrographs of sections of brush-border preparations usually show portions of lateral
membranes still attached to the luminal membrane and the cores of these clumps often
contain a collection of small vesicles of undefined origin. Preparations of basal-lateral
membranes have only recently been investigated. Their resolution from microvilli has
been effected by the use of free-flow electrophoresis, the basal-lateral membrane fraction, characterized by (Na++K+)-activated ATPase, migrating more rapidly towards
the anode than the microvillus-rich fraction (Heidrich et al., 1972). Such preparations
have been reported to contain a Caz+-activatedATPase and a parathyroid-hormonesensitive adenylate cyclase in addition to (Na++K+)-activated ATPase (Kinne, 1975).
A speciiic transport system forp-aminohippurate has been attributed to the basal-lateral
membranes (Berner & Kinne, 1976).
In contrast, the microvillus membrane is much more complex in the variety of enzymes
associated with it, although this apparent difference may be due to the paucity of enzymic studies on preparations of basal-lateral membranes. So far, the following have been
characterized as microvillus-membrane enzymes: alkaline phosphatase (EC 3.1.3.1))
aminopeptidase M (EC 3.4.1 1.2) (Kinne & Kinne-Saffran, 1969); y-glutamyltransferase
(EC 2.3.2.2), maltase (EC 3.2.1.20), 5‘-nucleotidase (EC 3.1.3.5), phosphodiesterase I
(EC 3.1.4.1) (Glossmann & Neville, 1972); aminopeptidase A (EC 3.4.11.7), neutral
endopeptidase (EC 3.4.24.-), trehalase (EC 3.2.1.28) (George & Kenny, 1973);dipeptidyl
peptidase IV (Booth & Kenny, 1974); a cyclic AMP-dependent protein kinase (Kinne,
1975). Structural proteins are also present in microvillus preparations:filamentsresembling actin are readily seen in electron micrographs (Rostgaard & Thuneberg, 1972),
and we have purified and characterized an actin from such preparations (Booth &
Kenny, 19766). A phlorizin-sensitive glucose-binding protein has been isolated from
brush-border preparations (Thomas, 1973), but so far this appears to be the only microvillus protein clearly identified with the transport of small molecules. The proposal that
y-glutamyl transferase is involved in amino acid transport in the kidney (Meister, 1973)
is now questionable. It is clear that the ‘y-glutamyl cycle’ does not function as an amino
acid-transport system. In particular, the activity of one of the enzymes required for the
cycle, 5-oxoprolinase, is too low to operate thecycleat aratecommensurate with normal
tubular function (Orlowski & Wilk, 1975). Even a more limited role in which y-glutamyl
transferase released the amino acid and glutamic acid rather than 5-oxoproline seems
unlikely following the report of a patient deficient in y-glutamyl transferase whose biochemical abnormalities included glutathionuria, but who showed no defect in amino
acid reabsorption (Schulman et al., 1975). It may well be that the primary role of this
microvillus enzyme is that of a glutaminase concerned with NH4+secretion.
The luminal pole of the proximal tubule has been long known to be capable of pinocytosis of proteins (see, e.g., Straus, 1964). Proteins substantially larger than 60000 mol.wt.
are excluded by the glomerular filter, but the filtrate has been variously estimated to contain between 0.1 and lOmg of protein/100ml (Strober & Waldmann, 1974), and almost
* Abbreviation :ATPase, adenosine triphosphatase.
1976
565th MEETING, STIRLING
1013
all of this is reabsorbed. It is generally accepted that the absorbed protein is digested by
lysosomes, but the possibility of reabsorption of intact proteins into the blood has been
suggested (Maack et al., 1971) and also disputed (Just et al., 1975). It is clear that the
initial step in the process involves invagination of the luminal membrane producing
apical pits, which pinch off to form the small apical vacuoles that are a prominent feature
of the apical cytoplasm. Bode et al. (1974, 1976) have attempted to purify a pinocytic
vesicle fraction by free-flow electrophoresis. This fraction was somewhat depleted in
microvillus-marker-enzyme activities and the authors concluded that the pinocytic
vesicle membrane was different in protein composition from that of the microvillus.
However, the homogeneity of the vesicle preparation must remain in doubt in the
absence of an authentic marker for pinocytic vesicles.
Our interest has focused on the proteins of the microvillus. Two of the peptidases
have been purified and characterized in this laboratory. One is a Znz+-metalloenzyme,
which we have referred to as neutral endopeptidase (Kerr & Kenny, 197&,6). It has the
ability to attack peptides such as insulin B chain and glucagon, but not more highly
structured proteins such as insulin, albumin or casein. The specificity shown by the endopeptidase is similar to that of the group of microbial neutral proteinases typified by
thermolysin :the peptide bonds attacked are those involving the or-amino group of hydrophobic residues. When purified the enzyme is a monomeric glycoprotein of 93 000mol.W.
The other peptidase is dipeptidyl peptidase IV (Kenny et al., 1976), an enzyme that
typically releases dipeptides from the N-terminus of peptides with the sequence X-Pro-Y
or X-Ala-Y. It also possesses endopeptidase activity in that it attacked certain N-blocked
peptides, e.g. Z-Gly-Pro-Leu-Gly-Pro. This enzyme is a serine proteinase that is extremely sensitive to inhibition by di-isopropyl phosphorofluoridate (Dip-F). Titration
of the active site with t3*P]Dip-F confirms that the purified enzyme is a dimer comprising
two glycoprotein subunits of mol.wt. 130000. Aminopeptidase M (also a Zn2+-protein)
has a rather broad specificity in removing N-terminal residues, whereas aminopeptidase
A, a protein that is difficult to purify without substantial contamination with aminopeptidase M, is specific for N-terminal glutamic acid and aspartic acid residues.
As a group these four enzymes are capable of hydrolysing most peptides of 35 or fewer
residues to their constituent amino acids. Indeed an extrapolation from the specific
activities of these enzymes in the rabbit kidney (Booth & Kenny, 1974) suggests that they
could dispose of about 20mg of such peptides/min in the whole animal, a potential
that is several orders of magnitude greater than the expected filtered load of peptide
substrates.
When purified microvilli are treated with mercaptoethanol and analysed by sodium
dodecyl sulphate/polyacrylamide-gelelectrophoresis and stained for protein about 20
bands are consistently observed (Booth & Kenny, 19766). Five are major bands, designated 180, 160, 130,95 and 42 (indicating that they have apparent mol.wts. of 180000,
160000, etc.). Apart from two high-molecular-weight peptides (>350000), there are other
bands designated 240, 220, 90,75, 70, 65, 55 and 50 and some incompletely resolved
bands of mol.wt. less than 40000. Bands 180,160, 130 and 95 also stained for carbohydrate. It is now possible to identify some of the microvillus proteins with four of the five
bands. The use of Dip-F as an affinity label has confirmed that band 130 is dipeptidy1 peptidase IV. Aminopeptidase M has been assigned to band 160 and is the major
glycoprotein of the membrane. Neutral endopeptidase (a glycoprotein), together with a
structural protein (lacking carbohydrate residues), a-actinin, have been assigned to band
95, and actin, the major protein of the microvillus, is identified with band 42. The contribution of each of these proteins (except a-actinin) to the total protein of the microvillus
has been determined, in the case of the three enzymes by comparison of the specific
activities of the purified enzyme with that of the microvillus preparation, and in the case
of actin by a computer analysis that is capable of resolving overlapping peaks of the
stained gel pattern. Actin comprises 9 % of the microvillus protein and the three peptidases together are another 12%. Alkaline phosphatase can be labelled with 32P,and this
approach has shown that the monomeric enzyme has a rno1.W. of 80000 and contributes only 0.04%of the microvillus protein and is not detectable as a stained band. Band
VOl. 4
Extrinsic glycoprotein
Aminopeptidase M
Dipeptidyl peptidase IV
Neutral endopeptidase
a-Actinin
Alkaline phosphatase
Actin
y-Glutamyl transferase
AminopeptidaseA
PhosphodiesteraseI
5’-Nucleot idase
Trehalase
Maltase
-
-
60oOo(?)
-
@oOo( ?)
180000
160000
130000
95 000
95 000
80000
42 000
Subunit mo1.w.
Location
Membrane
Membrane
Membrane
Membrane
Core
Membrane
Core
Membrane
Membrane
Membrane
Membrane
Membrane
Membrane
Class
Extrinsic
Intrinsic
Intrinsic
Intrinsic
Extrinsic
Intrinsic
Extrinsic
Intrinsic
Intrinsic
Intrinsic
Intrinsic
Intrinsic
Intrinsic
-
0
0
+
+
+
+
+
+-
Glycosylation
+
0
0
0
+
+
0
0
0
0
+
+
+
Solubilization
by papain
-
-
4.1
3.9
4.1
2.5(?)
0.04
9.0
-
Proportion of
microvillus
protein (%)
-
-
450
530
760
470( ?)
10
3800
-
No. of
molecules per
microvillus
Subunit mol.wts. are for the native rather than the purified proteins, where these are known to be significantly different. The term extrinsic
indicates a protein extractable by simple salt media. -, Not known; +, detectable; 0 , not detectable.
Table 1. Kidney microvillusproteins
4
?I
Ez
F
8
a
8
8
565th MEETING, STIRLING
1015
fa)
Fig. 1. Proposed model of a kidney microvillus showing the organization of the core and
membraneproteins
(a)Projection of the view along the axis of a microvillus. (b) Projection of a longitudinal
section of the central region between arrows in (a). The designation of the membrane
proteins is as follows: (A) extrinsic proteins; (B) intrinsic proteins released by papain;
(C) intrinsic proteins not released by papain. The drawing is approximately to scale, the
diameter of the microvillus is 100nm, the lipid bilayer is 7nm in thickness and the longitudinal spacing of the cross-bridges is 36nm, a value dictated by geometry of the actin
double helix.
180, also a glycoprotein, may be classified as an ‘extrinsic’ protein, since it is extractable
by simple salt solutions. Extraction of microvilli with NaCl or EDTA releases about
15-20 % of the protein. None of the known enzymes are detectable in the extract, which
appears to contain proteins corresponding to bands 180,95 and 42 and attributable to
the extrinsic glycoprotein, a-actinin and actin respectively.
The other known components of the microvillus membrane are all intrinsic membrane
proteins that resist release by simple salt extraction and require treatment with either
proteinases or detergents for their solubilization. Papain is effective in releasing aminopeptidases M and A, as well as dipeptidyl peptidase IV and y-glutamyl transferase.
Neutral endopeptidase, alkaline phosphatase, trehalase and phosphodiesterase I are
wholly resistant to papain treatment and are arguably more deeply inserted into the
VOl. 4
1016
BIOCHEMICAL SOCIETY TRANSACTIONS
lipid bilayer of the membrane. Dipeptidyl peptidase IV has been purified from Triton
X-100-extracted kidney microsomal fraction (R.D. C. Macnair & A. J. Kenny, unpublished work).
The various strands of information derived from electron microscopy, enzymology
and protein chemistry now permit a somewhat speculative model of the microvillus
organization to be suggested. The core is composed of actin filaments arranged in a 1+6
array (Rostgaard & Thuneberg, 1972; Booth & Kenny, 19766). Knowing the length of a
microvillus (l.5,um) and the half-pitch of the actin double helix (36nm) containing 13-14
subunits, we can calculate that there are some 3800 molecules of actin/microvillus,
amounting to 9.0% of the microvillus protein. We can therefore estimate the number
of monomers of some of the other microvillus proteins (Table 1). The role of
a-actinin (Mooseker & Tilney, 1975; Booth & Kenny, 1976~)may well be to provide
cross-bridges between actin filaments, and between actin and the membrane. In intestinal
microvilli the ratio of actin: a-actinin is 8: 1 (Mooseker &Tihey, 1975), and if the same
is true of kidney microvilli one would predict a-actinin to constitute 2.5% of the
microvillus protein. One way in which the a-actinin cross-bridging might be accommodated is shown in Fig. 1. The dimensions of the a-actinin molecule are 30nmx2nm
(Podlubnaya et al., 1975) and in this model one-third are involved in actin-actin links
and the remainder in actin-membrane bridges.
The functional role of the microvillus peptidases is a question now demanding attention. We have already indicated that their potential as peptide hydrolases for substrates
in the glomerular filtrate seems greatly in excess of requirements. This degree of ‘overkill’, coupled with the fact that three of the peptidases are major intrinsic proteins in the
membrane, makes an alternative role a strong possibility. Two questions require attention. (1) D o these outward-facing enzymes of the microvillus membrane become inwardfacing peptidases in a pinocytic vacuole? Although unable to initiate attack on proteins
this group of enzymes could complement the subsequent attack by the lysosomal
cathepsins. (2) Are these proteins concerned primarily with amino acid transport?
There are a number of attractive features in this suggestion: they are large intrinsic
proteins(93000-160000mol.wt.);twoofthem aredimeric, at least when released from the
membrane (though not necessarily dimeric in situ); there is some similarity between their
specificities as peptidases and the specificities of the group-specific transport systems
(for review, see Young & Freedman, 1971) such that aminopeptidase A resembles the
acidic system, aminopeptidase M or neutral endopeptidase the neutral system, dipeptidyl
peptidase IV the iminoglycine system. The scheme fails to account for the basic system
that transports diamino acids, and clearly the correlation of the specificities is far from
perfect. Nevertheless we believe that the hypothesis is one that can be tested in isolated
tubules or microvillus vesicles and in the perfused kidney.
We thank the Medical Research Council for support.
Berner, W. & Kinne, R. (1976) Pfltigers Arch. 361,269-277
Bode, F., Pockrandt-Hemstedt,H., Baumann, K. & Kinne, R. (1974) J. Cell Biol. 63,998-1008
Bode, F., Baumann, K. & Kinne, R. (1976) Biochim. Biophys. Acta 433,294-310
Booth, A. G. & Kenny, A. J. (1974) Biochem. J. 142,575-581
Booth, A. G. & Kenny, A. J. (1976~)J. Cell. Sci. 21,449-464
Booth, A. G. & Kenny, A. J. (19766) Biochem. J. 159,395407
George, S. G. & Kenny, A. J. (1973) Biochem. J. 134,43-57
Glossmann, H. &Neville, D. M., Jr. (1972) J. Biol. Chem. 247,7779-7789
Heidrich, H.-G., Kinne, R., Kinne-Saffran, E. & Hannig, K. (1972) J. Cell Biol. 54,232-245
Just, M., Rockel, A., Stanjek, A. & Bode, F. (1975) Naunyn-Schmiedeberg’sArch. Pharmacol.
289,229-236
Kenny, A. J., Booth, A. G., George, S. G., Ingram, J., Kershaw, D., Wood, E. J. & Young,
A. R. (1976) Biochem. J. 157,169-182
Kerr, M. A. & Kenny, A. J. (1974~)Biochem. J. 137,477-488
Kerr, M. A. & Kenny, A. J. (19746) Biochem. J. 137,489-495
Kinne, R. K. H. (1975) Med. Clins. N . Am. 59,615-627
U m e , R. K. H. & KinneSaffran, E. (1969) Pftigers Arch. 308,l-15
1976
565th MEETING, STIRLING
1017
Maack, T., MacKensie, D. D. S. & Kinter, W. B. (1971) Am. J. Physiol. 221,1609-1616
Meister, A. (1973) Science 180, 33-39
Mooseker, M.S. & Tilney, L. G. (1975) J. Cell Biol. 67,725-743
Orlowski, M. & Wilk, S. (1975) Em. J. Biochem. 53,581-590
Podlubnaya, Z . A. Tskhovrebova, L. A., Zaalishvili, M. M. & Stefanenko, G. A. (1975) J. Mol.
Biol. 92,357-359
Rostgaard, J. & Thuneberg, L. (1972) Z . Zellforsch.Mikrosk. Anat. 132,473-496
Schulman, J. D., Goodman, S. I., Mace, J. W., Patrick, J. D., Tietze, F. &Butler, E. J. (1975)
Biochem. Biophys. Res. Commun. 65,68-74
Straw, W. (1964) J. Cell Biol. 21,295-308
Strober, W. & Waldmann, T. A. (1974) Nephron 13,35-66
Thomas, L. (1973) Biochim. Biophys. Acra 291,454-464
Young, J. A. & Freedman, B. S. (1971) Clin. Chern. 17,245-266
Biochemical Differentiation of the Plasma Membrane of the Intestinal
Epithelial Cell
ROBERT H. MICHELL, ROGER COLEMAN and BARBARA A. LEWIS*
Department of Biochemistry, University of Birmingham, P.O. Box 363,
Birmingham B15 2TT, U.K.
The brush border and basolateral regions of the surface of the intestinal epithelial
cell (enterocyte) are exposed to environments which differ even more than those which
bathe the different functional regions of the surface of the hepatocyte or the renal tubular
cell, and therefore the biochemical differentiation of the enterocyte plasma membrane is
also more striking. In addition, it has been easier to study, since the highly adapted
brush-border membrane of the enterocyte is one of the easiest of plasma membranes to
isolate pure and in quantity (see Isselbacher, 1974).
From the study of brush-border preparations a picture has emerged of a membrane
with a variety of terminal digestive hydrolases (especially aminopeptidase, disaccharidases and alkaline phosphatase) attached to its luminal surface (Louvard et al., 1975).
These are large proteins (Maestracci et al., 1975), which are heavily glycosylated (Kelly
& Alpers, 1973; Weiser, 1973; Quaroni et al., 1974) and are moored to the membrane by
relatively small hydrophobic polypeptide segments, probably at their N-terminal ends
(Maroux & Louvard, 1976). Some of the end-products of digestion that these enzymes
generate are then passed across the membrane by transport systems which are closely
coupled to the hydrolases and are Na+-independent (Ramaswamy et al., 1976). Some
enter the cell via one of several electrogenically driven Na+/solute symport systems
(Crane, 1965,1974; Schultz & Curran, 1970; Berger et a[., 1972; Murer & Hopfer, 1974;
Hopfer et al., 1973, 1976; Sigrist-Nelson et al., 1976; Alvarado, 1976) and others are
taken into the cell by specific transport systems that are independent of Na+ (e.g.
fructose; Sigrist-Nelson & Hopfer, 1974).
At the cytoplasmic surface of the brush-border membrane, and attached to it by aactinin bridges, is an actin array which forms the microvillus core that also interacts with
myosin-like proteins in the terminal web (Mooseker & Tilney, 1975). This raises the
exciting, but still largely unexplored, possibility that the brush border is motile (Thuneberg & Rostegaard, 1969; Sandstrom, 1971; Mooseker, 1974) and that this somehow
facilitates its digestive and absorptive function, perhaps by preventing the formation of
unstirred layers of fluids at either surface of the brush-border membrane.
The remainder of the enterocyte plasma membrane (the basolateral region) is separated
from the brush border by the terminal bar. Thus, in contrast with the brush border, it
interacts with the protective and controlled internal environment of the body and might
be expected to be much more similar to the plasma membranes of other cells. It is this
region of the plasma membrane which plays the key role in such processes as generating
* Present address: Department of Biochemistry, University of Bristol, Medical School,
Bristol BS8 1TD.U.K.
VOl. 4