- Email: [email protected]
Uptake of processing of glycoproteins by isolated rat hepatic endothelial and Kupffer cells
211
Uptake of processing of glycoproteins by isolated rat hepatic endothelial and Kupffer cells
Mannose- (Man) and N-acetylglucasamine(GlcNAc)-terminated glycoproteins are cleared tiom blood by carbohydratespecific receptors present on both bepatic endothelial and Kupffer cells. It k not known whether the same receptors are present on each cell type or the relative contributions to glycoprotein metabolism made by Kupffer and endadwlial cells. Here we report experiments where data fmm glymprotein metabolism by purified populations of isokted rat hepatic endatlmliil and Kupffer cells have been analyzed by mathematical modelliog and parameter estimatica. Kupffer cells had signifhntly higher binding rate mostants (k’,,) than endotbeliat cells for agalactwmsomuwid (.AG@R) and hyalumnida, bm &ier /q2 (%-rate’) indicating that Kuy~~:cl CC% :.a; ;6&ci affinities for MatVGlcNAc-termittatzd glycqxo. tebts than endathelial receptors. Fmthetmore, &bough endotbelial cells bad similar affinities (k’ll and k,d for AGOR and hyaluronidase, the ‘off-rate’ of Kupffer celk was significantly greater for AGOR than for hyabmmidase, indicating that Kupfkr cell receeptors have lower afIinity for AGOR. Internalization and ligand catabolic rates also differed between the two cell types. ‘I’be data indicate that Kupffer and mdotbelial cells appear to have diffaeot MardGlcNAc recepton and that the destination of a glycoprotein and its subsequent processing is determined by the structure of a glywprotein’s oligosaccharide.
Kupffer
cells (4-6)
and tb,t
iron loading
ana diabetes
Mannose-(Man) and N-aeetylgIumsamine (GkNA+ tetminated glyqmteins are cleared from blood by te-
mellims impair ligaod uptake and processing
ceptms
Most
ceptors (7,8). We have shovm, in the intact rat, that the
studies on ManiCilcNAc receptors have been performed using extrahepatic macrcpbagea, principally because pleural and peritoneal macmphages are easier to prepare than liver sinusoidal cells. These experiments have shown that Man/GlcNAc rece~tots are reclycled (1). down-regulated by high glucose concentrations (2) and up-regulated
‘itmcture of a glycoprotein tigand ini%mces its uptake and praxessing ‘myhepatic ManlGlcNAc recepton (9). In these experiments we have sought to define tUtIter the prqvxties of hepatic sbmsoidal Man/GlcNAc receptors by analyzing glycopmtein me;aboIism by isolated Kupffer and endotbelial cells using mathematical modelling and
by denamethmnne (3). In contrast, there k little information on the properties of the ManlGlcNAc receptors on endothelial and Kupffer cells. It is known that Man/ GlcNAc receptors are present on bath endothelial and
parameter estimatkm. The data cmdirm that @and strw we determines the rams of uptake and processing and show that Kupffer u!k are most active in metaboliing the ligands tested.
on liver sinusoidal
cells and macmphages.
by these re-
A. SAN0
212 Experimental
et at.
pmceduw
The materials and methods have been described iu detail elsewhere (5). Male Sprague-Dawley rats (250-300 g), bred in the Comparative Biology Unit of the Royal Free Hospital, were used. Liver cells wete isolated by e01lagenase perfusion followed by differential centrifugation and centrifugal elutriaticm. Mixed sinusoidal cells were loaded into a Beckman JE6.B elutriator rotor running at 2590 rprn at 15 OC in a Beckman JZ-21 centrifuge (Beckmsn, High Wycombe, U.K.) at u flow rate of 13 mlitnin, and drblis ad lymphocytes were removed by collecting 150 ml. Endothelial cells were collected at 22 udlmin and Kupffer cells at 46 mUmin. The cell fractions were counted and sized using a Coulter Counter and Channelyser (Gndter Electronics, Luton. U.K.). The modal volume was 116 + 2 urn’ (mean & S.E.1 for endothelial cells end 160 i 3 jun’ for Kupffer cells. The proportion of Kupffer cells in each fraction was assessed by histochemical staining for tartrate-resistant acid phosphatase, which is restricted to Kupffer cells (10). We have previously shown that tartrate-resistant acid phosphatase staining gives similar results to endogenous peroxidare staining (5). Kupffer cells accounted for 93 f 5% (n = 5) of the ccllr ic !he Kupffer cell franion and 13 Ir 3% (II = 5) of the cells in the endothelial cell Era&n. The &bility of isolated hepatocytes, assessed by trypan blue exclusion, war 88 f 4% (n = 5). Trypan blue exclusion is not a reliable indication of cell viabiliw so the viabilitv of sinusoidal cells was judged to be satisf&tory on the basis of results for trypan blue exclusion oC!lspatocytes (5).
sinusoidal
Mannose,-Al-BSA (Man-BSA), prepared by the method of Lee et al. (ll), was the generous gift of ProfesSOTY.C. Lee (Baltimore). A highly purified preparstion (240 fXB IIJimg) of bovine testicular hyahtronidase was kindly donated by Biorex Laboratories, London, U.K. Agulactoorosomucoid (AGOR) wus prepared from human orosomucoid by incubation with neuraminidase to yield arialoorosomuwid followed by incubation with @galactosidase (12). The efficiency of removal of the terminal carbohydrate residues was assessed by measuring released sialic acid using a thiobarbitulic acid assay (13) and galactore using galactose dehydmgenase. Seventeen mol of t&h sialic acid and galactose were removed per mol of ororomucoid. Cross-inhibition studies, where the iuhibition by AGOR of [“‘f]asialoorosomucoid (ASOR) uptake by isolated rat he$&acytes was measured, showdd that onlv 1.4% of the AGOR owDaratian was ASOR. Glycopr&ins were rsdioiodinat;d ;ith carrier free Na’“l by the chloramine T method (14). The specific activities of the ligands ranged between 10 ana 2Oj0i/tg.
&pensions of endothelial and Kupffer cells (5-10. lob/ml1 were incubated in Hunks’ hulanced salt solution (HESS) pH 7.4 containing 5 mM glucorr and 25 mM Heper at 37 ‘C for 30 min before the assay to pennir replenishment of cell surface receptors from the intraceliula pool. 5 mM glucose does not inhibit ligund uptake by the Man/GlcNAc reczptor (15). Buffers were gassed with 95% 045% CO, bdore the assay. Assays were stated by adding labelled l&and (14 pmollml) or, to parallel tubes, labelled ligand plus an excess of the inhibitor maunan (500 J&I) and incubated at 37 ‘C with agitation for 90 min. At sequential times of 2.5, 10,15,30,45,60 and 90 miu duplicate sumples (ZDO#l) were removed, placed in 4C@#l micmfuge tubes coutainiug 100 pi silicone oil (5) and 50~1 phasphotungstic acid solution (PTA, 2% w/v in 1 M HQ) and centrifuged for 30 s in a Beckman minofuge B. The cell pellets were removed by slicing through the silicone oil layer and radioactivity iu the cell pellet and supemate was measured in a gamma wunter. Acid precipitabie radioactivity (intact ligand) and acid soluble mdioactivity (ligand catabolite) were determined hy adding 1Klpl 1% (w/v) BSA and 800@ PTA to 100/d of the supernate or the ce!l pellet, cooling the mixture for 10 miu on ice, celltlifuging at 3500 x g for20 min aud measuring the radioactivity in the super&e and precipitate. Specific lfgaud uptake and catabolism were estimated by subtracting the counts in the tubes containing menuan (non-specific uptake) from counts in the tubes containing &and alone.
The mathematical model developed for this study is compartmental as shown in Fig. 1. The model is based on that developed in previous in viva work (9), but is simpler than other models developed by our group (16) and by others, e.g., reference 17. The model equations arc derived as follows.
213 Suppose there are a?, free surface recepton/cell and that this number remains constant over the experiment. (Experiments with reduced ligand d33e did not signifiunlly change the speed of l&and upta’ce and processing. it was thus assumed that there was no aturation of receptors.) The binding flux of ligand ontt the cell (mmollcell per min) is given by
where [L] is the ligand concentration
@moUrnI),
RL is
the amount of surface bound ligand (mmoUcell), and k,, is the binding constant. If there are n cells/mi. then the binding tlux (mmollml) is
where [RL.] is llow a concentration
(mmoUml). Writing x,
= [L], x, = [RL] gives the cawentional mental equation
k,, = “k+,R,
linear cornpart-
(4)
(It is important to note that k,, depends on the three paramctem wbfch cannot be distinguished separately.) k,, is the figand-receptor diswciation rate constant. All kii have the dimension miti’. The equation for bound ligattdx, is
The model paramete.~ were estimated by fitting the model to the data for free ligand fy, = q). cell-asstiated ligand (yz = x2 + x,) and free catabolite fyi = &) obtained in the assays described previously. The weighted residual sum of squares function describiig
the goodies-of-fit
is
given by
where y&) and y&r’, are the data and model values for the ith data poinr on thejth outpat. q,(i) is the standard deviation of each data point. Taking accmmt of ass&y precision and other expetimental errors, a ‘wnstcase value of o&i) = O.ly,,(i) (i.e., a 10% cwf6cientoirtintion of the error) was used. A non-linear optimisation program MINUET (24) was used to find values of the parameters v/&h minim&d the W function, giving a weighted-least squares estimate. These values are taken as the parameter estimates I?,*. Since W has the fotttt of a 2 function, the change in a single parameter which causes W to fncrease by unity giver the 67% confidence on that parameter (a;$ The cc&icient of variation of a parameter estimate is defined as cv = lcc+,/~,,%
The mode, paramaers vlere obtabled for each aeu and ligand type (E, endotbelial; K, Knpffer) and replicated in a number of experimemu. Fwameterdifferences between E and K cells were tested using a one-sample Wikoxou signed-rank test (PlW) of the paired differences, since E and K cells were nut in a paired experiment. Differences between ligands of a given cell type were compared tith a two-sample ?.iann-Whihtey-Wilcoxon rank tat (hw). Care has to b
where k,t is the endqtmic, or intemafisation rate constant (mitt-t). The equations for intemalised and cat&olised (free) ligand are given by
where k,l is the ligand catabolic rate perimental results (Table 2) showed ible catabolite in cells (clO% total represatrcatabcdite in the medium
constant (miti’). Exthat there was negligcatsboli:e), hence x, (as in Fig. 1).
taken when comparing
values of 41,
since from Eqos. 2 and 3 k2, effects the uptake kinetics. but isitselfdepen&nton thefellttumbern. (Nootherpammeters depend on n). In the experiments, cell rmmber varied between 5 and 2!l. ?p cell%tl. Tbe 4, values were therefore sealed to eliminate the effect of this variation
where no is S-106 cells/ml. In this way, comparisons were only being made between differences in the pmdua k+ ,R, (the association constant Kh). From Eqns. 4 and 9 it follows that
214 Res”,ts and
Dlscusion
In this study the in vitro metabolism of tw Van/ tilcNAc-terminated giycoprot+ts by isolated and endothelial cells have beeo canpared. By the USC01 mathematical modelling and paratneter estimation v?:
R~pifeer
have shown that there are diiferenms in the way the two cell types handle theseglycoproteins. All three ligandswere trken up and catabolised by both endothelial and Kuoffer tolls (Fig. 21. For each lieand nnd cell type, cell-assobated ligand rose to a maximum between 20 and 30 min. Ligand catabolite appeared in the tnedium after about 13 min and increased up to 90 min. Less than .O% of lidand catabolite was foeod ssrociated with thecell pellet at any time (Table 1). A mathematical model (based on the compartmental structureshown in Fig. 1) was developed from the dynamic data and parameters of ligand binding, internalization and catdbolism were estimated as described in Experi-
_
mental procedures. The model provided good fits to the data obtaioed for both cell types with AGOR and hyaluronidase (Fig. 2 showsrepresentativefits). The coefficients of variation of the parameter valu:s were all less than 30%. In contrast the fits (not shown) obtained with Man-k&4 were poor and the results were not used in the analysis. Model parameters estimated for the ligands are summatised as their mean and standard error in Table 2. Collagenase
2 5 I” 20 F :
o.as(o.2, 0.98(0.4) 2.9(0.8, 5.Ol1.3) 6.5(1.1) 5.7(0.6) 4.1(rJ.3, 2.7lO.4)
l.l(O.7) 3.2(1.6) 6.1(2.7) 8.2(2.7) 7.4(1.5) 5.2(0.6) 4.2(0.5) 2.0(0.4)
0.98(0.3) 1.7(0.6) 3.8&w) 1.0(1.0, 5.*(w) 3.2(0.14, t.7 (0.32) Lu(O.3)
1.3(0.8, 3.4(1.6) 7.0(3.6, 1”.2(4.0, fx(1.7) 3.1(O.B, l.8Q3.7) l.l(O.3)
Percentage (+ S.E.M.) of&and catablite pre5ent in cellpelletclutbtgthe assay.The remainderof the catabalitewasin the medium. Fordetails we Experimental preced”leS.
perfusion will damage some surface recepton and a 30 min incubation will probably be insufficient to restore them all. However, since the affinity of a receptor is independent of the number of receptors present on the cell the estimater given below are likely to be valid. We cannot, however, exclude the possibility that partial degradatiott of a receptor or damage to tteighbouring molecules might have intloenced their affinity. Kupffer cells had significantly higher binding rata constants (k’,,) than ettdothelial cells for both AGOR @ = 0.05, PIW, see experimental procedures for definition) and hvaluronidasefo = O.W. PlWI. Thisfindine_ could Rfleet &her a greater number of receptors per cell (I?,, on Kupffer cells or that the ‘cm-rate’ (k,,) &,E highe; for Kwffer cell receotors. In contrast k., (the ‘off-rate’) of hyaluronidase & significantly gre&; for eodotIt&l than for Kupffer cells f.p = 0.053, PlW). This difference was not observedwith AGOR. Since the parameter ktl is independent of receptor number these data indicate that ”
.I
Kupffer cell receptors have a higher affinity for glympro-
UPTAKE AND PFXXESSING OF GLYCOPROTEINS
215 similar rates by alveolar macrophages indicating that receptor-ligand internalization is independent of ligand binding in a particular cell type (IS). A” in viva study of glvwprotei” uptake by whole liver also indicated that inwmalizatio” tates were independent of ligand binding rates (9). Cieariy .rperime”b on ahok !iver ~+ch IIICEsorzd the combined internalization rates cd endotneiial and Kupffer cells could not distill&h differences between the two cell types. However this study has shown that endothelial and Kupffer cells intemalized ligands at different rates, suggesting that the mechanism for internalisation is difkrent. It is unclear whether the differences in endothelial and Kupffer cell internaEration rates die a further reflection of differences in receptor stntctme or are due to other factors such as cell surface receptor distribution or receptor coupling to the cell’s endocytotic mechanism. Endothelial cells catabolised AGOR and hyaluroni-
ieins (at IeaSt b~altuonidase) than endothelial ceil recap tars and suggest that although the receptors share similar carE&ydrate specificities they are probably structurally different. Further,wre. sltbo.,gb e”doth&zl cell ieccptors bed similar &i”ities (i.e. similar Ir’,, and similar k,, values) for both AGOR end hytduronidase, the ‘o&rate parameter (k& of the Kopffer cell receptors was significantly greeter for AGOR than for hyaluronidase @ = 0.066, MW). This finding indicates that KupKer cell receptors
dase at a similar rate (k,). Kupffer cells catabolized AGOR at a similar to endothelialcells but hyaltuonidase was more resistant to degmdatiw in these cells (p = 0.0.53, PIW). Futhemtore hyeluronidase was catabolized at a significantly slmer rate than AGOR by Kupffeer cells@ = 0.02, UW). ‘The 6”diag that the lysosmnal e”zyme hyahwonidase was catabolised more slowly the” AGOR by Kupffer cells is consistent with the findings of a” in viva study in which three neoglympmteios were found to be catabolized more rapidly thao the rzmxel xlvcomotei” AGOR (9). Many ridi& have show” that deglyc&ylated proteins k cataboliaed “tore quickly than the corresponding holopro. teins (lE-22) indicating that deglycosylation is the rate limiting step in their catabolism and there is also evidence that proteolysis ~an”ot wxr until the oligosacchfuide
have B lower affinity for AGOR than for hyehuonidax sod provide tioiber evidence that two populations of re-
chains have been removed (23). It follows that hyaluronioligosacchadase, a glycuprotein bearing a high-mantose
rate
ceptors
are probably
different.
This result L ctmsirtent
ride. which is normally resident in the lysommer catabolised
btg rate eatstem for the Ma” terminated gtycoprotei” MatvSSA wes found to be higher than that for the GkNAc-terminal glycopmtein AGOR. This study has also shown tbc higher binding rate co”stant of tbe whole liver for Man-terminated glymproteins is probably due to the contribution of Kupffer cells which are more active
somucoid from which the protective sialic acid raidues have been removed. It is unekar why these differences in catabohc rates are restricted t@Kupffer celix. In conclusian we haveshow” that Kupik-, and endothelial cells appear to have different Mao/GlcNAc receptors (as judged by theft &i”ities for glymproteh ligretds) and
than endothelial
cells in the uptake
of Ma”-
and tlte
G!cNAc-terminated ligands. The rate of intemalisation of AGOR (k& and hyalwonidruc by eodothetial &Is was greater than by Kupffer sellr. This difference reached significance for AGOR @ = 0.05. PlW). IO contrast each eel1 ‘yp internalized the two ligands et similar rates. Another study has show” that different Man-terminated ligand; are intemalizd at
mope slowly the” AGOR a derivative
would be
with the fittdbtgs of a” in tivo study (9) in which the bind-
that the destination
of a glycopmtein
or ore-
and 115subsequent
pacessing %eems to be determined by the stroc~re of the carbohydrate that the glycoprotei” bears. One of the fonctioos imposed for these “pton is the hepatk clearance of potentially harmful proteins or microorganisms such es lysosomal enzymes, bacteria and yeasts. However, it is noreworthy that mannow-receptor mediated clearanfe of microorga”isms has only bee” ob-
216
served by Kupffer
cells (25.26)
and no, by endotbelial
cells. It is possible that the fine differences in ,be carbabydraw specificities of %%~GlcN&
receptors reponed
in
fbis paper mediate the preferential
uptake of different
li-
gands by either Kupffer
orendotbelial
cells.
an heparic binding protein specific far N-aeetylglueoramine-termina,edglyespm,eins.3 Biol Chcm ,977: 252: 65364543.
These sh,dies were supported by the W&come
Trust.
Uptake of processing of glycoproteins by isolated rat hepatic endothelial and Kupffer cells
Mannose- (Man) and N-acetylglucasamine(GlcNAc)-terminated glycoproteins are cleared tiom blood by carbohydratespecific receptors present on both bepatic endothelial and Kupffer cells. It k not known whether the same receptors are present on each cell type or the relative contributions to glycoprotein metabolism made by Kupffer and endadwlial cells. Here we report experiments where data fmm glymprotein metabolism by purified populations of isokted rat hepatic endatlmliil and Kupffer cells have been analyzed by mathematical modelliog and parameter estimatica. Kupffer cells had signifhntly higher binding rate mostants (k’,,) than endotbeliat cells for agalactwmsomuwid (.AG@R) and hyalumnida, bm &ier /q2 (%-rate’) indicating that Kuy~~:cl CC% :.a; ;6&ci affinities for MatVGlcNAc-termittatzd glycqxo. tebts than endathelial receptors. Fmthetmore, &bough endotbelial cells bad similar affinities (k’ll and k,d for AGOR and hyaluronidase, the ‘off-rate’ of Kupffer celk was significantly greater for AGOR than for hyabmmidase, indicating that Kupfkr cell receeptors have lower afIinity for AGOR. Internalization and ligand catabolic rates also differed between the two cell types. ‘I’be data indicate that Kupffer and mdotbelial cells appear to have diffaeot MardGlcNAc recepton and that the destination of a glycoprotein and its subsequent processing is determined by the structure of a glywprotein’s oligosaccharide.
Kupffer
cells (4-6)
and tb,t
iron loading
ana diabetes
Mannose-(Man) and N-aeetylgIumsamine (GkNA+ tetminated glyqmteins are cleared from blood by te-
mellims impair ligaod uptake and processing
ceptms
Most
ceptors (7,8). We have shovm, in the intact rat, that the
studies on ManiCilcNAc receptors have been performed using extrahepatic macrcpbagea, principally because pleural and peritoneal macmphages are easier to prepare than liver sinusoidal cells. These experiments have shown that Man/GlcNAc rece~tots are reclycled (1). down-regulated by high glucose concentrations (2) and up-regulated
‘itmcture of a glycoprotein tigand ini%mces its uptake and praxessing ‘myhepatic ManlGlcNAc recepton (9). In these experiments we have sought to define tUtIter the prqvxties of hepatic sbmsoidal Man/GlcNAc receptors by analyzing glycopmtein me;aboIism by isolated Kupffer and endotbelial cells using mathematical modelling and
by denamethmnne (3). In contrast, there k little information on the properties of the ManlGlcNAc receptors on endothelial and Kupffer cells. It is known that Man/ GlcNAc receptors are present on bath endothelial and
parameter estimatkm. The data cmdirm that @and strw we determines the rams of uptake and processing and show that Kupffer u!k are most active in metaboliing the ligands tested.
on liver sinusoidal
cells and macmphages.
by these re-
A. SAN0
212 Experimental
et at.
pmceduw
The materials and methods have been described iu detail elsewhere (5). Male Sprague-Dawley rats (250-300 g), bred in the Comparative Biology Unit of the Royal Free Hospital, were used. Liver cells wete isolated by e01lagenase perfusion followed by differential centrifugation and centrifugal elutriaticm. Mixed sinusoidal cells were loaded into a Beckman JE6.B elutriator rotor running at 2590 rprn at 15 OC in a Beckman JZ-21 centrifuge (Beckmsn, High Wycombe, U.K.) at u flow rate of 13 mlitnin, and drblis ad lymphocytes were removed by collecting 150 ml. Endothelial cells were collected at 22 udlmin and Kupffer cells at 46 mUmin. The cell fractions were counted and sized using a Coulter Counter and Channelyser (Gndter Electronics, Luton. U.K.). The modal volume was 116 + 2 urn’ (mean & S.E.1 for endothelial cells end 160 i 3 jun’ for Kupffer cells. The proportion of Kupffer cells in each fraction was assessed by histochemical staining for tartrate-resistant acid phosphatase, which is restricted to Kupffer cells (10). We have previously shown that tartrate-resistant acid phosphatase staining gives similar results to endogenous peroxidare staining (5). Kupffer cells accounted for 93 f 5% (n = 5) of the ccllr ic !he Kupffer cell franion and 13 Ir 3% (II = 5) of the cells in the endothelial cell Era&n. The &bility of isolated hepatocytes, assessed by trypan blue exclusion, war 88 f 4% (n = 5). Trypan blue exclusion is not a reliable indication of cell viabiliw so the viabilitv of sinusoidal cells was judged to be satisf&tory on the basis of results for trypan blue exclusion oC!lspatocytes (5).
sinusoidal
Mannose,-Al-BSA (Man-BSA), prepared by the method of Lee et al. (ll), was the generous gift of ProfesSOTY.C. Lee (Baltimore). A highly purified preparstion (240 fXB IIJimg) of bovine testicular hyahtronidase was kindly donated by Biorex Laboratories, London, U.K. Agulactoorosomucoid (AGOR) wus prepared from human orosomucoid by incubation with neuraminidase to yield arialoorosomuwid followed by incubation with @galactosidase (12). The efficiency of removal of the terminal carbohydrate residues was assessed by measuring released sialic acid using a thiobarbitulic acid assay (13) and galactore using galactose dehydmgenase. Seventeen mol of t&h sialic acid and galactose were removed per mol of ororomucoid. Cross-inhibition studies, where the iuhibition by AGOR of [“‘f]asialoorosomucoid (ASOR) uptake by isolated rat he$&acytes was measured, showdd that onlv 1.4% of the AGOR owDaratian was ASOR. Glycopr&ins were rsdioiodinat;d ;ith carrier free Na’“l by the chloramine T method (14). The specific activities of the ligands ranged between 10 ana 2Oj0i/tg.
&pensions of endothelial and Kupffer cells (5-10. lob/ml1 were incubated in Hunks’ hulanced salt solution (HESS) pH 7.4 containing 5 mM glucorr and 25 mM Heper at 37 ‘C for 30 min before the assay to pennir replenishment of cell surface receptors from the intraceliula pool. 5 mM glucose does not inhibit ligund uptake by the Man/GlcNAc reczptor (15). Buffers were gassed with 95% 045% CO, bdore the assay. Assays were stated by adding labelled l&and (14 pmollml) or, to parallel tubes, labelled ligand plus an excess of the inhibitor maunan (500 J&I) and incubated at 37 ‘C with agitation for 90 min. At sequential times of 2.5, 10,15,30,45,60 and 90 miu duplicate sumples (ZDO#l) were removed, placed in 4C@#l micmfuge tubes coutainiug 100 pi silicone oil (5) and 50~1 phasphotungstic acid solution (PTA, 2% w/v in 1 M HQ) and centrifuged for 30 s in a Beckman minofuge B. The cell pellets were removed by slicing through the silicone oil layer and radioactivity iu the cell pellet and supemate was measured in a gamma wunter. Acid precipitabie radioactivity (intact ligand) and acid soluble mdioactivity (ligand catabolite) were determined hy adding 1Klpl 1% (w/v) BSA and 800@ PTA to 100/d of the supernate or the ce!l pellet, cooling the mixture for 10 miu on ice, celltlifuging at 3500 x g for20 min aud measuring the radioactivity in the super&e and precipitate. Specific lfgaud uptake and catabolism were estimated by subtracting the counts in the tubes containing menuan (non-specific uptake) from counts in the tubes containing &and alone.
The mathematical model developed for this study is compartmental as shown in Fig. 1. The model is based on that developed in previous in viva work (9), but is simpler than other models developed by our group (16) and by others, e.g., reference 17. The model equations arc derived as follows.
213 Suppose there are a?, free surface recepton/cell and that this number remains constant over the experiment. (Experiments with reduced ligand d33e did not signifiunlly change the speed of l&and upta’ce and processing. it was thus assumed that there was no aturation of receptors.) The binding flux of ligand ontt the cell (mmollcell per min) is given by
where [L] is the ligand concentration
@moUrnI),
RL is
the amount of surface bound ligand (mmoUcell), and k,, is the binding constant. If there are n cells/mi. then the binding tlux (mmollml) is
where [RL.] is llow a concentration
(mmoUml). Writing x,
= [L], x, = [RL] gives the cawentional mental equation
k,, = “k+,R,
linear cornpart-
(4)
(It is important to note that k,, depends on the three paramctem wbfch cannot be distinguished separately.) k,, is the figand-receptor diswciation rate constant. All kii have the dimension miti’. The equation for bound ligattdx, is
The model paramete.~ were estimated by fitting the model to the data for free ligand fy, = q). cell-asstiated ligand (yz = x2 + x,) and free catabolite fyi = &) obtained in the assays described previously. The weighted residual sum of squares function describiig
the goodies-of-fit
is
given by
where y&) and y&r’, are the data and model values for the ith data poinr on thejth outpat. q,(i) is the standard deviation of each data point. Taking accmmt of ass&y precision and other expetimental errors, a ‘wnstcase value of o&i) = O.ly,,(i) (i.e., a 10% cwf6cientoirtintion of the error) was used. A non-linear optimisation program MINUET (24) was used to find values of the parameters v/&h minim&d the W function, giving a weighted-least squares estimate. These values are taken as the parameter estimates I?,*. Since W has the fotttt of a 2 function, the change in a single parameter which causes W to fncrease by unity giver the 67% confidence on that parameter (a;$ The cc&icient of variation of a parameter estimate is defined as cv = lcc+,/~,,%
The mode, paramaers vlere obtabled for each aeu and ligand type (E, endotbelial; K, Knpffer) and replicated in a number of experimemu. Fwameterdifferences between E and K cells were tested using a one-sample Wikoxou signed-rank test (PlW) of the paired differences, since E and K cells were nut in a paired experiment. Differences between ligands of a given cell type were compared tith a two-sample ?.iann-Whihtey-Wilcoxon rank tat (hw). Care has to b
where k,t is the endqtmic, or intemafisation rate constant (mitt-t). The equations for intemalised and cat&olised (free) ligand are given by
where k,l is the ligand catabolic rate perimental results (Table 2) showed ible catabolite in cells (clO% total represatrcatabcdite in the medium
constant (miti’). Exthat there was negligcatsboli:e), hence x, (as in Fig. 1).
taken when comparing
values of 41,
since from Eqos. 2 and 3 k2, effects the uptake kinetics. but isitselfdepen&nton thefellttumbern. (Nootherpammeters depend on n). In the experiments, cell rmmber varied between 5 and 2!l. ?p cell%tl. Tbe 4, values were therefore sealed to eliminate the effect of this variation
where no is S-106 cells/ml. In this way, comparisons were only being made between differences in the pmdua k+ ,R, (the association constant Kh). From Eqns. 4 and 9 it follows that
214 Res”,ts and
Dlscusion
In this study the in vitro metabolism of tw Van/ tilcNAc-terminated giycoprot+ts by isolated and endothelial cells have beeo canpared. By the USC01 mathematical modelling and paratneter estimation v?:
R~pifeer
have shown that there are diiferenms in the way the two cell types handle theseglycoproteins. All three ligandswere trken up and catabolised by both endothelial and Kuoffer tolls (Fig. 21. For each lieand nnd cell type, cell-assobated ligand rose to a maximum between 20 and 30 min. Ligand catabolite appeared in the tnedium after about 13 min and increased up to 90 min. Less than .O% of lidand catabolite was foeod ssrociated with thecell pellet at any time (Table 1). A mathematical model (based on the compartmental structureshown in Fig. 1) was developed from the dynamic data and parameters of ligand binding, internalization and catdbolism were estimated as described in Experi-
_
mental procedures. The model provided good fits to the data obtaioed for both cell types with AGOR and hyaluronidase (Fig. 2 showsrepresentativefits). The coefficients of variation of the parameter valu:s were all less than 30%. In contrast the fits (not shown) obtained with Man-k&4 were poor and the results were not used in the analysis. Model parameters estimated for the ligands are summatised as their mean and standard error in Table 2. Collagenase
2 5 I” 20 F :
o.as(o.2, 0.98(0.4) 2.9(0.8, 5.Ol1.3) 6.5(1.1) 5.7(0.6) 4.1(rJ.3, 2.7lO.4)
l.l(O.7) 3.2(1.6) 6.1(2.7) 8.2(2.7) 7.4(1.5) 5.2(0.6) 4.2(0.5) 2.0(0.4)
0.98(0.3) 1.7(0.6) 3.8&w) 1.0(1.0, 5.*(w) 3.2(0.14, t.7 (0.32) Lu(O.3)
1.3(0.8, 3.4(1.6) 7.0(3.6, 1”.2(4.0, fx(1.7) 3.1(O.B, l.8Q3.7) l.l(O.3)
Percentage (+ S.E.M.) of&and catablite pre5ent in cellpelletclutbtgthe assay.The remainderof the catabalitewasin the medium. Fordetails we Experimental preced”leS.
perfusion will damage some surface recepton and a 30 min incubation will probably be insufficient to restore them all. However, since the affinity of a receptor is independent of the number of receptors present on the cell the estimater given below are likely to be valid. We cannot, however, exclude the possibility that partial degradatiott of a receptor or damage to tteighbouring molecules might have intloenced their affinity. Kupffer cells had significantly higher binding rata constants (k’,,) than ettdothelial cells for both AGOR @ = 0.05, PIW, see experimental procedures for definition) and hvaluronidasefo = O.W. PlWI. Thisfindine_ could Rfleet &her a greater number of receptors per cell (I?,, on Kupffer cells or that the ‘cm-rate’ (k,,) &,E highe; for Kwffer cell receotors. In contrast k., (the ‘off-rate’) of hyaluronidase & significantly gre&; for eodotIt&l than for Kupffer cells f.p = 0.053, PlW). This difference was not observedwith AGOR. Since the parameter ktl is independent of receptor number these data indicate that ”
.I
Kupffer cell receptors have a higher affinity for glympro-
UPTAKE AND PFXXESSING OF GLYCOPROTEINS
215 similar rates by alveolar macrophages indicating that receptor-ligand internalization is independent of ligand binding in a particular cell type (IS). A” in viva study of glvwprotei” uptake by whole liver also indicated that inwmalizatio” tates were independent of ligand binding rates (9). Cieariy .rperime”b on ahok !iver ~+ch IIICEsorzd the combined internalization rates cd endotneiial and Kupffer cells could not distill&h differences between the two cell types. However this study has shown that endothelial and Kupffer cells intemalized ligands at different rates, suggesting that the mechanism for internalisation is difkrent. It is unclear whether the differences in endothelial and Kupffer cell internaEration rates die a further reflection of differences in receptor stntctme or are due to other factors such as cell surface receptor distribution or receptor coupling to the cell’s endocytotic mechanism. Endothelial cells catabolised AGOR and hyaluroni-
ieins (at IeaSt b~altuonidase) than endothelial ceil recap tars and suggest that although the receptors share similar carE&ydrate specificities they are probably structurally different. Further,wre. sltbo.,gb e”doth&zl cell ieccptors bed similar &i”ities (i.e. similar Ir’,, and similar k,, values) for both AGOR end hytduronidase, the ‘o&rate parameter (k& of the Kopffer cell receptors was significantly greeter for AGOR than for hyaluronidase @ = 0.066, MW). This finding indicates that KupKer cell receptors
dase at a similar rate (k,). Kupffer cells catabolized AGOR at a similar to endothelialcells but hyaltuonidase was more resistant to degmdatiw in these cells (p = 0.0.53, PIW). Futhemtore hyeluronidase was catabolized at a significantly slmer rate than AGOR by Kupffeer cells@ = 0.02, UW). ‘The 6”diag that the lysosmnal e”zyme hyahwonidase was catabolised more slowly the” AGOR by Kupffer cells is consistent with the findings of a” in viva study in which three neoglympmteios were found to be catabolized more rapidly thao the rzmxel xlvcomotei” AGOR (9). Many ridi& have show” that deglyc&ylated proteins k cataboliaed “tore quickly than the corresponding holopro. teins (lE-22) indicating that deglycosylation is the rate limiting step in their catabolism and there is also evidence that proteolysis ~an”ot wxr until the oligosacchfuide
have B lower affinity for AGOR than for hyehuonidax sod provide tioiber evidence that two populations of re-
chains have been removed (23). It follows that hyaluronioligosacchadase, a glycuprotein bearing a high-mantose
rate
ceptors
are probably
different.
This result L ctmsirtent
ride. which is normally resident in the lysommer catabolised
btg rate eatstem for the Ma” terminated gtycoprotei” MatvSSA wes found to be higher than that for the GkNAc-terminal glycopmtein AGOR. This study has also shown tbc higher binding rate co”stant of tbe whole liver for Man-terminated glymproteins is probably due to the contribution of Kupffer cells which are more active
somucoid from which the protective sialic acid raidues have been removed. It is unekar why these differences in catabohc rates are restricted t@Kupffer celix. In conclusian we haveshow” that Kupik-, and endothelial cells appear to have different Mao/GlcNAc receptors (as judged by theft &i”ities for glymproteh ligretds) and
than endothelial
cells in the uptake
of Ma”-
and tlte
G!cNAc-terminated ligands. The rate of intemalisation of AGOR (k& and hyalwonidruc by eodothetial &Is was greater than by Kupffer sellr. This difference reached significance for AGOR @ = 0.05. PlW). IO contrast each eel1 ‘yp internalized the two ligands et similar rates. Another study has show” that different Man-terminated ligand; are intemalizd at
mope slowly the” AGOR a derivative
would be
with the fittdbtgs of a” in tivo study (9) in which the bind-
that the destination
of a glycopmtein
or ore-
and 115subsequent
pacessing %eems to be determined by the stroc~re of the carbohydrate that the glycoprotei” bears. One of the fonctioos imposed for these “pton is the hepatk clearance of potentially harmful proteins or microorganisms such es lysosomal enzymes, bacteria and yeasts. However, it is noreworthy that mannow-receptor mediated clearanfe of microorga”isms has only bee” ob-
216
served by Kupffer
cells (25.26)
and no, by endotbelial
cells. It is possible that the fine differences in ,be carbabydraw specificities of %%~GlcN&
receptors reponed
in
fbis paper mediate the preferential
uptake of different
li-
gands by either Kupffer
orendotbelial
cells.
an heparic binding protein specific far N-aeetylglueoramine-termina,edglyespm,eins.3 Biol Chcm ,977: 252: 65364543.
These sh,dies were supported by the W&come
Trust.