Discussion

advertisement

5 Discussion

5 D

ISCUSSION

The main subjects in this thesis were to study the secretion of a recombinant serglycin in the

MDCK II cell line and in an endothelial cell line. To enable easy detection and isolation of the serglycin molecule, a C-terminal His-flag was added to the protein core. After cloning of this recombinant serglycin-His-flag, both the MDCK II cell line and two different endothelial cell lines (HUVEC and HAEC), were chosen for transfection experiments. Due to the non-successful transfection of the endothelial cells, the project was reduced to study the secretion of serglycin in the MDCK II cell line only. The use of the recombinant serglycin-His-flag required both characterisation of the transfected MDCK II cell clones, at both mRNA and proteoglycan secretion levels, and an optimisation of a Ni

2+

Hi-Trap chelating chromatography system for isolation of the recombinant serglycin-His-flag.

5.1 Expression of serglycin-His-flag in endothelial cells

5.1.1 Transfection of endothelial cells

It was not possible to obtain stably transfected endothelial cells. Both a HUVEC cell line and a

HAEC cell line were transfected, with several transfection methods. The HUVEC cell line was more difficult to transfect than the HAEC cell line, but it was more tolerant against the different transfection reagents. For instance, it tolerated the calcium phosphate method, which killed almost all HEAC cells. Another lipid transfection reagent, lipofectamine, was also tested, but it was found to be more toxic to the HUVEC cell line than the lipofectin reagent (result not shown). Lipofectamine was not tested on HAEC cells.

The initial experiments revealed that the HAEC cell line was easier to transfect than the HUVEC cell line. Thus, since it seemed easier to get stably transfected HAEC cells, only this cell line was chosen for the further transfection experiments. In the optimised experiment, approximately 10 percent of the HAEC cells were transiently transfected with the lipofectin reagent, which should be enough for obtaining stable transfectants. In this, and in the repeated experiments, some transfected cells even survived considerably longer than control cells, which indicated that they were stably transfected. However, it was not possible to expand these single cell colonies, and stably transfected subclones were not obtained.

There are several possibilities why the endothelial cells does not grow after selection. A common observation was that the surviving cells often were alone, without contact with other neighbouring cells. This cell to cell contact may be important for the endothelial cells. During culturing, the endothelia secretes various products, which stimulate growth of neighbouring cells.

With only a few living cells in the well, the concentration of these products may be too low, and thus the cells die, due to lack of growth stimulation and cell communication. This is supported by the observation that, when the wild type HUVEC cells became too diluted after splitting, they often fail to divide, or divided at a very low rate. Interestingly, the transfected cells also often died the day after exchange of the culture medium, which further indicates that this may be a relevant factor. A way to overcome this might be to culture the cells in conditioned medium, or maybe, not to exchange the culturing medium at all, before the cells expand to moderate cell colonies. However, due to the extensive work on characterisation and isolation of serglycin-Hisflag secreted by the transfected MDCK II subclones, we decided to give priority to that work.

129

5 Discussion

Although it was not possible to stably transfect an endothelial cell line, the transfected MDCK II cells have given important information about the GAG attachment on the serglycin core protein.

The proteoglycan synthesis in primary HUVEC cells has been investigated previously (Bergli,

1997). Of the total

35

(S)macromolecules secreted by primary HUVEC cells, 30-40 % is CSPG and 30-40 % is HSPG, equally secreted to both the apical and the basolateral compartments.

Since these endothelial cells also synthesise HSPG, it is possible that the serglycin core protein can be modified with both CS-GAG and HS-GAG in HUVEC cells, as found for the transfected

MDCK II cells. Whether this attachment is regulated, and if the different variants are secreted to different compartments, would be interesting to find out.

5.2 Expression of serglycin-His-flag in MDCK II cells

5.2.1 Characterisation of transfected MDCK II cells

The MDCK II cell line was successfully stably transfected with the recombinant serglycin-Hisfalg, and several different subclones were obtained. The mRNA analysis revealed that the different transfected subclones varied in serglycin-His-flag mRNA expression (figure 4-11.B).

Some of the cells did not express the recombinant serglycin-His-flag molecule, even though they survived the selective marker. It is not determined if these subclones have introduced the pcDNA3.1(serglycin)/Myc-His vector into the genome, but it is unlikely that they could survive the selection pressure marker without expressing the neomycin resistance gene. Most likely the

DNA sequence required for expression of the serglycin-His-flag has been ejected, or divided during the insertion of the pcDNA3.1(serglycin)/Myc-His vector. An insertion of the vector without serglycin mRNA expression is proven for subclone 1-6, since this subclone has altered phenotype. It resembles more a muscle cell, by stretching out, and does not grow to confluency as normal MDCK II cells. One plausible explanation for this, is integration of the vector into a gene that has been inactivated.

The variable serglycin-His-flag mRNA expression levels among the clones expressing the recombinant molecule, is probably due to the site of incorporation into the genome. Factors as chromatin structure and methylation of the area surrounding the inserted vector, are likely to influence the accessibility for transcription of the inserted gene. It could also be that some subclones have integrated multiple copies of the transfected vector into the genome.

There is considerable evidence that the 250 kDa PG analysed represents the recombinant serglycin-His-flag. The serglycin-His-flag mRNA expressing clones 1-7, 1-10 and 4-2 were shown to secrete a 250 kDa PG both to the apical and basolateral media (figure 4-12.B), which is absent in non-transfected clones. This 250 kDa PG also binds with high affinity to the Ni

2+ immobilised Hi-Trap chelating column (figures 4-20.C&D). Tram Thu Voung (1998), has also previously shown that this band is precipitated in immmunoprecipitation experiments, with antibodies against the His-flag. These results together highly favours that the 250 kDa PG represents the serglycin-His-flag gene product. An attempt to identify the 250 kDa PG band with an anti-His-flag antibody did not succeed, due to problems with transfer of the 250 kDa PG onto a blotting membrane, since the highly negatively charged GAG chains do not bind to most protein binding membranes. Instead a cationic membrane (Hybond-N+) was used for electricalblotting in SSC. Unfortunately, the background was to high, and specific detection failed (result not shown). Recently, a new dot blot assay for quantitation of GAGs has been published

(Bjornsson, 1998), where binding of proteoglycans can be increased by wetting the membrane in cationic detergents prior to blotting (Karlsson and Bjornsson, 1998). This dot blot assay can be used to optimise a western blot system for detection of the serglycin-His-flag.

130

5 Discussion

The serglycin-His-flag mRNA expression level for the different subclones, did not correlate with the serglycin-His-flag secretion to apical and basolateral media. The relative secretion of serglycin-His flag, as percentage of the total secretion of

35

(S)proteoglycans, was determined to be 45 percent (apical) and 43 percent (basolateral), respectively, for clone 1-7. The same values for clone 4-2 were 32 percent (apical) and 35 percent (basolateral), respectively (table 4-9). This result does not correlate with the mRNA expression levels for the transfected clones, since clone

4-2 was shown to express more serglycin mRNA than clone 1-7 (figures 4-11.B and 4-12.A).

The reason for this is not investigated in detail. However, it is important to note that the MDCK

II cell line consists of a heterogeneous cell suspension, and that the different isolated subclones therefore may show minor differences in their phenotypes. The different clones can vary in proteoglycan core protein synthesis, in basal GAG synthesis, and sulphation efficiency. A variation in proteoglycan synthesis is indeed observed between the two established MDCK cell strains. The MDCK I cell line secretes 58 % CS and 44 % HS, whereas the MDCK II cell line secretes 31 % CS and 69 % HS (Svennevig et al., 1995). Since the MDCK I cell line is derived by selective culturing of the MDCK II cell line (Simmons, 1981), the MDCK II sub clones may also vary in proteoglycan expression.

Another factor, which may influence on the secretion of the recombinant serglycin-His-flag, is the culturing time prior to metabolic labelling. In the beginning the cells were usually labelled for

20 hours after three days growth on Transwell polycarbonate filter membranes (performed by

Tram Thu Voung). However, in all labelling experiments present in this report, the cells were labelled after four days (all other procedures were identical). Thus, it is interesting that Tram Thu

Voung found clone 1-10 to have the highest 35 (S)macromolecule secretion, while clone 1-7 shows the highest secretion level in all experiment in this report. In an experiment where the proteoglycan synthesis in clone 1-7 was tested by labelling with

35

(S)sulphate for 20 hours 2, 3 and 4 (normal) days after seeding onto the filter membranes, the proteoglycan synthesis was reduced gradually (result not shown). Thus, it might be that the different subclones have variable reduction in proteoglycan synthesis during development of the polarised cell layer. To characterise proteoglycan synthesis in the subclones more, the experiment above should be performed for all subclones, including the wild type MDCK II cell line.

The different subclones did also show minor differences during the culturing, which may be an effect of the transfected serglycin-His-flag. Subclone 1-10 was usually found to reach confluency first, when cultured in 75 cm

2

flasks, followed by clones 4-2, 1-7 and 7-4. This was not due to a higher cell number in the culture flask, which actually was found to be lower in a confluent layer for clone 1-10 than for the other clones (result not shown). It thus seems that a high serglycin synthesis stimulates cell division and formation of the confluent cell layer. Clone 1-10, also seems to have a reduced endogenous proteoglycan synthesis, compared to the other tested subclones (figure 12.B). It may be that a high expression level of serglycin-core protein reduces the possibility for GAG attachment and sulphation of the wild type proteoglycans. The serglycin core protein could in this clone function similarly to xylosides, which prime GAG synthesis and reduce the attachment of GAGs to the endogenous proteoglycan core proteins (Kolset et al.,

1990). In that chase, the serglycin-His-flag core protein is likely to influence both HS- and CS-

GAG synthesis. Thus, the growth stimulating effect may be caused by reduced GAG attachment/ sulphation of wild type proteoglycans, and/or the high presence of serglycin-His-flag. For the reasons above, clone 1-10 was found to differ from the wild type MDCK II cell line, and was not used to study the polarised secretion of the serglycin-His-flag.

131

5 Discussion

The presence of serglycin-His-flag in intracellular fractions was not analysed, since the aim of the study was to study the secretion of the serglycin-His-flag. However, the high presence of recombinant serglycin-His-flag extracellularly, indicated that most was secreted. A pulselabelling experiment could confirm this, but was not performed. Even though serglycin is stored in intracellular granules (with bound proteases) and released by stimuli in mast cells (Hunt et al.,

1997), it is unlikely that the epithelial cell line MDCK II could store the serglycin-His-flag intracellularly in a similar way. In one experiment, the total

35

(S)macromolecules in the apical, the basolateral and the cell fractions was counted (appendix 1.2). This experiment revealed that the intracellular PG in serglycin-His-flag mRNA producing clones, increased proportional with the apical and the basolateral secretion.

5.2.2 Optimising the method for Ni 2+ chelating chromatography

When starting to optimise the method for purification of serglycin-His-flag by metal chelating chromatography, only two studies were found that used his-flag fusion protein for isolation of proteoglycans. A recombinant decorin fusion protein (Ramamurthy et al., 1996) and a recombinant biglycan fusion protein (Hocking et al., 1996) were both isolated by nickel chelating chromatography. However, purification of these two proteoglycans was considered to be simpler, because these small proteoglycans have only one and two GAG-attachment sites, respectively, and both are modified with chondroitin/dermatan sulphate glycosaminoglycan chains in vivo. In addition, the proteoglycans were expressed in HT-1080 cells, derived from fibrosarcoma, which has a low HS synthesis. Serglycin, on the other hand, contains 8 attachment sites, arranged in close position in the middle of the core protein (Humphries et al., 1992), and was found to be modified with both CS and HS when expressed in the MDCK II cells (figure 4.13). The domination of glycosaminoglycan chains, compared to core protein in the serglycin-His-flag, could make the isolation of this proteoglycan more difficult. The basal HS synthesis in the

MDCK II cell line is also high, as mentioned before, which further could influence on the isolation of the recombinant protein.

Previous studies had shown that CSPG does not attach to either Co

2+

or Cu

2+

immobilised matrix during isolation of perforin and granzymes from cytotoxic lymphocytes (Winkler et al., 1996), and that re-injected decorin after removal of the His-flag by protease, did not attach to the Ni 2+ column (Ramamurthy et al., 1996). CS-GAG chains were thus not expected to interfere with the system. Meanwhile, it was not known if the high density of the GAG chains, or the presence of

HS-GAG chains attached to the core protein, would influence on the purification of the recombinant serglycin-His-flag. Thus, there was a need to optimise the conditions for proper purification, to be able to purify the recombinant serglycin-His-flag.

In the pilot experiments, where elution of the bound serglycin-His-flag was facilitated with reduction in pH or increasing imidazole concentrations in the buffer, it was found that some

35

(S)macromolecules secreted from the moc-transfected clone 7-4 bound with low affinity to the column (figure 4-15). The unspecifically bound

35

(S)macromolecules were later characterised as

HS proteoglycans normally expressed in the wild type MDCK II cell line, and were found to have a large molecular weight, since they did not enter the gel during SDS-PAGE (figure 4-

20.C&D), and eluted in V

0

during the Superose 6 gel filtration analysis (figures 4-21 and 4-22).

It is not fully determined if the HS-GAG chains in this proteoglycan mediate the unspecific binding. This could be tested by

-elimination, where the GAG chains are chemically cleaved off from the core protein, prior to purification with the Hi-Trap chelating system. However, we

132

5 Discussion found this analysis unnecessary, since other HS proteoglycans did not attach to the column. Thus, it seems unlikely that HS-GAG chains alone are able to mediate binding to the column, but rather the proteoglycan binds by another motif, or several motifs working together. This is also supported by the observation that it was not possible to separate the unspecifically bound proteoglycan from the specifically bound serglycin-His-flag by the reduction in pH, even if it could be achieved with increasing imidazole concentration. Most likely, the unspecifically bound proteoglycan contains two or more histidines in a proximal position, which are able to mediate a weak interaction with the column compared to the high affinity mediated by the six repeated histidines in the recombinant serglycin-His-flag. The histidines in both molecules are likely to accept protons within the same pH range, and thus both molecules will elute with the same pH.

In an attempt to reduce the unspecific binding, the Ni

2+

ions immobilised on the Hi-Trap chelating column were exchanged with Co

2+

. The bound material was not analysed, but less material in the high affinity fraction indicated that recombinant serglycin-His-flag either was not bound to the column, or was eluted in the low affinity fraction (figure 4-17). Thus, Co

2+

ions are not suitable for purification of serglycin-His-flag. Other ions were not tried out, but it may be that other ions are better than Ni

2+

ions, for instance Cu

2+

ions, which are frequently used in metal chelating chromatography. Immobilised Cu

2+

ions should mediate stronger binding, according to the common Cu

2+

> Ni

2+

pattern (Sloane et al., 1996). However, increased specific binding can also be followed by increased unspecific binding, and thus, it will not necessarily increase the purity of the isolated serglycin-His-flag.

The serglycin-His-flag bound with both low and high affinity to the Hi-Trap chelating column.

Except for more CS-GAG attached to the serglycin-His-flag in the high affinity peak, other structural differences between the low and high affinity fractions were not found. The retention values are nearly identical for both variants (peak 1: K av

= 0.25, peak 2: K av

= 0.28), which should be considered as not significantly different, due to the broad peaks (K av

= 0.1 to 0.5 for both, see figures 4-21 to 4-24). The molecular weight of the eluted serglycin-His-flag, in both the low and high affinity fractions is thus nearly identical. Nevertheless, for unknown reasons, the serglycin-

His-flag modified with HS seems to mediate a weaker binding to the Hi-Trap chelating column than the variant modified with CS. This may also be considered as another prove that the heparan sulphate chain in the unspecific bound proteoglycan, not are responsible for the unspecific binding to the Hi-Trap chelating column, as indicated above.

The observation that the serglycin-His-flag isolated from polarised MDCK II cells seems to have a lower affinity for the Hi-Trap chelating column is not investigated further (compare figures 4-

15.D and 4-18). However, a reasonable explanation is that the heparan sulphate synthesis is more active in non-polarised cultures of MDCK II cells.

Some of the experiments done during the optimising of the Hi-Trap chelating column system, also indicated that the binding of serglycin-His-flag varied with the ion concentration in the running buffer. The binding of HS-GAG modified serglycin-His-flag may be reversly related to the ion concentration in the running buffer. In an experiment identical to the one presented in chapter 4.5.2, with the exception that the running buffers contained 500 mM NaCl instead of 150 mM, more serglycin-His-flag eluted in peak 1 and less serglycin-His-flag eluted in peak 2. In this experiment, the serglycin-His-flag eluted in the peak 1 fraction was modified with HS-GAG only

(both apical and basolateral), whereas serglycin-His-flag eluted in peak 2 was modified with both

CS- and HS-GAG (result not shown). This indicates that increased ion concentration, favours separation of CS- and HS-serglycin-His-flag. However, unspecifically bound proteoglycans were also eluted in peak 1, when 500 mM NaCl was present in the running buffer. Thus, low ionic

133

5 Discussion strength in the running buffer seems to facilitate more pure serglycin-His-flag eluted in peak 2, whereas high ionic strength seems to increase the ratio of CS- compared to HS-serglycin-His-flag in peak 2.

In the initial experiments, low recovery after Hi-Trap chelating chromatography was observed when aliquots of the samples was counted in a scintillation counter. This was not observed when the samples were collected directly in scintillation tubes and the whole fraction counted. It was later found that the proteoglycans adhered to the collecting tube wall when eluted from the Hi-

Trap chelating chromatography (detected by the experiment in appendix 3). Binding of PGs to the tube wall was not observed for eppendorf tubes, which were used for collecting of sample in all experiments presented in his thesis.

5.2.3 Glycosaminoglycan attachment on serglycin-His-flag core protein

The MDCK II cells were transfected with the recombinant serglycin-His-flag to study the secretion of an easily detectable chondroitin sulphate proteoglycan in MDCK II cells. Serglycin has almost exclusively been reported to be a CS proteoglycan (Kolset and Gallagher, 1990), with the exception of a CS and heparin hybrid in mouse mastocytoma cells (Lidholt et al., 1995). We therefore expected the serglycin-His-flag to be modified with chondroitin sulphate chains when expressed in MDCK II cells. The glycosaminoglycan analysis, however, revealed that the serglycin-His-flag was modified with both CS and HS. In clone 1-7, the secreted serglycin-Hisflag was modified with ~40 percent CS and ~60 percent HS, whereas the same values for clone

4-2, were ~30 percent CS and ~70 percent HS (table 4-10).

The HS-GAG attachment on the serglycin-His-flag core protein in the MDCK II cells, indicates that the type of GAG chain attached to a GAG attachment site in a proteoglycan is dependent of several factors. In which way the cell is able to decide the attachment of a particular type of GAG to an attachment site, is not fully clarified. Several factors have been postulated to influence what type of GAG that is elongated on the linkage region. These include (a) the amino acid sequence flanking the serine residue, (b) the access of UDP-sugars and the presence of GAG synthesising enzymes in the Golgi apparatus, and (c) phosphorylation or sulphation of the linkage region (see introduction). The presence of both CS- and HS-GAG chains attached to the same type of core protein, also makes it possible that the proteoglycan is secreted as a hybrid, consisting of both

CS- and HS-GAG chains on the same core protein. As mentioned above, the serglycin core protein is secreted as a CS proteoglycan in hematopoietic cells, or as a hybrid consisting of both

CS and heparin in mouse mastocytoma cells, which proves that such hybrids can be synthesised.

Serglycin contains a typical HS assembly site

To try to find out why the serglycin-His-flag core protein was modified with HS-GAG chains in

MDCK II cells, we examined the GAG attachment site in serglycin and compared it with HS assembly sites in other HSPG. The GAG attachment region of serglycin, which contains a multiple Ser-Gly region flanked by acidic acids (complete amino acid sequence for serglycin-

His-flag in appendix 1), was indeed similar to what is suspected to be the consensus GAG attachment site for heparan sulphate assembly (Zhang et al., 1995). Thus, since serglycin contains a preferred HS attachment site, the observation that the serglycin core protein is found to be modified with CS-GAG in most cells studied, indicates that the attachment of HS-GAG may be a cell specific phenomenon.

134

5 Discussion

Attachment with another type of GAG than CS on the serglycin core protein is described only once, when secreted as hybrid, consisting of both CS and heparin, by mouse mastocytoma cells

(Lidholt et al., 1995). However, since heparin and HS only differ in the amount of epimerisation and sulphation, the polymerisation of the disaccharides is probably initiated by the same transferases, and both initially synthesised as HS-GAG chains. This, together with our results, clearly demonstrates that the GAG attachment region alone, can not be the only factor that determines which type of GAG to be synthesised on a core protein. In some way, also the GAG synthesising machinery in the Golgi apparatus determines the type of GAG chains to be attached on the core protein.

The CS- and HS-GAG attachment on the serglycin-His-flag shows that the same attachment region can be modified with both types of GAG chains in the same cell. However, this is not observed for all GAG attachment sites. Syndecan-1 contains two GAG attachment regions, where the N-terminal primes both CS-GAG and HS-GAG attachment, whereas the other primes

CS-GAG attachment only, when expressed in Chinese hamster ovary cells (Zhang et al., 1995).

The observation that HS sites always are capable to prime CS synthesis, but that some sites primes CS synthesis only, is probably an important control point. The cell can avoid HS assembly and prime CS assembly on an HS site, simply by repressing the HS synthesis. This may also explain why serglycin mostly has been described as a CSPG. The proteoglycan has so far, with one exception, not yet been studied in cells able to prime HS synthesis. If the hematopoietic cells control the repression of the HS synthesis with absence of heparan sulphate transferases in the Golgi apparatus, or other factors, for instance inhibition of transport of UDP-GlcNAc into the

Golgi lumen, should be interesting to know.

Serglycin is secreted either as a CS or as a HS proteoglycan

The observation that serglycin-His-flag is modified with both CS- and HS-GAG also render it possible that the same core protein is secreted as a hybrid, modified with both GAG chains. A fraction of the serglycin is secreted as a hybrid, consisting of both CS and heparin, in mouse mastocytoma cells (Lidholt et al., 1995), which indicates that the core protein may prime hybrids.

Hybrid proteoglycans can be detected by degradation of either the CS-GAG or the HS-GAG chains attached to the protein core, which results in reduced molecular weight of the remaining molecule. This shift in the molecular weight of the remaining proteoglycan can sometimes be detected by both gel filtration and PAGE analysis, as observed for a cell surface hybrid proteoglycan made by mouse mammary epithelial cells (Rapraeger et al., 1985). This hybrid proteoglycan, with a total molecular weight ranging from 130-260 kDa, was estimated to contain one or two CS-GAG chains for every fourth HS-GAG chains.

The experimental procedures used in this thesis are not extensive enough to fully conclude if the serglycin-His-flag synthesised and secreted by MDCK II cells is a hybrid consisting of both CS- and HS-GAG chains, or separate CS and HS proteoglycans. So far, however, no results favour the existence of a hybrid, and thus it seems that the core protein is modified with either CS- or

HS-GAG chains. The serglycin-His-flag eluted in peak 1 consists of 20-30 % CS and 70-80 %

HS (apical) and 100 % HS (basolateral) (table 4-8), which proves that some serglycin-His-flag core proteins are attached with HS chains only. The serglycin-His-flag eluted in peak 2 contains

55-65 % CS and 45-35 % HS, and thus a hybrid serglycin-His-flag would probably contain approximately the same amount of both CS- and HS-GAG chains attached to the same core protein. This would reduce the molecular weight from 250 kDa to about the half after degradation of one type of GAG chains. Neither the SDS-PAGE (figure 4-13) nor the Superose 6 gel filtration analysis (figures 4-21 to 4-24) of chondroitinase ABC lyase or nitrous acid treated samples, showed indication of a reduced molecular weight of the remaining proteoglycan after

135

5 Discussion degradation of one type of GAGs. The serglycin-His-flag eluted with K av

0.25 both before and after treatment. In comparison, gel filtration analysis of free GAG chains attached to the serglycin-His-flag (released by

-elimination) eluted with K av

=0.5-0.6 (result not shown). These free GAG chains are estimated to have a molecular weight larger than 11 kDa, by comparison with the GAG chains attached on the serglycin secreted by U937 cells (Mustorp, 1997).

However, the possibility of a hybrid serglycin-his-flag can not yet be excluded, and should be examined further, for instance by anion exchange chromatography. Since the CS- and HS-GAG show different affinity for anion exchangers, the purified serglycin His-flag is expected to elute in two peaks if it contains separate GAG chains, and in a single peak if it is a hybrid proteoglycan. Moreover, degradation of the chondroitin sulphate chains in a hybrid, with chondroitinase ABC lyase treatment, is expected to reduce the remaining proteoglycans affinity for the anion exchanger

1

.

As mentioned previously, the serglycin core protein contains 8 attachment sites, arranged in close position in the middle of the core protein (Humphries et al., 1992). It would be interesting to know if some serines in the attachment site in serglycin-His-flag show a preferred attachment for either CS- or HS-GAG in MDCK II cells. Even if all serines are potential attachment sites, it is not fully documented that all serines are modified with GAG chains. However, no known good technique exists, which could be used to examine this, since the repeating Ser-Gly attachment region is shown to be resistant to proteases (Stevens et al., 1985). For the same reason, determination of preferred CS- and HS-GAG attachment sites in the GAG-attachment region is a great challenge.

Factors with may determine either CS- or HS-GAG synthesis on the core protein

The fact that serglycin is secreted as a hybrid by mouse mastocytoma cells (Lidholt et al., 1995), proves that such hybrids exist. Since hybrids seems to be absent in MDCK II cells, there must be some differences in the glycosaminoglycan synthesis machinery between MDCK II cells and mouse mastocytoma cells.

It could be that the synthesis of one type of GAG chain excludes the possibility of modification with the other type of GAG chain in MDCK II cells. The cell could control this by having separate membrane locations for the CS and HS synthesis machinery in the Golgi apparatus, as hypotheses by Silbert and Sugumaran (1995). Transport of the proteoglycan core protein to a site, which contains GlcNAc transferase I, or alternatively, GalNAc transferase I, would be a critical process. Certain sequences in the core protein, or modifications such as sulphation and phosphorylation of the linkage region, could regulate this transport. The linkage region in proteoglycans from various sources has been studied, which has shown that C-2 of xylose is a major site for phosphorylation in both chondroitin sulphate (Oegema et al., 1984; Sugahara et al.,

1992) and heparan sulphate proteoglycans (Fransson et al., 1985). Structural studies have also revealed that sulphation of C-6 of both Gal residues and C-4 of Gal adjacent to GlcA was characteristic for chondroitin sulphate (de Waard P. et al., 1992; Sugahara et al., 1988; Sugahara et al., 1991; Sugahara et al., 1992), whereas, sulphation of the two Gal residues in the linkage region was not found in heparin or HS proteoglycans (Sugahara et al., 1992; Sugahara et al.,

1994). Sulphate and phosphate has never been found in the same linkage region. Thus,

1 For unknown reasons, the chondroitin sulphate chain mediates a stronger binding to an anion exchanger than the heparan sulphate chain, even though the chondroitin sulphate chain is generally less negatively charged.

136

5 Discussion sulphation of Gal or phosphorylation of xylose in the linkage region can be signals for specific

GAG synthesis, by targeting the core protein to the right membrane location.

The secretion of the recombinant serglycin-His-flag in MDCK II cells would be an excellent tool for studying this, by investigating potential differences in the linkage region between the CS- and the HS-GAG modified serglycin-His-flag. It would be interesting to know if any of the galactoses in the linkage region are sulphated, if the xylose is phosphorylated, and especially if there are any differences in the linkage region of the CS- and the HS-GAG modified serglycin-His-Flag. For instance, phosphorylation may be required for HS-GAG elongation, which could be prevented by sulphation of the Gal residues (Figure 5-1). Xylose is generally non-phosphorylated in proteoglycans derived form tissues with an abundant extracellular matrix (HS proteoglycans)

(Cheng et al., 1996; Sugahara et al., 1995a). However, only the secreted proteoglycans are analysed, and it is not unlikely that the phosphate is removed prior to secretion, since addition of the first GlcA in the biosynthesis of decorin is followed by rapid dephosphorylation in fibroblasts cells (Moses et al., 1997). Phosphorylation could then be necessary for guiding of the proteoglycans to a site for GAG elongation.

GlcA

Gal

Gal

Xyl

O

Ser

S P

CS-synthesis CS-synthesis

HS-synthesis

Figure 5-1: Possible role of sulphation and phosphorylation of linkage region.

The transfer of the core protein to separate membrane locations for CS and HS synthesis machinery in the Golgi apparatus, is not necessarily determined by signals in the linkage region.

The linkage region in both types of GAGs could also be identical, and selection for CS or HS synthesis mediated by other factors. If so, the selection for GAG synthesis must lie in the HS synthesis machinery because it is shown that the HS synthesis requires specific structures on the xylosides, while the CS synthesis is primed on almost all xylosides (Fritz et al., 1994; Fritz et al.,

1997). In addition, one of the GAG attachment sites in syndecan-1 seems to never be modified by

HS-GAG. This site is modified with CS-GAG, both in wild type CHO cells, and in a mutant

CHO cell line (CHO 1d1 D) (Zhang et al., 1995). This mutant CHO cell line has a deficiency in the 4'-epimerase that catalyses GlcNAc

GalNAc formation, and thus makes less CS when deprived of endogenous GalNAc (Esko et al., 1988). Still, with minimal CS synthesis, one of the

GAG attachment regions in syndecan-1 never primes HS synthesis, which indicates that HS synthesis seems to be more controlled than CS synthesis.

137

5 Discussion

Without signals in the linkage region, one possible way to prevent CS-GAG assembly on HS-

GAG sites, is to locate the site for HS synthesis more proximal in the Golgi apparatus than that for CS synthesis. As the core protein is transported through the Golgi apparatus, the linkage region is modified with HS-GAG if it contain an appropriate HS-GAG assembly site, and if not, the core protein is passed on to the site for CS synthesis. Indeed, treatment with BFA has revealed that the HS assembly is primed more proximal in the Golgi apparatus than CS synthesis in both rat ovarian granulosa cells (Uhlin-Hansen and Yanagishita, 1993) and in melanoma cells

(Spiro et al., 1991). Separate Golgi localisation of the CS and HS machinery has not been determined in MDCK cells, because MDCK cells and PtK cells (also a kidney epithelial cell line) are resistant to the drug BFA (Ktistakis et al., 1991; Sandvig et al., 1991). However, it is thought that the epithelial cell line MDCK II also has separate Golgi localisation of the CS and HS synthesis machinery, since sulphation of CS and HS has different sensitivity to the inhibitor chlorate, which inhibits the formation of PAPS (Prydz, personal communication).

Thus, the different membrane localisation of the CS and HS synthesis machinery, may directly function to avoid hybrid modification of typical HS attachment regions. In this context, it is very interesting that the mouse mastocytoma cell line, found to secrete hybrid serglycin with both CS-

GAG and heparin chains (Lidholt et al., 1995), is found to have co-localised CS and HS synthesis machinery (Uhlin-Hansen et al., 1997). Even with different membrane localisation of the CS and HS synthesis enzymes, a hybrid could be formed, however, if the core protein has a poor HS-GAG attachment region, which then will be non- modified at the HS assembly site, and first modified when the core protein reaches the site for CS assembly. A proximal localisation of the HS synthesis enzymes in the Golgi apparatus, may thus be an important control factor determining the type of GAG to be synthesised on the core protein.

This model does not, however, explain why the serglycin-His-flag core protein is modified with both CS-GAG and HS-GAG in MDCK II cells. Since serglycin contains a HS-GAG assembly site, it would be expected that all the serglycin-His-flag core proteins were primed with HS-

GAG. However, it could be that, in addition to have a strict location of the enzymes involved in the GAG synthesis in the Golgi apparatus, the enzyme concentrations, or the amount of the UDPsaccharides in the Golgi lumen, also are strictly controlled. The MDCK II cell could control the efficiency of the GAG synthesis machinery by regulating the transport of the different UDPsaccharides into the Golgi lumen by the activity of the UDP-translocators (Hirschberg et al.,

1998). Thus, with a high core protein content in the Golgi apparatus, the availability of either

UDP-GlcNAc or the different HS synthesising enzymes may be less than required to assemble

HS-GAG on all HS sites.

Thus, it is very interesting that, even if the analysis of the GAG-attachment in clone 1-7 and 4-2 revealed that clone 1-7 secretes more serglycin-His-flag than clone 4-2, the HS-GAG modification in both clones was nearly identical (table 4-9). It could be that the CS synthesis machinery is more dominant clone 1-7, but it is also possible that the HS synthesis machinery is nearly saturated in both clones, and that the excess core protein is transported to the site of CS-

GAG attachment. A saturation of the HS synthesis machinery in clone 1-7 and 4-2 is not unlikely, since the wild type PG synthesis was reduced in the high serglycin-His-flag expressing clone 1-10, even if this clone secreted less

35

(S)macromolecules than both clone 1-7 and 4-2

(figure 4-12.B). The CS- versus HS-GAG attachment in clone 1-10 is not analysed, but it should be interesting to know if this clone primes even more CS-GAG on the serglycin-His-flag than clone 1-7. Another factor facilitating a possible saturation of the HS synthesis machinery, is that the MDCK I cell line, which is found to secrete more proteoglycans into the medium than

138

5 Discussion

MDCK II, also primes more CS proteoglycans (Svennevig et al., 1995). A tight control of the HS synthesis machinery may thus be important for the CS- versus HS-GAG assembly in MDCK cells. If this is controlled by the different chain elongating enzymes, or the concentration of

GlcNAc in the Golgi apparatus should be interesting to know.

With a nearly saturated HS synthesis machinery, it could be anticipated that this would increase the possibility of hybrid formation, and thus hybrid serglycin-His-flag should be present.

However, this is not necessarily true. The serglycin core protein has a highly concentrated glycosaminoglycan attachment site, with the eight sites within an 18 amino acid peptide sequence

(Humphries et al., 1992). Recognition and capture of the core protein to the location of the HS synthesis machinery is thus likely to be followed by GlcNAc transfer to all linkage regions synthesised on the core protein, followed by further elongation of HS-GAG. So, if all HS assembly sites in the Golgi apparatus are occupied, the core protein will escape and reach the site for CS synthesis, and thus no hybrids are formed. The different factors, which may determine the type of GAG attached to the core protein is summarised beneath (Figure 5-2).

TGN

4 3

CS-synthesis

HS-Synthesis

2

1

Formation of linkage region

Figure 5-2: Model for factors involved GAG attachment in MDCK II cells.

The linkage region is formed, as the core protein is transported through the Golgi apparatus. (1): During this process, the linkage region may be modified with signals that facilitate either CS or HS polymerisation on the linkage region (phosphorylation of Xyl, sulphation of Gal). (2): When the proteoglycan reaches the site for HS synthesis, the linkage region may be polymerised with HS-GAG. HS-GAG assembling requires that the core protein contains a HS assembling site, phosphorylation of Xyl may be required, and UDP-GlcNAc, UDP-GlcA and HS polymerising enzymes must be available. (3): Then the core protein reach the site for CS synthesis, where all linkage regions not modified with HS-GAG, are polymerised with CS-GAG. Sulphation of the linkage region may prevent HS assembly or facilitate CS synthesis. (4): After the polymerisation is initiated, the core protein is sulphated (and epimerised for HS) in various positions. These modifications occur simultaneously with the polymerisation reactions.

139

5 Discussion

140

5 Discussion

5.2.4 Polarised secretion of the recombinant serglycin-His-flag in MDCK II cells

It is important to understand the molecular mechanisms that govern sorting of cell proteins, since the distribution of the proteins is important for the function of the cells. The importance of receptor localisation and functions in signal transduction, is observed in several diseases, which are caused by changes in receptors localisation or mutations in sorting determinants. A lack of sorting signal may have fatale consequences. In familiar hypercholesterolemia, the low density lipoprotein receptor is unable to internalise, and can not transport cholesterol for storage as cholesterol esters (Brown and Goldstein, 1986). Lysosomal storage disease is caused by genetic defects that effect one or more of the lysosomal hydrolases, and results in accumulation of their undigested substrates in the lysosomes. In inclusion cell (I-cell) disease, the missorting of the hydrolytic enzymes has been traced to a defective or missing GlcNAc-transferase, which phosphorylates the hydrolytic enzymes in the cis-Golgi. Thus, the non-phosphorylated lysosomal enzymes are not recognised by the M6P receptor, and not packaged into lysosomal transport vesicles in the trans-Golgi network (Alberts et al., 1994).

The study of mechanisms underlying the molecular sorting of some molecules, has been intensified during the last decade, due to the discovery of the first sorting signals. The now well established apical targeting signal, the GPI-anchor, was first demonstrated to have this function by Brown et al. (1989). GPI-anchored proteins are mainly localised to the apical plasma membrane in MDCK II cells (Lisanti et al., 1988), and are also included in the detergent-resistant sheets ("rafts") formed in the TGN (Brown and Rose, 1992). Recently, it has however become clear that the GPI-anchor may not be an apical targeting signal, as previously expected.

Examination of GPI-anchored proteins found to be apically secreted in previous experiments, has revealed that all of them contained N-liked saccharides, the another characterised apical targeting signal (Gut et al., 1998). Removal of these saccharides from the core proteins, result in nonpolarised secretion of the GPI-anchored proteins (Benting et al., Simons lab, unpublished data).

The role of GPI-anchor in apical targeting is thus unclear, but it may be important for anchoring of the protein near the membrane, which then could facilitate the packaging into the apical transport vesicle.

Recent models for apical transport, should thus be restricted to employ N-glycosylation as the sorting determinant, and sphingolipid-cholesterol "rafts" as platforms (Simons and Ikonen,

1997). "Rafts"-association is found for several apically targeted molecules, like influenza virus

HA, in MDCK II cells (Scheiffele et al., 1997). A lectin-like protein, VIP-36, was isolated from apical transport vesicles and found to associate with "rafts" (Fiedler et al., 1994).

Characterisation of VIP-36 revealed that the molecule was able to bind Ca 2+ and GalNAc

(Fiedler and Simons, 1996), and was thus, initially by us, suspected as a candidate for binding and targeting of CS to the apical surface in MDCK II cells. However, further experiments have revealed that the binding-coefficient for complex-formation between VIP-36 and GalNAc is too high for having any biological relevance (Fullekrug et al., unpublished data).

In MDCK II cells, chondroitin sulphate chains are secreted to the apical medium, but not to the basolateral medium. The observed apical targeting of CS and a dominating basolateral targeting of HS in MDCK II cells, indicates that the proteoglycans have functions on opposite sides of the epithelial layer. This may be due to the physiological function of the kidney epithelium. As mentioned in the introduction, the epithelium of the distal and proximal tubules is involved in the filtration and re-absorption of nutrients in the kidney (figure 1.2). One of the major properties of the basal lamina is to prevent leakage of charged molecules between the blood and the filtrate.

Reduced sulphation of HS proteoglycans, or altered GAG attachment in the basal lamina of the

141

5 Discussion kidney, often leads to dysfunctional basal lamina (Kasinath, 1995; Kasinath et al., 1996; Kolset et al., 1994). Increased delivery of CS proteoglycans, which are less sulphated, and thus, less negatively charged than HS proteoglycans, may also reduce the charge specific repulsion of the basal lamina. Thus, exclusion of CS proteoglycans from basolateral secretion may be important for the formation and the maintenance of the basal lamina.

The wild type proteoglycans secreted by the MDCK II cell line have not been well studied, but most HS proteoglycans secreted basolaterally are presumably matrix components. Antibodies against perlecan and versican revealed that both proteoglycans were secreted by MDCK II cells

(Svennevig et al., 1995). Perlecan (Mw ~470 kDa) is a basement membrane proteoglycan (HS), and is in part responsible for the charge specific ultrafiltration of this extracellular matrix.

Versican (CS) (Mw ~260 kDa) contains domains that are highly homologous to aggrecan, and may play a role in intracellular signalling, cell recognition and to connect extracellular matrix components and cell surface glycoproteins (Ayad et al., 1994). Most versican was found to be cell associated, while perlecan was mostly secreted. However, apical transport of both of these proteoglycans was lower than the basolateral (Svennevig et al., 1995).

In the first experiments, to characterise the proteoglycans secreted by the two MDCK strains, if was found that the MDCK II cell line secreted a low amount of CSPG to the basolateral compartment (Svennevig et al., 1995). Later experiments, with

35

(S)sulphate labelling of polarised cultured MDCK II cells, have, however, revealed that the secreted CS proteoglycan(s) is sorted to the apical compartment exclusively (Vuong, 1998). Variations in growth conditions were found to not influence on the apical sorting of CS proteoglycan(s), which indicated that the apical sorting was a stable phenomenon. Why CS was found basolateral in the first experiments, is still not known.

Treatment of MDCK II cells with hexyl-thio-

-D-xylosides, which primes CS synthesis efficiently in MDCK II cells, also results in apical secretion of the major free CS-GAG synthesised (Kolset et al., 1999). Thus the CS-GAG chain may contain a signal for apical targeting in MDCK II cells, which is not unlikely since also other saccharides may facilitate in apical targeting. It is now well documented that N-glycans presumably function as an apical sorting signal (Scheiffele et al., 1995), and in addition that O-glycosylation also might facilitate apical targeting in MDCK II cells (Zimmer et al., 1997). It is still not known if the apical secretion of CSPG in MDCK II cells is independent of "rafts" and glycolipids. Some experiments indicate this, since treatment with mannosamin and fumonisin B

1

, which inhibit GPI-anchor formation and glycolipid synthesis, respectively, did not affect the apical secretion of the CS proteoglycan(s) (Vuong, 1998). Apical secretion without "rafts"-association has also been found for other molecules. Both the N-linked and O-glycosylated bovine enteropeptidase (Zheng et al.,

1999), and the non-glycosylated CD3-

are delivered to the apical surface without integration into "rafts" in MDCK cells (Alonso et al., 1997). The previous results, thus indicate that the sorting of CSPG to the apical surface, probably is mediated by an alternative non-characterised sorting machinery. It should however be considered that detection of "rafts"-association is difficult, and thus the involvement of "rafts" in apical targeting of CS should still not be excluded.

To study the role of the CS-GAG chain in apical targeting, we transfected the MDCK II cell line with serglycin-His-flag to determine the secretion of an easily detectable CS proteoglycan. As mentioned already, the serglycin-His-flag core protein was modified with both CS- and HS-GAG chains, which complicated the study. Purification of the two variants of serglycin-His-flag followed by glycosaminoglycan analysis was therefore necessary to determine the secretion of

142

5 Discussion the CS- and the HS-GAG modified variants of the serglycin-His-flag. This revealed that the serglycin-His-flag secreted by clones 1-7 and 4-2 were modified with 30-40 percent CS and 60-

70 percent HS. Approximately 20 % (10 % CS and 10 % HS) was secreted apically, while approximately 80 % (20-25 % CS and 55-60 % HS) was secreted basolaterally (table 4-10). Even if clone 1-7 secreted more CS-serglycin-His-flag than clone 4-2, the sorting of the synthesised

CS-serglycin-His-flag was 1/3 apical and 2/3 basolateral for both clones, and the secretion of HSserglycin-His flag was 1/6 apical and 5/6 basolateral.

The glycosaminoglycan analysis presented in table 4-10 is determined from one experiment only.

An exact determination of the CS- versus HS-GAG attachment, and apical versus basolateral secretion of the CS- and HS-GAG modified form of serglycin-His-flag has not been carried out, since this would require several repetitions of the experiments for statistical analysis. However, we found this unnecessary, since both clones 1-7 and 4-2 showed similar secretion distribution, and a clear indication of basolateral targeting of CS-serglycin-His-flag. The experiment was repeated once for clone 4-2, which gave a secretion pattern nearly identical to the results presented in tables 4-9 and 4-10. Basolateral targeting of CS serglycin-His-flag was also consistently observed, during the optimisation of the Hi-Trap chelating chromatography system.

Thus, basolateral secretion of the CS-GAG modified serglycin-His-flag is a reproducible phenomenon.

A non-sorted secretion of macromolecules by the MDCK II cells, results in approximately 1/3 apical and 2/3 basolateral transport, depending to some extent on the growth conditions, as observed for growth hormone (GH) (Scheiffele et al., 1995). Thus, CS-serglycin-His-flag seems to be unsorted, while the HS-serglycin-His-flag may have a regulated sorting to the basolateral compartment. An enrichment of HS proteoglycans to the basolateral compartment was also expected. The distribution of endogenous secreted

35

(S)macromolecules from MDCK II cells is

25 percent (apical) and 75 percent (basolateral) (Svennevig et al., 1995). This distribution was also found for the moc-transfected clone 7-4 (figure 4-12.A). Approximately 70 % of the

35

(S)macromolecules found apically and basolaterally represents CS and HS proteoglycans

(apical: 30 % CS and 70 % HS, basolateral: 100 % HS) (Vuong, 1998). With the values above, it can thus be estimated that approximately 1/5 of the secreted HS is located apically, and that 4/5 is located basolaterally (Table 5-1). The transport of HS modified serglycin-His-flag is thus similar to the secretion of wild type HS proteoglycans in the MDCK II cell line. This indicates that the HS modified serglycin-His-flag follows the normal secretion pattern for HS proteoglycans synthesised by the MDCK II cell line.

GAG type

Compartment

Chondroitin sulphate Heparan sulphate

Apical Basolateral Apical Basolateral

Wild type secreted GAG *

Secreted serglycin-His-flag

1

~1/3

0

~2/3

~1/5

~1/6

~4/5

~5/6

Table 5-1: The distribution of endogenous and serglycin-His-flag proteoglycans, modified with

CS- and HS-GAG, secreted by the MDCK II cell line.

(*) Values for wild type proteoglycans are determined by the results in (Svennevig et al., 1995) and

(Vuong, 1998).

143

5 Discussion

It is still not known why the CS modified serglycin-His-flag is non-sorted in MDCK II cells.

However, very little is known about the secreted CS proteoglycan(s) in MDCK II cells, apart from that their molecular weight are above 250 kDa. Thus, structural differences between the wild type CSPG and the serglycin-His-flag can be responsible for the non-sorted secretion of the serglycin-His-flag. The most plausible explanation, is that the tight assembly of CS-GAG chains on the serglycin-His-flag core protein in some way either disturbs packaging, or is not recognised by the apical packaging machinery. For instance, apically enriched CS-GAG chains secreted by xyloside-treated MDCK II cells (Kolset et al., 1999), are much smaller (~14 kDa) than the nonsorted CS-serglycin-His-flag (above 200 kDa), and thus probably also easier to package into vesicles. Alternatively, the chains can be unavailable for binding to a sorting factor that functions to guide CS-GAG and CSPG to the apical surface. This factor may contain a GalNAc binding domain, as the "raft"-associated lectin VIP-36 (Fiedler and Simons, 1996). It is unlikely that VIP-

36 itself is involved in the apical targeting of CS, as explained previously.

Disturbance of the apical secretion signal is not unlikely, because it is likely to be a weak signal.

It is shown that the amyloid precursor-like protein-2 (APLP2) is transported basolaterally in transfected MDCK II cells, whether the core protein contains a CS-GAG chain or not (Lo et al.,

1995). This basolateral targeting was caused by a cytoplasmic sorting signal in the core protein, which demonstrates that the possible apical signal in the CS-GAG chain must have been entirely restrained. The observation that basolateral sorting signals are stronger than the apical, is general.

In three different membrane proteins, with basolateral sorting signals, also containing N-linked glycans, the presence or absence of the N-linked sugars did not affect the basolateral delivery.

However, in the presence of N-linked glycans and an absence of basolateral signals, all proteins were targeted apically (Gut et al., 1998).

Another factor, which can explain the non-sorted secretion of the CS-serglycin-His-flag, is the possibility of structural differences between the GAG chains attached to the wild type CS proteoglycans and the GAG chains attached to the CS-serglycin-His-flag core protein. The chondroitinase ABC lyase enzyme used in the analysis, degrades all modified forms of CS-GAGs

(both CS and DS). Thus, a specific modification, as for instance, a distinct sulphation pattern or a

GlcA to IdoA epimerisation (DS) can be required for binding to the apical transport-factor(s). To exclude this possibility, the exact glycosaminoglycan composition in both the wild type CSPG(s) and the serglycin-His-flag must be characterised.

The wild type CS proteoglycan could also be dependent on other signals for apical targeting. For instance, the core protein may contain N-linked and O-linked glycans, which are thought to be involved in apical targeting (Scheiffele et al., 1995; Zimmer et al., 1997). In the lack of other basolateral signals, the proteoglycan is thus likely to be apically secreted. Svennevig et al. (1995) detected the chondroitin sulphate versican in MDCK II cells. This proteoglycan contains many potential N-linked sites, and thus versican is likely to contain N-linked sugars in MDCK II cells.

If this proteoglycan is the major CS-proteoglycan secreted by the MDCK II cells, it could be possible that N-glycosylation contributes to the apical targeting, and that the CS-GAG chain may not contain an apical targeting signal at all. The N-glycans can, however, not be the only signal that sorts chondroitin sulphate chains to the apical medium in MDCK II cells, since inhibition of

N-glycans synthesis with tunicamycin, had no effect on the apical secretion of CSPG (Prydz, personal publication). This, together with the result that CS-GAGs assembled onto xylosides are apically secreted in MDCK II cells, the chondroitin sulphate chain is still hypothesised to be an apical sorting signal in MDCK II cells. It should, however, be interesting to know why endogenous proteoglycans and CS-GAGs are apically secreted, while CS serglycin is not.

144

5 Discussion

Besides the predominant secretion of endogenous CSPG to the apical medium, it is also interesting that the MDCK II cells transport HSPG mainly to the basolateral medium. The HS-

GAG chain has been assigned a possible role in basolateral transport by other experiments. It is shown that HS-GAG chain attachment of the glypican core protein, antagonised the apical delivery by the N-linked glycans

2

(Mertens et al., 1996). Whether this effect is caused by a dominant basolateral signal in the HS-GAG chain, or by an interruption of incorporation into apical transport vesicles, is still unknown.

If the serglycin-His-flag is synthesised as a hybrid, the presence of HS-GAGs on the core protein is thus likely to interfere with the apical targeting signal in the CS-GAG. In that case, introduction of HS-GAG on the core protein can redirect the molecule from apical to basolateral destination. It is therefore important to further investigate if the CS-GAG attachment on the serglycin-His-flag secreted to the basolateral compartment is found on a hybrid proteoglycan, since a hybrid serglycin-His-flag could explain why CS is found attached to the serglycin-Hisflag secreted to the basolateral compartment. However, due to the lack of indications of a hybrid serglycin-His-flag, this is probably not the explanation for the randomised secretion of the CSserglycin-His-flag.

5.2.5 Stimulation of MDCK II cells with TGF-

Transforming growth factor-

(TGF-

) is a potent growth factor, which is shown to increase the level of CS-GAG on syndecan in mouse mammary epithelial cells (Rapraeger, 1989). High glucose induced TGF-

1 production, is shown to cause a dose-dependent increase in the production of matrix proteins in porcine glomerular mesangial cells (Kolm-Litty et al., 1998). In

MDCK II cells, TGF-

seems to inhibit the formation of tubular structures, and to inhibit epithelial cell division (Santos and Nigam, 1993), thereby suggesting that TGF-

modulates tubulogenesis, as well as branching, in the kidney. The growth factor, TGF-

, may therefore play an important role in the formation of the nephrones and the basal lamina in the kidney.

To try to find out more about the GAG attachment on the serglycin-His-flag core protein, clones

1-7 and 4-2 were stimulated with TGF-

during labelling with 35 (S)sulphate. Stimulation with

TGF-

increased the secretion of

35

(S)proteoglycans to the apical medium, whereas the secretion to the basolateral medium was unchanged (figure 4-27.A). Glycosaminoglycan determination of the GAG chains attached to the serglycin-His-flag secreted by clone 1-7, showed that the amount of chondroitin sulphate type chain was increased from ~40 percent (for non-stimulated cells) to

~55 percent (for stimulated cells) (table 4-12). Moreover, after stimulation with TGF-

, the distribution of the HS-GAG modified serglycin-His-flag was altered from 1/6 apical and 5/6 basolateral to 1/3 apical and 2/3 basolateral.

The experiment was not repeated, so the observed alternation of the glycosaminoglycan attachment by TGF-

stimulation is not validated. However, the experiment indicates that TGF-

 is able to alter the proteoglycan synthesis in MDCK II cells, by decreasing the HS-GAG attachment and increasing the CS-GAG attachment on the non-endogenous core protein.

Increased GAG chain length on the serglycin-His-flag was not observed, as was observed for the

CS- and HS-GAG chains attached to syndecan produced by mouse mammary epithelial cells,

2 In the summary, the authors conclude that: "…. glycanation of the core protein antagonizes the activity of the apical sorting signal conveyed by the GPI anchor of this proteoglycan." However, since the studied proteoglycan, glypican, also contains N-glycans, it is more likely that N-glycan chains mediate the apical signal in this proteoglycan.

145

5 Discussion after stimulation with TGF-

(Rapraeger, 1989). The alteration of the GAG attachment to serglycin-His-flag, is thus probably caused either by a decreased efficiency of the HS synthesis machinery in the Golgi apparatus, or alternatively, by an increased HS attachment on endogenous core proteins. The latter may be the case, if the attachment of CS-GAG chains on the serglycin-

His-flag core protein is due to a nearly saturated HS synthesis machinery in MDCK II cells, as described previously. With increased endogenous core protein concentration in the Golgi apparatus, even more serglycin-His-flag will be able to escape from the HS-attachment site, and reach the site for CS synthesis.

The glycosaminoglycan chains attached to endogenous proteoglycans were not analysed, which is necessary to understand more about the effect of TGF-

stimulation. However, the increased apical delivery of

35

(S)proteoglycans, must be due to an altered delivery of endogenous proteoglycans in the MDCK II cell. Even if more serglycin-His-flag is secreted to the apical medium after stimulation with TGF-

, the total secretion of serglycin-His-flag apical and basolateral is about the same (~42 percent) in both non-stimulated and in TGF-

stimulated cells

(compare tables 1-10 and 1-12). Thus, the increased apical delivery is caused by increased secretion of both serglycin-His-flag and endogenously synthesised proteoglycans.

Since the HS-serglycin-His-flag was more randomly secreted after stimulation with TGF-

, the increased apical delivery may be an effect of decreased basolateral targeting of also endogenous

HS-proteoglycans. This should indeed be investigated further, since a decreased basolateral delivery of HS-proteoglycans, will probably reduce the contents of HS-proteoglycans in the basal lamina, which may disturb the formation and the maintenance of the basal lamina. The diabetic nephropathy observed in some diabetic patients, may thus be mediated by high blood glucose levels, which stimulates TGF-

production (Kolm-Litty et al., 1998), which by an unknown mechanism, stimulates the epithelial cells in the nephrones to reduce the secretion of HSproteoglycan to the basal lamina. Glycosaminoglycan chain attachment to proteoglycans secreted by MDCK II cells, cultured in low and high glucose medium, is now under investigation.

146

Download