-MHC-VEGF-C-FL-MHC-VEGF-C-FL55 potential founders were obtained for the -MHC-VEGF-C-FL
construct. Both PCR and Southerns were repeatedly negative. A probable explanation
would be that VEGF-C overexpression in the heart is lethal in utero. It was
considered to determine time point and cause of the potential developmental
arrest by examining the embryos, but the project was not pursued due to capacity
reasons.
-MHC-VEGF-B16742 potential founders were obtained for the -MHC-VEGF-B167
construct, out of which 6 were positive. Four of the 6 founders were positive
in both Southern blotting and PCR, one founder was positive only in PCR and
another only in Southern blotting analysis. It became clear that the PCR-positive,
but Southern-negative founder had undergone a rearrangement in the transgene,
which resulted in the loss of the -MHC
promoter for which the Southern probe was specific. A possible reason, why one
other potential founder was Southern positive, but PCR negative, could have
been a minor deletion resulting in the loss of one primer annealing site. However,
this was not confirmed (a pedigree of the mice is included in the appendix).
All Southern-positive founders were mated and the pups were screened as described for the transgene. One mouse did not transmit the transgene, whereas the others transmitted to 37.5 - 53% of the pups. The copy number was not exactly determined, but ranged between 2 and 25 copies/diploid genome, as judged by the Southerns.
Positive pups from all founders and their non-transgenic littermates were sacrificed for histological and in situ hybridization analysis. Especially the heart was subject of examination, where the transgene was expected to be expressed. Surprisingly, no transgene expression could be detected by in situ hybridization. In agreement with this, the histological findings are normal, but analysis is still going on. In this respect it is noteworthy, that much more founders were positive in nested PCR, but semiquantitative analysis revealed apparent copy numbers significantly lower than 0.1/diploid genome. These mice might have been either mosaics or false positives.
Of 32 potential founders one contained an high copy number (35-50/diploid genome) of the transgene, two a copy number between 5-10/diploid genome, and one a copy number near to the sensitivity border of the dot blot assay. The mouse with high copy number was reduced in size, had faint cranial malformations and seemed to be blind on the right eye. It was killed at four weeks of age to avoid natural death and loss of material. Its pancreas was histologically analyzed by both hematoxylin-eosin staining and immunostaining for insulin, glucagon, pancreatic polypeptide and somatostatin. The islets appeared to be normal. Whether the transgene was expressed at all, remains to be determined. Instead, the observed phenotype might result from the disruption of an unrelated gene by the insertion of the transgene. If the transgene is transmitted, the F2 generation of the other two founders (both females, which appear completely normal) will be subject of more detailed analysis.
Two positive mice were identified out of 46 potential founders, both with approximate copy numbers between 5-10/diploid genome. Similarly to the RIP-VEGF-B167 transgenic mice, detailed analysis is going on.
50 potential founders were obtained, one of which died shortly after birth and was not analyzed. 5 of them had unmistakably more than 1 copy of the transgene in their genomes, although signals approximately equivalent to the 0.1 copy number control were obtained by PCR for 7 other potential founders, probably mosaics. For the five undoubted founders the copy numbers were determined as approximately 2-5 (in two cases), 10-20 and 30-50 (in two cases). All of them were mated, but three of them failed repeatedly to transmit the transgene (altogether 17, 16 and 6 pups were born, respectively). The founder which had 2-5 copies in its genome transmitted to 5 out of 13 pups (38%). The other founder which transmitted (to 3 out of 21 pups, 14%) had 30-50 copies in its genome.
Positive pups and negative littermates were sacrificed for analysis. Preliminary
data shows a distinct phenotype in both mice, which is difficult to interpret
and which requires further investigation. The most striking in a first comparison
to the negative control was the reduced number of hair follicles in the dermis,
which was itself thickened due to an increase in extracellular matrix forming
collagen. The reduction of hair follicles was frequently reported from mice
over-expressing cytokines under the keratin 14 promoter, e.g. TGF-,
PTHrP and KGF (Guo et al., 1993; Vassar and Fuchs, 1991; Wysolmerski et al.,
1994) and might be therefore unspecific, perhaps due to cytotoxic effects. However,
in contrast to VEGF-B167 over-expressing mice, in these mice no increased
collagenization was observed. The possibility was considered, whether VEGF-B
would only be active in the presence of both isoforms, probably as heterodimer,
since recent unpublished data indicates a constant 1:1 ratio of these isoforms.
Consequently, we decided to direct the expression of a VEGF-B186 cDNA
under the same keratin 14 promoter. Since the genomic structure was meanwhile
sufficiently explored, a genomic fragment capable of generating the two different
mRNAs was considered as well.
 |
 |
| A | B |
Out of 27 founder mice analyzed at 3 weeks of age, 3 were positive, having approximately 40-50, 20 and 4-6 copies of the transgene in their genome.
The mouse with the high copy number transgene was small and developed slower than its littermates. Further examination showed a swollen, red snout, poor fur and abnormally long incisors. It had difficulties in breast feeding and breathing. Although nourished with special liquid diet, it suffered from oedema of the upper respiratory and digestive tracts and died eight weeks after birth. It was immediately processed for histology, immunohistochemistry and in situ hybridization.
The other two positive mice, both males, exhibited a similar, although milder phenotype and in spite of initial difficulties they were mated successfully. One transmitted the gene to 6 out of 11 pups (55%) before it died at the age of 4 months; the other mouse transmitted the gene to 1 out of 17 pups (6%), although two mice were born dead and not included in the analysis. A pedigree of the mice is given in the appendix.
Histological examination showed that, in comparison with the skin of littermates, the dermis of K14-VEGF-C-FL transgenic mice was atrophic and the connective tissue of the skin was replaced by large lacunae devoid of red cells, but lined with an attenuated endothelial cell layer. At the walls of the distended vessel-like structures remnants of valves were identified, a distinguishing feature between initial lymphatics and blood vessels in the upper dermis (Ryan et al., 1986). The vessel-like structures resembled those seen in human lymphangiomas and were most remarkable in the snout and dorsal region.
In situ hybridizations of skin samples show that the major VEGF-C receptor FLT4 is highly upregulated in these transgenic mice, and that this overexpression is confined to the endothelial lining of the vessel-like structures. Notably, FLT4 is the most promising nominee as a marker for lymphatic vascular endothelial cells in adults (Kaipainen et al., 1995). It is interesting that the effect of VEGF-C was completely confined to the skin. Probably VEGF-C sticks not only in vitro, but also in vivo to various surfaces. This quality might be mediated through its silk homology domain and is probably significant in the spatial limitation of VEGF-C activity.
Thus it appears, that VEGF-C induces selective angiogenesis of the lymphatic vessels in vivo. Few entirely specific lymphatic endothelial cell markers have been established to date. Antibodies against collagen type XVIII gave weak or no staining of the vessels, while the basement membrane of blood vessels was stained prominently. The expression of desmoplakin I and II and the less developed basement membranes compared to blood capillaries strongly suggest that the abnormal vessels were derived from lymphatic vessels (Casley-Smith, 1980; Schmelz et al., 1994). Platelet-endothelial cell adhesion molecule (PECAM-1) was used as a general marker for vascular endothelium (Dejana et al., 1995).
The lymphatic structures were mainly increased in size, but lymphatic sprouting was absent or subordinate. According to preliminary data (measuring cell proliferation of in vitro explants by BrdU incorporation), the observed lymphatic vascular endothelium is proliferating and the increased size of the vessels might be therefore due to an absent sprouting signal. An alternatively explanation is that the oversized capillary network in the skin cannot be drained efficiently by the larger lymphatic vessels, which are normally sized. In this case the enlarged vessel lumen would be a secondary effect due to congestion of lymphatic fluid. Accordingly, the reduced number of skin adnexal organs and hair follicles might be merely due to a physical displacement of the dermal connective tissue. The first explanation would suggest that at least two distinct signals are required for the establishment of a lymphatic network: one signal that would stimulate migration and sprouting and another, distinct signal, that would stimulate the proliferation of the involved endothelium.
As a conclusion, VEGF-C delivers a growth signal rather to lymphatic endothelial cells than to non-lymphatic endothelial cells, thus being distinct from VEGF (Wilting et al., 1996). The VEGF subfamily of growth factors is still expanding, and the present work tried to address the question, why different VEGFs have emerged during evolution and to what degree their functions are still redundant.
Figure 12. K14-VEGF-C-FL transgenic mouse with littermate |
 |
 |
 |
| A | B |
 |
 |
| C | D |
 |
 |
| E | F |
 |
 |
| G | H |
 |
 |
| A | B |
 |
 |
| C | D |
 |
 |
| E | F |
 |
 |
| A | B |
 |
 |
| C | D |
 |
 |
| E | F |
NC-VEGF-C, K14-VEGF-B186No data is available yet.
Although the full length cDNA led to the expression of biologically active VEGF-C, total amount, processing and secretion seemed not to be optimal. Different constructs were generated to meet the different purposes (see Tables 3 and 4). Firstly, the melittin signal peptide was used instead of the endogenous one to enhance secretion. Secondly, the protein was trimmed at both N- and C-termini to circumvent the proteolytical processing, which was inefficiently done in insect cells, but which is inevitable for biological activity. Unlike most other members of the VEGF family VEGF-C shows no affinity towards heparin, on which the purification procedures could have been based similar to VEGF (Ferrara et al., 1991). Therefore a histidine tag was incorporated for affinity purification. Since the protein was secreted into serum-free media a one-step affinity purification was enough to obtain VEGF-C clean enough for most applications. melSP-NC-C1HIS-VEGF-C mostly seems to be secreted as a non-covalently linked dimer or monomer, although some of it migrates even in reducing SDS-PAGE as a dimer.
Interestingly, from all four tested locations, the N-terminal histidine tag seemed to interfere with receptor activation. The C-terminal histidine-tag was designed to purify the C-terminal proteolytical cleavage product and the uncleaved propeptide. Both histidine tags that were placed behind the VEGF homology domain functioned in the affinity purification using Ni2+-NTA Superflow resin (Qiagen), although the more N-terminally located histidine-tag (C1HIS) seemed to be slightly superior.
The short splice variant of VEGF-C was produced much less efficiently than all other forms. In the used concentrations it seems neither to stimulate FLT4 nor KDR phosphorylation. Inhibition experiments were also negative; the short splice variant seems not to interfere with receptor activation by biologically active VEGF-C.
The expression of VEGF-B in insect cells appeared to be intrinsically difficult. Differently from mammalian cells, VEGF-B167 is seemingly not secreted from Sf9 and HF cells. The efficiency of secretion could not be altered by exchanging the endogenous signal peptide against the honey bee melittin signal peptide. Alternatively, VEGF-B167 might be very strongly bound to the cell surface unable to be released by heparin concentrations < 200 
g/ml.
VEGF-B186 is secreted, but N-terminally histidine-tagged VEGF-B186 tends to aggregate (personal communication by Birgitta Olofsson). The C-terminally histidine-tagged VEGF-B186 was secreted into the medium, probably because the histidine-tag compensated for the otherwise hydrophobic C-terminus. It showed on SDS-PAGE the same size and dimerization as the mammalian protein. Apparently the O-glycosylation is performed by insect cells. Sf9 cells infected with the same virus produced a protein with a much lower molecular weight. This protein cannot be affinity-purified, probably due to proteolytical cleavage which removes the histidine-tag.
Opposite to VEGF-C, the VEGF-B receptor(s) is/are unknown and hence no efficient way exists to confirm the biological activity and specificity of modified versions of VEGF-B.