Immunolocalization of hevin in normal mouse cerebellum. IHC analysis of methyl Carnoy’s-fixed, paraffin-embedded sections of mouse brain with MAb 12-155 (red). Two ϫ 200 magnification fields ( A,B ) and two ϫ 400 magnification fields ( C,D ) are shown. Reactivity was visualized after incubation with Cy3-conju- gated anti-rat IgG secondary antibody. The slides were counterstained with Hoechst 22258 (blue). MAb TUJ1, specific for neuronal ␤ -actin, was used to identify neurons and was visualized after incubation with FITC- conjugated anti-mouse IgG (green). Arrows in C indicate reactivity with Bergmann glia and/or Purkinje cells in the molecular layer. pl, Purkinje layer; gl, granular layer; ml, molecular layer; w, white matter. 

Context in source publication

Context 1

... heterogeneity in posttranslational modification of the full-length protein. Recoveries of recombinant hevin are shown in Table 1. Based on scanning densitometry of SDS-polyacrylamide gels, which tends to overestimate the con- tribution of impurities, preparations of recombinant hevin were ف 80% pure after elution from an isoelectric focusing column. Greater yields, however, with less purity, were obtained from size-exclusion chromatography compared with isoelectric focusing. Figures 3 and 4 confirm that recombinant hevin (rhevin), expressed and purified as described above, is biologi- cally active. Human hevin has been described as antiadhesive for cultured human umbilical vein endothelial cells (HUVECs) (Girard and Springer 1996). We have confirmed these data and have extended the findings to define an anti-adhesive activity of mouse recombinant hevin on BAECs in vitro. As shown in Figures 3A and 3B, purified hevin, added at 10 ␮ g/ml to cells in suspension, inhibited the spreading of BAECs on tissue culture plastic (i.e., was anti-adhesive) in the presence of 0% FBS. This effect was apparent up to 6 hr after plating, after which hevin-treated BAECs completed adhesion and resembled non-treated or BSA-treated controls. Quantification of these data as Rounding Indices is shown in Figure 4. These results were also obtained in the presence of 1% FBS (not shown). We also measured the anti-adhesive activity of recombinant hevin presented as a substrate to trypsin- released BAECs. As shown in Figure 5, BAECs in the absence of FBS adhered to substrates coated with fibronectin (Figures 5E and 5H), but were rounded and clumped on hevin-coated wells (Figures 5F and 5I). This effect was maintained over ف 12 hr, after which most of the cells began to spread. Intermediate levels of adhesion that were time-dependent were seen with substrates coated with 10% FBS, and rounding similar to that shown in Figures 5C, 5F, 5I, and 5L was observed with substrates coated with recombinant SPARC (not shown). Cell viability was maintained under all conditions. Partially purified hevin was used to immunize rats to produce MAbs specific for mouse hevin. This protocol resulted in over 200 hybridoma clones that bound strongly to hevin by ELISA (data not shown). These reagents were acutely needed because there were no anti-hevin antibodies that reacted reliably with mouse tissues or with paraformaldehyde-fixed tissue, that blocked any activity of native hevin, or that could immunoprecipitate native antigen. For selection of hybridomas, tissue culture supernatants from the clones were screened for reactivity with recombinant hevin in a variety of assays. A partial list of the most ex- tensively characterized rat anti-hevin hybridomas is shown in Table 2. The majority of MAbs are IgG 2a , and each clone reacts with mouse hevin by indirect ELISA. In contrast, many MAbs failed to react with hevin in solution, as determined by a capture ELISA (Table 2). Several of the MAbs described in Table 2 were used to identify hevin in mouse brain lysates (Figure 6A, Lane 1). In each lysate, a band of M r 105,000–110,000 was the predominant immunoreactive species detected by three different MAbs. Minor bands most likely re- flect differences in posttranslational modification of the hevin chain (Hambrock et al. 2003) because tissue lysates contain secreted as well as cellular (endoplas- mic reticulum and Golgi-associated) hevin, with dif- fering degrees of glycosylation. Immunoprecipitation of hevin from extracts of mouse brain by MAb 12-18 also showed a protein of M r 105,000–110,000 by SDS-PAGE (Table 2; Figure 6B, Lane 2). In no cases was immunoreactivity obtained with proteins in lysates from hevin-null brains (Figure 6A, Lanes 2). The disparity in M r of hevin observed in Figures 1 and 2 vs Figure 6 is likely due to the differences in polyacrylamide gel composition (10% in Figures 1 and 2 vs 4–12% gradient in Figure 6), as well as in buffer conditions (Tris-HC1 vs Bis-Tris). We found that MAb 12-51 was particularly useful for IHC localization of hevin in frozen or paraffin-embedded tissue because it also crossreacts with human hevin. Figure 7 shows the reactivity of 12-51 with human skin and mouse kidney. Strong reactivity was apparent in the dermal layers of the skin, as well as in muscle and the epithelia of glands (Figures 7A and 7B). In the brain, 12-51 highlighted specific cells in the parenchyma of the cerebellum and also vessels in other regions (not shown). Arterioles, venules, and capillaries in the dermis were also reactive with MAb 12-51 (Figure 7B). In paraffin-embedded mouse kidney, hevin was localized by MAb 12-155 to certain glomeruli and tubule epithelial cells (Figure 7C). These data are consistent with previous results in which hevin mRNA was localized to vessels, epidermis, and selected populations of renal tubules by ISH (Soderling et al. 1997). Staining for hevin in mouse cerebellum with MAb 12-155 revealed patterns as shown in Figure 8. Hevin, shown in red, localized conspicuously to the Purkinje cell layer (Figures 8A and 8B) and to the Bergmann glia and Purkinje cells throughout the molecular layer (Figures 8C and 8D). Neurons, shown in green, were localized with MAb TUJ1, which is specific for neuronal class III ␤ -tubulin; nuclei, shown in blue, were stained with Hoechst 22258. These results extend ear- lier reports of hevin mRNA expression in the Purkinje cell layer (McKinnon and Margolskee 1996; Sullivan and Sage 2004). We also evaluated the production of hevin in a limited panel of cancerous tissues (Figure 9). Of the tumor tissues that we screened (human colorectal, lung, breast, pancreatic, and endometrial adenocarcinoma), reactivity was apparent only in endometrial carcinoma associated with the luminal duct epithelial layer (Figure 9G). Although there was no reactivity in two samples of human pancreatic adenocarcinoma, we found tumor as well as stromal reactivity with MAbs 12-51 and 12-155 in human pancreatic tumor xenografts grown in immunocompromised mice (not shown). Xe- nografts of human non-small-cell lung carcinoma from immunocompromised mice are shown in Figures 9A–9F), after incubation with several additional anti- hevin MAbs (identified on each panel) that have not been listed in Table 2. Each of these staining patterns reflects a stromal/vascular component that was identified by the respective anti-hevin MAb. These data pro- vide evidence for hevin as a stroma-associated protein of neoplastic tissue, an observation consistent with findings by other investigators from serial analysis of gene expression (SAGE) analysis of invasive pancreatic carcinomas (Ryu et al. 2001; Iacobuzio-Donahue et al. 2003). Matricellular proteins regulate the interaction of cells with the ECM, a critical function in the maintenance of tissue homeostasis. Hevin, originally described in rodents as SC-1, is a matricellular protein belonging to the SPARC family. In the present study we have produced recombinant mouse hevin and have purified it to near homogeneity, developed MAbs specific for hevin, and identified a function of hevin in vitro. The MAbs described herein will be useful for elucidating the functions of hevin in vivo. For example, hevin has recently been shown to be identical to a neuronal guidance protein, the antigen for the MAb RAGS1 (Gongidi et al. 2004). RAGS1 (Radial Glial Stop Sig- nal molecule 1) recognizes an antigen on the surface of radial glial cells that is distributed where neurons ter- minate their migration. Gongidi et al. (2004) found that the MAbs RAGS1 and 12-155 react on immuno- blots with the same protein. Moreover, inhibition of hevin with RAGS1 allowed neurons to migrate past the normal termination point in the cerebral cortex, thereby demonstrating that hevin is critical for appropriate neuron migration. Among the many cell lines that have been tested for the production of hevin, the few that produce hevin at detectable levels by immunoblotting are tumor lines. Lysates of hevin-producing tumor cells yield a primary immunoreactive band of M r 130,000 on SDS- PAGE using 10% polyacrylamide gels (not shown). We found hevin of similar M r in lysates of human and mouse brain tissue, or of M r 105,000 on 4–12% Bis- Tris gradient gels. The major immunoreactive band of recombinant mouse hevin migrates slightly faster than the M r 130,000 band characteristic of cell and tissue lysates, a possible consequence of the inability of insect cells to produce N -glycoproteins containing a sialic acid cap, a modification of many complex carbohydrate chains of secreted proteins synthesized by mammalian cells. Note that the apparent M r of recombinant hevin is 105,000 when estimated by separation on a NuPAGE 4–12% Bis-Tris gel with NuPAGE MOPS SDS running buffer (Invitrogen) (Figure 6). Estimates of the M r of human hevin in the liter- ature vary from 95,000 (Hambrock et al. 2003) to 150,000 (Bendik et al. 1998) by SDS-PAGE. Therefore, we consider acceptable the apparent size varia- tion that we have observed in recombinant murine hevin and explain it as a possible consequence of the gel resolution system and/or interaction of the acidic N-terminal domain with buffers of varying pH and/or ionic strength. The de-adhesive effects of hevin that we report are consistent with a prior study. Girard and Springer (1996) found that hevin (in contrast to fibronectin, thrombospondin 1, and tenascin C) was non-permis- sive for attachment and spreading by HUVECs and BAECs. Furthermore, it diminished adhesion to fibronectin, an effect correlated with reduced formation of focal adhesions. Initially, it was suggested that hevin supported lymphocyte extravasion through high endothelial venules via the diminishment of cell–cell and/or cell–ECM contacts in a dynamic, chemokine- responsive manner (Girard and Springer 1995,1996). We speculate that hevin might play a comparable role in the desmoplastic endothelium associated with tumor ...