Intercellular transfer of β-glucuronidase in chimeric mice

A protein in a eukaryotic cell is usually thought to be the product of that cell’s synthetic machinery. However, several lines of evidence suggest that, occasionally, a cell may acquire a functional protein from extracellular sources. Binding of a polypeptide hormone to its cell surface receptor can be followed by internalization of the hormone (e.g. Carpenter & Cohen, 1977; Maxfield et al. 1978). There is good evidence that nerve growth factor is taken up by the nerve terminals of responsive adrenergic and sensory neurons and that the internalized protein is capable of eliciting a biochemical (Paravincini, Stoeckel & Thoenen, 1975) and morphological (Hendry, 1977) response in that neuron. Kreutzberg & Toth (1974) have shown that acetylcholinesterase, secreted from neural dendrites is taken up by nearby capillary endothelial cells where it retains activity. Lasek, Gainer & Barker (1977) gave evidence that an entire spectrum of peptides synthesized in glial cells, are subsequently found in the squid giant axon. Cultured fibroblasts from patients suffering from a variety of enzyme deficiencies show correction of the metabolic defect if the missing enzyme is added exogenously (see Dorfman & Matalon, 1976; Neufeld, Lim & Shapiro, 1975; Sly, 1979, for review). There is also evidence to suggest that uptake of extracellular lysosomal enzymes occurs in vivo as part of the normal physiology of mammalian cells. The liver and other organs are known to accumulate exogenous β-glucuronidase (Thorpe, Fiddler & Desnick, 1974) and specific recognition sites for several lysosomal enzymes have been identified on liver cells (Stahl et al. 1976). Hickman & Neufeld (1972) have hypothesized that enzymes are packaged into lysosomes by a secretion-reuptake mechanism. The purpose of this report is to examine whether transfer of endogenous β-glucuronidase occurs among the cells of a single organism.

Mouse chimeras made by aggregating the embryos from strains which differ in their levels of β-glucuronidase activity offer a means of exploring intercellular enzyme transfer in vivo. The structural gene for β-glucuronidase is denoted Gus. Of the known alleles at this locus, the most common is denoted Gus^b and is found in C57BL/6J mice as well as in most other inbred strains. A second allele, Gus^b, is an electrophoretic variant described by Lalley & Shows (1974). A third allele, denoted Gus^a, is found in decreased amounts, relative to Gus^b enzyme, in all tissues tested (Morrow, Greenspan & Carrell, 1949; Paigen, Swank, Tomino & Ganschow, 1975). In addition, the Gus^h enzyme is more heat sensitive than the Gus^b enzyme (Paigen, 1961; Herrup & Mullen, 1977).

Glucuronidase mosaicism in chimeras was first demonstrated histochemically by Condamine, Custer & Mintz (1971) in liver tissue. More recently, Feder (1976) reported increased staining in the Gus^h cells of Gus^h/Gus^h↔ Gus^b/Gus^b chimeras compared to cells of Gus^h/Gus^h homozygous animals and suggested that this occurred by in vivo intercellular transfer of enzyme. Enzyme activation or induction of enzyme synthesis, however, would result in the same observations. In the present work we describe a method of measuring rates of heat inactivation on sections of tissue fixed for β-glucuronidase histochemistry. The evidence suggests that the light-staining Gus^h/Gus^h cells of the chimera contain heat-stable Gus^b Mike enzyme in their cytoplasm. These results show induction or activation is not an explanation for the increased staining and imply that intercellular transfer of β-glucuronidase molecules is occurring as part of the normal physiology of the mouse.

The histochemical enzyme substrate, naphthol AS-BI-β-D-glucuronide was obtained from Sigma (St Louis, Mo.) as was the chromophore, pararosanaline, and polyvinylpyrrolidone (PVP-10). All other chemicals were of reagent grade.

The mice used to make chimeras in this study were of 2 types. The first was homozygous for the Gus^b allele. These mice were from a highly inbred line of C57BL/6 and C57BL/10 origins. The second was homozygous for the Gus^k allele. These mice were either C₃H/Hej or C₃CBAF₁. The females used as host mothers were B6C₃F₁.

Chimeras were produced by the embryo aggregation technique of Tarkowski (1961) and Mintz (1962, 1965) with the variations described by Mullen & Whitten (1971). Briefly, 8-cell embryos were flushed from their mother’s oviducts on day 2·5 of gestation. The protective zona pellucida was digested away with pronase. At this stage, the embryos were naturally sticky and adhered to each other when placed in contact at 37 °C. In the experiments described here, one embryo of the pair was homozygous agouti (A/A) and Gus^k/Gus^K, the other nonagouti (a/a) and Gus^b/Gus^b. The coat pigmentation differences allow the identification of the successful chimeras. After ensuring the embryos had adhered, they were cultured overnight. Under these conditions, the cells of the 2 embryos mix and develop into a single double-sized blastocyst. The next day, these were transplanted to the uterus of a host female made pseudopregnant by mating with a vasectomized male. The embryos then implanted in the uterus, adjusted in size, and finished development normally.

The histochemical methods employed improved techniques devised by Feder (1976) except that the animals were perfused with cold (2–4 °C) fixative. The animals were anaesthetized with Avertin and perfused through the heart with cold 4 % phosphate-buffered formaldehyde. After extensive rinsing (4–6 days) in 6 % sucrose/10% polyvinylpyrrolidone and incomplete dehydration in 98 % acetone-2 % water (1 day), the livers were embedded in polyester wax (with 2 % water) that had been treated with NH₄HCO₃. All chimeric and control tissues for each experiment were embedded in a single block, allowing identical treatment during heat inactivation and staining. Sections were cut at 7 μm. The histochemical staining followed essentially the procedure worked out by Hayashi, Nakajima & Fishman (1964) except that 5 % gelatin was added to the stain. No counterstain was used.

Photometric measurements were used as indices of enzyme activity. The measurements were of homogeneous fields of cells (e.g. cells of one genotype in the chimeras) that filled the entire field of an oil-immersion objective. A photocell, mounted on the trinocular head of the microscope, measured the amount of light transmitted. The photocell was connected to a galvanometer (Photovolt, Inc., N.Y.) with a 0–100 scale, which corresponded closely to a scale of percent transmission. A constant filament temperature was used so that measurements made on different days could be compared. The response of the system was not perfectly linear with respect to the log of the optical density, so a standard curve was constructed with the use of a graded series of neutral density filters (Kodak, Rochester, N.Y.). All results were expressed as optical density. The system’s ‘dark current’ was determined with the microscope prism directing the light into the normal binocular eyepieces. In selecting fields for measurement in the chimeras, extreme care was taken to find areas which were uniformly Gus^k/Gus^k or Gus^b/Gus^b genotype. An area of the microscope slide a few microns off the tissue section was used as the 100 % transmission blank. Only small adjustments were needed to set this figure and these were performed using the substage diaphragm.

Heat inactivations were performed in 0·1 M citrate pH 5·0 (22 °C). After dewaxing, the slides were presoaked in this buffer for a minimum of 60 min. A reservoir of buffer was heated to 73 °C in a Tecam circulating bath (Techne, Princeton, N.J.). The slides were immersed in the bath with the use of a slide holder such that the entire glass area was covered with buffer and allowed to incubate for the appropriate time. To stop the inactivation, the slides were plunged into an ice-water/slurry after which they were stored in distilled water until ready for staining.

Following inactivation, slides were stained in the same vessel for identical times (usually 6–8 h). Controls were presoaked in buffer but not exposed to heat. To control for tissue background, slides were boiled for 5 min in buffer to destroy all enzyme activity. The optical density of these tissue sections was considered the ‘blank’ value and was subtracted from all points before calculations were performed.

The histochemical staining for β-glucuronidase in liver tissue from various sources is shown in Fig. 1. None of these sections was heat treated. The appearance of the 2 non-chimeric, homozygous controls are shown in Fig. i A, B. The differences between the specific enzyme activity in Gus^b/Gus^b (Fig. 1A) and Gu/Gu (Fig. 1B) animals are clearly reflected in the staining intensities of the 2 tissues. Fig. 1C-E illustrate the staining pattern observed in Gus^b/Gus^b↔ Gu^h/Gus^h chimeric liver. The staining reveals a mosaic of unambiguous light- and dark-staining cells with several significant features. In the context of its immediate neighbours, every cell can be assigned with confidence to either the light-staining or dark-staining group. In agreement with the observations of Condamine et al. (1971) we conclude that this cannot be a random event, but rather that the dark-staining cells are genotypically Gus^b/Gus^b, while the light-staining cells are Gus^h/Gus^h. The boundaries between dark-staining (Gus^b/Gus^b) and light-staining (Gus^h/Gus^h) cells are quite sharp. This is true whether one is observing a low activity cell with predominantly high-activity neighbours (Fig. 1 D) or a high activity cell with predominantly low-activity neighbours (Fig. 1E). Finally, in accord with previous observations (Feder, 1976), we find the Gud^h cells in the chimera to be more darkly stained than their homozygous Gus^h/Gus^h controls. For ease of reference, we have chosen the term Gus^h-Χ cells to describe the light-staining, Gus^h/Gus^h, cells in the chimera and Gus^b-Χ cells to describe the darkstaining, Gus^b/Gus^b cells in the chimera.

The heat denaturations were performed in 0·1 M Na-citrate buffer, pH 5·0 at 73 °C. The appearance of the subsequently stained tissue from the normal Gus^b/Gus^b and GUS^H/GUS^H animals is shown in Fig. 2 A, D, respectively. After 20 min of heat treatment, the staining intensities in Fig. 2E, H are obtained. The Gus^b/Gus^b sections lost noticeable amounts of staining activity but were, nonetheless, distinctly stained.

The identical heat treatment of Gus^h/Gus^h sections, on the other hand, lead to the complete loss of staining activity. This is true even for the relatively dark-staining Kupffer cells.

Fig. 2B, c show untreated Gus^b-Χ and Gus^h-Χ cells, respectively, while Fig. 2F, G show similar fields after 20 min at 73 °C. The Gus^b-Χ cells, like their normal Gus^b/Gus^b counterparts, lose some, but not all of their staining activity. The Gus^hΧ cells, however, do not behave as the normal Gus^h/Gus^h cells. They lose considerable staining activity, but even after 20 min of heat they still show reaction product deposited in their cytoplasm (arrows in Fig. 2G). Thus, although untreated Gus^h/Gus^h Kupffer cells often stain more darkly than Gus^h-Χ hepatocytes, the staining activity of the former is completely inactivated, while the staining activity in the latter is not. These data suggest that heat-stable enzyme activity is located in cells whose nucleus codes for a heat-labile protein.

Photometric measurements, obtained as described in Materials and methods, were undertaken to quantitate the visual impressions described above. The optical density of a section, expressed as the percent optical density of an untreated section and plotted against time at 73 °C is shown in Fig. 3. The values from 10–50 fields were averaged to obtain each point. In both Fig. 3 A and B, •–• are the results obtained with normal (non-chimeric) Gus^b/Gus^b liver tissue in 2 separate experiments; ○–○ are the results with normal Gus^h/Gus^h tissue. Despite fixation and subsequent histological processing the results parallel earlier biochemical work (Herrup & Mullen, 1977) well. Under the conditions described, the rates of inactivation differed by approximately 10-fold. Note that after 20 min, the activity remaining in Gus^h/Gus^h sections was well below 10% of that found in an untreated section, whereas the activity in the Gus^b/Gus^b tissue was still around 70% of its unheated control.

Like their homozygous Gus^b/Gus^b counterparts, the dark-staining Gus^b-Χ cells showed a relatively slow loss of enzyme activity (Fig. 3 A, ▾,♦). The rate of decay was similar to that of the Gus^b/Gus^b control. The light-staining Gus^h,-Χ cells showed more complicated heat denaturation curves as indicated by ▿, ◊ in Fig. 3 A. There was a rapid loss of around 65 % of the activity. This rapid loss was similar to that seen for the cells of the homozygous Gus^h/Gus^h control. The later time points showed that the remaining activity was relatively stable, similar to that of the Gus^b/Gus^b controls and the dark-staining Gus^b-Χ cells. Fig. 3 B shows the results from a second experiment using 2 additional controls and the light-staining cells from a third chimera.

The combined photographic and photometric data suggest that β-glucuronidase is transferred intercellularly as part of an animal’s normal physiology. In agreement with Feder (1976), the photomicrographs in Figs. 1 and 2 show more staining in the Gus^h-Χ cells (the Gus^h/Gus^h cells in the chimera) than in the control, Gus^h/Gus^h cells. The simple comparison of absolute staining intensities, however, cannot eliminate the possibilities of enzyme induction or activation. The concentration of enzyme activators (e.g. Ho & Light, 1973) might be higher in the chimera than in the Gus^h/Gus^h strain since all the cells share the same circulation. In addition the levels of low-molecular-weight inducers such as testosterone (Paigen et al. 1975) might be higher in the chimera than in the low-activity strain.

Activation or induction, however, would mean that all the enzyme in the cells of one genotype should have the same physical properties. The Gus^b and Gus^b enzymes differ not only in activity levels, but also in thermostability. The Gush enzyme is less thermostable than the Gus^h enzyme (Paigen, 1961). Herrup & Mullen (1977) examined the relative rates of heat inactivation of Gus^b and Gus^h enzyme under various ionic conditions. That the extra staining observed in Gus^h-Χ cells results from transfer rather than activation of enzyme or induction of enzyme synthesis is suggested by the comparison of the patterns of heat inactivation of Gus^b/Gus^b, Gus^h/Gus^h, Gus^b-Χ and Gus^h-Χ enzyme activity. The photographs in Fig. 2 form part of this evidence. In all animals, the Kupffer cells of the liver are more darkly stained than the hepatocytes. The staining intensities of normal (non-chimeric) Gus^h/Gus^h Kupffer cells often appear darker than those of the Gus^h-Χ hepatocytes. After 20 min at 73 °C, all of the enzyme activity in the Gus^h/Gus^h tissue has been lost (Fig. 2H), including that in the Kupffer cells. Yet staining activity is still present in the Gus^h-Χ hepatocytes even though they were treated identically.

The photometric data make the same point in a more quantitative fashion. The dark-staining Gus^b-Χ cells have a staining activity which decays at a rate similar to that seen for the activity in normal Gus^b/Gus^b cells. The light-staining Gus^h-Χ cells show a more complicated heat-decay pattern. The initial loss of activity occurs rapidly and is similar to the rate of activity loss of non-chimeric Gus^h/Gus^h cells. Following this the cells lose their remaining activity more slowly at a rate similar to that seen for the activity in Gus^b/Gus^b cells or Gus^b-Χ cells. This suggests the presence of some heat-stable enzyme in their cytoplasm.

The quantitative observations are consistent with the hypothesis that the extra staining in Gus^h-Χ cells compared to Gus^h/Gus^h controls is due mainly to transferred Gus^b protein. The Gus^h-Χ cells in the chimeras we examined showed x-5-fold more staining than their Gus^h/Gus^h counterparts (range 1·44–1·65). If one fits a straight line to the later time points of the Gus^h-Χ cell decay curves, it is roughly parallel to the Gus^b/Gus^b line, and intersects the y-axis at around 50%. Within the limits of this system then, the extra staining in the light-staining Gus^h-Χ cells can be accounted for by the transferred Gus^b enzyme. Whether the reciprocal transfer occurs (Gus^h enzyme to Gus^b-Χ cells) cannot be determined by these methods since if the same absolute amount of staining activity were transferred, its presence would be masked by the higher background.

While technical artifacts might be involved in a study of this sort we feel the possibility is a remote one in this work for several reasons. (1) There are sharp boundaries between ‘patches’ of cells of different genotype. (2) There are no gradients of staining intensity from the edges toward the centre of any one patch. These findings rule out diffusion of enzyme or reaction product during fixation and processing. (3) The section thickness (7 μm) relative to the average cell thickness (ca. 20 μm) makes overlapping of Gus^b-Χ and Gus^h-Χcells unlikely in general, and (4) in most instances a single low patch could be followed through many serial sections. Hence the cells in these fields had no contiguous high cells which might cause sectioning artifacts.

The possibility that some sort of enzyme stabilizer is being transferred rather than the enzyme itself also seems remote. The Gus locus is known to be the structural locus for β-glucuronidase (see Introduction) and thus the reduced stability of the Gus^h enzyme is believed to be due to its altered structure. Since the properties of the Gus^h enzyme are the same regardless of the strain of mice carrying the gene, including a congenie C57BL/6J strain, it is unlikely that there is any ‘stabilizer’ of genetic origin or, if there is, it must be closely linked to the Gus locus. Furthermore, it there were a diffusible stabilizer, one would expect that enzyme from Gus^b/Gus^h hybrids would be all heat stable, which it is not.

A final question which arises is whether it is the enzyme protein itself or, rather, messenger RNA which is transferred. Mintz (1962, 1965) has examined enzymes whose isomers vary in the 2 embryos of the chimera. She observed no hybrid enzyme bands in chimeric tissue and concluded that cells do not fuse during development (striated muscle being the sole exception). The lack of hybrid band also suggests that molecules such as messenger RNA are not transferred intercellularly and hence the transfer we obseive is probably the protein itself.

This phenomenon is also not unique to liver. In several areas of the central nervous system using an identical protocol, we have observed heat-stable enzyme activity in Gus^h-Χ nerve cells (Herrup & Mullen, unpublished). This finding is noteworthy since the CNS is separated from the general circulation by the blood-brain barrier. Thus, it would be incorrect to assume that the serum is the only possible mediator of the transfer. This also suggests that the mechanism may differ from the well documented ‘clearing’ of lysosomal enzymes which is done almost exclusively by the liver (Thorpe et al. 1974; Stahl et al. 1976).

The mechanism of this phenomenon and its role in the physiology of a normal animal remain unknown. However, from the experiments mentioned above as well as those reported here it would appear that intercellular protein transfer is a significant in vivo event. The chimeric mouse offers a good model system in which to conduct further studies. Other mutations are known which affect processes such as secretion of enzymes like β-glucuronidase (beige (bg) -Brandt, Elliott & Swank, 1975) as well as its intracellular localization (egosyn (Eg) - Paigen et al. 1975) and developmental appearance (glucuronidase-temporal (Gut) - Paigen et al. 1975). Thus other chimeric combinations are possible.

This research was supported by a Basil O’Connor Starter Research Grant from the National Foundation -March of Dimes (R. J.M) and by a grant from the Jane Coffin Childs Memorial Fund for Medical Research (K.H.).