Fine Structure of Millipore Filters

Millipore filters and membranes are being used with increasing frequency in biological and chemical investigation. However, the structure of these filters, as well as some of the problems associated with histological techniques involved in their use, do not appear to be widely known. Early electron microscope studies (5) of Millipore filters were restricted to replicas of surface texture and to the measurement of average pore size. In a work on transfilter mesonephrogenic tubule induction, Grobstein and Dalton (2) suggested that morphological and dimensional changes might occur in the filters during specimen embedding and sectioning. Recently McCombs et al. (3) deliberately swelled or dissolved the filter in acetone to facilitate sample preparation. In this note, we wish to comment on the fine structure of and some properties of the filters in order to facilitate the work of subsequent users of these materials. The observations reported here were made during a study of bone induction across Millipore filters, the details of which are described elsewhere (1). To summarize, filters composed of mixed esters of cellulose, bearing the manufacturer's designation HA, VC, or VF, with respectively, 0.45, 0.1, and 0.01 Ap pores, formed the walls of diffusion chambers which enclosed fragments of tumor tissue (osteosarcoma). It was found that new bone was induced to grow along the outer surfaces of the chamber walls under certain in vivo conditions. This growth resulted in a layer of tumor tissue separated from a layer of induced bone by the filter. Thin sectioning and electron microscopy were then carried out in order to study the structures which grew into the filter pores. From the observations made, insight into the mechanism of bone induction was obtained. Initial problems encountered in the preparation for Araldite embedding of the composite specimens were overcome by introducing minor modifications into the commonly employed embedding procedure. These were the substitution of isopropyl alcohol for ethanol in the dehydration step and the use of toluene in place of propylene oxide as the transitional solvent between isopropyl alcohol and the resin mixture. These steps were necessary since the usual procedure swelled or even dissolved the filters. Furthermore, methacrylate embedding proved less desirable than Araldite embedding. The first specimens examined in the electron microscope proved to be confusing. These specimens were sections of tissue and filter and were confusing in that the filter matrix appeared electron-lucent, producing the illusion that cytoplasmic extensions were growing into the filter material rather than into the pores (Fig. 1). Therefore, it became necessary to examine the filter without tissue growth. Further samples were prepared to ascertain the effect of fixatives and embedding techniques on filter structure. Thus, some segments of filters were fixed in 6% buffered glutaraldehyde, postfixed in 1% buffered osmium tetroxide, and dehydrated in graded dilutions of isopropyl alcohol, while others received no treatment prior to Araldite embedding. Sectioning was carried out on a Reichert ultramicrotome fitted