Techniques are described for selectively removing the carbohydrate moieties from the high mobility group (HMG) 14 and 17 proteins of Friend erythroleukemia and calf thymocyte cells without causing degradation of these nonhistone proteins. Investigations were therefore undertaken to investigate the possible significance of this secondary biochemical modification by comparing the activity of the HMGs with and without their glycosyl side chains in various functional assays of the proteins. For example, these HMGs have been found to be equally effective in the partial inhibition of the histone deacetylase enzymes of mammalian cell nuclei whether or not they contain covalently bound carbohydrate residues. We also investigated the possibility that the glycosyl modifications might be involved in the reported ability of these HMGs to specifically cause reconstitution of the DNase I-sensitive structure of "active" genes in salt-depleted chromatin. Unexpectedly, in experiments with both the active beta-globin gene of Friend cells and the total complement of active genes in these same mouse cells (which can be preferentially labeled by nuclear nick translation) we have been unable, using purified HMG 14 and 17 preparations, to reconstitute the active DNase I structure of genes in salt-depleted chromatin preparations. Therefore, the possible role, if any, played by HMG glycosylation in such reconstitution experiments remains unknown. However, in the course of these experiments we did find that HMG 14 and 17 proteins covalently linked to carbohydrate side chains bind preferentially to the nuclear protein matrix of mammalian cells. Furthermore, this association appears to be mediated through the glycosyl side chains since enzymatic removal of these modifications from the HMGs greatly reduced their binding to the nuclear matrix. Since numerous workers have implicated the nuclear matrix as the site of both RNA transcription and DNA replication in eukaryotic cells, the finding of HMG association with this nuclear structure may have significance for our understanding of the overall architectural organization of the active domains of chromatin in cells.