Nuclease-deactivated Cas (dCas) proteins can be used to recruit epigenetic effectors, and this class of epigenetic editing technologies has revolutionized the ability to synthetically control the mammalian epigenome and transcriptome. DNA methylation is one of the most important and well-characterized epigenetic modifications in mammals, and while many different forms of dCas-based DNA methyltransferases (dCas-DNMTs) have been developed for programmable DNA methylation, these tools are frequently poorly tolerated and/or lowly expressed in mammalian cell types. Further, the use of dCas-DNMTs has largely been restricted to cell lines, which limits mechanistic insights in karyotypically normal contexts and hampers translational utility in the longer term. Here, we extend previous insights into the rational design of the catalytic core of the mammalian DNMT3A methyltransferase and test three dCas9-DNMT3A/3L variants across different human cell lines and in primary donor-derived human T cells. We find that mutations within the catalytic core of DNMT3A stabilize the expression of dCas9-DNMT3A/3L fusion proteins in Jurkat T cells without sacrificing DNA methylation or gene-silencing performance. We also show that these rationally engineered mutations in DNMT3A alter DNA methylation profiles at loci targeted with dCas9-DNMT3A/3L in cell lines and donor-derived human T cells. Finally, we leverage the transcriptionally repressive effects of dCas9-DNMT3A/3L variants to functionally link the expression of a key immunomodulatory transcription factor to cytokine secretion in donor-derived T cells. Overall, our work expands the synthetic biology toolkit for epigenetic editing and provides a roadmap for the use of engineered dCas-based DNMTs in primary mammalian cell types.