The nervous system is composed of a variety of neuron types. It is therefore intriguing to understand how neuronal diversity is achieved from a developmental biology perspective. Mounting genetic analyses indicate that terminal selectors directly control the expression of genes tightly related to the functions of fully differentiated neurons, which are termed terminal identity genes (TIGs). Nevertheless, it is still unclear how the maintenance of TIG expression over such long time is achieved as neurons are post-mitotic cells and they remain stable throughout the life of the animal. Addressing this question will not only help us get a comprehensive model of how neuronal identity remains stable, but also provide clues to the etiology of late-onset neurodegenerative diseases where neurons fail to maintain their identity and eventually die. Therefore, the central goal of this thesis is to investigate how the continuous expression of terminal identity genes is achieved by focusing on the motor neurons of C. elegans. To investigate the maintenance of TIG expression, I focus on one example which is the glutamate receptor subunit gene glr-4. First, I characterized the endogenous expression pattern of glr-4, showing it is only expressed in specific but not all cholinergic MNs. I found that a novel AT-rich interaction domain (ARID) family TF CFI-1 acts as a repressor and prevents the cholinergic terminal selector UNC-3 and the midbody Hox proteins LIN-39 and MAB-5 from activating glr-4 in all cholinergic MNs. Using a combination of inducible protein degradation system, genetic analysis, and promoter bashing analysis, I showed that CFI-1 is continuously required to maintain the repression of glr-4 and it directly represses glr-4 by binding to the glr-4 promoter via a conserved binding motif for the ARID protein family. Furthermore, I showed the genome-wide binding profile of CFI-1 which displays significant overlap with that of the terminal selector UNC-3, suggesting that CFI-1 plays a critical role in co-regulating terminal identity genes with UNC-3 in late developmental stages thereby maintaining neuronal identity in adulthood. Second, I explored the upstream regulatory mechanisms of the expression of cfi-1, which appears to be crucial from the results mentioned above. With promoter bashing analysis, I found that although cfi-1 is broadly expressed in the nervous system and muscle cells in the head, a distal enhancer of cfi-1 is sufficient to drive its expression in the MNs. Careful examination of this enhancer showed that cfi-1 expression in the MNs is regulated by UNC-3 and LIN-39 and MAB-5. Interestingly, the Hox proteins are only required for the initiation of cfi-1 expression during larval stages but are dispensable in adult worms. On the other hand, UNC-3 is only important for the maintenance of cfi-1 in adulthood but is not necessary to drive the initiation of cfi-1 early on. This finding provides novel perspectives to the functions of the terminal selector UNC-3, as it may indirectly contribute to the maintenance of neuronal identity by maintaining the expression of other TFs which then act on the terminal identity genes.