Memory and learning are significantly influenced by long-term changes in synaptic strength, though the exact molecular mechanisms remain unclear. The postsynaptic density (PSD) is a critical component for storing and organizing synaptic long-term changes. The PSD itself is organized into complex nano-clusters, which are modulated during synaptic plasticity. However, the underlying organizational principles yielding such changes and maintaining stability otherwise are still to a large degree unknown. Recent research assumes that liquid-liquid phase separation (LLPS) might be crucial for PSD formation and maintenance, necessitating further investigation. To investigate the mechanisms regulating PSD organization, we developed a minimal coarse-grained patchy model of the PSD with interacting multivalent proteins. Using the simulator PyRID, we simulated PSD-95 proteins to explore the collective phase behavior of the PSD. Our molecular dynamics simulations revealed the formation of multi-clusters characteristic of PSD only if we assume a fixed concentration boundary condition, reflecting the ongoing exchange of proteins between spines and dendritic shaft. Additionally, we observed that the number of clusters can be tuned by changes in the boundary concentration as well as in the protein interaction strength. Our study highlights that these two parameters can explain the organization of PSDs across different phases. To verify our results, we apply the Cahn-Hilliard equation (cite) incorporating boundary reactions. The results from this model support our findings, indicating that boundary concentration and interaction strength play a crucial role in the emergence of diverse phases. Furthermore, we confirmed our findings by comparing them with data from a MINflux imaging study (cite). This dataset mapped PSD95 protein positions in spines across varied treatment conditions. Our analysis showed that PSDs demonstrate phase organization resembling our simulations, indicating that neuronal activity can impact PSD states, such as nanoclusters. Overall, our results show that the concentration of proteins in the dendritic shaft can determine the organization of the PSD into LLPS-like clusters, providing a new possibility about how a neuron can control the state of its synapses.