In this paper, we study the clustering problem for crossroads in Vehicular Ad hoc Networks (VANETs). Considering the load balancing of both the whole network and each cluster based on the multiple metrics, an Enhanced Low Overhead and Stable Clustering (EnLOSC) scheme is presented to ensure the stability and security of clusters and to reduce the communication overhead in this case. The proposed capability metric, designed to find the vehicles with similar direction and better channel quality, is exploited in the processes of formation and maintenance to determine which node is suitable for a cluster head. Based on this, a Cluster Head Electing in Advance Mechanism (CHEAM) is developed in order to fairly select a new head for "isolated" vehicles that may not belong to a cluster. Meanwhile, other metrics are related to the node density and cluster size, which are exploited in the Cluster Merging and Splitting Mechanisms to keep the system load balancing and to improve the communication quality. Furthermore, the proposed Discovery and Elimination Scheme (DES) is designed to tackle the malicious nodes that may hurt the cluster communication. Accordingly, an enhanced cluster maintenance strategy with multi-metrics and a secure scheme is proposed so as to reduce the number of isolated vehicles, keep appropriate loading for each cluster head, and protect the whole link over cluster communication. Numerical results and discussion indicate that the cluster stability, communication overhead, load balance, and security can be significantly enhanced by our proposed scheme.