It is well-known that probabilistic constellation shaping (PCS) provides a shaping-gain of 1.53dB asymptotically as signal-to-noie (SNR) increases. This is however, under ideal assumptions that the considered system operates in optimal sense and can achieve the Shannon capacity. While in practice the system can operate below the capacity, and in particular when block-error rate (BLER) and throughput are considered which are the most relevant merits, the benefits of PCS are unclear. In this paper, we propose a PCS-transceiver for the fifth-generation new-radio (5G-NR) that supports quadrature-amplitude-modulation (QAM) constellations shaped with PCS, namely, PCS-QAM. We put a special interest in the throughput achieved under a stringent BLER constraint, and validate the effectiveness of PCS with practical detection and decoding algorithms, in comparison to conventional uniform QAM constellations, namely, Uniform-QAM. We also analyze the properties of power-gain, entropy-loss, and peak-to-average power-ratio (PAPR), in connections to the PCS design. Further, we derive a necessary and sufficient condition for a PCS-QAM to outperform a Uniform-QAM in throughput, and prove that the normalized entropy-loss with the PCS-QAM must be less than the BLER obtained with the Uniform-QAM. Furthermore, we demonstrate that the PCS-transceiver is flexible in rate-adaption without affecting the encoder, and yields a better throughput-envelope by mitigating the SNR-gaps between two adjacent modulation and coding scheme (MCS) indexes in 5G-NR.