Hyperluminous quasars ($L_{rm bol}gtrsim 10^{47}$ erg s$^{-1}$) are ideal laboratories to study the interaction and impact of extreme radiative field and the most powerful winds in the AGN nuclear regions. They typically exhibit low coronal X-ray luminosity ($L_{rm X}$) compared to the UV and MIR radiative outputs ($L_{rm UV}$ and $L_{rm MIR}$) with a non-negligible fraction of them reporting even $sim$1 dex weaker $L_{rm X}$ compared to the prediction of the well established $L_{rm X}$-$L_{rm UV}$ and $L_{rm X}$-$L_{rm MIR}$ relations followed by the bulk of the AGN population. We report in our WISE/SDSS-selected Hyperluminous (WISSH) $z=2-4$ broad-line quasar sample, the discovery of a dependence between the intrinsic 2-10 keV luminosity ($L_{rm 2-10}$) and the blueshifted velocity of the CIV emission line ($v_{rm CIV}$) indicative of accretion disc winds. In particular, sources with fastest winds ($v_{rm CIV}gtrsim 3000~rm km s^{-1}$) possess $sim$0.5-1 dex lower $L_{rm 2-10}$ than sources with negligible $v_{rm CIV}$. No similar dependence is found on $L_{rm UV}$, $L_{rm MIR}$, $L_{rm bol}$, photon index and absorption column density. We interpret these findings in the context of accretion disc wind models. Both magnetohydrodynamic and line-driven models can qualitatively explain the reported relations as a consequence of X-ray shielding from the inner wind regions. In case of line-driven winds, the launch of fast winds is favoured by a reduced X-ray emission, and we speculate that these winds may play a role in directly limiting the coronal hard X-ray production.