\@doendnote{endnote62}{For an analysis of LFV $B$ decays with leptoquarks see Ref.\protect \nobreakspace {}\cite {Varzielas:2015iva} and for a model independent analysis see Ref.\protect \nobreakspace {}\cite {Boucenna:2015raa}.} \@doendnote{endnote63}{Here we assumed that the $Z'$ boson is a $SU(2)$ singlet and not the neutral component of a $SU(2)$ triplet. In the second case, the relation $\Gamma ^L_{\ell _i\ell _j}=-\Gamma ^L_{\nu _i\nu _j}$ would hold.} \@doendnote{endnote64}{Our predictions are for $B^0\to K^{(*)0}\ell ^+ \ell ^{\prime -}$, those for the charged modes $B^+\to K^{(*)+}\ell ^+ \ell ^{\prime -}$ can be found by multiplying by the ratio $\tau _{B^+}/\tau _{B^0} $ of $B$-meson lifetimes.} \@doendnote{endnote65}{As stated before, we assume that the $Z'$ is a $SU(2)_L$ singlet. The same upper bound from $B\to K\nu \protect \mathaccentV {bar}016{\nu }$ would also apply if the $Z'$ would be the neutral component of a $SU(2)_L$ triplet, but would not hold anymore if it is a mixture of different representations.} \@doendnote{endnote66}{This limit would be even slightly stronger if one would assume a vanishing NP contribution in the $ee$ sector and a small contribution to $\mu \mu $ (as preferred by the global fit) together with a maximally destructive interference in $\tau \tau $.} \@doendnote{endnote67}{Similar results are obtained by the CKMfitter collaboration \cite {Charles:2004jd}.} \@doendnote{endnote68}{Likewise there is no upper limit on $\Gamma _{sb}$ if the $Z'$ does not couple to muons, as constraints from $b\to s \mu ^+ \mu ^-$ transitions do not apply in this case.}