Let $G$ be a finite non-Dedekind $p$-group with non-cyclic center, where $p$ is an odd prime.
By $[\langle x\rangle]_G=\{g^{-1}\langle x\rangle g\ |\ g\in G\},$
I mean the conjugacy class of the subgroup $\langle x\rangle$ of $G$ and
$U=\{[\langle x\rangle]_G\ |\ \langle x\rangle\ntrianglelefteq G\}$ is the set of all
conjugacy classes of non-normal cyclic subgroups of $G$.
Also $\Omega_1(G)$ denotes the subgroup generated by all elements of $G$ of order at most $p$.
Let $z\in \Omega_1(Z(G))\setminus\{1\}$ and consider the following subsets of $U$
\begin{eqnarray*}
&&U_1=\{[\langle x\rangle]_G\in U\ |\ \langle z\rangle\not\leq \langle x\rangle,
\ \ [\langle x\rangle]_G=\{ \langle xz^i\rangle\ |\ 0\leq i\leq p-1\}\}\\
&&U_2 =\{[\langle x\rangle]_G\in U\ |\ \langle z\rangle\not\leq \langle x\rangle,\ \
[\langle x\rangle]_G\cap\{ \langle xz^i\rangle\ |\ 1\leq i\leq p-1\}=\phi\}
\end{eqnarray*}
I want to give an affirmative or negative answer to the follwing question
Question: If for each $z\in \Omega_1(Z(G))\setminus\{1\}$, we have $U_1=U_2=\phi$ , then is $\mbox{exp}(Z(G))=p$ and $G^{\prime}$ an elementary abelian $p$-group?
I could not find any counter example but in the following minor case I know the answer to the first part of the Question is yes,
Let all non-normal subgroups of $G$ be cyclic and let there exists an element $x\in \Omega_1(G)\setminus Z(G)$. In the contrary case, if there exists $z\in Z(G)$ such that $o(z)>p$, then $\Omega_1(\langle x,z\rangle)\cong C_p\times C_p$ and by assumption $\Omega_1(\langle x,z\rangle)\lhd G$. Now since $\langle x\rangle\leq \Omega_1(\langle x,z\rangle)$, it follows that $[\langle x\rangle]_G\in U_1$, which is a contradiction.
Any help will be greatly appreciated.