In set theory, if $R$ is a relation, it is considered a set such that if $aRb$, then $(a,b)\in R$. However, if $\in$ (the "element of" symbol) is considered to be a relation in the same regard, then I can prove that this is inconsistent with ZF.
Theorem: If $\in$ is a relation, then there exists $U$ in ZF such that $\forall x(x\in U)$.
Proof: Let $\in$ be a relation. Let this relation also be denoted $S$. By the axiom of the power set, for any set $x$ there exists a family $y$ such that $x\in y$. By the definition of a relation, $x\in y \implies (x,y)\in S$. Using the Kuratowski's definition of an ordered pair $$(x,y):=\big\{\{x\},\{x,y\}\big\}.$$ Therefore, $x\in y \implies \big\{\{x\},\{x,y\}\big\}\in S$. By the axiom of the union, there exists the arbitrary union $T=\cup S$. So $\big\{\{x\},\{x,y\}\big\}\in S\implies\{x\}\in T$. Once again, there must exist the arbitrary union $U=\cup T$. Hence, $\{x\}\in T\implies x\in U$. Therefore, if $\in$ is a relation, there exists $U$ such that $\forall x(x\in U)$ in ZF. $\square$
Of course, this immediately becomes inconsistent, as then Russel's Paradox arises.
Does this mean that ZF is inconsistent, or that $\in$ is not considered to be a relation in the same regard as other relations? If the latter, how is $\in$ then defined?