The problem seems like a variant of Håstad's broadcast attack, except that is usually stated with $e$ ciphertexts, when here there is $2<e$ ciphertexts.
We know $m^e$ modulo $n_1$ and modulo $n_2$ (that's the givens $c_1$ and $c_2$), thus we can compute $m^e\bmod(n_1\,n_2)=y$ using the Chinese Remainder Theorem, per $y\gets(n_2^{-1}(c_1-c_2)\bmod n_1)\,n_2+c_2$
If we are lucky enough that $m<\left\lfloor\sqrt[e]{n_1\,n_2}\right\rfloor$ (roughly, $m$ of bit size less than $1/e$ of the sum of the bit sizes of the moduli), then $y$ will be exactly the $e^\text{th}$ power of some integer $x$, we can get that $x$, and it will be $m$.
If not, we can try, for incremental $i$, if any of the $y_i=i\,c_1\,c_2+y$ is exactly an $e^\text{th}$ power; that slightly extends the attack.
Note: none of this is RSA as correctly practiced. What's raised to the power $e$ modulo $n$ should not be the message, but a randomly padded message (per e.g. RSAES-OAEP) computed separately for each ciphertext sent, and almost as wide as the public modulus.