Here is another proof: it's fairly elementary, but it's also a bit long-winded, I apologize for that. But hey, sometimes the scenic route is interesting.
We start with what we know, $G = \langle x\rangle$, and that $x$ has order $m = 2k$, where $k$ is a positive integer.
We'll do the "easy part" first:
$G$ has at least one element of order $2$:
Consider $x^k$. We see that $(x^k)^2 = x^{2k} = x^m = e$. Thus $x^k$ has order dividing $2$, so either $1$, or $2$. If $x^k$ had order $1$, we would have $x^k = e$, and since $0 < k$, then $k < 2k$, contradicting the fact that $2k = m$ is the LEAST positive integer $n$ such that $x^n = e$. So $x^k$ does not have order $1$, and thus has order $2$.
Now the "harder part", here is where we will take our scenic detour:
Claim: any subgroup of $G$ is also cyclic, with generator $x^r$ for some $0 \leq r < m$. Let's call our subgroup $H$. If $H = \{e\}$, there is nothing to prove, the trivial group is clearly cyclic (the identity is a generator). So assume $H$ has some non-identity element in it. So we can write:
$H = \{e,x^{r_1},x^{r_2},\dots,x^{r_t}\}$, for some positive integers $r_1,r_2,\dots, r_t$. Choose $r = \min(r_1,r_2,\dots,r_t)$. I claim $x^r$ generates $H$. For suppose it did not. Then we have some $r_j$ such that $x^{r_j}$ is not a power of $x^r$, that is $r_j$ is not a multiple of $r$. In other words, we can write:
$r_j = qr + u$, where $0 < u < r$.
It follows that $x^u = x^{r_j - qr} = x^{r_j}[(x^r)^q]^{-1}$.
Now $x^{r_j} \in H$, and $x^r \in H$, so any power $(x^r)^q \in H$ (by closure), and thus (since $H$ is a subgroup, and contains all inverses) $[(x^r)^q]^{-1} \in H$, and thus $x^u \in H$, being the product of two elements in $H$.
But $0 < u < r$, contradicting our choice of $r$ as the least positive power of $x$ in $H$ (dramatic pause).
So...since assuming $x^r$ does not generate $H$ leads to a contradiction,...can't have that.
Now...where were we? Oh yeah, we want to show that $G$ has AT MOST one element of order $2$. Again, let's suppose it has at least $2$, we'll call them $a$, and $b$.
Now cyclic groups are abelian (I hope you already know this, or I will certainly try any reader's patience proving that, too), so in particular $ab = ba$. Hence:
$(ab)^2 = (ab)(ab) = a(ba)b = a(ab)b = (aa)(bb) = a^2b^2 = ee = e$. So either $ab$ also has order $2$, or order $1$.
If $ab = e$, then $b = a^{-1} = a$ (since $a$ has order $2$), but we started out assuming $a$ and $b$ are two different elements of $G$. So this can't be right, it must be that $ab$ has order $2$ as well.
And now, the punch line:
It is easy to see that $\{e,a,b,ab\}$ is closed under multiplication, and possesses all inverses, and is thus a subgroup of $G$. But it's not cyclic! C'est impossible! (we just wasted 10 minutes of our lives proving it above).
And that contradiction shows more than one element of $G$ of order $2$ cannot be.