Reading the Covariance Matrix

In the covariance matrix $\Sigma$ of a multivariate Gaussian over features $x_1,\ldots,x_d$ , what does the off-diagonal entry $\Sigma_{ij}$ (for $i \neq j$ ) represent?

The joint mean of feature $i$ and feature $j$

How feature $i$ and feature $j$ vary together (their covariance)

The variance of feature $i$ alone

The Pearson correlation coefficient between feature $i$ and $j$ , bounded in $[-1, 1]$