MLE of multinomial distribution from multiple realizations

Question

I observe multiple histograms as data points, each denoted as $n_{i}$. The total number of points is $N$, i.e., $i=1:N$. Each $n_i$ is $V$ dimensional, and $L_i = \sum_{v=1}^V n_{iv}$. $L_i$ can be different for each data point. If I assume this data is sampled from $n_{i} \sim Mu(n_{i} | L_i, \Theta)$, how can I find ML estimate of $V$ dimensional probability vector $\Theta$ given $\{n_{i}\}_{i=1:N}$?