There are several approaches you could use depending on what you intend to do with the sum.
Simplest would be to simply scale up the sum according to the proportion of missing values.
If there are ten variables, V1 to V10, you could do this (computes could be combined).
compute nmiss=nmiss(V1 to V10).
compute mean = mean.1(V1 to V10).
compute imputedsum = (mean * (10 - nmiss) + (nmiss * mean)).
If you have reason to believe that the missing values are related to other variables, you might use the imputation procedure to calculate estimated means.
--