If none of the independent variables predict the dependent variable when they are entered simultaneously into a multiple regression model, then that is simply a result of the analysis. It means that that set of independent variables does not help you predict the DV, at least not based on additive linear effects. There's nothing else to do unless you also have a hypothesis about interactive effects (moderation) and/or non-linear effects.

Also, if you had multiple indicators for some or all of your independent variables, you could try structural equation modeling (SEM, latent regression) as an alternative. Sometimes, relationships between variables are masked by measurement error, which SEM could address.

This sounds like you are trying a forward selection approach to building a regression model. This is a bad choice for starting the analysis. I would start by plotting the data. Can you see any relationships between any of the independent variables and the dependent variable? Are all the relationships that you see linear? Do the univariate models conform to the statistical assumptions? Are any of the independent variables correlated with one another? The answers to these questions will help you decide on the next step. Also consider whether your sample size is sufficient to support a more complex analysis.

If no independent variables in your model and only the DV variable here you can run a regression but this will be a NULL model. Later after including independent variables in your model you can make a comparison between a NULL model VS a model with a list of predictors to see if there is a significant improvement in the model fit

Thank you Johnson. This was very helpful.