An improper transfer function has a numerator with higher order than the denominator. In your case, the numerator is second order and the denominator is first order.
If you want to approximate the step response, you can augment the transfer function with an additional high frequency pole so that the numerator and denominator are both second order:
sys = ((-1.0024*(s+55.86)*(s+0.4987))/(s));
aporoximatesys = sys*tf(1,[.0001 1]);
Now you can plot the step response of approximatesys (make sure you pick a reasonably large final time), but it's only an approximation.
If you want the exact solution, you'll need to work it out by hand or use the Symbolic Math Toolbox. But the solution will be difficult to plot exactly.