All variables in Matlab are double by default unless you make them something else explicitly. All the formats you've used are also floating point and only affect the displayed value, not internal storage.
That said, integer values up to the size that can be held in a double-precision float (roughly 15 decimal digits) are also exact, so certainly the value 15440 can and will be stored exactly in memory.
You don't show actual code and data normally aren't stored in .m files, so not sure just how you wrote it to get it converted to floating representation with rounding--show us the code that created the file and how you subsequently read/displayed the value and we can then show you how to avoid the problem.
ADDENDUM
Also, format only affects the command window display form, not internal nor external storage. So, if the value is 1.5449999999999e+05, the value was obtained from either computation that had some floating point rounding to be not quite an integer value or read from a file written with a value of the same situation and that's what the value actually is.
Observe the following:
>> x=1.5449999999999e+05; % store the value...
>> format rat % display as ratio of nearest integers...
>> disp(x) % looks like an integer
154500
>> format short % revert to normal
>> disp(x) % but isn't integer, really...
1.5450e+05
>> format long % what is full value???
>> disp(x) % what we stored and what is in memory is this...
1.544999999999900e+05
>> disp(round(x)) % make it look like an integer...but it really isn't
154500
>> x=round(x) % now make it so...
x =
154500
>>
