You can use a bandpass filter, or a notch filter in the frequency domain.

I do not understand your notion of cleaning without filtering.

Any type of cleaning similar to spatially invariant convolution can be written as a filtering operation. If you do spatially variant operations, it is still adaptive filtering.

because digital filtering affect past processing especially change causality between time series and cause granger causality become wrong. in the other words studies shown that filtering, particularly low pass filtering, can have a damaging impact on Granger causality and other connectivity computations.

The power line interference is at 60Hz +/- 5% , or +/- 1% according to the regulation of your Country. At wards, it is not unusual to find 5% variation from the ideal value. In your data, I suppose that this was the case. An harmonic.

Thanks for your consideration

Our Data were recorded and I should clean it by signal processing approaches.

I used a multi-taper decomposition approach for identifying and removing line noise components while minimizing background signal distortion. It is suggested by Mitra and Pesaran (1999)  [link is attached]. This approach measures artifacts in The frequency domain and then subtracts this fitted signal from the data. it is desired for my post process and could clean 60 Hz without Digital filtering but it can't work for 113Hz as fine as 60Hz.

I change its parameters but can't take a  good result.

can this method work better?

 know anybody another method?

https://www.ncbi.nlm.nih.gov/pubmed?Db=pubmed&Cmd=ShowDetailView&TermToSearch=9929474