Any compound that has less absorbance than the mobile phase will appear as negative peak

u may have to modify sample preparation method

chance of accumulation of non polar constituents from previous injections

use hplc grade solvents wash column many times

check detector and column life span also

As you say, You are getting negative peak even after injecting ACN or Water which are part of your mobile phase it means there is residue of previous injection as explained  above by Dr Arun. May be you are injecting more concentrate solution so please make appropriate dilution and before that Wash  your column thoroughly .

You are getting a negative peaks because there is no TFA in your samples! Try injecting your mobile phases (containing 0.1% TFA) you should not see a negative peak. The  TFA you are using is probably impure. Purchase and use an HPLC grade or better TFA sealed in glass ampules to make your mobile phases. 

In addition to the above explanations, try preparing sample in mobile phase to avoid the different composition between sample solvent and mobile phase solvent.Also if the mobile phase is absorbing wave length, you need to change mobile phase .pH also plays an important role in overcoming the negative peaks. Use HPLC grade reagents , buffers as suggested by Bruce.

we have to check one by one parameter

1.0 check quality of TFA , if some body open longer time, there may chance of impurity

 if problem not solve then try

Inject mobile phase composition  one time and observe base line is clear

If negative peak is not seen  their then conclude problem in sample preparation

try preparing sample in mobile phase to avoid the different composition between sample solvent and mobile phase solvent as per mentioned above

Keep needle wash Mobile phase composition 

2.0 solubility is one most important Factor your sample should be clearly soluble in your (0-70% ACN in 35 min) gradient

try to use gradient grade ACN.

Several issues may cause what you see in the chromatography.  Like all the above have indicated, using HPLC grade solvents, prepare your sample in mobile phase, check pH, check TFA, etc. are all valid suggestions.  I'd suggest that you use water with a resistance of 18.2 mega-ohm/cm or better.  If you can get the Millipore desktop water system or something alike to produce your HPLC water, that should help.  

One of the things that you may want to consider is that since your sample is scorpion toxin, is it possible that you still have some protein residue that is not completely removed in sample preparation?  This little residue may stick on C18 column, and get gradually flushed out along your chromatographic run and causes havoc.  Hence changing your sample preparation method is something you may want to think about.  If you still get the same interference even after changing all the above and switched to another column of the same packing material, you may want to consider the possibility of contamination in your injector assembly.  You may want to clean that part of the HPLC as well.

To trouble shoot your problem, you may need to change one parameter at a time.  It takes some patience, and wish you good luck for that (you really need it!).

1. Check if our re-equilibration phase of Your gradient is sufficiently long (You may not be completely back at the start conditions of You gradient, which means that the concentrations of different components in Your mobile phase may be different, thus causing the negative peaks. However, this should be fairly repeatable).

2. Also, viscosity and and differences in density will cause the detector to behave a bit like a refractive index detector.

3. Can You run at a higher wavelength, 215 nm is quite low and causes detection tp be  very sensitive to everything, including solvent effects.

Your uv detector can have refractive index effects which make the baseline excursions go in the negative direction..

Arun's answer is very correct, that any compound that has a less absorbance than the substance will give you a negative peak. 

I support the answer made by Jayram VG, However, I would like to know which answer does (did) work.

The phenomenon you are observing may be due to the increasing concentration of acetonitrile during your gradient which has a suppressing effect on the ionization of TFA and its UV absorbance compared to that in aqueous solution.  See the following link. 

The answers above are correct. Think of your problem as the fact that the uv absorbance of your mobile is greater than the uv absorbance of whatever is causing the negative peaks. So one one approach isto reduce the uv background of your mobile phase - so good quality reagents far UV grad CH3CN, high quality water (possible put it down a column before use, and the best TFA you can get. Buy it in 1 mL sealed amples for example. Consider other acids. Some of these peaks may be material add has adsorbed and then elute as the gradient progresses hence the need for good quality water but also consider the sample preparation.

Negative peak are analyte with half ionized and half unionized state